Vegf signaling promotes vascular endothelial differentiation by modulating etv2 expression

Abstract

Vasculogenesis involves the differentiation of vascular endothelial progenitors de novo from undifferentiated mesoderm, their migration and coalescence to form the major embryonic vessels and the acquisition of arterial or venous identity. Vascular Endothelial Growth Factor (Vegf) signaling plays multiple roles during vascular development. However, its function during embryonic vasculogenesis has been controversial. Previous studies have implicated Vegf signaling in either regulating arteriovenous specification or overall vascular endothelial differentiation. To clarify the role of Vegf in embryonic vasculogenesis and identify its downstream targets, we used chemical inhibitors of Vegf receptor (Vegfr) signaling in zebrafish embryos as well as zebrafish genetic mutants. A high level of chemical inhibition of Vegfr signaling resulted in the reduction of overall vascular endothelial marker gene expression, including downregulation of both arterial and venous markers, ultimately leading to the apoptosis of vascular endothelial cells. In contrast, a low level of Vegfr inhibition specifically blocked arterial specification while the expression of venous markers appeared largely unaflected or increased. Inhibition of Vegfr signaling prior to the initiation of vasculogenesis reduced overall vascular endothelial differentiation, while inhibition of Vegfr signaling starting at mid-somitogenesis stages largely inhibited arterial specification. Conversely, Vegf overexpression resulted in the expansion of both arterial and pan-endothelial markers, while the expression of several venous-specific markers was downregulated. We further show that Vegf signaling affects overall endothelial differentiation by modulating the expression of the ETS transcription factor etv2/etsrp. etv2 expression was downregulated in Vegfr-inhibited embryos, and expanded in Vegfaa-overexpressing embryos. Furthermore, vascular-specific overexpression of etv2 in Vegfr-inhibited embryos rescued defects in vascular endothelial differentiation. Similarly, vegfaa genetic mutants displayed a combination of the two phenotypes observed with chemical Vegfr inhibition: the expression of arterial and pan-endothelial markers including etv2 was downregulated while the expression of most venous markers was either expanded or unchanged. Based on these results we propose a revised model which explains the different phenotypes observed upon inhibition of Vegf signaling: low levels of Vegf signaling promote overall vascular endothelial differentiation and cell survival by upregulating etv2 expression, while high levels of Vegf signaling promote arterial and inhibit venous specification.

1. Introduction

Vasculogenesis, or the assembly of the first embryonic blood vessels from undifferentiated mesoderm, involves multiple steps. During intraembryonic vasculogenesis, vascular endothelial progenitors originate within the lateral plate mesoderm, migrate and coalesce into the major embryonic vessels such as the dorsal aorta (DA) and the posterior cardinal vein (PCV), and acquire their arteriovenous identity prior to the initiation of circulation (Hong et al., 2006; Zhong, 2005). While it is dificult to study these early steps of vascular development in a mammalian embryo, the zebraflsh has emerged as an excellent model for studying vasculogenesis. Similar to what is observed in mammalian embryos, vascular endothelial progenitors of the major axial vessels originate within the lateral plate mesoderm in zebrafish embryos (Childs et al., 2002; Eriksson and Lofberg, 2000; Fouquet et al., 1997; Jin et al., 2005; Lawson and Weinstein, 2002; Williams et al., 2010). Arterial and venous progenitors are known to emerge at different times and distinct locations, with arterial progenitors originating earlier and closer to the midline than venous progenitors. These lines of endothelial cells migrate to the midline in two waves, with the medial arterial progenitors migrating first, followed by the laterally located venous progenitors, and they subsequently coalesce into the dorsal aorta and posterior cardinal vein, respectively (Kohli et al., 2013; Williams et al., 2010). However, it is not known when the arterial and venous progenitors commit to their respective fates, and the signals influencing these fate decisions are not well understood.

The Vascular Endothelial Growth Factor (Vegf) signaling cascade has been implicated in multiple steps of vascular endothelial cell development, both in mammals and zebrafish (Carmeliet et al., 1996; Ferrara et al., 1996; Lawson et al., 2002). However, there is contradictory evidence regarding its involvement in general vascular endothelial cell differentiation versus arteriovenous specification during embryonic vasculogenesis. Mice deficient in just one VegfA allele form abnormal blood vessels, have an overall reduction in vascularization, and die between 10 and 12 days of embryonic development. These mice also have complications in cardiac development and dorsal aorta morphogenesis (Carmeliet et al., 1996; Ferrara et al., 1996). Additionally, homozygous null mice for VegfR1, VegfR2, or VegfR3 all die between embryonic days 8.5 and 9.5 due to impaired vasculogenesis including disorganized vessel growth (Dumont et al., 1998; Fong et al., 1995; Hamada et al., 2000; Shalaby et al., 1995). Chemical inhibition of Vegf signaling using tyrosine kinase inhibitor PTK787 in zebrafish embryos resulted in the loss of overall endothelial marker expression, including the loss of expression of pan-endothelial marker fli1a, arterial marker ephb2a and venous marker flt4 (Chan et al., 2002). A similar phenotype was observed in Vegf receptor knockdown embryos (Kim et al., 2013). Conversely, overexpression of vegfaa, one of the two zebrafish VegfA homologs, resulted in the expansion of vascular endothelial marker expression (Liang et al., 2001). These results argue for a requirement of Vegf signaling in overall vascular endothelial differentiation, defined as inducing or maintaining the expression of multiple markers specific to vascular endothelial cells. However, Vegf signaling has also been implicated in arteriovenous specification. In Vegfaa morpholino (MO) injected zebrafish embryos (morphants), a reduction of arterial and an expansion of venous markers was observed (Lawson et al., 2002). Vegf signaling is thought to function downstream of Hedgehog (Hh) and upstream of Delta-Notch signaling during arteriovenous specification of the major axial vessels (Lawson et al., 2002). Because these previous studies used different methods to inhibit Vegf signaling, and no comprehensive analysis of both arteriovenous specification and overall vascular endothelial differentiation has been performed in the same study, it remains unclear if Vegf signaling indeed plays multiple roles during vasculogenesis. Furthermore, the mechanism by which Vegf signaling regulates the two distinct processes of arteriovenous specification and overall vascular endothelial differentiation is unknown.

The transcription factor Etv2/Etsrp/ER71 has been identified as a critical regulator of vasculogenesis in multiple vertebrates including zebrafish and mice (Ferdous et al., 2009; Lee et al., 2011, 2008; Pham et al., 2007; Sumanas et al., 2005; Sumanas and Lin, 2006). Etv2 belongs to the family of ETS transcription factors, which shares a conserved DNA binding domain and plays multiple roles in vasculogenesis and angiogenesis (Craig and Sumanas, 2016; Lelievre et al., 2001). Loss of Etv2 in zebrafish embryos results in the near-complete loss of many vascular endothelial markers including the major Vegf receptor vegfr2/kdrl, while overexpression of etv2 mRNA is sufficient to induce ectopic expression of multiple vascular endothelial markers. In an epistasis experiment, Etv2 was shown to be required for the induction of kdrl expression by Vegf signaling (Sumanas and Lin, 2006). Other data show that Etv2 expression can be induced by Vegf and VegfR2/Flk1 signaling in mice (Rasmussen et al., 2012). Mice null for Flk1 show a reduction in Etv2 reporter expressing cells. Additionally, increasing amounts of Vegf and Flk1 correlated with an increase in Etv2 activity (Rasmussen et al., 2012). Furthermore, in vitro studies have shown that Vegf signaling can directly activate Etv2 transcription (Rasmussen et al., 2013, 2012). Together, these results suggest that Etv2 can both regulate and be regulated by Vegf signaling. However, the relationship between Vegf signaling and the transcriptional regulation by Etv2 during vasculogenesis in vivo is not well understood.

