Angpt1 binding to Tie1 regulates the signaling required for lymphatic vessel development in zebrafish

Lymphatic vessels, together with heart and blood vessels, constitute the vascular system and play pivotal roles in the maintenance of fluid balance, fat absorption, and immunosurveillance (Alitalo, 2011; Donnan et al., 2021; Escobedo and Oliver, 2016; Francois et al., 2021; Petrova and Koh, 2020). During development, most lymphatic vessels derive from pre-existing blood vessels. At the onset of lymphangiogenesis, a subpopulation of venous endothelial cells (ECs) within the large veins, such as the cardinal vein (CV) in mice and the posterior cardinal vein (PCV) in zebrafish, express the transcription factor Prox1, a well-known marker of lymphatic ECs (LECs), and thereby commit to a lymphatic endothelial cell fate (Koltowska et al., 2015; Nicenboim et al., 2015; Wigle and Oliver, 1999). These Prox1-positive LEC progenitors sprout from the vein and undergo migration, proliferation and differentiation, eventually giving rise to the primary lymphatic vessels (Escobedo and Oliver, 2016; Suárez and Schulte-Merker, 2021). The endothelial-specific receptor tyrosine kinase (RTK)­­ Vegfr3 (also known as Flt4) and its ligand Vegfc play a central role in lymphangiogenesis in both zebrafish and mice (Hogan et al., 2009; Jeltsch et al., 2013; Karkkainen et al., 2004; Le Guen et al., 2014; Tammela and Alitalo, 2010; Villefranc et al., 2013). Vegfc/Vegfr3 signaling is essential for lymphatic differentiation and the initial sprouting of lymphatic progenitors from the vein (Karkkainen et al., 2004; Koltowska et al., 2015; Küchler et al., 2006; Shin et al., 2016; Srinivasan et al., 2014; Tammela and Alitalo, 2010). This process of primary lymphatic vasculature formation is conserved from fish to mammals (Hägerling et al., 2013; Mauri et al., 2018).

The angiopoietin/Tie system constitutes another endothelial-specific RTK signaling complex important for mammalian cardiovascular development. Of the two TIE receptors, only TIE2 (also known as TEK) has been reported to bind its ligands, angiopoietins, in mammals (Augustin et al., 2009; Eklund and Saharinen, 2013). Therefore, TIE2 is thought to be a key mediator of Ang/Tie signaling. Among angiopoietins, angiopoietin 1 (ANG1 or ANGPT1) can serve as a constitutive agonist for TIE2 phosphorylation, which is crucial for blood vessel integrity (Fukuhara et al., 2008; Puri et al., 1995; Saharinen et al., 2008), whereas ANG2 (ANGPT2) is a context-dependent TIE2 agonist or antagonist (Daly et al., 2006; Kim et al., 2016; Maisonpierre et al., 1997). During lymphatic development, ANG2 acts as a TIE2 agonist (Gale et al., 2002). In contrast, until recently (Hußmann et al., 2023; Sato-Nishiuchi et al., 2023) mammalian TIE1 was considered an orphan receptor that functions by forming a heterodimer with TIE2, thereby acting as a context-dependent modulator of TIE2 activity (Korhonen et al., 2016; Saharinen et al., 2005; Savant et al., 2015; Seegar et al., 2010; Yuan et al., 2007). However, it is noteworthy that some studies show a unique role of TIE1 in lymphatic vessel formation and maturation. Hypomorphic mice with reduced expression of Tie1 show lymphatic vascular abnormalities and prominent edema (D'Amico et al., 2010; Qu et al., 2010). Furthermore, postnatal deletion of Tie1 causes abnormalities in the lymphatic capillary network, whereas deletion of Tie2 does not (Korhonen et al., 2022; Shen et al., 2014). These reports highlight the crucial role of TIE1 as the dominant TIE receptor required for lymphatic vessel formation. However, how TIE1 regulates lymphangiogenesis is not fully understood.

Recent studies unexpectedly reveal that, of the two Tie receptors, the tie1 gene is more commonly present in vertebrates, but the tie2 (tek) gene is absent in most of the ray-finned fish, including Medaka (Oryzias latipes) (Jiang et al., 2020). Although the tie2 gene is present in zebrafish, no obvious phenotypes have been reported in tie2 mutants (Gjini et al., 2011; Jiang et al., 2020). In contrast, zebrafish tie1 mutants die, primarily as a result of cardiac defects (Carlantoni et al., 2021). These mutants exhibit vascular defects, such as reduced EC numbers, reduced blood flow, impaired caudal vein plexus formation, and impaired lymphatic development (Carlantoni et al., 2021; Hußmann et al., 2023). Therefore, Tie1 may be responsible for Ang/Tie signaling in cardiovascular development in most fish species that lack functional Tie2. However, there have been no reports on the roles of angiopoietins in these species. In addition, Tie1-driven signaling pathways, including ligands, downstream signaling, and cellular functions, remain unclear in any animal models.

In this study, we performed comprehensive functional analyses of angiopoietins and Tie receptors in zebrafish. Our in vitro binding assay shows that, among angiopoietins, only Angpt1 (the zebrafish ortholog of Ang1) binds to Tie1 as a ligand. Similarly, angpt1 mutants, unlike angpt2a and angpt2b mutants, phenocopy tie1 mutants, which show defects in lymphatic vessel development. Angpt1/Tie1 signaling thus functions in early lymphangiogenesis in the trunk, where Tie1 negatively regulates nuclear translocation of Foxo1a downstream of phosphatidylinositol-3-OH kinase (PI3K) and plays an essential role in the full activation of Vegfc/Vegfr3 signaling. We present evidence that Tie1 functions as a receptor for angiopoietin and constitutes an essential signaling pathway for lymphatic development in zebrafish.

