Zebrafish Scube1 and Scube2 cooperate in promoting Vegfa signalling during embryonic vascularization

Abstract

Aims

The secreted and membrane-anchored signal peptide-CUB-EGF domain-containing proteins (SCUBE) gene family composed of three members was originally identified from endothelial cells (ECs). We recently showed that membrane SCUBE2 binds vascular endothelial growth factor (VEGF) and acts as a co-receptor for VEGF receptor 2 to modulate EC migration, proliferation, and tube formation during postnatal and tumour angiogenesis. However, whether these SCUBE genes cooperate in modulating VEGF signalling during embryonic vascular development remains unknown.

Methods and results

To further dissect the genetic interactions of these scube genes, transcription activator-like effector nuclease-mediated genome editing was used to generate knockout (KO) alleles of each scube gene. No overt vascular phenotypes were seen in any single scube KO mutants because of compensation by other scube genes during zebrafish development. However, scube1 and scube2 double KO (DKO) severely impaired EC filopodia extensions, migration, and proliferation, thus disrupting proper vascular lumen formation during vasculogenesis and angiogenesis as well as development of the organ-specific intestinal vasculature. Further genetic, biochemical, and molecular analyses revealed that Scube1 and Scube2 might act cooperatively at the cell-surface receptor level to facilitate Vegfa signalling during zebrafish embryonic vascularization.

Conclusions

We showed for the first time that cooperation between scube1 and scube2 is critical for proper regulation of angiogenic cell behaviours and formation of functional vessels during zebrafish embryonic development.

Angiogenesis, Embryonic vascularization, Genome editing, Vasculogenesis, Vegfa, Zebrafish

1. Introduction

The signal peptide-CUB (complement C1r/C1s, Uegf, Bmp1)-EGF (epithelial growth factor) domain-containing protein (SCUBE) gene family was originally identified from endothelial cells (ECs).¹ However, these SCUBE genes were also found in other cell types, such as platelets,^2–6 mammary ductal epithelium,^7,8 or osteoblasts⁹ and overexpressed in pathological conditions, such as acute myeloid leukaemia,^10,11 breast,^7,12,13 or lung cancers.¹⁴ To date, three different members (SCUBE1 to 3) have been described and appear to be evolutionarily conserved in vertebrates from zebrafish and mice to humans.^{1,8,9,15–19} These genes encode polypeptides of about 1000 amino acids structured in a modular manner with 5 distinct protein domains: an amino-terminal signal sequence, 9 tandem copies of EGF-like repeats, a spacer region, 3 cysteine-rich motifs, and 1 CUB domain at the carboxy-terminus.

SCUBEs are multi-functional proteins depending on their subcellular localization and distribution. For instance, plasma SCUBE1 released from activated platelets not only serves as a biomarker for acute coronary syndrome and acute ischaemic stroke² also actively participates in platelet aggregation by crosslinking adjacent activated platelets expressing the surface-exposed SCUBE1 during thrombus formation.⁶ Furthermore, zebrafish genetic, biochemical, and molecular studies suggested that soluble SCUBE2 might act in a non-cell-autonomous fashion on hedgehog (HH) ligand-producing cells to promote the secretion of dual-lipidated HH ligands for distant signalling.^{16–18,20–22} However, when these SCUBEs express as peripheral membrane proteins tethered on the cell surface, they can function as co-receptors in promoting signal activity of growth factors, such as vascular endothelial growth factor (VEGF).^23,24

Our recent studies demonstrated that endothelial SCUBE2 acting as a co-receptor for VEGF receptor 2 (VEGFR2) regulates VEGF-induced tube formation and proliferation of ECs: it fine-tunes VEGFR2-mediated signalling during postnatal angiogenesis induced by ischemia²³ or during pathological tumour angiogenesis under hypoxic conditions.²⁴ Nevertheless, ablation of any single Scube gene resulted in no overt vascular phenotype during development,^5,25,26 which suggests potential functional redundancy or genetic compensation with other Scube genes during embryogenesis. Although SCUBE proteins could biochemically form homo- or heterodimeric complexes,^1,9 whether these genes genetically interact with or functionally cooperate during embryonic vascular development remains largely unknown.

In this study, we further dissected the genetic cooperation of these scube genes in zebrafish. Transcription activator-like effector nuclease (TALEN)-mediated genome editing was used to generate insertion/deletion null [knockout (KO)] alleles for each scube gene. Again, no apparent vascular phenotypes were seen in any single scube KO mutant, which reiterates the redundancy or compensation among these scube genes during zebrafish development. Of note, scube1 and scube2 double KO (DKO) markedly impaired EC filopodia extension, migration, and proliferation, thus disrupting vascular formation in different vascular beds during embryogenesis. In addition, our genetic, biochemical, and molecular analyses support that scube1 and scube2 additively function in vascularization by modulating Vegfa signalling during embryonic development.

2. Methods

2.1 Ethics statement

Animal handling protocols used in this study were reviewed and approved by the Institutional Animal Care and Utilization Committee of Academia Sinica (Protocol RMiIBMYR2010063), and conform to the guidelines from Directive 2010/63/EU of the European Parliament on the protection of animals used for scientific purposes or the Guide for the Care and Use of Laboratory Animals published by the US National Institute of Health (NIH Publication No.85-23, revised 1996).

2.2 Zebrafish lines and maintenances

Wild-type AB strain, Tg(fli1a: EGFP)^y,1 Tg(gata1: dsRed), Tg(fli1a: nEGFP)^y7, and Tg(kdrl: mCherry) transgenic zebrafish were maintained in a 14-h light/10-h dark cycle at 28.5°C as described.^15,19 Zebrafish embryos were euthanized by immersion in 0.8 mM of Tricaine (MS-222, Sigma-Aldrich, St. Louis, MO).

