Distinct Requirements for Zebrafish Angiogenesis Revealed by a VEGF-A Morphant

Angiogenesis is a fundamental vertebrate developmental process that requires signalling by the secreted protein vascular endothelial growth factor-A (VEGF-A). VEGF-A functions in the development of embryonic structures, during tissue remodelling and for the growth of tumour-induced vasculature. The study of the role of VEGF-A during normal development has been significantly complicated by the dominant, haplo-insufficient nature of VEGF-A-targeted mutations in mice. We have used morpholino-based targeted gene knock-down technology to generate a zebrafish VEGF-A morphant loss of function model. Zebrafish VEGF-A morphant embryos develop with an enlarged pericardium and with major blood vessel deficiencies. Morphological assessment at 2 days of development indicates a nearly complete absence of both axial and intersegmental vasculature, with no or reduced numbers of circulating red blood cells. Molecular analysis using the endothelial markers fli-1 and flk-1 at 1 day of development demonstrates a fundamental distinction between VEGF-A requirements for axial and intersegmental vascular structure specification. VEGF-A is not required for the initial establishment of axial vasculature patterning, whereas all development of intersegmental vasculature is dependent on VEGF-A signalling. The zebrafish thus serves as a quality model for the study of conserved vertebrate angiogenesis processes during embryonic development.


Introduction
Signalling by members of the vascular endothelial growth factor (VEGF) gene family is implicated in the formation of vasculature during embryogenesis, during wound healing and for the growth of tumour-induced vasculature (for reviews, see Carmeliet and Collen, 1997;Ferrara, 1999). Pioneering work in mice with VEGF-A demonstrates the extreme dose responsiveness of the mouse embryo to VEGF-A signalling during development. Loss of a single copy of the VEGF-A gene induces haplo-insuf®cient lethality by day 9.5 pc (Ferrara et al., 1996;Carmeliet et al., 1996). This biological hurdle to the genetic investigation of VEGF-A requirements during later development has resulted in a series of experiments using conditional knock-out strategies (Gerber et al., 1999;Haigh et al., 2000) or dominant negative proteins (Gerber et al., 1999). A more recent approach to address this problem used intravenous injection of antisense oligonucleotides in pregnant mice to reveal loss of function requirements of VEGF-A function during murine embryogenesis (Driver et al., 1999).
Zebra®sh embryos develop externally and have only limited requirements for a functioning circulatory system during early development. For example, embryos with no circulating red blood cells due to porphyria live through the ®rst 3 days of development (Ransom et al., 1996), a time period which includes all of segmentation and organogenesis in the ®sh embryo. Multiple mutations in cardiovascular development were isolated in the initial largescale chemical mutagenesis screens (Stainier et al., 1996;Chen et al., 1996;Weinstein et al., 1995). The zebra®sh has the potential to rapidly assess the biological role of angiogenic factors required for this essential vertebrate process.
We used morpholino-based oligonucleotides (morpholinos, MO;Summerton, 1999) to create a VEGF-A loss of function developmental model to circumvent potential haplo-insuf®cient genetic complications in zebra®sh. These VEGF-A morphant embryos display an enlarged pericardium and a modest body size decrease and have severe de®ciencies in vascular development. We show that the establishment of the axial and intersegmental vasculature has distinct requirements for VEGF-A signalling, as revealed by analysis of the expression of two endothelial markers, the tyrosine kinase VEGF receptor,¯k-1 (Sumoy et al., 1997;Fouquet et al., 1997;Thompson et al., 1998) and the early vascular marker, the transcription factor¯i-1 (Thompson et al., 1998;Brown et al., 2000). Initial axial expression of these markers is not altered in VEGF-A morphant embryos, while no intersegmental expression of these markers is detected. Both axial and intersegmental vasculature fail to function in VEGF-A morphant embryos, however, indicating a role for VEGF-A beyond the establishment of¯k-1 and¯i-1 expression in blood vessel formation. Similar phenotypes were observed in some mutations found in chemical mutagenesis screens, suggesting a possible role for these genes in VEGF-A signalling in the zebra®sh embryo.

FITC±Dextran injections
Microangiography was performed similarly to the method described in Weinstein et al. (1995). Fluorescein isothiocyanate±Dextran (FITC±Dextran) with a molecular weight of 2 000 000 Da (SIGMA, catalogue #FD-2000S) was used for these studies. The Dextran was solubilized in 1rDanieau solution (58 mM NaCl, 0.7 mM KCl, 0.4 mM MgSO 4 , 0.6 mM Ca(NO 3 ) 2 , 5 mM HEPES, pH 7.6) at 2 mg/ml concentration. 10 ml of the prepared solution was injected into the sinus venosa/cardinal vein of the anaesthetized 48 h embryo. The visualization and photography was performed on a ZEISS Axioskop 2 microscope using a standard FITC ®lter set.