In this study, we investigate the roles of Ang/Tie signaling in zebrafish by analyzing angiopoietin genes (angpt1, angpt2a and angpt2b) and Tie genes (tie1 and tie2). To accomplish this, we generated mutants of angpt1, angpt2a, angpt2b and tie1 using transcription activator-like effector nucleases (TALENs) (Fig. S1) and compared possible phenotypes with an established tie2 mutant (Gjini et al., 2011). As reported previously (Gjini et al., 2011; Jiang et al., 2020), homozygous tie2 mutants did not show any obvious phenotypes in lymphatic or blood vessel development (Fig. 1A,C). In contrast, homozygous tie1 mutant larvae showed a loss of lymphatic structures in the trunk (Fig. 1A). In tie1 mutants, angiogenesis from the dorsal aorta (DA) showed minor defects, such as slight delay and reduced number of ECs (Fig. S2A-D), but eventually formed functional trunk vessels (Fig. 1A). In contrast, formation of the thoracic duct (TD), the first functional lymphatic vessel along the DA, was severely disrupted in these mutants (Fig. 1A,B and Fig. S2E,F). In addition, tie1 mutants exhibited cardiac edema from 2.5 days post-fertilization (dpf) (Carlantoni et al., 2021) and severe edema around eyes and intestine from 4 dpf (Fig. 1C,D and Fig. S2G). Finally, all homozygous tie1 mutants died before reaching adulthood.

The function of mammalian TIE1 is considered to depend upon its binding to TIE2 (Saharinen et al., 2005; Yuan et al., 2007). To determine whether the function of zebrafish Tie1 depends on Tie2, we analyzed tie1-tie2 double mutants. Impaired lymphatic development and severe edema formation, which were observed in tie1 mutants, were similarly observed in tie1-tie2 double mutants, with no differences between tie1 single mutants and tie1-tie2 double mutants (Fig. S2H-K). Considering that tie2 single mutants did not show any obvious phenotypes (Fig. 1A-D) (Gjini et al., 2011), these results suggest that Tie2 is not required to mediate Tie1 signaling in zebrafish.

Among angiopoietin mutants, angpt1 mutants exhibited cardiac edema from 2.5 dpf and impaired TD formation at 5 dpf (Fig. 1E-H and Fig. S3A,B), similar to tie1 mutant larvae, although their phenotypes were milder than tie1 mutants (see Fig. 1B,D). In contrast, neither angpt2a mutants nor angpt2b mutants exhibited these phenotypes (Fig. 1E-H). In addition, the impaired TD formation observed in angpt1 mutants was not significantly exacerbated in triple-homozygous angpt1-angpt2a-angpt2b mutants (Fig. 1F), indicating that Angpt2a and Angpt2b do not compensate for the loss of Angpt1 in lymphatic development. Whereas ang1 mutants exhibited defects in lymphatic development in the trunk, they did not show a distinct phenotype in arterial angiogenesis from the DA (Fig. S3C-F). Our results indicate that among angiopoietins and Tie receptors, Angpt1 and Tie1 function in lymphatic development in the zebrafish trunk.

In zebrafish, tie1 is expressed in all ECs in developing and mature blood vessels (Carlantoni et al., 2021; Lyons et al., 1998; Pham et al., 2001). We confirmed that tie1 transcripts are present in the DA, PCV and intersegmental vessels (ISVs) at 56 hours post-fertilization (hpf) by RNAScope analyses (Fig. S4A). By contrast, angpt1 is predominantly expressed in the hypochord (Fig. S4B), a single layer of cells immediately dorsal to the DA, as well as in ventral mesenchyme surrounding the major trunk vessels (Lamont et al., 2010; Pham et al., 2001). Thus, angpt1-expressing cells are in close proximity to tie1-expressing ECs in the trunk; however, it has not been demonstrated whether Angpt1 directly binds to Tie1 as a ligand. To examine this, we performed an in vitro binding assay. We tested the association of immunoglobulin Fc-domain-tagged soluble extracellular domain of zebrafish Tie1 (hereafter referred to as zTie1) (zTie1-Fc) or zTie2 (zTie2-Fc) with Angpt1, Angpt2a or Angpt2b (Fig. 2A). zTie1-Fc specifically bound to Angpt1 but not to Angpt2a or Angpt2b (Fig. 2B), suggesting that Angpt1 serves as a ligand for zTie1. This specific binding of Angpt1 and zTie1 is consistent with the fact that angpt1 mutants phenocopied tie1 mutants (Fig. 1). We also detected interaction of zTie2-Fc with both Angpt1 and Angpt2a (Fig. 2B), consistent with Tie2 binding to both Angpt1 and Angpt2 in mice.

Next, we examined phosphorylation of zTie1 by angiopoietin. Of note, upon stimulation with cartilage oligomeric matrix protein (COMP)-ANG1, a potent human angiopoietin 1 variant (Cho et al., 2004), Tyr phosphorylation of full-length zTie1 but not kinase-dead zTie1 (zTie1^K854R) was significantly enhanced (Fig. 2C,D). Therefore, zTie1 can function as an RTK that responds to angiopoietin. Collectively, our results strongly suggest that Angpt1/Tie1 signaling functions in zebrafish.

To determine which event in lymphangiogenesis is regulated by Angpt1/Tie1 signaling, we monitored trunk lymphatic development in live tie1 mutants. In zebrafish, lymphangiogenesis starts with endothelial sprouting from the PCV, which is called secondary sprouting (Fig. S5A, Movie 1) (Hogan and Schulte-Merker, 2017; Isogai et al., 2003). About half of the sprouts from the PCV connect to the primary ISVs to form veins [venous ISVs (vISVs)]. The other half of the sprouts do not connect to the ISVs, but instead migrate to the horizontal myoseptum to form parachordal lymphangioblasts (PLs), a pool of lymphatic precursors, which will form the entire trunk lymphatic vasculature including the TD (Fig. S5A, Movie 1). When we observed lymphatic development prior to TD formation, we found that PLs were mostly absent in tie1 mutants at 54 hpf (Fig. 3A,B). Therefore, Tie1 is essential for proper development of lymphatic precursors in the trunk. We then examined the secondary sprouting from the vein. In tie1 mutants, the number of secondary sprouts was markedly reduced in the trunk (Fig. 3A,C) as well as in the tail (Fig. S5B). Because the secondary sprouts contribute to both lymphatic and venous blood vessels, we examined whether Tie1 only regulates lymphatic development or also affects venous development. In tie1 mutants, both vISV and PL formation were blocked (Fig. 3A,B,D), indicating that Tie1 is essential for the secondary sprouting that contributes to both lymphatic and venous vessel development. Moreover, arterial-venous (A-V) connections consisting of an arterial ISV (aISV) and a secondary sprout remained in the mutants at 54 hpf (Fig. S5C-E), indicating that Tie1 is also important for vessel remodeling to form vISVs. Similar to the tie1 mutants, angpt1 mutants also exhibited impaired PL formation and a reduced number of secondary sprouts (Fig. S5F-I). These results demonstrate that Angpt1/Tie1 signaling regulates secondary sprouting from the large vein at the onset of lymphangiogenesis.