2.3 Embryo dissociation and fluorescence-activated cell sorting (FACS)

Wild-type embryos of Tg(fli1a: EGFP)^y;1 Tg(gata1: dsRed) double transgenic fish were dechorionated by pronase treatment at 3 dpf. Embryos were rinsed in calcium-free Ringer’s buffer (58 mM NaCl, 4.02 mM KCl, and 4.76 mM NaHCO₃, pH 7.0, with double-distilled water) and passed several times through a 200-μL pipette tip to remove their yolk. Embryos were washed with 1 mL 1× phosphate-buffered saline and centrifuged for 5 min at 1600 rpm. For cell dissociation, embryos were incubated with 0.25% trypsin (Gibco, Life Technologies, Grand Island, NY) for 30 min at room temperature during which they were triturated with a 200 μL pipette tip every 10 min. The digestion was stopped by a wash with 3 mL 20% foetal bovine serum (FBS) through a 40 μm cell strainer (Falcon, BD Biosciences, San Jose, CA). Cells were centrifuged for 5 min at 2000 rpm, washed and resuspended with HBSS (Gibco, Life Technologies, Grand Island, NY) without phenol red. Fluorescence-activated cell sorting (FACS) of single-cell suspensions was performed at room temperature under sterile conditions by using a FACSAria IIIu system (Becton Dickinson, San Jose, CA); GFP(+) and GFP(−) cells were separately collected into 20% FBS. After sorting, cells were collected for total RNA isolation.

2.4 TALEN genome editing

A TALEN expression plasmid including scube1, 2, or 3 genome-editing targeted sites was generated with the golden gate assembly system (ZGene Biotech Inc., Taipei). The TALEN mRNAs were synthesized for microinjection with the mMESSAGE mMACHINE SP6 in vitro transcription kit (Invitrogen, Thermo Fisher Scientific, Waltham, MA). Mutant zebrafish at the one-cell embryo stage were injected with forward and reverse TALEN mRNA with 150 pg for each gene. Zebrafish mutant allele genotyping involved using the LightCycler 480 High Resolution Melting Master kit (Roche Diagnostics GmbH, Mannheim, Germany) in LightCycler480 Instrument II (Roche Diagnostics GmbH). For primers, see Supplementary material onlineTable S1.

2.5 mRNA and morpholino-oligonucleotide microinjection

Sense mRNA encoding full-length scube genes and vegfaa²⁷ were transcribed in vitro from the pCS2+- or pCDNA3.1-expression vectors by using the mMESSAGE mMACHINE SP6 kit (Invitrogen, Thermo Fisher Scientific, Waltham, MA). kdrl-MO, 5’-CCG AAT GAT ACT CCG TAT GTC AC-3’, vegfaa-MO, 5’-GTA TCA AAT AAA CAA CCA AGT TCA T-3’ and tnnt2a-MO, 5’-CAT GTT TGC TCT GAT CTG ACA CGC A-3’ (Gene Tools LLC, Philomath, OR, USA) were used as described.^27,28 For the ectopic or knockdown experiments, 150 pg of each scube mRNA, 300 pg vegfaa or 150 pg shh mRNA, and 5 ng kdrl-MO, 0.625 ng vegfaa-MO or 2.5 ng tnnt2a-MO were used for microinjection at the one-cell stage.

2.6 Confocal image analysis

To reduce pigmentation, embryos (24 hpf) were treated with 0.003% phenylthiourea (Sigma-Aldrich, St. Louis, MO) in 0.3× Danieau’s buffer and mounted in 1% low melting agarose including 0.4 mM Tricaine. Images were obtained with an LSM 510 confocal microscope (Carl Zeiss MicroImaging, Thornwood, NY, USA). All results are shown by the maximal intensity projection of confocal z-stacks images.

2.7 Phenotype quantification

The size or formation patterns of dorsal aortas (DAs), posterior cardinal veins (PCVs), intersomitic vessels (ISVs), dorsal longitudinal anastomotic vessels (DLAVs) and subintestinal vessels (SIVs) were measured in confocal images at indicated stages. DA width was measured at the mid-point between two ISV sprouting sites above the yolk extension. ISV width was measured at the mid-point of an entire ISV above the embryo’s yolk extension as well. Zebrafish embryonic ISV blood circulation was examined for the first 14 ISVs of double transgenic Tg(fli1: EGFP; gata1: DsRed) embryos at 3 dpf.

2.8 EC proliferation analysis

Human umbilical vein ECs (HUVECs) were purchased from the Bioresource Collection and Research Center (BCRC, Hsinchu, Taiwan) and cultured according to the supplier’s recommendations. HUVEC proliferation was determined by 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide assay as described.¹² Briefly, ECs were trypsinized and plated onto 96-well cell culture plates at 2000 cells/well in 100 µL complete media. The next day, the cells were stimulated with VEGF (100 ng/mL) or control media for 4 days and cell number was counted.

2.9 EC tubulogenesis analysis

HUVECs (3×10³ cells per well) were seeded on Matrigel (Becton Dickinson Biosciences, San Jose, CA) in 15-well µ-slide angiogenesis plates (ibidi GmbH, München, Germany) with VEGF (100 ng/mL). After 16 h, tubulogenesis was determined by counting vessels in three random fields per well.