In addition to Etv2, other ETS factors have been implicated in regulating vasculogenesis or arteriovenous specification. Recently, the ETS factor Erg was shown to regulate the expression of Dll4 and arteriovenous specification downstream of Vegf signaling (Wythe et al., 2013). It has also been demonstrated that ETS factors can directly bind and regulate Notch4 expression (Wythe et al., 2013). Interestingly, it has been shown that Etv2 together with FoxC transcription factors can directly bind to Delta and Notch promoters (De Val et al., 2008). Yet, Etv2 has been implicated in overall vascular endothelial differentiation rather than arteriovenous specification, and its function is required for both arterial and venous marker expression (Kohli et al., 2013; Sumanas and Lin, 2006).

Here, we used chemical inhibitors and genetic mutant analysis to investigate the role of Vegf signaling during endothelial differentiation and arteriovenous specification in zebrafish embryos. We also explored the interaction between Vegf signaling and the transcriptional regulation of vasculogenesis by Etv2. Our results argue that a low level of Vegf signaling during late gastrulation/early somitogenesis stages is required to promote overall vascular endothelial differentiation and cell survival while a high level of Vegf signaling during mid- to late somitogenesis stages promotes arterial and inhibits venous specification. We further demonstrate that Vegf controls endothelial differentiation by modulating etv2 expression, and that etv2 overexpression is sufficient to rescue defects resulting from the inhibition of Vegfr signaling. Similar to the results obtained with chemical inhibitors, vegfaa mutant embryos display downregulation of overall vascular endothelial markers including etv2, while venous markers show either a slight reduction or an upregulation. These results argue that Vegf signaling plays two distinct roles in vasculogenesis: overall endothelial differentiation and arteriovenous specification, which differ in the level and timing of Vegf signaling.

2. Materials and methods

2.1. Zebrafish strains and staging

Tg(kdrl:GFP)^s843 (Jin et al., 2005), Tg(fli1:nGFP)^y7 (Roman et al., 2002), Tg(kdrl:mCherry)^ci5 (Proulx et al., 2010), Tg (2.3 kb etv2:GFP)^zf372 (Veldman and Lin, 2012) and Tg(stab1l:YFP)^hu4453 (Hogan et al., 2009) zebrafish lines were used for all experiments in this study unless otherwise noted. A vegfaa mutant line (bns1) was created containing a 10 base pair deletion in exon 3 using TALEN technology (Rossi et al., 2016) and used for noted experiments. vegfaa mutants were crossed into Tg(etv2:GFP)^ci1 or Tg(kdrl:mCherry)^ci5 background (Proulx et al., 2010) for easy identification of mutant embryos. Embryos were raised at 28.5 °C, or 32 °C. Embryonic staging was performed according to established criteria (Kimmel et al., 1995). Developmental delay in morphant or chemically treated embryos was accounted for by allowing embryos to develop until morphological staging criteria matched that of control embryos.

2.2. Morpholinos

A VegfA morpholino (Gene Tools, Inc) with sequence (5′-GTAT-CAAATAAACAACCAAGTTCAT) (Nasevicius et al., 2000) was injected into embryos at the 1–2 cell stage. A final dose of 10 ng was used for all experiments.

2.3. Chemical treatments

For SU5416 experiments, Tg(kdrl:GFP) embryos were treated with either 1% DMSO or 20 μM, 10 μM, 5 μM, or 2.5 μM SU5416 (Fisher #502307978) dissolved in DMSO. Treatments began at the 50% epiboly stage unless otherwise noted, and embryos were kept in chemical treatment solution until the time of analysis. For PTK787 experiments, embryos were continuously treated with either 1% DMSO or 200 μM PTK787 (Fisher #NC0540125) dissolved in DMSO beginning at the 50% epiboly stage.

2.4. Real time PCR

Pools of 15–25 embryos were frozen on dry ice at 24hpf. Embryos were homogenized in lysis buffer using a 23-gauge needle and extraction of RNA from the embryos was carried out using the RNAqueous 4-PCR kit (Ambion). Quantification of purified RNA was carried out using a Nanodrop spectrophotometer machine. Next, cDNA synthesis was performed using SuperScript^® VILO cDNA Synthesis Kit or Superscript III cDNA synthesis kit (Invitrogen). Quantitative real-time PCR (qRT-PCR) was performed using PowerUp™ SYBR™ Green Master Mix (Applied Biosystems) in a StepOnePlus™ Real-Time PCR System. Quantification was performed using the relative standard curve method; controls/DMSO treatment were assigned a value of 1 and experimental values were computed by the software based on the C_T values, relative to the control standard curve. Three separate runs were performed with duplicates in each run and ef1α was used as an endogenous control. The relative quantity of cDNA in each sample of each gene was normalized to the value of ef1α. Specificity of amplication was confirmed by analyzing the size of PCR product using gel electrophoresis. In addition, melting curves indicated a single peak for all genes analyzed (see Suppl. Fig. S1). Data was analyzed using StepOne™ Software-version 2.3 (Life Technologies).

The following primers were used:

ef1α (5′-TCACCCTGGGAGTGAAACAGC) and (5′-ACTTGCA-GGCGATGTGAGCAG),etv2 (5′-GAGCTGTTGCACAAAGGTCA) and (5′-CAGAGAGGGACGAGGTTCTG),flt4 (5′-GAGTCCACCCACCCTTG-ACAAC) and (5′-GCCGTGGTGCCATCAATAAC),stab1l (5′-CATCT-TGTTCAAGCTGCCCAC), and (5′-GAGTTCACAGGTGGTGGAAG), cldn5b (5′-GGACTCTGCTTGACCGTCAC) and (5′-CGATGGTTATCAT-GAGAGCC),dab2 (5′-GCTCTTGCTGTCTCGTTCCT) and (5′-CATCTG-CAAGAGCAGCATTC),cdh5 (5′-GGTGCCTCCGACAAGGATGA) and (5′-AACACTCTTTTGCTCTGGCGT).lyve1b (5′-TCTTGGCTGGGGGAG-GAATA) and (5′-CGGACCTTCTTGTCACCCAT).mrc1a (5′-TGGAT-GGTTCGCCTGTTGTA) and (5′-TCCTGGTCGCACAGTAGGTA).aqp8a (5′-TCTTGGCTGGGGGAGGAATA) and (5′-CGGACCTTCTTGTCA-CCCAT).

2.5. Overexpression of vegfaa and etv2

To overexpress vegfaa, an equal ratio of vegfaa₁₂₁ and vegfaa₁₆₅ RNA was injected at a final amount of 50 pg into zebrafish embryos at the 1–2 cell stage. mRNA was synthesized as previously described (Liang et al., 2001). The same amount of mRFP was injected as a control.

For vascular specific etv2 overexpression, a construct containing etv2 coding sequence under control of the fli1ep promoter and tagged with 2A-mcherry was inserted into a pCSDest vector using Gateway cloning technology (Invitrogen) (Villefranc et al., 2007), and injected at a final dose of 18 pg/embryo in conjunction with tol2 mRNA at a dose of 24 pg/embryo. Embryos with vascular-specific mCherry expression were selected at 24 hpf and fixed for further analysis.

2.6. Imaging and cell counting

Embryos were whole mounted in 3% methylcellulose on glass slides. Images were captured using a 10× objective on an AxioImager Z1 (Zeiss) compound microscope with an Axiocam ICC3 color camera (Zeiss). Images in multiple focal plans were captured individually and combined using the Extended Focus module within Axiovision software (Zeiss). For confocal imaging, anesthetized embryos were mounted in 0.6% low-melting point agarose and imaged using a 20× or 40× objective on a Nikon A1R confocal microscope. Cell counts in fli1:nGFP embryos were performed using Imaris sofware package (Bitplane) by cropping an area in 3D that corresponded to 2 segments within the trunk region (the area was defined based on the location of sprouting intersegmental vessels). In order to define a size for the cells, the diameters of 5 cells of varying size were measured while scrolling through the Z-stack image and an average size was calculated. Then, using the ‘Spots’ tool in the Imaris software, an automated count of the number of cells within the selected region was performed based on the average size of cells and the highest intensity of the fluorescent signal. Following the automated count performed by the software, the 3D rendering of the selected region was analyzed in detail to ensure that only true nuclear GFP signals were taken into account for the automated count. Cells that were not counted by the software due to a weak signal were manually marked and rare GFP signals which were falsely marked as cells were unmarked. In this manner, cell counts were performed in at least 6 control and experimental embryos and an average cell count was calculated for each group. For vegfaa mutants, cell counts were also obtained by defining an area that corresponds to a two somite region within the embryo trunk using Fiji software, and then manually counting GFP-positive cell nuclei while scrolling through the Z-stack image. Both approaches yielded similar results.

