Role of synectin in lymphatic development in zebrafish and frogs

The lymphatic vasculature regulates interstitial fluid homeostasis, fat resorption, immune defense, inflammation and cancer metastasis. The molecular basis of the lymphatic development remains incompletely understood.¹  Lymphatic endothelial cells (LECs) derive from venous blood endothelial cells (BECs).^2-5  Sox18⁶  and Prox-1,^2-5  as well as miRNAs,⁷  play a key role in this process. Intriguingly, despite their venous origin, formation of lymphatics relies in part on signals that participate in arterial development. For instance, the forkhead transcription factors Foxc1/2 are required for arterial specification and sprouting of LECs.^8,9  In addition, the PDZ interaction site in EphrinB2, itself a marker of arterial endothelial cells (ECs), is essential for lymphatic development.¹⁰  Similarly, Dll4/Notch signaling regulates arterial and lymphatic development.¹¹ 

Recently, we identified the PDZ domain–containing scaffold protein synectin (GIPC1) as a regulator of arterial but not venous growth,¹²  at least in part through control of Vegfa signaling.¹³  Prompted by the finding that similar factors control arterial and lymphatic development, we studied whether synectin also regulated lymphatic development. Although PDZ domain–dependent signaling is crucial for lymphatic development,¹⁰  a role for synectin in lymphatic development has not been documented yet. We therefore explored a role of synectin in this process using zebrafish and Xenopus tadpole models.

Fli1:eGFP^y1,¹⁴ Gata-1:DsRed¹⁵ zebrafish and PLCγ^y1016  zebrafish were maintained under standard conditions. All morpholinos were previously reported and purchased from Gene Tools (supplemental Table 1; available on the Blood Web site; see the Supplemental Materials link at the top of the online article). Different doses of morpholinos were injected into single- to 4-cell stage embryos, as previously described.¹²  Phenotyping is described in supplemental material. All animal experimentation was approved by the Katholieke Universiteit Leuven institutional ethical committee.

The generation and characterization of transgenic Flk1:eGFP Xenopus laevis frogs will be reported elsewhere (manuscript in preparation). Eggs were obtained by natural mating of hormonally induced Flk1:eGFP females and wild-type males and injected with different doses of synectin-specific or control morpholinos (Gene Tools, supplemental Table 1) into the 2-cell stage.¹⁷  Nonoverlapping translational blocking synectin morpholinos were designed based on published GenBank Xenopus laevis sequences (NM_001088594) and morpholino efficiency was tested using an in vitro luciferase reporter assay¹⁷  and immunoblotting (supplemental Methods). Tadpoles were kept in tadpole growth medium (0.1 × MMR) at 18°C until gastrulation was completed and from then on at 22°C.¹⁷  Phenotyping is described in supplemental Methods.

Fli1:eGFP^y1 zebrafish embryos were fixed overnight in 4% paraformaldehyde/PBS at 4°C and processed for paraffin- or cryosectioning. Sections (7 μm) were stained using a polyclonal chicken anti–human synectin antibody (Abcam) and a rabbit anti-GFP antibody (Torrey-Pines Biolabs), which were then detected using biotinylated anti-chicken (Abcam) and Alexa Fluor 488–conjugated anti-rabbit secondary antibodies, respectively. Sections were mounted with Vectashield DAPI (4′,6-diamidino-2-phenylindole; Vector Laboratories).

For Xenopus in situ hybridizations, embryos were fixed in Memfa fixative and whole-mount in situ hybridization using antisense probes for Prox-1¹⁷  and synectin (primer sequence, supplemental Table 1) was performed as described.¹⁷  Analysis of Prox-1–stained areas in the tadpole tail was performed as reported (supplemental Methods).¹⁷  For zebrafish whole-mount in situ hybridization, dechorionated embryos were fixed overnight in 4% paraformaldehyde at 4°C. In situ hybridization was performed as described,¹²  using antisense probes for EphrinB2a,¹² Gridlock (Grl),¹⁸ EphB4,¹⁹ Flt4,¹²  and Tie-2.²⁰  Both sections and whole-mounts were visualized on a Zeiss Axioplan 2 imaging microscope.