2.10 Statistical analysis

In all experiments, n numbers refer to biological replicates and were repeated at least three times. Data are presented as mean±SD and were analysed by Student’s t-test (for two groups) or one-way ANOVA (for ≥3 groups) with Tukey multiple comparison analysis by using Prism 7 (GraphPad Software, La Jolla, CA). P < 0.05 was considered statistically significant.

3. Results

3.1 Expression of zebrafish scube genes in vascular ECs

Because mammalian SCUBE genes were originally identified from ECs,¹ we first examined whether zebrafish scube genes were also expressed in vascular ECs. For this purpose, we took advantage of the Tg(fli1: EGFP) transgenic zebrafish line in which the fli1 promoter drives expression of an enhanced green fluorescent protein (EGFP) transgene in ECs and haematopoietic cells.²⁹ To minimize contamination of red blood cells, another Tg(gata1: DsRed) line driving red fluorescent protein expression from the gata1 promoter mostly expressed in erythroid progenitor cells³⁰ was used as a negative selection along with Tg(fli1: EGFP). Following proteolytic dissociation of double transgenic Tg(fli1: EGFP; gata1: DsRed)³⁰ embryos, FACS was used to isolate GFP(+) and GFP(−) DsRed(−) double-negative cell populations (Figure 1A). Diagnostic FACS of the sorted cell population showed that the double-negative GFP(−) DsRed(−) or GFP(+) cells showed 99.4% (not shown) or 96.5% purity (Figure 1B). Quantitative PCR analysis verified that a gene known to be expressed in ECs, the Vegf receptor-2 orthologue kdrl, was expressed higher in GFP(+) than GFP(−) cells (Figure 1C). In line with their mammalian homologues,¹ in zebrafish embryos, the mRNA expression of these scube genes was relatively high in GFP-positive EC populations (Figure 1C). Of note, scube genes were also expressed in GFP(−) cells, which is consistent with previous whole-mount in situ RNA hybridization studies showing expression of scube genes in tissues other than ECs.^16–18,31 Thus, these scube genes might play important roles in blood-vessel formation during zebrafish embryogenesis.

Figure 1

Endothelial expression of scube genes and generation of scube KO mutants. (A) Zebrafish GFP (+)-ECs isolated from Tg(fli1: EGFP; gata1: DsRed) embryos by FACS at 3 days post-fertilization (dpf). (B) Post-sorting from FACS shows 96.5% of GFP (+) ECs. (C) Scube1-3 mRNA expression in GFP (+) ECs by quantitative PCR. kdrl (the Vegf receptor-2 orthologue), an EC marker, was a positive control. Data are mean±SD (n of each genotype was shown from at least three experiments). ns, not significant; *P < 0.05; and ***P < 0.001 by one-way ANOVA. (D–F) Scube KO mutants generated by TALEN-mediated genome editing targeting exon 1. The TALEN protein binding sites in scube genes are marked by red fonts. Null mutant alleles were produced by nucleotide insertion (blue words) in scube2^Δ1/Δ1 (2-dp insertion) or deletion (dotted lines) in scube1^Δ1/Δ1 and scube3^Δ1/Δ1 (left panels). These mutations result in a frame shift into a premature stop codon and translate a truncated protein lacking most of the functional domains (right panels). SP, signal peptide sequence; E, EGF-like domain; Cys-rich, cysteine-rich repeats.

3.2 Generation of the zebrafish scube KO mutants

To investigate the function and genetic cooperation of scube genes during embryogenesis and vascular development, we first generated scube null mutant alleles by using TALEN-mediated genome editing^32–34 targeting exon 1 (Figure 1D–F). Two mutant alleles for each scube gene were generated and further characterized. Sequencing analyses of genome DNA and cDNA verified a 19-bp deletion in scube1^Δ1, a 2-bp insertion in scube2^Δ1, or a 7-bp deletion in scube3^Δ1 (Figure 1D–F) and an 8-bp deletion in scube1^Δ2, a 7-bp deletion in scube2^Δ2, or a 5-bp deletion in scube3^Δ2 (see Supplementary material online, Figure S1). These insertions or deletions result in a frame shift into a premature stop codon and truncate the Scube1 protein at amino acid 13 or 23, Scube2 protein at amino acid 17 or 14, and Scube3 protein at amino acid 40 or 67, i.e. encoding only a partial signal sequence (Figure 1D–F and see Supplementary material online, Figure S1). Furthermore, these mutant alleles were further backcrossed with wild-type zebrafish for at least three generations before phenotyping analyses.

3.3 Analysis of the scube single KO mutant zebrafish

The mRNA expression of each scube gene was in general reduced although to various degrees in the corresponding scube mutant embryos (see Supplementary material online, Figure S2). In agreement with recent studies, the up-regulated expression of functional homologues is frequently observed in genetic mutants as part of a compensatory mechanism.^35–37 Indeed, scube2 and scube3 expression was significantly increased in the scube1^Δ1 mutant; scube1 and scube3 expression in the scube2^Δ1 mutant; and scube1 and scube2 expression in scube3^Δ1 mutant embryos (see Supplementary material online, Figure S2). The functional compensation of other scube genes in the single scube mutants might rescue and in part account for normal morphology, with mild or no overt vascular defects and viability as adults (see Supplementary material online, Figure S3). Of note, even with gene compensation, the scube2^Δ1 mutant had U-shaped somites and a curled body axis, implicating a defect in the HH signalling pathway³⁸ critical for proper somitogenesis. This was reminiscent of the phenotype of a scube2 mutant you^ty97 allele encoding a truncated form of Scube2 lacking the carboxy-terminal 366 amino acids.^16–18 However, unlike early embryonic lethality seen in you^ty97 null mutant embryos, ∼60% of scube2^Δ1 homozygous mutants, again possibly compensated by scube1 and scube3 up-regulation, featured a normal body shape at 5 days post-fertilization (dpf) and survived to adulthood. Moreover, because comparable vascular phenotypes were observed between the Δ1 and Δ2 mutant alleles of each scube gene, we only report the detailed characterization of embryonic vascularization in Δ1 mutants for simplicity and clarity.