Red blood cell (RBC)¯uorescence visualization in urod morpholino-injected embryos
The¯uorescence of red blood cells was observed and photographed as described in Nasevicius and Ekker (2000).

RNA localization
Whole mount in situ hybridization was performed as described in Jowett (1999). Hybridization was performed at 65uC. Riboprobes for¯i-1 and¯k-1 were synthesized using plasmids zf¯i-1 and zf¯k-1 (Thompson et al., 1998), digested with EcoRI and SmaI, respectively. T7 polymerase was used for riboprobe synthesis.

Tissue sectioning and visualization
Embryos with FITC±Dextran visualized vasculature were ®xed overnight, embedded into paraf®n using standard procedures, and sectioned. FITC± Dextran¯uorescence outlining blood vessels in unprocessed tissue sections was visualized using ZEISS Axioskop 2 microscope with a FITC ®lter set. Histological haematoxylin±eosin staining of the sections was subsequently carried out using a standard protocol.

Digital photography
Bright-®eld and in situ photography was performed on a ZEISS Axioplan 2 microscope using Nikon CoolPix 990 (bright-®eld) or Kodak DCS 420 (in situ) digital cameras. For¯uorescent photography, a ZEISS AxioCam or a Nikon CoolPix 990 digital camera was used.

Results and discussion
Zebra®sh VEGF-A is expressed during embryogenesis in the anterior nervous system, in mesoderm anking the prospective heart ®elds, and in somitic mesoderm that¯anks the developing endoderm (Liang et al., 1998). We generated morpholino (Summerton, 1999) antisense oligonucleotides against VEGF-A to analyse the requirements of this gene during embryonic development. Morpholinos have been recently shown to be effective at gene inactivation during the ®rst 2 days of zebra®sh development (Nasevicius and Ekker, 2000). We term the loss of function effects due to morpholinos a`morphant' phenotype to distinguish this assessment of gene function from that of classical mutant analyses.
VEGF-A morphant embryos develop with no overt phenotype during the ®rst day of development. The VEGF-A morphant phenotype at 2 days of embryogenesis, however, consists of an enlarged pericardium, no circulating red blood cells, a slight reduction in neural tube and overall body size, and little or no functioning vasculature ( Figure 1B). In a subset of embryos, red blood cell accumulation can be noted in the ventral tail ( Figure 1C).
We used two separate¯uorescent assays to assess vascular function in detail. First, we generated uorescent-labelled red blood cells through the inactivation of the uroporphyrinogen decarboxylase (urod; Wang et al., 1998) gene using morpholinos (Nasevicius and Ekker, 2000). Red blood cells quantitatively accumulate in the anterior hypochord ( Figure 2B). To analyse the vasculature directly, we injected¯uorescein isothiocyanate±Dextran (FITC±Dextran) into the sinus venosa/cardinal vein of an anaesthetized 48 h embryo ( Figure 2C). This microangiography assay labels the entire vasculature of the zebra®sh embryo, including the yolk sac, heart, head, axial and intersegmental blood vessels (Weinstein et al., 1995). These structures are differentially sensitive to VEGF-A signalling. At high dose injections of VEGF-A morpholino, the only vasculature detectable in these animals using this method is found in the heart and yolk ( Figure 2D). The vasculature either fails to form at all or contains no functioning connections to the heart in these embryos. To distinguish between these possibilities we performed histological analyses on these most severely effected embryos ( Figure 2G±J). Neither dorsal aorta nor axial vein were noted in the injected embryos ( Figure 2I, J). We also observed a frequent but less severe phenotypic class of embryos with heart, yolk and head blood vessels ( Figure 2E); no axial or intersegmental vasculature was observed in these embryos, however. The least severe phenotypic classi®cation was represented by embryos with reduced intersegmental vasculature and normal heart, yolk, head and axial blood vessels (Figure 2F). The penetrance of these phenotypic classes is very dose-dependent (Table 1), consistent with the strong dose dependence of VEGF-A function in mouse embryos (Ferrara et al., 1996;Carmeliet et al., 1996). Heterozygous mouse VEGF-A mutants showed a reduced dorsal aorta detected by histological analysis. Fewer intersegmental blood vessels were also detected by a tissue-speci®c lacZ expression (Carmeliet et al., 1996). Lack of the dorsal aorta was indicated by histological analysis in homozygous mouse VEGF-A mutants (Carmeliet et al., 1996), suggesting that the most severe zebra®sh morphant classes represent a nearly complete loss of function phenotype. However, while mouse VEGF-A mutants also display hearts with underdeveloped myoblasts (Ferrara et al., 1996;Haigh et al., 2000), the heart in zebra®sh VEGF-A morphants has an essentially normal appearance with a slightly enlarged atrium and ventricle, possibly due to higher cardiac pressure (histological analysis not shown).
