The vascular endothelium plays a critical role in the formation and permanence of blood vessels in vertebrates. A complete cardiovascular system able to transport blood requires the proper differentiation, proliferation, and migration of endothelial cells during both vasculogenesis and angiogenesis (Yancopoulos et al., 2000). During vasculogenesis, vascular endothelial cells differentiate from angioblasts and form a primary vascular lumen. Angiogenesis involves the production and maturation of new vessels by endothelial sprouting from the primary vasculature. This process is pivotal for various neovascularization events such as wound healing and pathological tumor proliferation.
The regulation of vascular endothelial development is complex and involves several known molecules and pathways (Folkman and D'Amore, 1996). The cell–cell signaling molecule vascular endothelial growth factor-A (VEGF-A) plays an essential and conserved role in vascular formation in humans (Ferrara, 1999), mice (Carmeliet et al., 1996; Ferrara et al., 1996), chickens (Wilting et al., 1993), and zebrafish (Nasevicius et al., 2000). Additionally, the VEGF receptor Flk1 (Fouquet et al., 1997, Habeck et al., 2002) and Syndecan-2 (Chen et al., 2004) have conserved angiogenic functions required for the transduction of VEGF signals. In mice, the receptor tyrosine kinases Tie-1 and Tie-2 are required for the structural integrity of vascular endothelial cells and angiogenic maturation, respectively (Sato et al., 1995). The expression of tie-1 and tie-2 in the developing vasculature is conserved in zebrafish (Lyons et al., 1998). The presence of the same genes and the common results of the knockout or knockdown of these genes in different vertebrates indicate conservation of the fundamental pathways for vasculature development.
Vascular endothelial cells also express a number of conserved cell surface genes such as vascular endothelial cadherin (VE-cadherin/Cdh5), neural cadherin, and several integrins that mediate cell–cell adhesion and cell–substrate adhesion. Of these genes, Cdh5 is specifically expressed in human (Lampugnani et al., 1992) and mouse (Breier et al., 1996) endothelial cells in both developing tissue as well as mature vasculature. Cdh5 is a member of the cadherin family of cell–cell adhesion molecules and aids in the calcium-dependent adhesion of endothelial cells (Heimark et al., 1990). Furthermore, as shown in mouse knockout models, Cdh5 is required for endothelial cell adhesion and survival beyond the initial formation of the vasculature (Carmeliet et al., 1999). The requirement for Cdh5 in cell survival revealed an additional cell signaling function for Cdh5 in the transmission of VEGF-A survival signals to Akt kinase and Bcl2. This effect is mediated by Cdh5 forming a complex with β-catenin, phosphoinositide-3-OH kinase (PI3-K), and VEGF receptor-2. Given its strong expression and essential roles in the vasculature in higher vertebrates and its conserved expression pattern, we investigated the applicability of Cdh5 as a vascular marker in zebrafish.
Investigation of the vascular system is practical using zebrafish as a developmental vertebrate model. Large-scale production of progeny, external embryogenesis, and visual transparency are a few great advantages of zebrafish as a model. Rapid development provides the zebrafish embryo with an operational primary cardiovascular system within 24 hr postfertilization (hpf), which becomes a detailed vascular network by 2 days postfertilization (dpf; Isogai et al., 2001). Therefore, the quick assessment of both vasculogenesis and angiogenesis is quite simple in the zebrafish embryo over the first 2 days of its development.
The expression of cdh5 in zebrafish embryos is advantageous for analyses of vascular endothelial cells during the establishment and maturation of its cardiovascular system. Before a functional circulatory system, cdh5 likely identifies angioblasts in the zebrafish embryo that migrate and localize to form vascular structures. cdh5 continues to clearly mark blood vessels throughout the embryo after blood circulation begins. Thus, cdh5 expression serves as a new tool to track the progression of endothelial cells during vasculogenesis and angiogenesis. To illustrate this utility, we have demonstrated the functional conservation of VEGF-A between zebrafish and humans. Human VEGF-A compensates for the lack of activity in zebrafish VEGF-A knockdown embryos, further demonstrating the value for using zebrafish to identify and study vascular genes in vertebrates.