tie1 mutants gradually lost blood flow from 36 hpf (Fig. 3E), most likely as a result of cardiac defects (Carlantoni et al., 2021). Because flow is implicated in lymphatic vessel development (Coffindaffer-Wilson et al., 2011; Ujiie and Kume, 2022), we examined whether the phenotypes in tie1 mutants might be caused by the loss of blood flow. With or without blood flow, tie1 mutants similarly exhibited impaired PL formation and a reduced number of secondary sprouts at 54 hpf (Fig. S6A,B). In addition, to examine the relationship between the phenotype and flow in tie1 mutants in more detail, we measured the flow velocity in the PCV by tracking polyethylene glycol (PEG)-coated fluorescent microspheres (PEG-coated FMs) (Movie 3). We noticed that some of tie1 mutants had normal flow comparable to wild type (WT) and heterozygous siblings at 45 hpf (Fig. 3F). Of note, tie1 mutants that had normal flow velocities (>360 µm/s) similarly exhibited a reduced number of secondary sprouts (Fig. 3G and Fig. S6C). These results indicate that impaired blood flow is not the main cause of the phenotypes in tie1 mutants, at least during the initial sprouting of lymphatic progenitors. We cannot rule out the possibility that lymphatic phenotypes at later stages might be enhanced by impaired blood flow. Given that secondary sprouting and PL formation occur in plcγ (plcg1) mutants that completely lack the DA and blood flow (Lawson et al., 2003), blood flow might be dispensable for the early stages of lymphatic development in the zebrafish trunk. In addition, blood flow was mostly unaffected in angpt1 mutants, which exhibit similar phenotypes to tie1 mutants (Fig. S6D). Collectively, these lines of evidence indicate that Angpt1/Tie1 signaling regulates lymphangiogenesis independently of flow regulation.

Next, we investigated cell behaviors controlled by Tie1 signaling during secondary sprouting. In tie1 mutants with reduced numbers of secondary sprouts (Fig. 3C), ECs extended their protrusions, but their nuclei remained mostly within the parental vein and rarely crossed the level of the DA (Fig. 4A,B and Movie 2). Therefore, Tie1 is important for the migration of EC nuclei from the PCV. In addition, even when their nuclei migrate dorsally, their migration distance and speed were significantly lower than those of WT siblings (Fig. 4C,D). Therefore, Tie1 is important for the migration of budding ECs from the PCV.

Secondary sprouting ECs proliferate to form the PL and the vISVs (Fig. 4E). In contrast, in tie1 mutants, such proliferation events were markedly reduced (Fig. 4E,F). Similarly, the number of ECs constituting each venous sprout was markedly reduced (Fig. 4G). Apoptosis was not enhanced in these mutants (Fig. S7). Therefore, Tie1 is important for the proliferation of budding venous ECs, but dispensable for their survival.

Next, we examined the role of Tie1 in lymphatic differentiation. Prox1 starts to be expressed in lymphatic precursors of the PCV in a manner dependent on ERK (extracellular signal-related kinase or mitogen activated protein kinase) activation downstream of Vegfc/Vegfr3 signaling (Koltowska et al., 2015; Shin et al., 2016). Prox1-positive ECs were found in the PCV and the secondary sprouts in the WT situation (Fig. 4H). In contrast, the number of Prox1-positive ECs was significantly reduced in tie1 mutants (Fig. 4H,I). Consistent with this, phosphorylation of ERK was reduced in the secondary sprouts of tie1 mutants (see Fig. 7A). Therefore, Tie1 regulates the induction of Prox1 expression, presumably by affecting ERK phosphorylation. Taken together, these results demonstrate that Tie1 signaling regulates multiple EC behaviors in lymphangiogenesis: EC migration, proliferation, and lymphatic differentiation (Fig. 4J).

To investigate changes in gene expression that are regulated downstream of Tie1 signaling, we performed RNA sequencing (RNA-seq) analyses of ECs from WT and tie1 mutant embryos during secondary sprouting. To this end, kdrl:EGFP/lyve1:DsRed double-positive venous and early lymphatic ECs were isolated from the trunk and tail of embryos at 47 hpf by fluorescence-activated cell sorting (FACS) and were subjected to RNA-seq analyses (Fig. 5A). Volcano plot analysis showed that the number of downregulated genes in the tie1 mutants was much greater than the number of upregulated genes (Fig. 5B). We identified 1942 differentially expressed genes (DEGs) with q-value <0.1 between homozygous tie1 mutant embryos and their WT and heterozygous siblings (Fig. 5C). Among the DEGs, 1917 genes were downregulated in tie1 mutants (Table S1). Genes downregulated in the tie1 mutants included those related to blood and lymphatic vascular development, such as aplnra, lmo2 and sox18 (Fig. 5D-F). Downregulated genes were also enriched for those that function in cell migration and cell cycle progression (Fig. 5E,F), consistent with impaired migration and proliferation in tie1 mutants (Fig. 4). These results suggest that Tie1 signaling controls a series of gene expression events involved in lymphatic development.

Of note, genes downregulated in tie1 mutants included those associated with the forkhead box O (FOXO) transcription factor FOXO1 (Fig. 5G), known as a gatekeeper of endothelial quiescence (Wilhelm et al., 2016). In mammals, endothelial FOXO1 suppresses signaling by MYC. MYC signature genes repressed by nuclear FOXO1 (Wilhelm et al., 2016) were similarly downregulated in the tie1 mutants (Fig. 5G). Therefore, we considered FOXO1 as a readout of Angpt1/Tie1 signaling.