Human microvascular lung endothelial cells (HMVEC-Lly; Lonza, Invitrogen) and adult dermal LECs (HMVEC-dLyAd; Lonza, Invitrogen), telomerase transfected dermal LECs (hTERT-HDLECs²¹ ), immortalized (i) LECs, and human umbilical vein ECs (Lonza, Invitrogen) were grown in endothelial growth medium (EGM)-2–MV medium (Lonza, Invitrogen) at 37°C. Human dermal LEC (HDLEC) cells were obtained from Promocell and cultured in endothelial medium provided by the supplier. Synectin expression was evaluated by quantitative reverse-transcription polymerase chain reaction (qRT-PCR; supplemental Table 1 primer sequences). The LEC spheroid and Sox18-mediated lymphatic reprogramming assays are described in supplemental Methods.

Absolute values were used to calculate mean ± SEM. Significance levels were calculated by unpaired Student t test, univariate, or multivariate analysis with the treatment groups as fixed factor and experiment as covariate. To determine the penetrance of the zebrafish and tadpole phenotypes, we counted the number of embryos, exhibiting the (different severities of) morphant phenotype. We used χ² analysis to determine whether the severity distribution differed between treatment groups.

Because the expression of synectin in lymphatics has not been documented yet, we analyzed its expression pattern during lymphangiogenesis in zebrafish embryos. For reasons of clarity, we will briefly explain first how lymphatics develop in zebrafish (supplemental Figure 1).^11,20,22-25  The thoracic duct (TD) is the first perfused lymphatic, located between the dorsal aorta (DA) and posterior cardinal vein (PCV). TD development initiates between 30 and 50 hours postfertilization (hpf) when secondary sprouts branch off dorsally from the PCV. Half of these sprouts are angiogenic, as they connect to primary (arterial) intersomitic vessels (aISVs), which thereby become venous ISVs (vISVs). The other half of these secondary sprouts are lymphangiogenic, as they migrate dorsally to the horizontal myoseptum, where they form (by 36-60 hpf) a transient, nonlumenized string of parachordal lymphangioblast (PL) cells, termed so because they are precursors of LECs forming the TD. Beyond 60 hpf, PL cells navigate ventrally and dorsally alongside aISVs, with ventral sprouts migrating to their location between the DA and PCV to fuse and establish the TD (3–6 days postfertilization [dpf]).

We and others previously reported that synectin is expressed in the PCV when secondary lymphangiogenic sprouts form,^12,26  but did not characterize synectin expression in lymphatic vessels. When using Fli1:eGFP^y1 embryos, which express enhanced green fluorescent protein (eGFP) in blood and lymph vessels,^14,25  we found that synectin was expressed in the DA, PCV, PL cells, and TD (Figure 1A-F) by double immunostaining for GFP and synectin in 3- and 7-dpf embryos. Furthermore, in agreement with previous findings,^12,26  synectin was detectable in the head (not shown) and in the neural tube, pronephric duct, somites, and gut (Figure 1A-F). Widespread expression of synectin has been observed in human, mouse, and Xenopus tissues.^27-29  Of note, synectin was also detectable in primary and immortalized human LECs (supplemental Figure 2), consistent with previous observations.³⁰ 

To study the role of synectin in lymphangiogenesis, we silenced its expression (synectin^KD) in Fli1:eGFP^y1 zebrafish embryos using previously characterized Syn^ATG1 and Syn^ATG2 morpholinos.¹²  As synectin regulates angiogenesis, we first determined a range of submaximal morpholino doses, which only minimally affected blood vessel development, to avoid that angiogenic defects secondarily caused lymphatic defects. At a dose of 9 ng Syn^ATG1 and 6 ng Syn^ATG2 per embryo, only 7% of synectin^KD embryos displayed subtle vascular defects, including a slightly thinner DA and a few malformed ISVs (Figure 2A-B and supplemental Table 2). The remainder of the blood vasculature in the head, trunk, tail and subintestinal vessels developed normally, had a normal size and shape, and exhibited comparable branching and density (supplemental Figure 3). Blood flow in large axial vessels and ISVs was normal in 2-dpf synectin^KDGata-1:DsRed embryos, harboring DsRed⁺ erythrocytes¹⁵  (Figure 2C-D). Furthermore, arterio-venous differentiation of large axial vessels and primary ISVs was normal in synectin^KD embryos, as evidenced by in situ hybridization for arterial and venous markers at 28 hpf (Figure 2E-L). Only morphant embryos with normal trunk circulation and body size and without developmental delay, tissue malformations, edema, or toxic signs were analyzed. (We refer to synectin^KD embryos without mentioning that silencing was incomplete.)