2.7. In situ hybridization

Whole mount in situ hybridization (ISH) was performed using DIG-UTP labeled probes synthesized with T3, T7, or SP6 polymerase as previously described (Jowett, 1999). Antisense RNA probes for the following genes were synthesized as previously described: notch3 (Lawson et al., 2001), ephrinb2a (Lawson et al., 2001), deltac (Lawson et al., 2001), cldn5b (Rost and Sumanas, 2014), etv2 (Sumanas et al., 2005), flt4 (Thompson et al., 1998), stab1l (Rost and Sumanas, 2014), mrc1a (Wong et al., 2009) and ephrinb4 (Lawson et al., 2001). Dab2 (Song et al., 2004) PCR fragment was amplified using primers with sequences GCTCTTGCTGTCTCGTTCCT and CTAATACGACTCACTATAGGGCATCTGCAAGAGCAGCATTC, and antisense RNA was transcribed using T7 RNA polymerase (Promega). Lyve1b (Hogan et al., 2009) PCR fragment was amplified from a plasmid using primers GCTCTTGCTGTCTCGTTCCT and CTAATACGACTCACTATAGGGCATCTGCAAGAGCAGCATTC, and antisense RNA was transcribed using T7 RNA polymerase (Promega).

2.8. Apoptosis assay

Embryos were fixed at the desired time point in 4% PFA at 4 °C overnight. Following fixation, embryos were washed four times with PBST at room temperature and then blocked in 10% lamb serum for two hours at room temperature. Detection of Caspase3 was carried out with purified rabbit anti-active Caspase3 antibody (1:100, Becton Dickinson catalog #559565) and goat anti-rabbit Alexa 594 (1:500, Thermofisher catalog# A-11037). Vascular GFP was detected using a chicken anti-GFP antibody (1:500, Abcam catalog# ab13970) and a goat anti-chicken Cy5-650 nm (1:200, Abcam catalog# ab97147). Vascular mCherry was detected using mouse anti-mCherry antibody (1:250, Abcam catalog# ab125096) and goat anti-mouse Alexa 555 (1:500, Thermofisher catalog# A-21422). Following the antibody treatment, embryos were mounted laterally in 0.6% low-melting agarose for imaging.

2.9. FACS analysis

A Tg (2.3 kb etv2:GFP)^zf372 (Veldman and Lin, 2012) transgenic fish line expressing GFP under a vascular-specific −2.3 kb etv2 promoter element was used to count vascular endothelial cells in embryos at the 15-somite stage. Wild-type (non-fluorescent) fish were used as GFP- controls. Embryos were treated with either DMSO or 20 μM SU5416 commencing at the 50% epiboly stage. At the 15-somite stage, embryos were collected and subjected to an embryo dissociation protocol for FACS analysis as described by Manoli and Driever (2012). Embryos were first dechorionated using pronase and then deyolked by mechanical disruption and centrifugation. Following the deyolking step, the pellet obtained was re-suspended in FACSmax and dissociated using a syringe plunger and strainer. The cell suspension was then transferred to a FACS tube. Cells from DMSO- and SU5416-treated wildtype embryos were first analyzed to obtain a threshold for auto-fluorescence. Thereafter, cells from DMSO- and SU5416-treated etv2:GFP embryos were analyzed for GFP fluorescence. The number of GFP+ single cells (endothelial cells) and the total number of cells were recorded. Analysis was performed twice for each sample using a BD LSRFortessa™ cell analyzer. In order to determine whether there was a statistically significant difference in endothelial cell number between DMSO and SU5416-treated embryos, an unpaired t-test was performed on the percentages of endothelial cells (of the total cell population) obtained from replicates of both treatments.

3. Results

3.1. Vegfr inhibition decreases arterial and venous marker expression

To examine the requirement of Vegf signaling in arteriovenous specification and overall vascular endothelial differentiation, we inhibited Vegfr function using chemical inhibitors. Embryos were treated with the pan-Vegf receptor inhibitor SU5416 (Fong et al., 1999) starting at the 50% epiboly stage (5.3 h post fertilization, hpf) and were examined for arterial and venous marker expression at 24 hpf (Fig. 1). The general morphology of embryos was undisturbed upon chemical treatment, but embryos displayed a severe reduction in intersegmental vessel (ISV) sprouting (data not shown), as reported previously (Cannon et al., 2010; Serbedzija et al., 1999). As expected, a decrease in arterial markers cldn5b, notch3, efnb2a and deltaC expression was observed in Vegfr inhibited embryos (Fig. 1A–H). In contrast to previous results obtained using a Vegfaa morpholino (Lawson et al., 2002), we observed a decrease in expression of venous markers flt4, dab2, ephb4, and stab1l (Fig. 1I–P). The observed decrease in both arterial and venous marker expression was also confirmed using real time qPCR analysis (Fig. 1Q). These results suggest that Vegf signaling is required for the expression of both arterial and venous markers.

Fig. 1.

Inhibition of Vegfr signaling by SU5416 chemical treatment decreases arterial and venous marker expression. (A–P) In situ hybridization analysis showing a decrease in arterial and venous vascular endothelial markers upon SU5416 treatment starting at the 50% epiboly stage (5.3 hpf), as compared to DMSO treated control embryos. Arterial markers cldn5b (A and B), notch3 (C and D), efnb2a (E and F), and deltaC (G and H) as well as venous markers flt4 (I and J), dab2 (K and L), ephb4 (M and N), stab1l (O and P), are downregulated in SU5416 treated embryos. (Q) Quantification of selected markers by qRT-PCR. Markers with high expression outside of the vasculature were excluded from qRT-PCR analysis. * indicates p < 0.05 and ** indicates p < 0.01, ± SEM is shown. Arrows indicate the dorsal aorta and arrowheads indicate the posterior cardinal vein. Lateral view of 24hpf embryos, anterior to the left. Numbers of embryos showing displayed phenotype are listed.

Data from in vitro studies have suggested that different concentrations of Vegf may have distinct effects on arterial and venous differentiation (Lanner et al., 2007; Zhang et al., 2008). To test if different levels of Vegfr inhibition had distinct effects on arterial and venous marker expression in zebrafish embryos, we treated embryos with varying doses of SU5416 and analyzed arterial and venous marker expression by in situ hybridization and qPCR at 24 hpf. Lower concentrations of 2.5 and 5 μM resulted in a specific downregulation of arterial markers cldn5b and aqp8a expression, while venous markers flt4, stab1l, lyve1b and dab2 either were not significantly changed (flt4, stab1l and lyve1b at 2.5 μM) or were upregulated (dab2), as analyzed by qPCR and ISH (Fig. 2). The expression of venous marker dab2 was clearly upregulated in the DA and not significantly changed in the PCV. The only exception was venous mrc1a expression which was downregulated at these concentrations of SU5416 (Fig. 2U). In contrast, higher SU5416 treatment doses of 10 μM and 20 μM resulted in significant downregulation of most (10 μM) or all (20 μM) arterial and venous marker expression. While venous marker expression was apparent in all vascular endothelial cells, the intensity of ISH staining was reduced in SU5416 treated embryos, which was further supported by qPCR analysis (Fig. 2). Pan-endothelial marker cdh5 expression was also significantly downregulated at all SU5416 doses tested, with the greatest effect observed at the highest doses of SU5416. These results suggest that different levels of Vegfr inhibition result in two distinct effects. A low level of Vegfr inhibition results in the decrease of arterial marker expression, while venous marker expression is either largely unaffected or increased. In contrast, a high level of Vegfr inhibition results in downregulation of both arterial and venous markers, likely due to the inhibition of overall vascular endothelial differentiation or a decrease in cell survival.

Fig. 2.