In ECs, FOXO1 is excluded from the nucleus by AKT-mediated phosphorylation downstream of PI3K signaling (Eijkelenboom and Burgering, 2013). To test whether Foxo1 is regulated downstream of Tie1, we examined the localization of Foxo1 by generating a transgenic (Tg) fish line, Tg(fli1:EGFP-foxo1a), which expresses EGFP-tagged zebrafish Foxo1a specifically in ECs. We simultaneously marked EC nuclei, by crossing with Tg(fli1:H2B-mCherry), in which EC nuclei are labeled by histone H2B-mCherry (Yokota et al., 2015). In the PCV and caudal vein plexus (CVP), endothelial EGFP-Foxo1a was predominantly detected in the cytoplasm in WT at 31 hpf before secondary sprouting (Fig. 6A). In contrast, in tie1 mutants, nuclear localization of EGFP-Foxo1a was clearly enhanced, especially in the CVP (Fig. 6A), indicating that Tie1 is important for nuclear exclusion of EGFP-Fox1a, at least in some venous ECs before secondary sprouting. During secondary sprouting, EGFP-Foxo1a was excluded from the nucleus in most of sprouting ECs of WT and heterozygous tie1^+/− embryos (Fig. 6B,C). Even when EGFP-Foxo1a was first localized in the nucleus prior to sprouting, it was exported from the nucleus before or during the sprouting (Fig. 6D). However, EGFP-Foxo1a accumulated in the nuclei of sprouting ECs in tie1^−/− embryos (Fig. 6B,C), where their nuclear migration was significantly blocked (Fig. 4A,B). These results indicate that Tie1 signaling induces nuclear exclusion of EGFP-Foxo1a in the sprouting venous ECs. To explore whether Tie1-dependent nuclear exclusion of EGFP-Foxo1a is mediated by the PI3K/Akt pathway, we examined the effect of PI3K signaling inhibition. Treatment with LY294002, a PI3K inhibitor, altered the localization of EGFP-Foxo1a in secondary sprouting ECs from cytoplasmic to nuclear (Fig. 6E). Our quantitative data revealed that EGFP-Foxo1a underwent a cytoplasmic-to-nuclear shift in 89.5% of secondary sprouting ECs after treatment with LY294002 (Fig. 6F). In addition, fluorescence intensity of nuclear EGFP-Foxo1a in secondary sprouting ECs was enhanced by LY294002 treatment but not by DMSO treatment (Fig. 6G). Therefore, nuclear exclusion of EGFP-Foxo1a during secondary sprouting is mediated by PI3K. Furthermore, after EGFP-Foxo1a entered the nuclei, these budding ECs retracted into the PCV (Fig. 6E; in 7/8 ECs), highlighting the importance of the PI3K pathway in secondary sprouting. Collectively, our results indicate that Tie1 signaling is activated in secondary sprouting ECs and induces nuclear exclusion of FOXO1 through the PI3K pathway.

Similar to tie1 mutants, homozygous vegfr3 mutants also show reduced vISV formation, reduced Prox1 expression, and loss of trunk lymphatic vessel development (Hogan et al., 2009; Shin et al., 2016). Therefore, we examined the relationship between Angpt1/Tie1 signaling and Vegfc/Vegfr3 signaling in lymphatic development. Vegfc/Vegfr3 signaling induces phosphorylation of ERK, thereby leading to lymphatic differentiation and secondary sprouting in the trunk of zebrafish (Koltowska et al., 2015; Shin et al., 2016). In tie1 mutants, phosphorylation of ERK was significantly reduced in secondary sprouting ECs (Fig. 7A,B). This result suggests that Vegfc/Vegfr3/ERK signaling is attenuated in tie1 mutants. To determine whether the tie1 mutant phenotypes could be explained by reduced Vegfc/Vegfr3 signaling, we examined whether enhancing Vegfc/Vegfr3 signaling could rescue the tie1 mutant phenotypes. When vegfc was overexpressed in the arterial ECs of tie1 mutant [using Tg(flt1:Gal4FF);Tg(10xUAS:Vegfc) double-transgenic embryos] (Koltowska et al., 2015), the phenotypes of decreased secondary sprouts, vISV formation, and PL coverage in tie1 mutants were partially rescued (Fig. 7C-F). Furthermore, the loss of secondary sprouts induced by inhibition of PI3K downstream of the Tie1 receptor was also partially rescued by overexpression of vegfc (Fig. S8A,B). These results indicate that the impaired lymphatic development due to loss of Tie1 signaling can be explained, at least in part, by reduced responsiveness of Vegfc/Vegfr3 signaling.

In the present study, we performed comprehensive analyses of angiopoietin and Tie mutants in zebrafish and found that Angpt1 and Tie1 are important for trunk lymphatic development. In contrast, we could not find any roles of Angpt2a, Angpt2b and Tie2, and mutants of any of the genes did not show any additional phenotypes. We also demonstrate binding between Angpt1 and zTie1 by in vitro binding assays. Thus, our results demonstrate that Angpt1 can serve as a ligand for zTie1. In addition, we show that zTie1 functions as an RTK that undergoes auto-phosphorylation upon angiopoietin stimulation. Therefore, our results show that Angpt1/Tie1 signaling functions in zebrafish lymphatic development.

angpt1 mutants and tie1 mutants exhibit similar phenotypes, including reduced secondary sprouting, decreased vISV and PL formation, as well as cardiac edema. However, all of those phenotypes are milder in angpt1 mutants than in tie1 mutants. Considering that Angpt2a and Angpt2b do not compensate for the loss of angpt1, this suggests that molecules other than angiopoietin might regulate Tie1 cooperatively or independently of Angpt1. It has been recently shown (Hußmann et al., 2023) that Tie1 has physical and genetic interaction with Svep1 (also known as Polydom), an extracellular matrix indispensable for lymphatic development in mice and zebrafish (Hußmann et al., 2023; Karpanen et al., 2017; Morooka et al., 2017; Sato-Nishiuchi et al., 2023). Similar to tie1 mutants, svep1 mutant fish exhibit defects in the trunk lymphatic development (Karpanen et al., 2017; Morooka et al., 2017). In addition, a recent report shows that mammalian SVEP1 can promote LEC migration via TIE1, but without apparent TIE1 phosphorylation (Sato-Nishiuchi et al., 2023). Therefore, Svep1 is a promising candidate for Tie1 activators other than Angpt1. Given that Ang1 binds to Svep1 in mice (Morooka et al., 2017), it is possible that Angpt1 and Svep1 coordinately regulate Tie1 signaling to control proper lymphatic development.