We studied TD formation by measuring its length at 7 dpf, when this vessel was completely formed in control embryos, using previous methods.¹¹  Briefly, the TD length was measured in 10 somites and expressed as percentage of this trunk fragment (supplemental Figure 4).¹¹  Because the penetrance of the lymphatic phenotype was variable (supplemental Note 1), we counted the fraction of embryos with severe, intermediate, or subtle lymphatic defects. For Syn^ATG1 and Syn^ATG2, the severity and penetrance of the knockdown phenotypes were dose-dependent (supplemental Tables 3–4); for reasons of brevity, only the most penetrant phenotype is shown (9 ng of Syn^ATG1). In addition, lymphatic development in zebrafish occurs in a metameric pattern, since lymphangiogenic sprouts, and hence the PL and TD, form, statistically, at every second unilateral somite segment, giving rise to discontinuous TD fragments that fuse with each other to establish a continuous perfused TD (supplemental Figure 1).¹¹ 

In control embryos, injected with a 5-base-pair mismatch morpholino (Syn^Ctr), the TD developed as a continuous lymphatic by 7 dpf (Figure 3A,C,E and supplemental Table 3). In contrast, 32% of synectin^KD embryos completely lacked a TD by 7 dpf (“severe” defects; Figure 3B,D-E and supplemental Table 3). Follow-up studies further revealed that TD development was not rescued at 10 dpf (supplemental Figure 3E-F), indicating that lymphatic formation was aborted and not merely delayed in these synectin^KD embryos. In another 14% of morphants, TD rudiments developed over only 1 to 3 somites into an incomplete string of disconnected segments (“intermediate” defects; Figure 3E and supplemental Table 3). Incomplete TD formation over 30%-90% of its normal length was further observed in 36% of synectin^KD embryos (“subtle” defects; Figure 3E and supplemental Table 3), while TD development was normal in the remainder of the morphant embryos. Because the TD develops in a metameric pattern (see above), the finding that the TD formed over for instance only 20% of its length implies that TD development was aborted in the 8 other of the 10 somites analyzed. Thus, silencing of synectin aborted lymphatic development completely in all somites in a third of all morphants, and partially in a fraction of the somites in another 50% of the morphants.

We then explored whether the TD defect in synectin^KD morphants was related to a defect in PL formation by quantifying its length in 10 somites in Fli1:eGFP^y1 embryos at 60 hpf (using a similar method as used for the TD). In control embryos, PL cells were detected in nearly every somite segment (Figure 4A,C and supplemental Table 4). By contrast, the PL was absent in 28% of synectin^KD embryos (Figure 4B-C and supplemental Table 4) or formed over only 10%-30% of its normal length in another 27% of morphants (Figure 4C and supplemental Table 4). An additional 40% of synectin^KD embryos displayed PL cells in only 3 to 9 somites, while in the remaining 5%, the PL string developed normally. The close correlation between TD and PL defects suggests that silencing of synectin impaired early lymphatic formation.

We then evaluated whether absence of PL cells resulted from lymphangiogenic sprout defects. In case lymphangiogenic but not angiogenic sprout formation would be defective, we expected that 50% of all secondary sprouts would be affected/absent. Several assays were used to address this issue. First, we used whole-mount in situ staining for Tie-2 to identify both angiogenic and lymphangiogenic secondary sprouts at 48 hpf. In 81% of control embryos (n = 66), 8 to 10 secondary sprouts developed in the 10 unilateral somites analyzed, (ie, in nearly every somite [Figure 4D-D′]); in another 17%, 6 to 8 secondary sprouts developed, while very rarely (in 2%), only half the number of secondary (4 to 5) sprouts formed. By contrast, in 22% of synectin^KD embryos, only half of the normal number of secondary sprouts formed (n = 74; P < .001; Figure 4E-E′), while in another 41%, only 6 to 8 Tie-2⁺ sprouts developed (N = 74; P < .001 vs control), while the remainder fraction displayed no secondary sprouting defects. Notably, comparable fractions of morphant embryos suffered severe secondary sprout defects, and lacked a PL and TD, suggesting that the latter and former processes were linked.

Second, we used the genetic PLCγ1^y10 model to visualize secondary sprouts in isolation of primary ISVs. Indeed, in these mutant embryos, aISVs fail to form, while secondary sprouts emanate normally.¹⁶  Synectin silencing in PLCγ1^y10xFli1:eGFP^y1 embryos showed that only half the normal number of secondary sprouts developed by 48 hpf (sprouts in 10 somites: 7.3 ± 0.9 in control vs 4.0 ± 0.8 in synectin^KD; N = 53-57 embryos, P < .01; Figure 4F-H). Thus, also with this method, approximately 50% fewer secondary sprouts branched off in synectin^KD embryos.