Inhibition of Vegfr signaling with a high dose of SU5416 decreases vascular endothelial differentiation while low doses of SU5416 inhibit arterial and promote venous specification. (A–T) In situ hybridization analysis of arterial (cldn5b), venous (stab1l, dab2) and pan-endothelial (etv2) marker expression in embryos treated with 2.5–20 μM doses of SU5416 starting at the 50% epiboly stage. Note that cldn5b expression is reduced in embryos treated with all doses of SU5416, while stab1l expression appears to be only mildly downregulated or unchanged at the lowest dose of SU5416. In contrast, dab2 expression is expanded into the artery in embryos treated with 2.5–10 μM doses (G,K,O), and reduced in embryos treated with a 20 μM dose (S). The expression of etv2 is reduced in embryos treated with medium and high doses of 5–20 μM SU5416 (L,P,T). Lateral view of 24 hpf embryos, anterior to the left. Arrows indicate the DA and arrowheads indicate the PCV. Numbers of embryos showing the displayed phenotype are listed. (U) qPCR analysis of arterial (cldn5b, aqp8a), venous (flt4, stab1l, dab2, lyve1b, mrc1a) and pan-endothelial (cdh5, etv2) markers in 24hpf-old embryos treated with varied doses of SU5416 starting at the 50% epiboly stage. Note that arterial cldn5b and aqp8a expression is reduced at all doses of treatment; pan-endothelial cdh5 and etv2 as well as venous flt4, stab1l and lyve1b expression is only mildly downregulated or not changed in embryos treated with low 2.5–5 μM doses of SU5416, and strongly downregulated in embryos treated with higher 10–20 μM doses; venous dab2 expression is upregulated in embryos treated with low doses, and downregulated only in embryos treated with the highest (20 μM) SU5416 dose. * indicates p < 0.05 and ** indicates p < 0.01, ± SEM is shown.

To confirm the results obtained with SU5416 treatment, embryos were treated with a different Vegfr chemical inhibitor, PTK787 (Chan et al., 2002; Lee et al., 2006; Wood et al., 2000) starting at the 50% epiboly stage and analyzed by in situ hybridization and qPCR at 24 hpf. PTK787-treated embryos showed no gross morphological defects, but did not form ISVs (data not shown). Similar to SU5416 treatments, arterial marker cldn5b, venous marker flt4 and pan-endothelial marker cdh5 expression was downregulated in PTK787 treated embryos (Suppl. Fig. S2). However, dab2 and stab1l expression was not significantly affected (Suppl. Fig. S2G). Therefore, this phenotype was comparable to the phenotypes obtained with low to intermediate doses of SU5416, when arterial and pan-endothelial markers were down-regulated, and venous markers were unaffected (10 μM and 2.5 μM SU5416 doses did not affect dab2 and stab1l expression, respectively). flt4 displays a significant expression in the artery, therefore it may not be as specific a venous marker as stab1l or dab2, which would explain different effects observed with these markers.

3.2. Vegf signaling is required to promote arterial and venous marker expression until the 20-somite stage

To investigate a potential time-dependent requirement of Vegf signaling for arteriovenous specification and overall vascular endothelial differentiation, we inhibited Vegfr function by 20 μM SU5416 chemical treatment beginning at different stages of development (Fig. 3, Suppl. Fig. S3). Embryos were treated with SU5416 beginning at the 50% epiboly (5.3 hpf), 5-somite (11.5 hpf), 15-somite (16.5 hpf), 20-somite (19 hpf), or 23-somite (20.5 hpf) stages, and analyzed for arterial, venous and pan-endothelial marker expression by in situ hybridization and qPCR at 24 hpf. The arterial markers cldn5b and aqp8a were strongly downregulated when treatments started between the 50% epiboly and 15-somite stages and moderately downregulated when treatments began at the 20- to 23-somite stages (Fig. 3A–D, I). Venous markers lyve1b and mrc1 followed a similar trend and were strongly downregulated when embryos were treated at the 50% epiboly and 5-somite stages (Fig. 3I), and moderately downregulated when embryos were treated at the 15-somite to 23-somite stages (Fig. 3E–I). In contrast, venous-specific flt4, stab1l and dab2 expression was downregulated when treatments started at or before the 15-somite stage, and mildly reduced or unaffected when treatments were started at the 20–23-somite stages; in fact, dab2 expression was upregulated when embryos were treated at the 23-somite stage (Fig. 3I). Pan-endothelial cdh5 expression was greatly reduced when treatments were started at or before the 20-somite stage while it was not significantly affected when the treatment was started at the 23-somite stage. Extended treatment starting at the 23-somite stage resulted in the reduced expression of all vascular endothelial markers, when analyzed at 28 hpf (Suppl. Fig. S4).

Fig. 3.

Analysis of arterial and venous marker expression in SU5416-treated embryos at 24hpf after commencing treatment at multiple developmental time points. (A–H) Embryos were treated with either 0.1% DMSO or 20 μM SU5416 beginning at stages ranging from 15 somites to 23 somites and analyzed at 24 hpf by in situ hybridization. (A–D) Arterial marker cldn5b was severely downregulated when treated with SU5416 beginning at the 15-somite stage (16.5 hpf, B) as compared to DMSO treated controls (A). A similar phenotype was observed when treated at the 20-somite stage (19 hpf, C) but was less apparent when treated at the 23 somite stage (20.5 hpf, D). (E–H) Venous marker flt4 was downregulated upon SU5416 inhibition at the 15-somite stage (F) as compared to DMSO treated controls (E). Milder downregulation was observed when treatment was started at the 20-somite and 23-somite stages (G,H) and expansion of flt4 expression into the DA was observed. Arrows indicate DA and arrowheads indicate PCV. Lateral view of 24 hpf embryos, anterior to the left. Numbers of embryos showing the displayed phenotype are listed. (I) qPCR analysis of arterial (cldn5b, aqp8a), venous (flt4, stab1l, dab2, lyve1b, mrc1a) and pan-endothelial (cdh5, etv2) marker expression in SU5416-treated embryos. The treatment was started at the stages indicated and embryos were analyzed at 24 hpf. Note that arterial marker expression is greatly reduced at all stages indicated, while venous marker expression is significantly reduced at the early treatment stages (50%-epiboly and 5-somite), and only slightly reduced or unaffected (flt4, stab1l) or even expanded (dab2, 23-somite) at the later treatment stages. Pan-endothelial markers show strong reduction at early treatment stages and only slight or no reduction at the later treatment stages. * indicates p < 0.05 and ** indicates p < 0.01, ± SEM is shown.

While it is clear that different markers have distinct responses to the timing of Vegf treatment, there are two major trends apparent from these results. Expression of all markers including pan-endothelial cdh5 was downregulated in early treatments, suggesting that overall endothelial differentiation was affected. However, later treatments resulted in two distinct responses. While arterial markers were still significantly downregulated, pan-endothelial cdh5 and the majority of venous markers were less affected, in some cases unchanged or even upregulated (dab2). This suggests that the early treatments inhibit overall endothelial differentiation, while later treatments inhibit arterial specification and have less of an effect on overall endothelial differentiation. However, prolonged treatment even at later stages can inhibit expression of multiple vascular endothelial markers.

3.3. High level inhibition of Vegf signaling leads to the apoptosis of vascular endothelial cells

Vegf signaling plays a known role in the survival of vascular endothelial cells (Alon et al., 1995). To test if apoptosis was increased in SU5416 treated embryos, immunostaining against Caspase3 was performed. Indeed, increased apoptosis was apparent within the axial vasculature in 10 μM SU5416 treated embryos at 24 hpf (Fig. 4A,B). 20 μM treatments resulted in increased Caspase3 staining which overlapped with fli1:GFP-expressing vascular endothelial cells as early as the 20-somite stage (Fig. 4C–H). However, there was no increased apoptosis apparent in SU5416 embryos at an earlier stage of 15-somites (Fig. 4I–N). These results show that the high doses of SU5416 treatment result in apoptosis of vascular endothelial progenitors after the 15-somite stage. It is possible that high doses of SU5416 and early treatments resulted in the loss of overall vascular endothelial marker expression due to cell apoptosis. However, prolonged SU5416 treatment starting at the 23-somite stage resulted in the downregulation of all endothelial marker expression by 28hpf, and no apoptosis was apparent yet (Suppl. Fig. S4). This suggests that the downregulation of endothelial cell markers occurs prior to apoptosis in Vegfr inhibited embryos and that endothelial cell apoptosis is a consequence of the inhibition of endothelial cell differentiation.