We assessed the phenotypic effects of two VEGF-A morpholinos of non-overlapping sequence and of a four-base mismatch sequence (see Materials and methods) to con®rm the speci®city of targeting ( Table 2). The four-base mismatch morpholino VEGF-A-D4 demonstrates the sequence-speci®c nature of the noted effects of the VEGF-A-1 morpholino (Table 2). To test independently for the speci®city of targeting to the endogenous VEGF-A gene, we used a second morpholino of completely independent sequence (VEGF-A-3). This very potent morpholino caused the same phenotypic effects on development, including a dose-dependent reduction of vascular function (Table 2), pericardial oedema and blood accumulation in the tail (data not shown). The observed differential ef®cacy might be due to the different secondary structure of the morpholinos or the targeted mRNA region. Alternatively, the effect might be caused by the higher VEGF-A-3 predicted melting temperature due to higher G/C content (48%) as compared to VEGF-A-1 (28%). We conclude that the observed effects are due to morpholino-based inactivation of VEGF-A gene function through the speci®c inhibition of VEGF-A transcript translation.
A number of genes whose mutation results in cardiovascular defects were observed in the chemical mutagenesis screens (Stainier et al., 1996;Chen et al., 1996). None of these mutations strongly resemble the VEGF-A morphant effects, although mutant embryos with overlapping phenotypes were  Table 1 and text). Blood accumulation in tail is also observed (C, black arrows; see Table 1  noted. Multiple loci result in embryos with cardiac oedema, and a similar accumulation of blood in the ventral tail ®n was noted due to disorganized endothelia in the scotch tape (sco) mutation . Several mutations with altered circulation were noted (Stainier et al., 1996;Chen et al., 1996), including gridlock, which encodes for a bHLH protein that is required only for arterial development (Weinstein et al., 1995;Stainier et al., 1996;Zhong et al., 2000). The role of these genes in VEGF signalling awaits molecular genetic characterization of the remaining loci. We analysed the expression of two endodermal vascular markers in VEGF-A morphant embryos. The transcription factor¯i-1 is a very early marker of vascular cell fate speci®cation (Thompson et al., 1998;Brown et al., 2000). In wild-type 26 h embryos,¯i-1 is expressed in the forming dorsal aorta and axial vein (axial vessels; arrowhead in Figure 3A) and in the intersegmental vasculature in overlying somites ( Figure 3A, arrows). In embryos that fail to complete vascular development, only a subset of the vascular expression pattern of these genes is altered, contrary to all vasculature sensitivity to VEGF-A knock-down. No detectable intersegmental expression is noted ( Figure 3B), Four (VEGF-A-D4 morpholino) or two (VEGF-A-3 morpholino) experiments were performed with each morpholino dose. The phenotype Figure 2. Visualization of vasculature defects in VEGF-A-1 morphants using microangiography (Weinstein et al., 1995). coinciding with the exquisite intersegmental vascular endoderm sensitivity to VEGF signalling ( Table 1). The cells are either not properly speci®ed or fail to migrate during formation of the intersegmental vessels. Similar results were obtained upon analysis of expression of the VEGF receptor,¯k-1. k-1 transcript distribution is very similar to that of i-1 in the trunk and tail of wild-type embryos ( Figure 3C). In VEGF-A morphant embryos, intersegmental but not axial expression is absent ( Figure 3D). A signi®cant reduction in¯k-1 gene expression was noted in mouse embryos with no VEGF-A activity (Carmeliet et al., 1996). A less extreme lack of¯k-1-expressing cells in the intersegmental vasculature was also observed in the mouse with the partial and conditional VEGF-A knock-out (Haigh et al., 2000). The results in Figure 3 represent work with 9 ng VEGF-A-1 injection-dose embryos; 18 ng injection-dose embryos display the same speci®c loss of expression only in the intersegmental regions for both¯i-1 and k-1 (data not shown).
The distinct responsiveness of the expression of the endothelial marker¯i-1 in intersegmental vessels to VEGF-A signalling demonstrates a dual role for VEGF during vascular development. First, VEGF-A is required for proper axial vessel formation but not for initial axial vessel patterning. Second, VEGF-A is required for intersegmental vessel cell speci®cation or migration and, presumably, for subsequent vascular formation. The lack of a requirement for VEGF signalling for¯k-1 expression is consistent with previous observations of paracrine modes of VEGF signalling (reviewed in Ferrara, 1999). The expression of the VEGF receptor¯k-1 is, however, VEGF-dependent during intersegmental vascularization. This latter observation suggests a possible autoregulatory loop, functioning during vasculogenesis of the intersegmental vessels.
The strong conservation of VEGF function from ®sh to mammals implicates this as a fundamental vertebrate biological pathway. The use of morpholino-based gene targeting represents a new tool in the genetic repertoire of vertebrate biologists and, combined with the excellent embryology of the zebra®sh, will be extremely powerful in the elaboration of gene function for similarly conserved developmental processes. This method will help further elaborate the role of VEGF during embryogenesis through the targeted knockdown of other players in this signalling pathway in this outstanding model system.