We then assessed whether synectin silencing impaired lymphangiogenic sprouting using an indirect method (ie, by counting the number of ISVs that were connected to the PCV [vISVs]), as the latter are established when angiogenic secondary sprouts connect to primary ISVs. Because a comparable fraction of lymphangiogenic and angiogenic sprouts normally emanates from the PCV, a normal number of vIVSs, in combination with a reduced total number of secondary sprouts, would be evidence for defective formation of lymphangiogenic sprouts. In control 4 dpf Fli1:eGFP^y1 embryos (N = 29), 55% of the ISVs were connected to the PCV and thus venous, while the remaining fraction was connected to the DA and hence arterial (Figure 4I). Notably, in the most severely affected synectin^KD embryos, lacking the PL completely or forming only a partial PL in 30% of the somites (N = 18), half of the ISVs was still connected to the PCV (Figure 4I), showing that angiogenic sprouts normally emanated from the PCV and implying that lymphangiogenic sprouting was impaired in synectin^KD embryos. As a lymphangiogenic sprouting defect might reflect a defect in migration and/or emergence of the lymphatic lineage, but the dynamics of Prox-1 expression during lymphatic differentiation can be better visualized in frog than zebrafish embryos, we resorted to tadpoles, a previously validated model to study lymphangiogenesis,¹⁷  to dissect the underlying mechanism in more detail.

We first analyzed synectin expression in tadpoles. In agreement with previous findings,²⁸  whole-mount in situ hybridization at stage 35-40 revealed expression of synectin in the neural tube, eye, brain, otic vesicle, somites, and branchial arches, as well as at locations of the DA, PCV, and dorsal longitudinal anastomosing vessel (DLAV), where the ventral caudal lymph vessel (VCLV) and dorsal caudal lymph vessel (DCLV) develop in the frog embryo (not shown and supplemental Figure 5A-F). We also used a novel ‘LEC labeling’ technique in transgenic Flk1:eGFP frogs, that express eGFP in blood and lymph vessels (generation of this line will be reported elsewhere). Upon intracardial injection and extravasation from blood vessels, tetramethyl rhodamine isothiocyanate (TRITC)-dextran is selectively taken up by LECs, allowing isolation of doubly labeled yellow LECs and singly labeled green BECs by flow cytometry. RT-PCR showed that sorted BECs and LECs expressed synectin (synectin/10⁵ β-actin mRNA copies: 874 ± 625 for BECs vs 1415 ± 822 for LECs; N = 4).

We then evaluated whether silencing of synectin impaired lymphatic development. In tadpoles, BECs from the PCV, expressing Prox-1, migrate ventrally to form the VCLV, or dorsally to the level of the DLAV to establish the DCLV.¹⁷  To knockdown synectin, Flk1:eGFP tadpoles were injected with the nonoverlapping morpholinos xSyn^ATG1 or xSyn^ATG2, targeting the ATG of the Xenopus laevis synectin mRNA. As both morpholinos blocked translation (supplemental Figure 6) and caused indistinguishable phenotypes (supplemental Table 5), only results for xSyn^ATG1 are shown. We first determined a range of submaximal morpholino doses, that minimally affected vascular development. Screening of GFP⁺ tadpoles at stage 45 (5 dpf) showed that, at 30 ng morpholino per embryo, 17% of GFP⁺ tadpoles had only subtle vascular defects (reduced complexity of capillary network interconnecting ISVs; Figure 5A-B and supplemental Table 5). In line herewith, blood flow in axial vessels was normal in most tadpoles (supplemental Table 5).