Fig. 4.

High concentrations of SU5416 result in apoptosis in trunk vasculature at 24hpf and 20-somite stages. (A,B) Caspase 3 immunofluorescent staining of 24hpf embryos treated with either DMSO or 10 μM SU5416 starting at the 50% epiboly stage. Strong Caspase staining, indicating apoptosis, was observed within trunk and tail vasculature in the 10 μM SU5416-treated embryos (B) as compared with the DMSO-treated control embryos (A). (C–H) Immunofluorescent staining of apoptotic marker Caspase3 in 20-somite stage embryos treated with either DMSO or 20 μM SU5416. Embryos treated with SU5416 showed strong Caspase3 staining which co-localized with vascular fli1:GFP fluorescence (F–H) compared to DMSO-treated embryos which did not have Caspase staining in the trunk vascular region (C-E). (I–N) 15-somite stage embryos treated with SU5416 did not appear to have apoptosis in the trunk region (L–N) and had comparable Caspase3 staining to the DMSO-treated controls (I–K).

We then analyzed the expression of arterial and venous markers in Vegfr-inhibited embryos at the 15-somite stage when no induction of apoptosis was observed. At this stage, 20 μM SU5416 treatment resulted in the reduction of arterial markers cldn5b and aqp8a, pan-endothelial marker cdh5 and venous markers lyve1b and mrc1, while the expression of venous markers flt4 and dab2 was not significantly affected and stab1l expression was upregulated (Fig. 5). The changes in marker expression are largely comparable to SU5416 treatment results observed with the low 2.5–5 μM doses at 24 hpf. Downregulation of pan-endothelial cdh5 expression supports the model that overall endothelial differentiation is inhibited prior to the initiation of apoptosis.

Fig. 5.

Analysis of vascular marker expression and endothelial cell number in 15-somite stage (15ss) embryos treated with high-dose SU5416. (A) qPCR analysis of arterial, venous and pan-endothelial markers in 15ss embryos treated with either DMSO or 20 μM SU5416 starting from the 50% epiboly stage. Note the significant reduction in both arterial markers (cldn5b, aqp8), while venous markers are either increased (stab1l), unchanged (flt4, dab2) or are reduced (lyve1b, mrc1a). Pan-endothelial marker cdh5 does not appear to be significantly changed, while etv2 is slightly downregulated. (B–D) FACS analysis of endothelial cell number in 15ss DMSO- or 20 μM SU5416-treated embryos expressing GFP under a vascular-specific −2.3 kb etv2 promoter. Representative images from FACS analysis depicting the population of GFP+ single cells in the total population of cells in DMSO-treated (B) and SU5416-treated embryos (C). There was no statistically significant difference in the percentage of GFP+ cells between DMSO controls and SU5416-treated embryos (D).

To test if Vegf inhibition by chemical treatment affected endothelial cell number, we analyzed the number of etv2:GFP-positive cells by FACS sorting at the 15-somite stage. No significant changes in the fraction of GFP-positive cells between SU5416 and DMSO treated embryos were detected, arguing that SU5416 treatment is unlikely to inhibit cell proliferation during these stages (Fig. 5B–D). Together, these results suggest that SU5416 inhibition results first in the reduction of overall endothelial differentiation (cdh5 expression), and inhibition of arterial specification, which is observed either at younger stages with a high SU5416 dose, or at later stages with a low SU5416 dose. The expression of venous markers is a combination of two opposing effects: reduced expression due to inhibited overall endothelial differentiation, and increased expression due to inhibited arterial specification. Therefore, the expression of some venous markers is either not significantly affected, or even increased in some cases. However, prolonged high-dose SU5416 treatment leads to a reduction of all vascular markers analyzed and the apoptosis of vascular endothelial cells.

3.4. Vegf overexpression results in the expansion of vascular endothelial markers

We then analyzed the effect of Vegfaa overexpression on arteriovenous specification and overall endothelial differentiation. A combination of vegfaa₁₂₁ and vegfaa₁₆₅ RNA isoforms was injected into zebrafish embryos at the 1–2 cell stage and embryos were analyzed by in situ hybridization at 24 hpf (Fig. 6). Arterial markers cldn5b, deltaC and notch3 were expanded into the region of the PCV as compared to uninjected control embryos (Fig. 6A–F). Venous markers stab1l and flt4 were also expanded into the DA as compared to uninjected control embryos (Fig. 6G–J). A similar expansion of venous flt4 expression was observed upon overexpression of vegfaa₁₆₅ or vegfaa₁₂₁ isoforms alone (data not shown). In contrast, the expression of venous markers dab2, lyve1b and mrc1a was downregulated in Vegfaa-overexpressing embryos (Fig. 6K–P). These changes were confirmed by qPCR analysis, which also showed upregulation of arterial apq8a and cldn5b, while the expression of venous dab2, lyve1b and mrc1 was downregulated (Fig. 6S). Pan-endothelial cdh5 expression was also elevated (Fig. 6Q,R,S). In addition, sprouting of intersegmental vessels was inhibited, as high levels of Vegfa are predicted to interfere with tip-stalk selection during angiogenic sprouting (Bentley et al., 2008). Injection of control mRFP RNA did not significantly affect the expression of these markers (Fig. 6S). These data corroborate the Vegfr inhibition data and suggest that Vegfaa overexpression results in two distinct effects, promoting overall vascular endothelial differentiation and arterial specification, while inhibiting venous specification. Out of the 5 venous markers used, dab2, lyve1b and mrc1 expression in wild type embryos is nearly exclusively restricted to the vein and largely excluded from the DA, which explains why their expression is downregulated in Vegfaa-overexpressing embryos. Both stab1l and flt4 display some expression in the DA, which could explain why their overall expression becomes upregulated in Vegfaa-overexpressing embryos.

Fig. 6.

Analysis of arterial and venous markers in vegfaa- overexpressing embryos. (A–R) In situ hybridization analysis of multiple arterial and venous endothelial markers upon vegfaa₁₂₁ and vegfaa₁₆₅ RNA injection as compared to uninjected controls. Arterial markers cldn5b (A and B), deltaC (C and D) and notch3 (E and F), venous markers stab1l (G and H) and flt4 (I and J), and pan-endothelial marker cdh5 (Q and R) are all expanded upon vegfaa RNA injection. In contrast, venous dab2, lyve1b and mrc1a expression is reduced in vegfaa overexpressing embryos (K–P). Note that dab2 and lyve1b expression is mostly confined to the PCV while stab1l and flt4 display some arterial expression in wt embryos (G,I,K,M). Arrows indicate the DA and arrowheads indicate the PCV. Lateral view of 24 hpf embryos, anterior to the left. Numbers of embryos showing the displayed phenotype are listed. (S) qPCR analysis of arterial (cldn5b, aqp8a), venous (flt4, stab1l, dab2, lyve1b, mrc1a) and pan-endothelial (cdh5, etv2) markers in vegfaa and control mRFP overexpressing embryos as compared to uninjected controls (wt). Note the significant increase in arterial and pan-endothelial marker expression and decrease in venous dab2, lyve1b and mrc1a expression in vegfaa overexpressing embryos. *, p < 0.05; ** p < 0.01 as compared to wt embryos, ± SEM is shown.