We here show that Angpt1/Tie1 signaling and Vegfc/Vegfr3 signaling cooperatively regulate the development of trunk lymphatic vessels, including PL migration, TD formation, and development of the dorsal longitudinal lymphatic vessel. Tie1 regulates trunk lymphatic development, at least in part, through the modulation of Vegfr3 signaling. By contrast, in facial lymphatic development, Tie1 and Vegfr3 signaling act separately in a different context of lymphatic vessel formation (Hußmann et al., 2023). Tie1, together with Svep1, regulates the facial collecting lymphatic vessel independently of Vegfc, whereas Vegfc is essential for development of the lymphatic branchial arches, lateral and medial facial lymphatics, and otolithic lymphatic vessels (Hußmann et al., 2023). These results indicate that regulation of lymphatic development varies in different lymphatic vascular beds, and additional efforts will have to be made to understand how Tie1 signaling and Vegfc/Vegfr3 signaling, sometimes in concert and sometimes independently, regulate lymphangiogenesis. During trunk lymphangiogenesis, Tie1 is required for nuclear exclusion of Foxo1a downstream of PI3K. It has also been reported that Vegfr3 signaling activates the PI3K pathway in mammalian ECs (Makinen et al., 2001). Therefore, Tie1 signaling and Vegfr3 signaling may coordinately regulate the PI3K-Foxo1 axis as a signaling hub for lymphangiogenesis in zebrafish. To clarify the relationship between Tie1 and Vegfr3 signaling, it will be important to elucidate the downstream molecular mechanisms of each signaling pathway in more detail.

Tie1 plays a crucial role in PI3K-mediated nuclear exclusion of Foxo1a in secondary sprouting ECs in zebrafish. Wilhelm et al. (2016) have reported that murine Foxo1 acts as a gatekeeper of endothelial quiescence, which decelerates metabolic activity through Myc inhibition. Consistent with this, our RNA-seq results identify that Tie1 signaling positively regulates the expression of Myc target genes and metabolic-related genes. Therefore, the Tie1/PI3K signaling pathway may regulate lymphatic development through changes in metabolic state of ECs. It would be interesting to investigate the importance of metabolic regulation in lymphangiogenesis that might be regulated by Angpt1/Tie1 signaling.

We show that Angpt1/Tie1 signaling functions in EC migration, proliferation, and lymphatic specification during early lymphangiogenesis. As reduced LEC numbers and defects in lymphatic sprouting have been reported in maternal-zygotic mutants of prox1a, an essential regulator of LEC differentiation (Koltowska et al., 2015), there is a possibility that the defects in sprouting and migration in tie1 mutants might be affected by an earlier defect in LEC differentiation. How these cellular behaviors are coordinately regulated by Tie1 signaling constitutes an interesting challenge for the future.

In zebrafish that lack functional Tie2, Tie1 plays an essential role in trunk lymphangiogenesis. Similarly, in mammals, TIE1 appears to have a unique role in lymphatic vascular development (D'Amico et al., 2010; Donnan et al., 2021; Qu et al., 2010, 2015; Shen et al., 2014); however, how mammalian TIE1 regulates lymphangiogenesis in concert with, or independently of, TIE2 is not fully understood. As many teleosts have lost the tie2 gene from their genome (Jiang et al., 2020), it appears likely that in teleosts the role of lymphatic endothelial Tie signaling resides solely with the Tie1 receptor.

Zebrafish of the AB strain (Danio rerio) were maintained and bred in 28°C water (pH 7.25 and conductivity 500 µS) with a 14 h on/10 h off light cycle. Embryos and larvae were incubated at 28°C in E3 medium. All zebrafish husbandry was performed under standard conditions according to institutional (National Cerebral and Cardiovascular Center) and national (Japan) ethical and animal welfare regulations. The experiments using zebrafish were approved by the animal committee of the National Cerebral and Cardiovascular Center (No.22054) and performed according to our institutional regulation.

The following transgenic and mutant zebrafish lines were used for this study: Tg(fli1:EGFP)^y1 (Lawson and Weinstein, 2002), Tg(kdrl:EGFP)^s843 (Jin et al., 2005), Tg(−5.2lyve1b:DsRed)^nz101 (Okuda et al., 2012), Tg(fli1:H2B-EGFP)^ncv69 (Ando et al., 2019), Tg(fli1:H2B-mCherry)^ncv31 (Yokota et al., 2015) and tie2^hu1667 (Gjini et al., 2011). Tg(flt1:Gal4FF)^ncv555 and Tg(fli1:EGFP-foxo1a)^ncv557 lines were generated by injecting the pTol flt1:Gal4FF plasmid and pTol fli1:EGFP-foxo1a plasmid with Tol2 transposase mRNA into one-cell-stage embryos, respectively, as described previously (Nakajima et al., 2023). Tg(10×UAS:Vegfc)^ncv556 line was established using the same plasmid used to generate Tg(10×UAS:Vegfc)^uq2bh (Koltowska et al., 2015). Throughout the text, all Tg lines used in this study are simply described without their line numbers. For example, Tg(fli1:H2B-GFP)^ncv69 is abbreviated to Tg(fli1:H2B-GFP).

The knockout alleles ncv110 for angpt1, ncv129 for angpt2a, ncv130 for tie1, and ncv131 for angpt2b genes were generated by TALEN techniques as described below.

TALENs targeting tie1, angpt1, angpt2a and angpt2b were designed using TAL Effector Nucleotide Targeter 2.0 software (Doyle et al., 2012) and constructed using the Golden Gate assembly method (Cermak et al., 2011). The TALENs were cloned into an RCIscript-GoldyTALEN vector. RCIscript-GoldyTALEN was a gift from Danial Carlson and Stephen Ekker (Addgene plasmid #38142) (Carlson et al., 2012). TALEN mRNAs were in vitro transcribed from SacI-linearized expression plasmids with T3 RNA polymerase using a mMessage mMachine mRNA kit (Thermo Fisher Scientific). Embryos, injected with 30-60 pg of the TALEN mRNAs at the one-cell stage, were raised to adulthood and crossed with WT AB to identify germline-mutated founders.

Screening for founders was conducted by genomic PCR and subsequent sequencing using the following primer sets: 5′-TCTTCCAGATGCTGTCATGG-3′ and 5′-ATCCCACTGTGGTCAAAACC-3′ for tie1; 5′-TGTGAGTTTTCCGTCCCATC-3′ and 5′-ATAACCGTGTAATCATCCAG-3′ for angpt1; 5′-GAGCAAATATGTTGAGATCATGGA-3′ and 5′-CTTTTGCAGCCACTGTGTGT-3′ for angpt2a; 5′-CTGCCGTGTCTGTGCTACCT-3′ and 5′-AAGCACCATGACTATTTTCCTTG-3′ for angpt2b. For genotyping mutants, PCR analyses of genomic DNAs were routinely performed using the same primer set and these amplified PCR products were analyzed using an MCE-202 MultiNA microchip electrophoresis system (Shimadzu) with the DNA-500 reagent kit (Shimadzu).