To characterize lymphatic development, we focused on the formation of the VCLV and DCLV. LECs in Flk1:eGFP tadpoles were labeled with TRITC-dextran to distinguish them from GFP⁺ BECs. Imaging of control tadpoles at stage 47 revealed that the DCLV and VCLV formed a regular continuous vessel over its entire length (Figure 5A,C,E). By contrast, in 57% of synectin^KD tadpoles, the DCLV was incompletely formed and dysmorphogenic, consisting of isolated LEC clusters, distributed discontinuously over the trunk (Figure 5B,D and supplemental Table 5). The VCLV appeared largely normal in most synectin^KD tadpoles, at least morphologically, while other morphants showed an irregular structure (Figure 5B,F). Lymphangiography at stage 45 revealed that the DCLV and VCLV were functional in all control tadpoles (not shown and Figure 5G). When analyzing the DCLV in 15 synectin^KD tadpoles, the dye was not drained at all or only over a short distance, indicating that this lymphatic was dysfunctional (not shown). Similarly, despite its apparent normal morphologic appearance, the VCLV failed drain the dye (Figure 5H). We previously noticed that the DCLV is more severely affected than the VCLV upon silencing of Vegfr3,^17,31  Vegfc (unpublished) or Liprinβ-1,³²  presumably because LECs, arising from the PCV, have to migrate further to form the distant DCLV than the nearby VCLV. In line with these findings, 32% of synectin^KD tadpoles suffered lymphedema around the heart and gut (Figure 5I-J and supplemental Table 5). Thus, synectin also regulates lymphatic development in the tadpole model.

The tadpole model allows study of the effects on LEC differentiation and migration.¹⁷  To explore whether lymphatic differentiation or migration were impaired in synectin^KD tadpoles, we stained stage 35/36 embryos by whole-mount in situ hybridization for Prox-1 to measure the accumulation of Prox-1⁺ cells in the ventral trunk as an index of lymphatic lineage emergence, and in the more dorsal trunk area as a measure of LEC migration using previously established methods.¹⁷  Morphometric quantification revealed a dose-dependent decrease of both Prox-1⁺ areas in synectin^KD tadpoles (by 40% at the highest morpholino dose; Figure 5K-N). Thus, synectin regulates both lymphatic lineage formation and LEC migration in tadpoles.

We then explored how synectin regulated LEC migration, using the Fli1:eGFP^y1 zebrafish model. Among the possible interacting partners of synectin, we focused on neuropilin-2 (Nrp2), as it contains a PDZ-domain³³  and regulates, as a coreceptor of Vegfr3,^34,35  lymphatic sprouting in mice.^36,37  Vegfc stimulates LEC migration through Vegfr3 signaling in mice,^1,38,39  zebrafish^20,40  and tadpoles.^17,31  Because Vegfr3 is the signaling receptor of the Vegfr3/Nrp2 complex, and silencing of synectin phenocopies the lymphatic defects induced by silencing of Vegfc or Vegfr3 in zebrafish^20,40  and tadpoles,^17,31  we first assessed whether synectin and Vegfr3 genetically interacted, by testing whether a combination of low morpholino doses of synectin (2.5 ng/embryo) and Vegfr3 (1.25 ng/embryo), which by themselves induced only a minimal effect, caused a more significant lymphatic defect (the low morpholino doses were denoted by L-KD, as in synectin^L-KD, to distinguish them from the dose used in the incomplete silencing studies). To silence Vegfr3, we used the previously characterized Vegfr3^ATG1 morpholino.⁴⁰  This analysis revealed that, compared with each single knockdown, the combined knockdown caused more severe defects in TD and PL formation in synectin^L-KDxVegfr3^L-KD morphants (P < .05 vs single knockdown; Figure 6A-B).

We also assessed whether synectin genetically interacted with Vegfr3 in regulating lymphangiogenic sprouting in the PLCγ^y10 zebrafish model. A low Syn^ATG1 dose did not reduce the number of secondary sprouts (7.4 ± 0.3 in control vs 7.7 ± 0.3 in Syn^L-KD; N = 49-62, P = .40; Figure 6C), while a low dose of Vegfr3^ATG1 slightly inhibited secondary sprouting (5.9 ± 0.3 in Vegfr3^L-KD; N = 49-62, P < .001; Figure 6C). However, the combination of both morpholinos impaired secondary sprouting significantly more (4.9 ± 0.2 in synectin^L-KDxVegfr3^L-KD; N = 70, P < .001 vs synectin^L-KD; P < .05 vs Vegfr3^L-KD; Figure 6C). We therefore assessed whether angiogenic or lymphangiogenic secondary sprouts were affected in synectin^L-KDxVegfr3^L-KD embryos, by counting the number of vISVs in 4 dpf Fli1:eGFP^y1 embryos (for rationale, see above).