3.4 Scube1 and scube2 DKO mutants showed the most severe defects in trunk circulation and vascular malformation

To investigate the possibility of genetic interactions and cooperation between these scube genes, we further analysed blood flow and vascular morphology of the scube single or DKO mutants in the double transgenic Tg(fli1: EGFP; gata1: DsRed) line to visualize the development of EGPF-labelled ECs (green) and circulation of DsRed-labelled red blood cells (red) in the same embryos. At 3 dpf, wild-type embryos displayed fully formed DAs and PCVs. In addition, ISVs were completely formed and were lumenized and appeared regularly at successive vertical myoseptal boundaries, joining together above the neural tube as paired DLAVs. Patent ISVs with proper circulation were observed within in the trunk region in wild-type, scube1^Δ1, or scube3^Δ1 single KO embryos, whereas scube2^Δ1 mutants showed a marginal circulatory defect (20% of ISVs in the trunk region), probably secondary to abnormal U-shaped somites and a curly body axis (see Supplementary material online, Figure S3). In contrast, many severe circulatory defects (70% of ISVs) were seen in scube1/2 DKO mutants as compared with scube1/3 or scube2/3 DKO mutants displaying a normal or mild vascular phenotype, respectively (Figure 2A). Of note, in scube1/2 DKO mutant embryos, most of these ISVs appear thin and fail to fully lumenize for erythrocytes to flow through (see Supplementary material online, Movie S1). Therefore, we further examined the embryonic vascular defects and investigated which signal pathway was critically involved in scube1/2 DKO mutants.

Figure 2

Genetic inactivation of both scube1 and scube2 severely impairs trunk vascular formation and circulation. (A) Quantification of ISV blood cell circulation in scube single and DKO embryos. The blood cell circulation of ISVs analysed in the ISV region of WT, scube1^△1, scube2^△1, scube3^△1, scube1^△1; scube2^△1 DKO, scube1^△1; scube3^△1 DKO and scube2^△1; scube3^△1 DKO mutant embryos at 3 dpf. Data are mean±SD (n of each genotype was shown from at least three experiments). ns, not significant; **P < 0.01 and ***P < 0.001 by one-way ANOVA. (B) Confocal imaging of trunk ISV and DLAV morphology in WT, scube1^△1, scube2^△1, and scube1^△1; scube2^△1 DKO in Tg(fli1: EGFP) embryos at 3 dpf. Lateral view of ISVs and DLAVs (a–d) and dorsal view of DLAVs (a’–d’). Arrows point to underdeveloped DLAV plexus and arrowheads to incomplete DLAV. Asterisks indicate poorly lumenized ISVs. Scale bar =50 μm. DA, dorsal aorta; PCV, posterior cardinal vein; dpf, days post-fertilization. (C–F) Quantification of ISV width (C), ISV length (D), and DLAV width (E) in WT, scube1^△1, scube2^△1, and scube1^△1; scube2^△1 DKO mutants. Data are mean±SD (n of each genotype is shown from at least three experiments). ns, not significant; *P < 0.05, **P < 0.01, and ***P < 0.001 by one-way ANOVA.

Consistent with circulatory obstruction, morphometric analyses showed severely compromised formation of ISVs and DLAVs via branching angiogenesis. The width and length of ISVs as well as diameters of DLAVs (Figure 2B–E) and the process of the DLAV plexus including the paired DLAVs and their interconnections (Figure 2B, d’) were significantly smaller or reduced within the trunk region of the scube1/2 DKO mutants as compared with single KO mutants or wild-type siblings. Notably, combined loss of scube1 and scube2 also affected the diameter of axial vessels including DA and PCV, which are formed by vasculogenesis (Figure 3C and D). Together, our results suggest that scube1 and scube2 genetically interact and cooperate for proper vascular development including vasculogenesis and angiogenesis in zebrafish embryos.

Figure 3

scube1 and scube2 act cooperatively in regulating EC proliferation in ISVs and maintain proper lumen formation in DAs and PCVs. (A) Confocal images of trunk vessels in WT (a), scube1^△1 (b), scube2^△1 (c), and scube1^△1; scube2^△1 DKO (d) embryos at 48 h post-fertilization (hpf). The numbers indicate total EC number of ISVs in the double transgenic Tg (fli1: nEGFP; kdrl: mCherry) line, which expresses a nuclear-localized nEGFP (green) and cytoplasmic mCherry (red) in ECs. White square brackets demarcate the lumen of the DAs and PCVs. Scale bar =50 μm. (B) Quantification of EC number and migration in ISVs at 48 hpf in WT, scube1^△1, scube2^△1, and scube1^△1; scube2^△1 DKO mutants. (C) Quantification of lumen width of DAs in WT, scube1^△1, scube2^△1, and scube1^△1; scube2^△1 DKO embryos at 48 hpf. (D) Quantification of lumen width of PCVs in WT, scube1^△1, scube2^△1, and scube1^△1; scube2^△1 DKO mutant embryos at 48 hpf. Data are mean ± SD (n of each genotype was shown from at least three experiments). ns, not significant; **P < 0.01, and ***P < 0.001 by one-way ANOVA.