For confocal imaging, the pigmentation of embryos was suppressed by the addition of 1-phenyl-2-thiourea (Sigma-Aldrich) in E3 media. Embryos were dechorionated and mounted in 1% low-melting agarose poured on a 35-mm-diameter glass-based dish (Asahi Techno Glass), and then anesthetized in 0.02% tricaine in E3 medium. Confocal imaging in Figs S2C, S3E, S6C, S8A and Movie 3 was performed on an Andor Dragonfly spinning disc confocal (Andor Technology Ltd) based on ECLIPSE FN1 upright microscope (Nikon), equipped with a water-immersion LWD 16×/0.80 NA lens (Nikon), Zyla4.2 PLUS USB 3.0 sCMOS cameras (Andor Technology Ltd) and P-725.4 PIFOC piezo nano-positioner (Physik Instrumente) regulated with Fusion software (Andor Technology Ltd). Other confocal images were taken with a FluoView FV1000, FV1200 and FV3000 confocal upright microscope system (Olympus) equipped with water-immersion XLUMPlan FL N 20×/1.00 NA objective lenses (Olympus) and a multi-alkali or GaAsP photomultiplier tube regulated with FluoViewASW software (Olympus). The 405 nm, 473 nm and 559 nm laser lines were used. Images were acquired sequentially to avoid cross-detection of the fluorescent signals. Image files were processed and analyzed with FV10-ASW4.2 viewer (Olympus), ImageJ (Schindelin et al., 2012), and IMARIS 9.2.1, 9.5.1, 9.9.1, or 10.1.0 software (Oxford Instruments). Stereomicroscopic images were taken with an Olympus SZX16 stereomicroscope with an Olympus DP2-BSW camera and CellSens imaging software (Olympus).

Whole-mount immunostaining of zebrafish embryo was performed in accordance with the methods of Le Guen et al. (2014) with some modifications. Briefly, embryos were fixed with 4% paraformaldehyde in PBS for 1 h at room temperature, and washed three times in ice-cold methanol, and then incubated 1 h in 3% H₂O₂ in methanol on ice. Embryos were washed three times in ice-cold methanol and kept in methanol at −20°C for 1 week. The fixed tissues were washed three times in PBS containing 0.2% Triton X-100 (PBS-T) for 10 min at room temperature, blocked in PBS-T containing 1% bovine serum albumin for 1 h at 4°C, and incubated with primary antibodies for 2 days at 4°C. After six washes in PBS-T for 30 min at 4°C, the samples were incubated with Alexa Fluor-conjugated secondary antibodies overnight at 4°C. After another six washes in PBS-T for 30 min at 4°C, the embryos were mounted in 1% low-melting agarose poured on a 35-mm-diameter glass-based dish (Asahi Techno Glass), and were analyzed with a Fluoview FV3000 (Olympus) confocal microscope. Primary antibodies used were: rabbit anti-Prox1 (11-002, Angiobio; 1:500), rabbit anti-phospho-p44/42 MAPK (ERK1/2) (4370, Cell Signaling Technology; 1:500) and mouse anti-RFP (M155-3, MBL; 1:500) or anti-GFP (632381, Clontech; 1:1000). Secondary antibodies used were: anti-rabbit Alexa Fluor 488 (A11034, Invitrogen; 1:1000) and anti-mouse Alexa Fluor 546 (A11030, Invitrogen; 1:1000). TUNEL (terminal deoxynucleotidyl transferase dUTP nick end labeling) assay was performed using an In Situ Cell Death Detection Kit, fluorescein (Roche) according to the manufacturer's instructions. The embryos were counterstained with DAPI during the fifth wash following secondary antibody staining. To quantify the fluorescence intensity of phosphorylated ERK (pERK), we set a spherical region of interest (ROI) in the center of lyve1:DsRed (anti-DsRed)/DAPI double-positive EC nucleus and measured the mean fluorescence intensity of anti-pERK staining in the spherical ROI.

Embryos were fixed with 10% neutral buffered formalin overnight at room temperature. After fixation, the solution was changed to 50% methanol/PBS-Tw (PBS with 0.1% Tween 20) for 10 min, then changed to 100% methanol at room temperature, and then stored in 100% methanol at −20°C overnight. After rehydration, embryos were washed three times for 10 min in PBS-Tw containing 1% bovine serum albumin. Embryos were treated with five drops of protease plus (ACDbio) for 40 min at 40°C, and subsequently washed three times for 10 min in PBS-Tw. For probe hybridization, embryos were treated in the tie1-C1 and angpt1-C3 probes (ACDbio; 1:50) overnight at 40°C. Embryos were washed and subjected to hybridization following the workflow in ACDbio. Embryos were stored in PBS-Tw at 4°C prior to confocal imaging.

A cDNA fragment encoding full-length zebrafish Tie1 was amplified by PCR using cDNA library from 72 hpf zebrafish embryo as a template, then inserted into pCS2+ vector (Clontech) fused with C-terminal HA tag to generate pCS2+-zTie1-HA plasmids. pCS2+-zTie1KD-HA vector encoding a kinase-deficient mutant of Tie1 (K854R) was generated using an inverse PCR method with KOD-plus DNA Polymerase (TOYOBO).

cDNA fragments encoding the extracellular region of zebrafish Tie1 (amino acids 1-747) and zebrafish Tie2 (amino acids 1-745) followed by the Fc region of human immunoglobulin G (amino acids 100-330) and a His tag were amplified by PCR, and inserted into the pCS2+ vector to generate pCS2+-zTie1-Fc-His and pCS2+-zTie2-Fc-His plasmids, respectively. The pCS2+-Fc-His plasmid was generated from the pCS2+-zTie1-Fc-His plasmid without the tie1 sequence after its signal sequence (amino acids 17-747).

cDNA fragments encoding zebrafish Angpt1, Angpt2a and Angpt2b without signal sequence were amplified by PCR using cDNA library from 72 hpf zebrafish embryo as a template, then inserted into the pCS2+ vector fused with N-terminal preprotrypsin leader sequence and FLAG tag to generate pCS2+-FLAG-zAngpt1, pCS2+-FLAG-zAngpt2a and pCS2+-FLAG-zAngpt2b plasmids, respectively.

pTol flt1:Gal4FF plasmid was constructed by inserting the Gal4FF cDNA into the pTol flt1 vector (Kwon et al., 2013). A cDNA fragment encoding zebrafish Foxo1a was amplified by PCR and subcloned into pEGFP-C3 vectors (Takara Bio Inc.) to generate the expression plasmids. Then, EGFP-foxo1a cDNA was subcloned into the pTol2-fli1 vector (Kwon et al., 2013) to generate the pTol fli1:EGFP-foxo1a plasmid.