In line with reports that Vegfr3 regulates angiogenic secondary sprouting,⁴⁰  the fraction of vISVs was reduced by a low dose of Vegfr3^ATG1 but not of Syn^ATG1 (% vISVs: 26 ± 12% in Vegfr3^L-KD vs 52 ± 7% in synectin^L-KD or 54 ± 9% in control; N = 17-29; P < .001). Notably, coknockdown of Vegfr3 and synectin did not further reduce the fraction of vISVs (% vISVs: 25 ± 12% in synectin^L-KDxVegfr3^L-KD; N = 27; P = .968 vs Vegfr3^L-KD), indicating that angiogenic sprouting was not under the control of a synectin/Vegfr3 interaction. Thus, the normal number of angiogenic sprouts, in combination with the reduced total number of secondary sprouts and impaired formation of PL and TD, in synectin^L-KDxVegfr3^L-KD embryos indicates that the genetic synectin/Vegfr3 interaction did not control angiogenic but lymphangiogenic sprouting. Because other experiments revealed that Vegf/Vegfr2 signaling did not affect lymphatic development in zebrafish (supplemental Note 2), we did not explore an interaction with synectin.

We then analyzed whether synectin genetically interacted with Nrp2. We first examined whether the zebrafish Nrp2a and Nrp2b coorthologues regulated lymphatic development, using previously validated morpholinos for Nrp2a (Nrp2a^ATG1) and Nrp2b (Nrp2b^ATG1).⁴¹  Consistent with previous findings,⁴²  a maximal dose of 10 ng Nrp2a^ATG1 caused hind brain edema and arteriovenous (AV) shunts in a small fraction of embryos, while a similar dose of Nrp2b^ATG1 induced cardiac edema formation and mild DA dysmorphogenesis in some morphants (not shown). As reported,⁴²  these blood vessel phenotypes were confined to the extreme caudal region of the trunk (not shown); morphant embryos with edema or morphologic anomalies were excluded. The residual Nrp2a^KD and Nrp2b^KD embryos developed normally by 6 dpf with no or only minimal blood vessel or circulatory defects in the mid-trunk (where we scored lymphatic development; ie, more rostral than the caudal region in which AV shunts/DA defects were present; Figure 6E,G). While lymphatic development was normal in Nrp2b^KD embryos (supplemental Figure 7A), Nrp2a^KD dose-dependently caused PL and TD defects (Figure 6D-K).

We then analyzed whether synectin genetically interacted with Nrp2a. A low dose of Nrp2a^ATG1 (Nrp2a^L-KD) or Syn^ATG1 (Syn^L-KD) minimally affected TD formation (Figure 6L). In contrast, knockdown of both genes by combined injection of a low dose of both morpholinos severely impaired lymphatic development, as the TD was absent in 30% of synectin^L-KDxNrp2a^L-KD morphants (P < .001 vs single knockdown; Figure 6L). Similar findings were obtained when analyzing the PL (not shown).

Notably, however, Nrp2a knockdown did not affect secondary sprout formation in PLCγ1^y10xFli1:eGFP^y1 embryos (supplemental Figure 7B). Coknockdown analysis did also not reveal a genetic interaction between synectin and Nrp2a in secondary sprout formation (supplemental Figure 7E; see below for discussion). Thus, our findings are consistent with a model, whereby synectin genetically interacts with Vegfr3 and Nrp2a during lymphatic development, but only beyond the initial stage of lymphangiogenic sprouting.

Because synectin regulates LEC migration in tadpoles and genetically interacts with Vegfr3, known to stimulate LEC migration,^1,31,39  we examined whether synectin knockdown inhibited sprouting of cultured HDLECs in response to vascular endothelial growth factor C (VEGFC). We silenced synectin expression in LECs using simultaneously 2 nonoverlapping synectin–specific siRNAs. qRT-PCR analysis showed that synectin mRNA levels in HDLECs were reduced by 80% (not shown). HDLECs treated with control or synectin-specific siRNA were aggregated as spheroids and lymphatic sprouts were visualized by CD31 immunostaining. VEGFC stimulated lymphatic sprouting in control spheroids (N = 17; Figure 7A,C,E), but synectin^KD HDLECs formed substantially fewer sprouts in response to VEGFC (N = 26, P < .001; Figure 7B,D-E).