Identification of Zebrafish cdh5
We identified the zebrafish cdh5 gene by searching genomic and EST databases for cadherin-related molecules. Based on these searches, a single gene was identified as a likely ortholog to Cdh5 in humans, mice, and chickens and consequently was named cdh5. The entire coding sequence for zebrafish cdh5 was established using reverse transcriptase-polymerase chain reaction (RT-PCR). Comparison of the predicted amino acid sequence of zebrafish CDH5 to the encoded amino acid sequences for human, mouse, and chicken CDH5 revealed a high degree of amino acid identity within the cytoplasmic-catenin binding region, which is common to all cadherins, as well as within the extracellular cadherin-repeat region (Fig. 1A). Based on the overall amino acid identities (Fig. 1B), the zebrafish cdh5 gene is the likely ortholog to Cdh5 in other vertebrate species. Amino acid analysis revealed other cadherin-like molecules with low-level homology in the core part of the protein (amino acids 100–500), but none of these molecules showed significant homology to CDH5 in the well-conserved c-terminus (data not shown). The Sanger Centre Web site (http://www.ensembl.org/Danio_rerio) states that version Zv3 of the assembly, released on November 27, 2003, represents roughly 5.7× coverage, and the supercontig coverage is estimated to be 96–97%. All versions of the zebrafish genome assembly were searched for a second sequence similar to the cdh5 cDNA, but no sequence of any relevance was identified. To further verify the identity of the zebrafish cdh5 gene, whole-mount in situ hybridization was performed to examine its expression.
Expression of cdh5 During Zebrafish Embryogenesis and Vasculature Formation
We observed the expression pattern of cdh5 throughout zebrafish embryogenesis, before and after the cardiovascular system is established. cdh5-expressing cells are first visible by 12 hpf in bilateral patches at the edge of anterior and trunk mesoderm (Fig. 2A,B). Between 12 and 18 hpf, the cells of the anterior lateral mesoderm converge medially (arrows in Fig. 2A) in the developing head to form the symmetric primordia of the anterior vasculature (Fig. 2C). The cdh5-expressing cells in the head organize into an intricate pattern of vessel primordia for the developing brain and eyes by 19 hpf (Fig. 2E). In parallel, the first arteries in the trunk are marked with cdh5 expression. Between 12 and 16 hpf, cdh5-positive cells in the trunk lateral mesoderm migrate to the midline (arrows in Fig. 2B) in an anterior-to-posterior manner (Fouquet et al., 1997). Arterial structures are present throughout the trunk by 18 hpf (arrows in Fig. 2D). During this time, cdh5 expression also marks the progression of vein formation. At 16 hpf, additional bilateral cells that border the trunk and tail bud endoderm express cdh5 (data not shown). By 19 hpf, these cells proceed to congregate into veins in the anterior trunk (arrowheads in Fig. 2D,F) and along the ventral midline of the posterior trunk (arrowhead in Fig. 2G). Intersegmental sprouts emerge from the tail artery primordium in an anterior-to-posterior pattern at 19 hpf (arrows in Fig. 2G) as the first wave of angiogenesis in the trunk.
As the vasculature forms throughout the zebrafish embryo, blood cells begin to circulate through a basic loop of primary blood vessels in the trunk, tail, and yolk sack at 24 hpf (Isogai et al., 2001). Vascular structures can be identified by cdh5 expression (Fig. 3A–F). At this stage, blood circulates from the heart into the ventral aorta and mandibular arch (Fig. 3E) and proceeds to flow through the arteries and veins of the trunk and tail. Bilateral intersegmental sprouts, secondary vessels formed by angiogenesis, are marked by cdh5 and span the length of the trunk and tail (Se in Fig. 3F) but do not yet allow blood flow. At least 16 separate vascular structures can be identified with cdh5 after 1 day of development.