3.5. Vegf signaling upregulates etv2 expression

The ETS transcription factor Etv2/Etsrp has been well characterized as a key regulator of vascular endothelial differentiation (Pham et al., 2007; Sumanas and Lin, 2006). Etv2 function is required for the induction of multiple vascular endothelial markers, including arterial and venous-specific genes. Our data suggest that the phenotype observed with the high dose of Vegfr chemical inhibition is similar to that observed with a downregulation of Etv2 function, resulting in the loss of overall endothelial differentiation and leading to endothelial cell apoptosis, which has been previously observed in etv2 mutants (Pham et al., 2007). Therefore, we tested if Vegf signaling regulates etv2 expression in zebrafish. Chemical inhibition of Vegfr with either SU5416 or PTK787 treatment resulted in a significant reduction in etv2 expression at 24 hpf (Fig. 7A–D,G,H). Additionally, treatment with SU5416 starting at the 50% epiboly stage resulted in a decrease in etv2 expression at the 10 and 15-somite stages, but not at the 2–3 somite stage as quantified by qRT-PCR (Figs. 5A, 7G). This argues that etv2 expression can be initiated independently of Vegf signaling, but Vegf signaling contributes to the upregulation of etv2 expression during subsequent stages. Treatment with SU5416 starting at the 50% epiboly or 5-somite stages resulted in an almost 2-fold downregulation of etv2 expression at 24 hpf, and a 20–25% downregulation when the treatment was started at the 15–23-somite stages (Fig. 3I). Treatment with SU5416 also affected etv2 expression in a dose-dependent manner, which correlated well with the inhibition of arterial and venous marker expression (Fig. 2). Conversely, overexpression of vegfaa resulted in an expansion of etv2 expression (Fig. 7E,F,H). Together, these results argue that Vegf signaling upregulates etv2 expression during endothelial differentiation.

Fig. 7.

Vegf signaling upregulates etv2 expression. In situ hybridization analysis showing a decrease or increase in etv2 expression upon Vegfr inhibition or vegfaa overexpression, respectively. (A–D) Chemical inhibition of Vegfr with either SU5416 (A,B) or PTK787 (C,D) treatment starting at the 50% epiboly stage results in a decrease in etv2 expression when analyzed at 24 hpf. (E,F) vegfaa RNA overexpression results in an expansion of etv2 expression. Lateral view of 24 hpf embryos, anterior to the left. Numbers of embryos showing the displayed phenotype are listed. (G) etv2 expression in embryos treated with SU5416 at the 50% epiboly stage and analyzed at 2–3-somite (11 hpf), 10-somite (14 hpf), and 24 hpf stages was quantified by qRT-PCR. Note that etv2 expression in control DMSO treated embryos is increased at the 10-somite and 24 hpf stages compared to the 2–3 somite stage and this upregulation is reduced in SU5416 treated embryos while its initial expression at the 2–3 somite stage is not affected. All expression values are shown relative to etv2 expression at 24 hpf in DMSO treated embryos. (H) Quantification of etv2 expression at 24 hpf by qPCR in PTK787 treated and vegfaa RNA injected embryos. ** denotes p < 0.01; * denotes p < 0.05, ± SEM is shown. Arrows indicate DA and arrowheads indicate PCV.

3.6. Etv2 overexpression partially rescues arterial and venous marker expression in Vegfr inhibited embryos

Because our data showed that etv2 expression is regulated by Vegf signaling, we tested if the overexpression of etv2 could rescue defects caused by Vegfr inhibition. An A mCherry-tagged etv2-2A-mCherry DNA construct under the control of the vascular endothelial-specific fli1a promoter (Villefranc et al., 2007) was injected into zebrafish embryos at the 1–2 cell stage. Coinjection of this construct with tol2 RNA resulted in a mosaic expression of etv2 and mCherry throughout the vasculature (Suppl. Fig. S5A–C). These embryos were then treated with SU5416 at the 50% epiboly stage and analyzed for arterial and venous marker expression by in situ hybridization at 24 hpf (Fig. 8). Venous expression of flt4 and arterial expression of cldn5b were largely unaffected by fli1:etv2 DNA injection in DMSO treated embryos, however some ectopic expression outside of the vasculature was observed (Fig. 8C,D, white arrowheads). Embryos treated with SU5416 but not injected with the etv2 DNA construct showed a downregulation of arterial cldn5b, venous flt4, dab2, stab1l and pan-endothelial cdh5 expression, similar to what was seen earlier (Fig. 8E,F,I). However, when embryos injected with fli1:etv2 DNA were treated with SU5416 they showed a marked increase in the expression of all vascular markers analyzed as compared to uninjected embryos treated with SU5416 (Fig. 8G–I). These results show that Etv2 is sufficient to partially rescue expression of all endothelial markers in Vegfr inhibited embryos. No significant change in the number of vascular endothelial cells was observed in fli1:Etv2-mCherry expressing embryos, when analyzed in the nuclear-specific fli1:nGFP background (Suppl. Fig. S5D–G). Therefore, these results suggest that Vegf signaling induces overall vascular endothelial differentiation as well as the expression of both arterial and venous markers by upregulating etv2 expression.

Fig. 8.

fli1:etv2 DNA injection rescues arterial and venous marker expression in Vegfr inhibited embryos. (A–H) In situ hybridization analysis showing expression patterns of arterial and venous markers upon Vegfr inhibition and etv2 DNA overexpression. (A,B) Uninjected embryos treated with DMSO displayed normal expression patterns of arterial marker cldn5b (A) and venous marker flt4 (B). (C,D) fli1:etv2 DNA injection does not significantly affect endogenous vascular endothelial marker expression, but did result in small patches of ectopic expression (white arrowheads). (E,F) SU5416 inhibition results in downregulation of arterial cldn5b (E) and venous flt4 (F) marker expression. Embryos injected with fli1:etv2 DNA and treated with SU5416 have increased cldn5b (G) and venous flt4 (H) vascular endothelial marker expression, as compared to SU5416 treated embryos (E,F) demonstrating a partial rescue of Vegfr inhibition phenotype. Arrows indicate DA and black arrowheads indicate PCV. Lateral view of 24 hpf embryos, anterior to the left. Numbers of embryos showing displayed phenotype are listed. (I) qPCR analysis of arterial (cldn5b), venous (flt4, dab2, stab1l) and pan-endothelial (cdh5) marker expression. Note that etv2 overexpression under fli1 promoter partially or completely rescues all marker expression in SU5416 treated embryos. *, p < 0.05; ** p < 0.01; significance of expression values in fli1:etv2+DMSO embryos was compared to wt +DMSO; ± SEM is shown.

3.7. Vegfaa mutants show a reduction of arterial marker expression while multiple venous markers are increased or unaffected

Both SU5416 and PTK787 inhibitors are expected to inhibit all Vegf tyrosine kinase receptors (Fong et al., 1999; Wood et al., 2000). Previously, defects in arteriovenous specification were observed upon morpholino (MO) inhibition of Vegfaa function, one of the two zebrafish VegfA homologs (Lawson et al., 2002). To confirm the previous results obtained using MO inhibition, we analyzed vegfaa genetic mutants obtained using the TALEN (Transcription activator-like effector nuclease) genome editing approach. The selected vegfaa allele (bns1) has a 10 bp deletion, resulting in a frame shift after amino acid Q60 and a premature stop codon, and is likely to represent a null allele (Rossi et al., 2016). Similar to previously reported Vegfaa MO results (Lawson et al., 2002; Nasevicius et al., 2000), vegfaa^−/−homozygous embryos display an absence of ISV development at 24 hpf, lack blood circulation, develop pericardial edema and die at 4–7 dpf (Fig. 9A,B, and data not shown)(Rossi et al., 2016). In addition, the DA and PCV fail to separate in vegfaa mutants and a single vessel is formed instead. Based on in situ hybridization and qPCR analysis, vegfaa mutant embryos showed a loss of expression of the arterial markers cldn5b, notch3 and ephb2a at 24 hpf (Fig. 9A–F,Q), similar to the phenotypes observed in Vegfaa MO embryos and Vegfr chemical inhibitor-treated embryos. The expression of venous markers flt4 and stab1l was downregulated but to a lesser extent than that of arterial markers, as apparent from qPCR analysis (Fig. 9Q). Based on ISH analysis, stab1l expression in vegfaa mutants was expanded into the artery, but its overall expression level was lower according to qPCR analysis (Fig. 9I, J, Q). Intriguingly, venous dab2 and lyve1b expression was slightly upregulated in vegfaa mutants, while mrc1a expression was not significantly changed. The expression of all these markers was expanded into the region of the dorsal aorta (Fig. 9K–Q). Similar to Vegfr inhibitor treatment results, cdh5 and etv2 expression was significantly downregulated in vegfaa mutants (Fig. 9G, H, Q). These results are largely comparable to the findings obtained with low doses of Vegfr inhibitor treatments and support the model that Vegfaa is involved in two different processes, vascular endothelial differentiation and arteriovenous specification (see Discussion). dab2, lyve1b and mrc1a expression is largely restricted to the vein, while flt4 and stab1l display a significant arterial expression, which explains why these markers respond differently in vegfaa mutants. However, in contrast to SU5416-treated embryos, no increased apoptosis was observed in vegfaa mutants (Fig. S6).