The reduced number of Prox-1⁺ cells in the ventral region in synectin^KD tadpoles suggested a role for synectin in lymphatic differentiation. In mice, Sox18 acts upstream of Prox-1 during this process and lymphatic reprogramming is mimicked in vitro by overexpression of Sox18 in venous ECs.^3,6  We therefore overexpressed Sox18 in murine H5V BECs, transfected them with control or synectin-specific siRNA at day 0, and analyzed lymphatic marker expression on day 7, as reported.⁶  The siRNA transfection procedure was repeated on day 3 to maintain prolonged synectin silencing until 7 days, when synectin mRNA levels were reduced by 60% (not shown).

As reported,⁶  Prox-1 transcript levels were higher in BECs transfected with Sox18-lentivirus (BECs^sox18) than control GFP-virus (BECs^GFP), when treated with control siRNA; expression of the blood vascular marker Nrp1 was not altered (Figure 7F). Upon synectin knockdown, induction of Prox-1, but not of Nrp1, expression in BECs^Sox18 was inhibited (Figure 7F). Knockdown of synectin did not alter gene expression in BECs^GFP (relative mRNA levels vs control: 1.07 ± 0.03 for Sox18; 0.84 ± 0.12 for Prox1; and 0.97 ± 0.09 for Nrp1; N = 3; P > .05). Thus, silencing of synectin impaired Sox18-driven lymphatic reprogramming in vitro.

The present study provides genetic evidence for a role of synectin in early lymphatic development. In zebrafish and frog embryos, knockdown of synectin dose-dependently impaired and, in the most severely affected embryos, even aborted lymphatic formation. Several lines of evidence suggest that TD defects in synectin^KD zebrafish were due to defects in lymphangiogenic sprout formation from the venous system. First, lymphangiogenic sprouts that give rise to PL cells (ie, precursors of LECs forming the TD) were underdeveloped or, in the most severely affected morphants, did not form at all in synectin^KD embryos. In the ccbe1, Vegfc, or Vegfr3 mutants, lack of formation of these lymphangiogenic sprouts also aborts further lymphatic development.^20,40  Second, the fraction of synectin^KD embryos without TD closely correlated with the fraction of morphants without lymphangiogenic sprouts and PL cells. Third, the missing secondary sprouts in synectin^KD embryos were lymphangiogenic and not angiogenic, since these embryos developed at the expected ratio vISVs, the formation of which requires normal angiogenic secondary sprouting.

So far, silencing of only a few other genes (Ccbe1, Vegfc, or Vegfr3) is known to abort lymphangiogenic sprouting, the formation of the PL and TD.^20,40  However, in these morphant embryos, branching of angiogenic sprouts from the PCV was also impaired.²⁰  By contrast, in Dll4/Notch hypomorphants, the fraction of angiogenic sprouts is increased at the expense of lymphangiogenic sprouts.¹¹  It is thus noteworthy that synectin silencing selectively impaired lymphangiogenic without affecting angiogenic secondary sprouting. The identification of synectin as the first selective regulator of lymphangiogenic sprouting suggests that formation of lymphangiogenic versus angiogenic sprouts is under distinct genetic control.

Synectin silencing inhibited migration of cultured LECs in response to VEGFC and reduced the dorsal Prox-1⁺ area, a parameter of LEC migration, in tadpoles. Similar LEC migration defects contribute to lymphatic impairment in tadpoles or mice lacking VEGFC or VEGFR3.^{1,17,31,38,39}  Moreover, silencing of synectin in zebrafish aggravated the lymphatic defects, induced by silencing of Vegfr3 as well as of Nrp2a, the coreceptor of Vegfr3,^34,35  that contains a PDZ-binding domain,³³  through which synectin interacts with its partners.²⁷  These data thus illustrate a genetic interaction of synectin with Vegfr3 as well as with Nrp2a. While Vegfr3/synectin participated already in lymphangiogenic sprouting, Nrp2a/synectin only regulated subsequent formation of the PL and TD. In mice, Nrp2 is also dispensable for initial lymphangiogenic budding from veins, but required for subsequent lymphatic development.³⁶  Together, these genetic studies suggest a model whereby synectin modulates Vegfr3 signaling in lymphatic development in a Nrp2a-independent manner (during lymphangiogenic sprouting) or Nrp2a-dependent manner (PL and TD formation).