As shown in Supplementary material online, Figure S3, there is no apparent cardiac defect that might underlie the circulatory defects in the scube1/2 DKO embryos. However, we did find a mild phenotype of slowing cardiac beating rate in the scube1/2 DKO mutants (Supplementary material online, Figure S4). To examine precisely which phenotypes are specific to Scube1/2 loss and what aspects might be secondary to a loss of blood flow, we included silent heart (cardiac troponin T, tnnt2) morpholino injected embryos as no-flow control (Supplementary material online, Figure S5). Because the scube1/2 DKO mutants showed marginal cardiac and circulatory defects and the scube1/2 DKO displayed more severe defects in DA and ISV (not DLAV) formation than no-flow controls, these data suggest that impaired blood flow might contribute very little to the vascular phenotypes seen in the scube1/2 DKO mutants. Nevertheless, additional investigation is needed to definitely separate these vascular phenotypes from the compromised blood flow.

Previous studies have demonstrated that Vegfa signalling or Vegfa-mediated EC proliferation and migration are required for dorsal ISV sprouting^39,40 or DLAV plexus morphogenesis.⁴¹ We then further visualized and scored the endothelial tip cell migratory behaviours and quantified the total EC number in ISVs in the double transgenic Tg(fli1: nEGFP; kdrl: mCherry),^42,43 which expresses nuclear-localized nEGFP (green) and cytoplasmic mCherry (red) in ECs. In wild-type embryos, ISVs started to sprout from DAs bilaterally at around 20 h post-fertilization (hpf), extended dorsally along each somite boundary, gradually migrated passing the notochord and the neural tube, and reached the dorsal roof to form DLAVs at 48 hpf (Figures 3 and 4). In scube1/2 DKO mutants, although the position of the initial sprout was not affected (see Supplementary material online, Figure S6A), ISVs grew only halfway through their ventral trajectory, and most ISVs (70%) failed to reach the dorsal roof to form DLAVs at 30 hpf (see Supplementary material online, Figure S6B), whereas we found only a slight ratio of ISV branching defects in wild-type or single KO mutant embryos (see Supplementary material online, Figure S6B). Consistently, in vivo time-lapse imaging revealed much slower migration for scube1/2 DKO endothelial tip cells from DAs to DLAVs than wild-type or single KO embryos (Figure 4A and B). Furthermore, at a later stage, 2 dpf, the EC number of paired ISVs was significantly lower in scube1/2 DKO mutant embryos (3.9±1.7, n=48) than wild-type (6.9±1.2, n=48), scube1 KO (6.2±1.3, n=56), or scube2 KO (5.8±1.5, n=40) mutant siblings (Figure 3A and B). Endothelial tip cells used dynamic filopodia to probe the surroundings and guide the outgrowth of capillaries. The number of tip cell filopodia extensions of ISVs was lower in scube1/2 DKO mutants than wild-type or single KO embryos (Figure 4C and D). These data suggest that scube1/2 collaborate in regulating both migration and proliferation of endothelial tip cells during ISV branching angiogenesis.

Figure 4

scube1 and scube2 co-regulate ISV tip cell behaviours. (A) Snapshot images from in vivo time-lapse analysis of ISV tip cell migration of WT (a), scube1^△1 (b), scube2^△1 (c) and scube1^△1; scube2^△1 DKO (d) in double transgenic Tg (fli1: nEGFP; kdrl: mCherry) embryos. Initial developmental stage (hpf) is noted in the first image frame and subsequent time progression (hour: minute) is indicated at the top. (B) Migration speed of ISV tip cells in WT, scube1^△1, scube2^△1 and scube1^△1; scube2^△1 DKO mutants. Data are mean ± SD (n of each genotype was shown from at least three experiments). ns, not significant; *P < 0.05, **P < 0.01, and ***P < 0.001 by one-way ANOVA. (C) Confocal imaging of ISV tip cell filopodia of WT (a), scube1^△1 (b), scube2^△1 (c) and scube1^△1; scube2^△1 DKO (d) in Tg(fli1: EGFP) embryos. Arrowheads indicate filopodia extensions. (D) ISV tip cell filopodia number per ISV in WT, scube1^△1, scube2^△1 and scube1^△1; scube2^△1 DKO mutants. Data are mean ± SD (n of each genotype was shown from at least three experiments). ns, not significant; ***P < 0.001 by one-way ANOVA.

Most importantly, single (see Supplementary material online, Figure S7A) or co-injecting (Figure 5A) scube1 and scube2 mRNA could rescue the impaired ISV circulation as an indication of angiogenesis defects in the trunk, which confirms the specificity of vascular phenotypes seen in the scube1/2 DKO mutant embryos. Furthermore, injection of vegfaa encoding the two most potent isoforms, Vegfa₁₆₅ and Vegfa₁₂₁, which are indispensable and predominantly expressed during embryonic vascular development,⁴⁴ restored the vascular deficits caused by loss of both scube1 and scube2 in embryos, so they may cooperate in modulating the vascular effects of Vegfa in zebrafish development (Figure 5B). Of note, we did not observe ectopic sprouting events on injection of vegfaa mRNA (400 pg) in scube1/2 DKO embryos (see Supplementary material online, Figure S8). In line with this notion, we tested the genetic interaction between scube1/2 and Vegfa signalling. Low-dose injection of the previously validated vegfaa translation-blocking antisense morpholino- oligonucleotide (tMO)²⁸ had no effect on axial vessel development or ISV circulation, but injection of vegfaa tMO in scube1/2 DKO mutants had a much stronger effect. That is, disruption of ISV circulation was more severe in vegfaa tMO; scube1/2 DKO embryos at 3 dpf than in single scube1 or scube2 KO injected with vegfaa tMO embryos (Figure 5C). Thus, inhibition of Scube1/2 and Vegfa function results in an additive genetic interaction leading to worsened vascular phenotypes. Together, these data indicate that Scube1 and Scube2 might cooperate in augmenting the Vegfa signalling essential for EC proliferation and migration during the formation of ISVs and DLAVs in zebrafish embryos.