To produce recombinant zTie1-Fc-His, zTie2-Fc-His, Fc-His, FLAG-zAngpt1, FLAG-zAngpt2a and FLAG-zAngpt2b, 293T cells were transfected with pCS2+-zTie1-Fc-His, pCS2+-zTie2-Fc-His, pCS2+-Fc-His, pCS2+-FLAG-zAngpt1, pCS2+-FLAG-zAngpt2a and pCS2+-FLAG-zAngpt2b plasmids, respectively, using 293fectin transfection reagent (Gibco), and cultured in DMEM (Wako) supplemented with 10% fetal bovine serum (FBS) for 2 days.

Binding of zTie to zAngpt was performed in accordance with the methods of Fukuhara et al. (2008) with some modifications (Fig. 2B). Initially, protein G sepharose beads (GE Healthcare Life Science) were incubated with supernatants of Fc-His, sTie1-Fc-His or sTie2-Fc-His for 2 h at 4°C. After washing four times with binding buffer (50 mM Tris-HCl at pH 7.5, 100 mM NaCl, 0.02% Triton X-100), protein-bound beads were incubated with supernatants of FLAG-Angpt1, FLAG-Angpt2a or FLAG-Angpt2b for 2 h at 4°C. After washing four times with binding buffer, the precipitates were subjected to western blot analysis with anti-penta-His (34660, QIAGEN; 1:2000) and anti-FLAG M2 (F1804, Sigma-Aldrich; 1:4000) antibodies to quantify the co-precipitated Tie and Angpt proteins, respectively. Proteins reacting with primary antibodies were visualized by the Immobilon forte Western HRP substrate (Millipore) for detecting peroxidase-conjugated secondary antibodies and analyzed with an ChemiDoc Touch (Bio-Rad).

293T cells were cultured until they were 90% confluent. Cells were transfected with pCS2+-zTie1-HA and pCS2+-zTie1KD-HA plasmids using 293fectin transfection reagent, and cultured in DMEM supplemented with 10% FBS for 9-12 h. After starvation in 0.5% FBS in DMEM overnight, the cells were stimulated with 500 ng/ml COMP-ANG1 (Cho et al., 2004) for 20 min. Cells were lysed in RIPA buffer containing 50 mM Tris-HCl at pH 7.5, 100 mM NaCl, 1% NP-40, 5% glycerol, 5 mM sodium orthovanadate, 10 mM sodium fluoride, 1× protease inhibitor cocktail, and 1 mM DTT at 4°C for 10 min, and centrifuged at 15,000 g for 15 min at 4°C. The supernatant was used for western blot analysis.

To detect the phosphorylated Tie1, aliquots of total cell lysate were subjected to SDS-PAGE and western blot analysis with anti-phosphotyrosine antibody (9411, Cell Signaling Technology; 1:3000). The total contents of Tie1-HA and Tie1KD-HA were assayed in a parallel run using anti-HA antibody (3F10, Roche; 1:2000). As a loading control, mouse anti-β-actin antibody (A5441, Sigma-Aldrich; 1:6000) was used. Proteins reacting with primary antibodies were visualized by the Immobilon forte Western HRP substrate (Millipore) for detecting peroxidase-conjugated secondary antibodies and analyzed with an ChemiDoc Touch (Bio-Rad). Image files were processed and analyzed with ImageJ.

The trunk and tail region from 47 hpf Tg(kdrl:EGFP);Tg(lyve1:DsRed) tie1^−/− embryos with blood flow and their WT and tie1^+/− siblings were manually dissected and collected into a low-binding 12-well plate. After washing in E3 media, samples were incubated with 1 ml of lysis solution containing collagenase P (Roche) and TrypLE Express Enzyme (Gibco) for 20 min under occasional pipetting. Digestion was terminated with 200 µl of stop solution (PBS with 30% FBS and 6 mM calcium chloride). The dissociated cells in suspension medium [Phenol Red-free DMEM (Life Technologies) with 1% FBS, 0.8 mM calcium chloride, 50 U/ml penicillin and 0.05 mg/ml streptomycin] were subjected to cell sorting using a FACS Aria III cell sorter (BD Bioscience). EGFP and DsRed double-positive cells were collected as venous and lymphatic endothelial cells for further RNA preparation.

Total RNA was prepared from kdrl:EGFP⁺/lyve1:DsRed⁺ ECs of tie1^−/− embryos and their WT and tie1^+/− siblings (tie1^+/?) using the NucleoSpin XS kit (Macherey-Nagel, 740902.50) according to the manufacturer's instructions. Reverse transcription and cDNA library preparation were performed with a SMART-Seq v4 Ultra Low Input RNA Kit for Sequencing (Clontech, 634888). cDNA was fragmented with a Covaris S 220 instrument (Covaris). Libraries for RNA-seq were prepared using a NEBNext Ultra II Directional RNA Library Prep Kit for Illumina (New England Biolabs, E7760) and sequenced on the NextSeq500 (Illumina) as 75 bp single-end reads. RNA-seq data were trimmed using Trim Galore version 0.6.6 (https://www.bioinformatics.babraham.ac.uk/projects/trim_galore/) and Cutadapt version 2.8 (Martin, 2011). The quality of reads was checked and filtered using FastQC version 0.11.9 (http://www.bioinformatics.babraham.ac.uk/projects/fastqc). The reads were mapped to a reference genome GRCz11 using HISAT2 version 2.2.1 (Kim et al., 2019), and the resulting aligned reads were sorted and indexed using SAMtools version 1.7 (Li et al., 2009). Relative abundances of genes were measured in TPM using StringTie version 2.1.4 (Kovaka et al., 2019; Pertea et al., 2015). DEGs were identified from TPM+1 values using an R package, TCC-GUI (Su et al., 2019; Sun et al., 2013). Genes with q-value <0.1 (10% false discovery rate, FDR) were considered to be differentially expressed. To identify the gene ontology (GO) biological process, enrichment analysis was performed on the list of 1942 differentially expressed genes using Metascape v.3.5 (Zhou et al., 2019).