How synectin regulates Vegfr3 signaling is an intriguing question. During arterial development, synectin promotes intracellular Vegfr2 trafficking to early endosomes, required for downstream signaling, via Nrp1-dependent and -independent mechanisms.^13,43  By analogy with Vegfr2/Nrp1, it is tempting to speculate that synectin might perhaps also participate in Vegfr3 endocytosis, at least partially, via an interaction with Nrp2. Supporting this model, Vegfr3 not only colocalizes with Nrp2 during endocytosis upon stimulation by Vegfc and Vegfd,³⁵  but internalization is also required for proper Vegfr3 signaling.⁴⁴  Because synectin modulates Vegfr3-driven lymphangiogenic but not angiogenic sprouting, it is tempting to speculate that synectin-mediated endocytosis regulates distinct Vegfr3 signaling pathways. Elucidation of the underlying molecular details will require future study.

Lymphatic defects were unlikely to be secondary to angiogenic alterations, as the PCV formed normally and expressed proper venous markers in synectin^KD embryos, consistent with findings that synectin is dispensable for venous morphogenesis.¹²  In addition, the DA and primary aISVs formed normally and expressed arterial markers upon incomplete silencing of synectin. In fact, alterations in arterial identity or morphogenesis are unlikely to perturb lymphangiogenic sprouting or cause the type of lymphatic defects, observed upon synectin knockdown, since secondary sprouts and PL cells formed normally in PLCγ1^y10 mutants, which even lack primary ISVs and exhibit an underdeveloped and improperly differentiated DA.^16,25  In addition, we and others²²  observed that lymphatic development is nearly normal upon silencing of Vegfaa- or Kdr-like, 2 well-established regulators of arterial morphogenesis and specification.^45,46 

A limitation of our study is that Prox-1 levels in the PCV at the time of lymphatic differentiation in zebrafish are below the threshold of detection (unpublished findings, P.C. and S.S.M.; personal communication, H. Gerhardt [London Research Institute, Cancer Research UK, London, United Kingdom] and F. Cotelli [University of Milan, Milan, Italy]), thereby precluding us from analyzing whether synectin silencing impairs LEC differentiation in zebrafish. We therefore used alternative models in vivo and in vitro. Indeed, our genetic analysis in tadpoles shows that silencing of synectin caused an underdevelopment of Prox-1⁺ cells at sites previously implicated in differentiation of LECs from venous ECs.¹⁷  In addition, silencing of synectin reduced lymphatic reprogramming of venous ECs in vitro upon overexpression of Sox18, an established regulator of lymphatic differentiation in mice.⁶  The role of Sox18 in lymphatic development in tadpoles and zebrafish has not yet been documented but, interestingly, silencing of Sox18 (in combination with Sox7) in zebrafish impaired venous ECs from which LECs differentiate.^47-49  Notably, because PDZ-dependent binding has been reported for some of the SoxA family members,⁵⁰  a possible interaction of synectin with Sox18 is an exciting possibility that remains to be further investigated.

In conclusion, synectin regulates lymphangiogenesis in zebrafish and tadpoles through genetic interactions with Vegfr3 and Nrp2a.

The authors thank K. Brepoels, A. Carton, M. De Mol, E. Janssens, S. Louwette, A. Manderveld, M. Peeters, J. Souffreau, B. Tembuyser, A. Van den Eynde, A. Van Nuffelen, B. Vanwetswinkel, S. Verstraeten, S. Vinckier, and S. Wyns for technical assistance.

K.H., W.V., K.A., I.G., and F.D.S. are sponsored by a PhD grant of the Institute for the promotion of Innovation through Science and Technology (IWT) in Flanders (IWT-Vlaanderen), Belgium. E.G. and S.S.-M. are supported by the Royal Netherlands Academy of Arts and Sciences (KNAW). This work is supported by long-term structural Methusalem Funding by the Flemish Government to P.C., Research Foundation Flanders (FWO) Research Project Funding by the Flemish Government to P.C., Interuniversity Attraction Poles–Belgian Science Policy (IUAP P6/20) to M.D., and European Commission Lymphangiogenomics Consortium Funding (LSHG-CT-2004-503573) to P.C.

Contribution: K.H., F.C., A.N., M.A., M.S., M.D., S.S.-M., E.D., K.A., and P.C. designed research; K.H., F.C., W.V., W.Z., I.G., F.O., A.N., F.D.S., E.G., K.A., and B.R. performed the experiments and analyzed the data; D.K. provided vital reagents; and K.H., F.C., M.D., and P.C. wrote the paper.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Peter Carmeliet, Vesalius Research Center, Vlaams Iustituut voor Biotechnologie, Katholieke Universiteit Leuven, Campus Gasthuisberg, Herestraat 49, B-3000, Belgium; e-mail: Peter.Carmeliet@vib-kuleuven.be.