Figure 5

Restoration of impaired ISV circulatory defect in scube1/2 DKO by injecting scube1/2 or vegfaa mRNA and genetic interaction between scube1/2 and vegfaa. (A) Co-injection of scube1 and scube2 mRNA could rescue the impaired ISV circulation. (B) Injection of vegfaa mRNA restored blood circulation defect in ISVs of scube1^△1; scube2^△1 DKO embryos. (C) ISV circulation was disrupted in vegfaa tMO; scube1/2 DKO embryos as compared with single scube1 or scube2 KO embryos injected with vegfaa tMO at 3 dpf. Data are mean±SD (n of each group is shown in parentheses from at least three experiments). ns, not significant; **P < 0.01 and ***P < 0.001 by one-way ANOVA.

3.5 Cooperation between scube1 and scube2 is essential for proper formation of intestinal vasculature

Besides the angiogenic formation of ISVs and DLAVs in the trunk, Vegfa signalling has been implicated in proper patterning of SIVs that vascularize the gut, liver, and pancreas.^45–47 These vessels generate a stereotypical basket-shaped structure composed of the subintestinal artery (SIA), subintestinal vein, and interconnecting vessels acquiring the mature form at about 3 dpf. Dorsal migration of ECs from the primary SIVs to up-regulate the arterial markers and generate the SIA was modulated by Vegfa signalling,^45,46 whereas the final step of remodelling the subintestinal plexus by retraction of the venous leading buds is controlled by Notch signalling.⁴⁷ In accordance with the cooperation of scube1/2 in Vegfa signalling, the overall SIV basket length regulated by Vegfa was much reduced in scube1/2 DKO mutants (78.07±34.3 μm, n=83) as compared with scube2 KO (95.18±31.1 μm, n=58), scube1 KO (127.8±41.4 μm, n=45), or wild-type (145.7±27.8 μm, n=49) embryos (Figure 6A and B). Again, single mRNA (see Supplementary material online, Figure S7B) or co-injection of both scube1 and scube2 mRNA (Figure 6D and E) as well as injection of vegfaa mRNA (Figure 6F and G) rescued the SIV basket length in scube1/2 DKO embryos. In contrast, scube1 and scube2 did not genetically cooperate in the Notch signalling-driven remodelling of the subintestinal plexus involving retraction of leading buds as the basket reaches its final shape because the number of remaining leading buds is comparable among scube1 (2.5±1.8, n=45) or scube2 single KO (1.9±1.6, n=59) vs. DKO mutant (2.3±1.6, n=84) embryos (Figure 6A and C). Together, these results reveal that scube1/2 co-regulating Vegfa signalling plays a critical role in different vascular beds during zebrafish embryonic development, i.e. required for proper trunk ISV and DLAV vessel formation and also essential for organ-specific SIV vascularization.

Figure 6

Both scube1 and scube2 are required for proper development of SIVs. (A) Confocal images of SIVs in WT (a), scube1^△1 (b), scube2^△1 (c), and scube1^△1; scube2^△1 DKO (d) in Tg(fli1: EGFP) embryos at 3 dpf. The dashed lines indicate the basket length of developed SIVs, and arrows mark the leading buds in SIVs. Scale bar =50 μm. (B) Quantification of the basket length of SIVs. Data are mean±SD (n of each genotype was shown from at least three experiments). ns, not significant; *P < 0.05 and ***P < 0.001 by one-way ANOVA. (C) Quantification of ectopic leading buds sprouting from SIVs. Data are mean±SD (n of each genotype was shown from at least three experiments). ns, not significant; **P < 0.01 and ***P < 0.001 by one-way ANOVA. (D) Confocal images of SIVs in scube1^△1; scube2^△1 DKO in Tg(fli1: EGFP) embryos restored by co-injection of scube1 and scube2 mRNAs at 3 dpf. The dashed lines indicate the basket length of developed SIVs. Scale bar =50 μm. (E) Quantification of the rescued basket length of SIVs in scube1^△1; scube2^△1 DKO mutants by scube1/2 mRNA treatment. Data are mean±SD (n of each genotype was shown from at least three experiments). ns, not significant; ***P < 0.001 by one-way ANOVA. (F) Confocal images of SIVs in scube1^△1; scube2^△1 DKO in Tg(fli1: EGFP) embryos rescued by vegfaa mRNA at 3 dpf. The dashed lines indicate the basket’s length of developed SIVs. Scale bar =50 μm. (G) Quantification of the rescued basket length of SIVs in scube1^△1; scube2^△1 DKO mutants by vegfaa mRNA. Data are mean±SD (n of each genotype was shown from at least three experiments). ns, not significant; **P < 0.01 and ***P < 0.001 by one-way ANOVA.