For the quantification of PL coverage, TD coverage, anastomosis, and the number of ECs, secondary sprouts, vISVs, and A-V connections across six segments in the trunk, we analyzed a 6-somite-long trunk area immediately anterior to the caudal end of the yolk extension. Primary and secondary sprouts emerge bilaterally from the DA and PCV, respectively. Thus, to focus on angiogenesis and lymphangiogenesis occurring on one side, our analyses were performed on the left side of the embryo.

PEG-coated FMs were synthesized to prevent clearance by ECs. Carboxylate-modified FluoSpheres [particle size: 0.48 μm, red fluorescent (580/605), Invitrogen] (250 μl, 20 mg/ml, 0.1575 meq/g of carboxylic groups) were dispersed in 1 ml of 100 mM MES buffer (pH 6) and centrifuged at 10,000 g for 10 min. The supernatant was removed, and the microspheres were resuspended in 900 μl MES buffer. This solution was mixed with 100 μl of MES buffer containing 43 mg/ml N-hydroxysulfosuccinimide (Sulfo-NHS, Thermo Fisher Scientific) and 15 mg/ml 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide hydrochloride (EDC HCl, Tokyo Chemical Industry). After stirring for 30 min at room temperature, 80 mg of methoxy-PEG-(CH₂)₂-NH₂ and 111 μl of 1 M sodium phosphate solution (pH 9.4) were added. After stirring for 12 h at room temperature, the solution was centrifuged at 15,000 g for 10 min and the supernatant was removed. The microspheres were washed three times by resuspending in 1 ml Dulbecco's PBS (D-PBS), centrifuging the solution at 15,000 g for 10 min and discarding the supernatant. The microspheres were resuspended in 1 ml D-PBS and sonicated for 5 min using a Taitec VP-050 ultrasonic homogenizer (1 s intervals, PWM 50%). The diameter of the microspheres was measured by dynamic light scattering using a Malvern Zetasizer Nano.

To measure the flow velocity in the PCV, PEG-coated FMs were injected into the common cardinal vein of Tg(kdrl:EGFP);Tg(Lyve1:DsRed) embryos at 45-46 hpf using an IM 300 Microinjector (Narishige). After injection, high-speed confocal imaging of the trunk region was performed using the Andor Dragonfly spinning disc confocal microscope (Andor Technology Ltd) at 200 frames per second for 10 s. Each PEG-coated FM was tracked in the PCV to measure the velocity of PEG-coated FMs using IMARIS 9.9.1 or 10.1.0 software as shown in Movie 3. The average velocity of all tracked PEG-coated FMs was quantified as the flow velocity in each embryo (Fig. 3F).

To analyze the secondary sprouting in the same embryo as shown in Fig. 3G and Fig. S6C, the trunk of the same embryo was observed by 3D confocal imaging before injection of the PEG-coated FMs. Embryos were genotyped by genomic PCR.

To quantify the fluorescence intensity ratio of nuclear EGFP-Foxo1a in secondary sprouting ECs as shown in Fig. 6G, Tg(fli1:EGFP-foxo1a);Tg(fli1:H2B-mCherry) embryos were treated with DMSO or 30 µM LY294002 (Merck) at 34 hpf. Confocal stack fluorescence images in the trunk regions were acquired before treatment (34 hpf) and 2 h after treatment (36 hpf) using an FV3000 confocal microscope in the same acquisition condition and were analyzed with IMARIS 9.9.1 or 10.1.0 software (Oxford Instruments). We focused on ECs that underwent secondary sprouting at 34 hpf and quantified their fluorescence intensity at 34 hpf and 36 hpf in the same ECs. To quantify the fluorescence intensity of nuclear EGFP-Foxo1a, we set a spherical ROI of 2 μm in the center of a fli1:EGFP-Foxo1a/fli1:H2B-mCherry double-positive EC nucleus and measured the mean fluorescence intensity in the spherical ROI. To calculate the fluorescence intensity ratio, the fluorescence intensity at 36 hpf was divided by that at 34 hpf in the same ECs across six segments in the trunk (Fig. 6G). Similarly in Fig. 6F, we focused on ECs that underwent secondary sprouting at 34 hpf. To examine the localization change of EGFP-Foxo1a after LY294002 and DMSO treatment, EGFP-Foxo1a localization was analyzed at 34 hpf (before treatment) and at 36 hpf (2 h after treatment) in the same ECs by tracking EC nuclei during time-lapse imaging. EGFP-Foxo1a localization was regarded as nuclear when accumulated in the nucleus, and as cytoplasmic when cytoplasmic signals were stronger than or equal to nuclear EGFP-Foxo1a signals.

Data were analyzed using GraphPad Prism software or Excel and are presented as mean±s.d. Sample numbers and the statistical methods are indicated in figure legends.

We thank Kenta Terai, Takefumi Kondo and Yukari Sando (NGS core facility of the Graduate Schools of Biostudies, Kyoto University) for supporting the RNA-seq analysis. We are grateful to B. Hogan (The University of Melbourne) and K. Koltowska (Uppsala University) for the plasmid for Tg(10×UAS:Vegfc); K. Koltowska (Uppsala University) for the detailed protocol for Prox1 immunostaining; K. Kawakami (National Institute of Genetics) for the Tol2 system; G. Y. Koh [Korea Advanced Institute of Science and Technology (KAIST)] for COMP-ANG1; F. le Noble [Karlsruhe Institute of Technology (KIT)] and T. Yamamoto (Kyoto University) for helpful advice; M. Sone, T. Satoh, K. Hiratomi, T. Babazono and E. Hanimura for technical assistance; and K. Shioya and K. Konno for fish care.

The peer review history is available online at https://journals.biologists.com/dev/lookup/doi/10.1242/dev.202269.reviewer-comments.pdf