Fig. 9.

Analysis of vascular marker expression in vegfaa mutant embryos. (A–P) In situ hybridization analysis of arterial marker cldn5b (A,B), ephb2a (C,D), notch3 (E,F), pan-endothelial marker etv2 (G,H) and venous marker stab1l (I,J), dab2 (K,L), lyve1b (M,N), and mrc1a (O,P) expression in vegfaa−/− and wild-type sibling embryos at 24 hpf. Note the absence of arterial marker expression in vegfaa mutants (B,D,F) while venous markers are expanded into the arterial region of the single vessel present in vegfaa mutants (J,L,N,P). etv2 expression is reduced in vegfaa mutants and intersegmental vessels are absent due to the failure of angiogenesis (G,H). (Q) qPCR analysis of vascular marker expression in vegfaa−/−embryos compared to their siblings at 24 hpf. Note that all arterial and pan-endothelial markers are strongly downregulated, while flt4 and stab1l are only mildly downregulated. In contrast, venous-specific dab2 and lyve1b expression is increased, while mrc1a expression is unchanged. * p < 0.05; ** p < 0.01, ± SEM is shown.

To determine if vascular endothelial cell number is affected in vegfaa mutant embryos, we performed confocal imaging of wild-type and vegfaa mutant embryos crossed into the nuclear fli1:nGFP reporter line. Cell numbers were counted at 24 hpf in a selected region of the DA and PCV, corresponding to the length of two somites. Because the DA and PCV were often indistinguishable in vegfaa mutants, only the combined arterial and venous cell number was obtained. There was no significant difference in endothelial cell numbers between wild-type and vegfaa mutants (Fig. 10A–C).

Fig. 10.

All vascular endothelial cells acquire venous identity in Vegfaa MO knockdown embryos. (A–C) Vascular endothelial cell numbers counted within a selected region of two-somites in wild-type siblings and vegfaa−/−; fli1:nGFP; kdrl:mCherry embryos at 24 hpf. Only cells in the axial vasculature (not ISVs) were counted in the control embryos. Because vegfaa mutants form only a single vessel, it was not possible to assess arterial and venous cell numbers separately. There was no significant difference in cell numbers between the vegfaa mutants (n=7) and their wildtype siblings (n=8). (D–I) All vascular endothelial cells acquire venous identity in Vegfaa MO-injected stab1l:YFP; kdrl:mCherry embryos, analyzed by confocal microscopy at 48–52 hpf. Note that stab1l:YFP expression is largely restricted to the PCV and is absent from the DA in control uninjected embryos (D–F). In Vegfaa morphants, all vascular endothelial cells exhibit stab1l:YFP fluorescence (G–I). Transverse optical cross section is shown in the insets on the right; the position of cross section is marked with a dashed line (F,I).

Vascular endothelial cells in vegfaa mutants may possess uniform identity where each cell has altered arteriovenous identity. Alternatively, arterial and venous progenitors may be mixed, and the numbers of arterial progenitors could be reduced, while the numbers of venous progenitors may be increased. To test if all endothelial cells in Vegfaa knockdown embryos display venous marker expression, embryos from venous-specific stab1l:YFP and pan-endothelial kdrl:mCherry reporter lines were injected with the previously validated Vegfaa MO (Nasevicius et al., 2000). In wild-type embryos, venous cells in the PCV displayed strong stab1l:YFP expression, while arterial cells in the DA had weak to no YFP expression (Fig. 10D–F). In Vegfaa morphants all endothelial cells showed strong stab1l:YFP expression, and no endothelial cells with weak or absent YFP expression were observed (Fig. 10G–I). Altogether, these results argue that vegfaa mutants have no significant change in endothelial cell numbers, and all vascular endothelial cells acquire venous identity. However, because overall endothelial differentiation is decreased, generic endothelial markers and some venous markers show downregulation of their expression.

Vegfaa mutants, similar to Vegfr inhibitor treated embryos, show loss of arterial marker expression. We therefore tested if Etv2 over-expression was sufficient to restore arterial marker expression in vegfaa mutant embryos. Indeed, a fli1a:Etv2 DNA construct injected into vegfaa−/− embryos was sufficient to partially restore arterial marker cldn5b expression (Fig. 11). This suggests that Etv2 may directly regulate arterial marker expression in the absence of Vegfaa function.

Fig. 11.

Arterial marker expression is restored by etv2 overexpression in vegfaa mutant embryos. In situ hybridization analysis of arterial cldn5b expression at 24 hpf. vegfaa mutant embryos and wild-type siblings were injected with fli1a:etv2-2A-mCherry DNA overexpression construct together with tol2 mRNA. Note the absent cldn5b expression in vegfaa−/− embryos (C) and restored and expanded cldn5b expression in (D). ISV sprouting was not rescued in fli1:etv2-mCherry; vegfaa−/− embryos.

4. Discussion

Previous studies have implicated Vegf signaling in either overall vascular endothelial differentiation or arteriovenous specification (Chan et al., 2002; Lawson et al., 2002). However, these studies used different approaches to inhibit Vegf signaling and described different defects in Vegf-inhibited embryos. Therefore, how Vegf signaling functions in vasculogenesis has remained unclear, and it has not been definitively shown if and how Vegf signaling results in two different outcomes. In this study we assayed the role of Vegf signaling in vasculogenesis using both chemical inhibition and mutant analysis. Our results argue that Vegf has two distinct functions in vasculogenesis. Both the early inhibition of Vegf signaling (starting at the early stages of gastrulation) and high-dose inhibition of Vegfr result in the downregulation of both arterial and venous as well as overall vascular endothelial markers, and ultimately leads to endothelial cell apoptosis. This argues that Vegf signaling at low levels promotes overall vascular endothelial differentiation and cell survival. However, inhibition of Vegf signaling starting at mid and late-somitogenesis or using lower doses of Vegf inhibitors has less of an effect on overall vascular endothelial differentiation. Instead, these treatments inhibit arterial specification while venous markers are largely unaffected or expanded (Suppl. Fig. S7). This suggests that late Vegf signaling at higher levels affects arteriovenous specification. Intermediate doses of inhibitors and vegfaa genetic mutants display combinatorial phenotypes, where overall vascular differentiation and arterial marker expression is reduced, while venous marker expression is increased or unaffected. Because venous marker expression can be affected by Vegf signaling in opposite ways, its net effect depends on the dose and timing of the inhibitor used, and on a specific venous marker analyzed; each venous marker may respond differently to the same inputs.

Vegfr-inhibited embryos displayed apoptosis of vascular endothelial cells. The apoptosis observed was limited to vascular endothelial cells, suggesting that this is a specific effect and not a consequence of inhibitor toxicity. It is possible that the reduction of both arterial and venous marker expression is simply a consequence of endothelial cell apoptosis. However, several findings argue that apoptosis is a consequence and not the cause of inhibited endothelial differentiation. First, SU5416 treatment starting at the 23-somite stage resulted in down-regulation of all endothelial marker expression by 28hpf without significant apoptosis. Downregulation of pan-endothelial marker cdh5 and etv2 expression was observed at the 15-somite stage in SU5416 treated embryos (when treatment was started at the 50% epiboly stage), which is prior to the initiation of apoptosis. And lastly, previous reports have shown that loss of etv2 function or combined inhibition of etv2 and fli1b function results in endothelial cell apoptosis (Craig et al., 2015; Pham et al., 2007). This suggests that endothelial cells undergo apoptosis in Vegfr-inhibited embryos due to the reduction in etv2 expression and inhibition of endothelial differentiation.