3.6 Cooperative interaction between SCUBE1 and SCUBE2 in modulating VEGF-induced proliferation and tube formation preserved in mammals

Because our studies revealed a novel cooperation between scube1 and scube2 for blood-vessel formation during zebrafish embryogenesis, we further investigated whether this SCUBE1/2 cooperation in regulating VEGF responses is evolutionarily conserved in mammals by knocking down SCUBE1 and/or SCUBE2 in cultured human ECs. After individual transduction of recombinant lentivirus encoding SCUBE1 or SCUBE2-targeting short hairpin RNAs alone or combined in HUVECs, knockdown specificity and efficiency were verified by RT–PCR (Figure 7A). Single knockdown of SCUBE1 or SCUBE2 individually decreased but double knockdown of both SCUBE1/2 additively and markedly impaired VEGF-induced human EC growth and capillary-like network formation on Matrigel (Figure 7B–D). In agreement with these findings, the VEGF-triggered phosphorylation of VEGFR2 at Tyr1059 was significantly repressed with SCUBE1/2 double vs. single-knockdown ECs (Figure 7E and F). Likewise, we also found that the number of tip cell filopodia extensions of HUVECs cultured on Matrigel is grossly decreased in SCUBE1/2 double-knockdown vs. single-knockdown ECs (see Supplementary material online, Figure S9). In addition, we further determined the biochemical interaction between zebrafish Scube1/2 and VEGFR2 by a pull-down assay. When co-expressed in HEK-293T cells, immunoprecipitation of VEGFR2 (Myc-tagged) resulted in a pull-down of Scube1 or Scube2 (FLAG-tagged) (see Supplementary material online, Figure S10), which generally agree with our previous studies of mammalian SCUBEs.^23,24 Together, these data support an evolutionarily conserved cooperation between SCUBE1 and SCUBE2 in modulating VEGF-stimulated proliferation and tubulogenesis in mammalian ECs.

Figure 7

SCUBE1 and SCUBE2 double knockdown had an additive effect on impairing VEGF-induced proliferation, tube formation, and VEGFR2 phosphorylation in HUVECs. (A) SCUBE1 and/or SCUBE2 expression was down-regulated by independent SCUBE1 and/or SCUBE2-targeting shRNA lentiviruses (SCUBE1, SCUBE2, or SCUBE1/2 shRNA) in HUVECs. A luciferase shRNA lentivirus was a negative control (Control shRNA). RT–PCR analysis confirmed the specificity and efficiency of knocking down SCUBE1 and SCUBE2 mRNA level in Control, SCUBE1, SCUBE2, or SCUBE1/2 shRNA-transduced HUVECs. (B) Effect of single or double knockdown of SCUBE1 and/or SCUBE2 on VEGF-induced HUVEC growth. Data are mean±SD (percentage increase relative to non-stimulated cells from three independent experiments). ***P < 0.001 by one-way ANOVA. (C–D) Effect of single or double SCUBE1 and SCUBE2 knockdown on VEGF-induced tube formation by in vitro Matrigel angiogenesis assay. Representative images are shown (C) and quantified by counting total number of tubules per field (D). Data are mean±SD (calculated from three independent experiments). **P < 0.01 and ***P < 0.001 by one-way ANOVA. Scale bar =500 μm. (E–F) SCUBE1 and SCUBE2 cooperate in modulating VEGFR2 phosphorylation in HUVECs. Western blot analysis of VEGF-induced activation of VEGFR2 in HUVECs (E) and quantification (F) of VEGF-induced phosphorylation of VEGFR2 at Tyr1059 with Control, SCUBE1, SCUBE2, or SCUBE1/2 shRNA knockdown in HUVECs. Phosphorylated VEGFR2 intensity was normalized with the total VEGFR2 protein intensity serving as a loading control. Data are mean±SD from three independent experiments. ***P < 0.001 by one-way ANOVA. (G) Working model for endothelial SCUBE1 and SCUBE2 cooperating in promoting VEGF signalling during embryonic vascularization. Under physiological conditions, such as during embryonic vascular development, endothelial SCUBE1 and SCUBE2 associating in a complex might facilitate VEGF ligand binding to VEGFR2, thus promoting the VEGF-induced VEGFR2 phosphorylation and its downstream signalling for proper vasculogenesis and angiogenesis.

4. Discussion

In this study, we generated novel scube KO mutant alleles by using TALEN-mediated genome editing and analysed the functional role of scube genes during zebrafish embryonic vascular formation. In line with mouse genetic studies,^5,25,26 single KO of each scube gene conferred mild or no overt vascular phenotype, which suggests redundancy or compensation among these scube genes during zebrafish development. While scube2 mutants show mild defects in sprouting angiogenesis, the vascular defects appear more significantly severe by co-inactivation of scube1. These data imply that functional redundancy might occur between these two genes. Because previous studies showed functional compensation of zebrafish genetic mutants that could up-regulate their homologous genes^35–37 and because scube1 expression is highly induced in scube2 mutants, it is likely that scube1 might partially compensate for the loss of scube2 function. Thus, strong vascular defects are only observed when both homologs are knocked out.

Of note, the genetic interactions between scube1/2 and Vegfa signalling are consistent with our previous biochemical evidence showing that SCUBE1 and SCUBE2 proteins can form heterodimeric complexes.¹ Also, SCUBE2 can bind VEGFA and VEGFR2 to act as a co-receptor for augmenting VEGF signalling in ECs during adult angiogenesis²³ or pathological tumour angiogenesis.²⁴ Because Vegfa plays multiple roles during vasculogenesis and angiogenesis,⁴⁸ impaired embryonic vascularization might result from reduced Vegfa signalling in scube1/2 DKO mutants. Together, our data support a model in which membrane Scube1 and Scube2 form a heterodimeric complex to facilitate Vegfa signalling during embryonic vascularization, at least in zebrafish (Figure 7G). Given that a co-operation between SCUBE1 and SCUBE2 in promoting VEGF signalling is conserved in mammalian ECs (Figure 7), further investigation is required to validate whether the genetic interaction between Scube1/2 and Vegfa indeed occurs during embryonic vascular formation in a mouse model system.