Based on these results, we propose the following model which describes the sequential roles of Vegf signaling and Etv2 function in vasculogenesis (Fig. 12). etv2 expression is initiated at the 1–3-somite stages independent of Vegf function, apparently by other upstream regulators such as npas4l/cloche (Reischauer et al., 2016). The earliest expression of etv2 in the trunk region appears in arterial progenitors, which originate earlier and closer to the midline than venous progenitors, although they may not have acquired arterial identity at this stage yet (Kohli et al., 2013). Etv2 directly upregulates the expression of multiple vascular endothelial genes including vegfr2 homologs which enables vascular endothelial progenitors to respond to Vegf signaling. Starting at approximately the 10-somite stage etv2 expression also appears in venous progenitors, which is also initiated independent of Vegf signaling (Kohli et al., 2013; Sumanas and Lin, 2006). During mid-somitogenesis stages, Vegfa signaling upregulates etv2 expression. Because low levels of Vegf are sufficient to upregulate etv2 expression, it is induced in both arterial and venous progenitors. Etv2 then promotes the expression of multiple vascular endothelial-specific genes including arterial and venous markers. However, only arterial progenitors experience sufficiently high levels of Vegf due to their proximity to the source of Vegfaa within the somites. A high level of Vegf promotes arterial marker expression and represses venous markers, possibly in an Etv2-independent manner, by upregulating Delta-Notch signaling in the arterial progenitors. As shown previously, Vegf can promote arterial differentiation by activating other ETS factors through MAP kinase pathways (Wythe et al., 2013). At the same time, Etv2 also promotes the expression of venous markers, while Vegf signaling represses it. Therefore, Vegf overexpression can result in two opposing phenotypes, expansion of both arterial and venous markers due to increased etv2 expression, and downregulation of some venous markers due to their repression by increased Vegf signaling through the Delta-Notch pathway.

Fig. 12.

A proposed model for two distinct functions of Vegf signaling in vascular endothelial differentiation and arteriovenous specification. A low concentration of Vegf is sufficient to upregulate etv2 expression in both arterial (red) and venous (blue) vascular endothelial progenitors which promotes vascular endothelial differentiation and expression of both arterial and venous markers. A high concentration of Vegf promotes expression of arterial and inhibits expression of venous markers in an Etv2-independent manner.

SU5416 and PTK787 chemical inhibitors were originally developed to inhibit signaling by Flk1/VegfR2 (Fong et al., 1999; Wood et al., 2000). They are among the most specific inhibitors of Vegf signaling currently available. Both compounds show highly specific binding to Vegf receptors Flt1/VegfR1, Flk1/VegfR2 and Flt4/VegfR3 (Itokawa et al., 2002; Tille et al., 2003; Wood et al., 2000). In addition, they can inhibit, to some extent, related tyrosine kinases Flt3, c-Kit and PDGFR-β (Smolich et al., 2001; Wood et al., 2000; Yee et al., 2002). They do not bind to other families of tyrosine kinases such as FGF receptors (Fong et al., 1999; Wood et al., 2000). Therefore, it is unlikely that the inhibition of overall endothelial differentiation observed with these inhibitors is caused by their toxicity or non-specific inhibition of unrelated proteins. Neither Flt3, c-Kit nor PDGFR-β have been previously implicated in regulating vascular endothelial differentiation. While we cannot rule out the possibility that the effect of these chemical inhibitors on endothelial differentiation is caused by their inhibition of other targets, this appears unlikely. Vegfaa mutants recapitulated most of the defects observed with both chemical inhibitor treatments, including a strong loss of arterial marker expression, and a reduction in etv2, cdh5 and flt4 expression. The most significant difference was in the expression of venous dab2, lyve1b and mrc1a markers, which were decreased in embryos treated with high doses of SU5416 and increased (dab2, lyve1b) or unaffected (mrc1a) in vegfaa mutants. A similar expansion of dab2 expression was observed when embryos were treated with low doses of SU5416 (lyve1b was unaffected in this case), or when the treatment was performed starting at the late somitogenesis stages. PTK787 treatment resulted in a phenotype comparable to the low or intermediate doses of SU5416 treatment. Therefore, it is likely that the high dose of SU5416 treatment results in a more severe phenotype than that seen in vegfaa mutants or PTK787 treated embryos. This effect is possibly due to the more complete inhibition of Vegf signaling in the high dose-treated embryos, while vegfaa mutants show milder levels of Vegf signaling inhibition. This suggests that vegfaa mutants, which are likely to represent a null allele based on the nature of the mutation, may not completely eliminate Vegf signaling. Other Vegf homologs may function redundantly in promoting vascular endothelial differentiation. In fact, upregulation of vegfab expression has been reported in vegfaa mutant embryos (Rossi et al., 2015), which also supports the possibility that additional Vegf homologs compensate for the loss of Vegfaa function during endothelial differentiation. Intriguingly, kdr and kdrl morphants have been reported to exhibit a reduction of venous marker expression (Kim et al., 2013), similar to SU5416 or PTK787 treated embryos, which suggests that there is redundancy between different Vegf receptors during vascular endothelial differentiation. Our findings also explain the discrepancy in the previous reports where VegfA MO injection resulted in the suppression of arterial markers and expansion of venous markers, while PTK787 treatment inhibited overall vascular endothelial differentiation (Chan et al., 2002; Lawson et al., 2002). It is likely that the VegfA MO caused only a partial inhibition of Vegf signaling and, therefore, only affected arteriovenous specification.

Our results further argue that Vegf signaling modulates etv2 expression. Vegf inhibition resulted in a decrease in etv2 expression, while Vegf overexpression increased etv2 mRNA levels. Additionally, endothelial-specific overexpression of etv2 in Vegfr inhibited embryos partially rescued vascular endothelial differentiation defects, suggesting that Vegf induces overall endothelial differentiation by upregulating etv2 expression. We have previously demonstrated that Vegf over-expression is not sufficient to induce endothelial markers in the absence of Etv2 function (Sumanas and Lin, 2006), which further argues that Etv2 functions downstream of Vegf during endothelial differentiation. These results are in agreement with in vitro studies demonstrating direct activation of Etv2 expression by Vegf-Flk1 signaling in embryoid bodies (Rasmussen et al., 2013). However, we and others also previously showed that Etv2 is required for the expression of Vegf receptors such as Kdrl/Flk1, and that it directly regulates their expression (De Val et al., 2008; Kataoka et al., 2011; Sumanas and Lin, 2006). Because the expression of Vegf receptors is absent in etv2 mutant embryos, it is likely that etv2 expression is initiated not by Vegf signaling but by other pathways instead. Indeed, our results show that the initial expression of etv2 at the 2–3-somite stage is not affected in Vegfr-inhibited embryos. A bHLH-PAS transcription factor npas4l/cloche has been recently implicated in regulating etv2 expression (Reischauer et al., 2016). After Etv2 initiates Vegf receptor expression, Vegf signaling functions as part of a positive feedback loop to upregulate the expression of etv2.

Intriguingly, Etv2 expression restored arterial and venous markers in Vegf-inhibited embryos. This argues that Etv2 can initiate arterial and venous marker expression in the absence of Vegf signaling, likely by direct transcriptional activation of these genes. Indeed, many vascular endothelial markers, including Delta and Notch homologs as well as flt4 are predicted to be direct transcriptional targets of Etv2, and ETS transcription factors have been implicated in direct regulation of arterial differentiation (De Val et al., 2008; Wythe et al., 2013). While Vegf may regulate arterial marker expression also in an Etv2-independent manner through other ETS transcription factors, increased Etv2 expression can clearly bypass this Vegf requirement.

In summary, we have shown that Vegf signaling results in two distinct outputs, arteriovenous specification and overall vascular endothelial differentiation, and demonstrated that Vegf signaling regulates vascular endothelial differentiation by modulating etv2 expression. Our results also argue that the zebrafish Vegfaa homolog is involved in both processes; however, additional Vegf homologs may contribute to the specification of vascular endothelial identity. Collectively, these findings provide new and critical insight into the intricacies of the well-established Vegf signaling pathway, and will help further our understanding of the molecular mechanisms of vasculogenesis.

Appendix A. Supporting information

Supplementary data associated with this article can be found in the online version at doi:10.1016/j.ydbio.2017.03.005.