Our results suggest a regulatory function of endothelial Scube1/2 in promoting Vegfa signalling. However, a cell non-autonomous function of Scube proteins may exist in facilitating vasculogenesis and angiogenesis because Scube genes appear to encode secreted proteins and are expressed in non-EC cell types. For instance, SCUBE2 has recently been shown to act cell—non-autonomously in mediating the release of the lipid-modified Sonic Hedgehog protein,²⁵ which functions genetically upstream of Vegfa to regulate arterial endothelial differentiation.²⁹ However, this notion might be excluded because expression of gli1 (a direct shh target gene) or vegfaa (an shh-regulated gene during arterial endothelial differentiation) remained unaltered and comparable between wild-type and scube1/2 DKO embryos, and injection of shh mRNA did not rescue the scube1/2 DKO vascular defects in ISVs and SIVs (data not shown). Furthermore, one shortcoming of global scube1/2 mutant zebrafish is that vessel phenotypes including ISVs, DLAVs, or the organ-specific intestinal vasculature could be caused by an unknown systemic scube1/2-deficient effect from other cell types, rather than the direct consequence of loss of scube1/2 in ECs. Additional investigations by using the EC-specific scube1/2-KO fish line with a CRISPR/Cas9 vector system^49,50 are warranted to further validate the specific contribution of ECs to the observed vascular abnormalities.

Alternatively, an additive genetic interaction between scube1/2 mutations and vegfaa tMO knockdown also indicates that scube1/2 might function in an independent and parallel signalling pathway that promotes angiogenesis. For instance, several other pathways including Delta-Notch signalling are involved in angiogenesis.⁴⁸ In line with this argument, our recent study showed that SCUBE2 is up-regulated in breast cancer stem cells and accelerated the epithelial to mesenchymal transition and tumour aggressiveness by activating NOTCH signalling.⁵¹ Furthermore, scube1 or 2 mutants showed impaired retraction of the venous leading buds, the final step of remodelling the subintestinal plexus controlled by Notch signalling.⁴⁷ Nevertheless, additional studies are necessary to clarify whether SCUBEs are indeed involved in modulating Notch signal transduction during embryonic vascular development.

We found that the width of venous ISVs is similarly smaller as arterial ISVs in scube1/2 DKO mutants when compared with wild-type siblings. These data suggest that Scube1/2 might also play a role in Vegfr3-medidated venous sprouting. However, further investigation is required to verify that compromised blood flow might have contributed these vascular phenotypes because mild reduction of cardiac contractility was observed in the scube1/2 DKO embryos. In addition, injection of scube3 mRNA (150 pg) could partially restore the scube1/2 DKO vascular defects in ISVs but not SIVs. These results suggest that Scube3 might also play a role in vascular morphogenesis overlapping with Scube1/2 function but in a vascular bed-specific manner.

Another intriguing finding of this study is that scube1 and scube2 appear to cooperatively function in maintaining proper vascular stability and maturation during vasculogenesis and angiogenesis. Indeed, formation of a patent vascular lumen within axial (DAs and PCVs) and trunk vessels (ISVs and DLAVs) was significantly impaired in scube1/2 DKO mutants: the lumen of some vessels was not sufficiently large to allow erythrocytes to flow through. Although the exact underlying mechanism remains to be determined, the cell–cell junctional expression of these SCUBE proteins^12,23,24 might stabilize the formation of EC junctional adherens to stabilize and fully lumenize the vasculature.

In summary, we report that a cooperative genetic interaction might exist between scube1/2 and Vegfa signalling in vasculogenesis and angiogenesis during zebrafish embryonic development. The zebrafish model system will be useful to genetically dissect the scube upstream regulators and downstream targets and future characterize the diverse EC signal pathways potentially modulated by scube genes. These studies will help shed light on how their cooperative activities might act to drive blood-vessel formation during physiological and pathological conditions.

Supplementary material

Supplementary material is available at Cardiovascular Research online.

Authors’ contributions

K.-C.T., S.-L.L., and R.-B.Y. conceived the study or contributed to the experimental design. K.-C.T., Y.-C.L., Y.-T.C., and R.-B.Y. performed experiments and data analysis. R.-B.Y. wrote the manuscript with help from K.-C.T. All authors discussed the results and commented on the manuscript. All authors approved the final version of the manuscript.

Acknowledgements

We thank the Taiwan Zebrafish Core Facility (TZCAS), which is supported by the Ministry of Science and Technology of Taiwan [MOST 108-2319-B-400-002] for providing fish lines, cultivation of mutant fish lines, and sperm cryopreservation. We also thank the Academia Sinica Core Facility and Innovative Instrument Project (AS-CFII108-113) for cell sorting service and the Light Microscopy Core Facility of the Institute of Biomedical Sciences for technical assistance. We are grateful for Chang-Yi Wu who provided Tg(fli1a: nEGFP; kdrl: mCherry) transgenic zebrafish and Chen-Hui Chen for careful reading and suggestions for the manuscript.

Funding

This work was supported by grants from Academia Sinica and the Ministry of Science and Technology of Taiwan [MOST 106-2320-B-001-017-MY3, MOST 107-2320-B-001-015-MY3, and MOST 109-2320-B-001-012-MY3 to R.-B.Y., MOST 107-2321-B-001-036-MY3 to Y.-C.L.].

Conflict of interest: none declared.

Data availability

The data underlying this article are available in the article and in its online Supplementary material.

Translational perspective

Our studies indicate that targeting SCUBE1 and/or SCUBE2 on modulating VEGF signalling might provide potential therapeutic treatments for VEGF-mediated proliferative pathological vascular diseases.