Soon after the 24 hpf time point, the intersegmental sprouts proceed to extend dorsally and split at the dorsolateral surface of the neural tube to form the dorsal longitudinal anastomotic vessels (DLAV) in an initial establishment of the angiogenic network (Isogai et al., 2003). This transient movement example of the sprouts to form the DLAV can be detected by cdh5 at approximately 28 hpf (arrowheads in Fig. 3G). This finding demonstrates that the in situ hybridization signal of cdh5 in conjunction with simple microscopy can identify the leading endothelial cells during primary angiogenic sprouting of the intersegmental structures.
By 2 dpf, the zebrafish embryo possesses a detailed vascular system throughout its body (Fig. 4). The blood vessels at 2 dpf can be visualized and identified with cdh5 expression throughout the head, trunk, and tail (Fig. 4A–E). Detailed inspection of the trunk (Fig. 4F) reveals localized cdh5-positive cells that surround primary vessels (dorsal aorta [DA] and posterior cardinal vein (PVC)) and mark the fully functional, secondary vessels (Se and DLAV) that form by angiogenesis. The expression of cdh5 in Figure 4 documents at least 30 distinct vascular structures in the zebrafish embryo. For ease of viewing the natural images in Figures 3A,C, 4A,C, respective negative images are used to label the areas of expression in Figures 3B,D, 4B,D. Table 1 lists the vascular nomenclature for particular vessels (Isogai et al., 2001). The expression pattern of cdh5 at 2 dpf is comparable to the functional vasculature visualized by microangiography (Fig. 4G).
Table 1. Identity of Vascular Structures Labeled With cdh5
First branchial arch
Second branchial arch
Third branchial arch
Fourth branchial arch
Anterior cerebral vein
Anterior cardinal vein
Caudal division of the internal carotid artery
Common cardinal vein
Cranial division of the internal carotid artery
Dorsal longitudinal anastomotic vessel
Inner optic circle
Lateral dorsal aorta
Middle cerebral vein
Posterior cerebral vein
Posterior cardinal vein
Primordial hindbrain channel
Primitive internal carotid artery
Primordial midbrain channel
cdh5 Is Expressed in Endocardial Cells of the Developing Zebrafish Heart
cdh5-positive cells begin to migrate toward the midline in the heart field by approximately 16 hpf (arrow in Fig. 5A) where they proceed to coalesce to form a ring of cells that will reside in the future heart cone (arrow in Fig. 5B). These cells lie between the bilateral cardiac tubes (data not shown) in the position of endocardial precursor cells (Stainier et al., 1993). At 26 hpf, cdh5 continues to mark the endocardium in the future atrium (arrow in Fig. 5C) and ventricle (arrowhead in Fig. 5C) of the heart tube. The cdh5 expression localized throughout the endocardium is sustained after the heart has looped (Fig. 5D). Intense staining was observed during this period in the endocardial cushions between the atrium and ventricle.
Zebrafish cdh5 Marks the Location of Endothelial Cells During Angiogenesis
Angiogenesis in the embryo tail and trunk achieves functional intersegmental vessels that sprout dorsally from primary vessels to the top of the neural tube (arrow in Fig. 6A). With microangiography at 36 hpf, we visualized the functional vascular system (Fig. 6B) during angiogenesis in the embryo trunk and tail and compared these vessels with the corresponding location of vascular endothelial cells with cdh5 expression at the same stage (Fig. 6C). The primary arteries and veins in the trunk and tail allow blood flow as assessed by microangiography (Fig. 6B) and are marked by cdh5 expression (Fig. 6C). Intersegmental vessels at this stage have full function (arrow in Fig. 6B), only a partial lumen (arrowheads in Fig. 6B), or no lumen at all. In comparison, cdh5 is expressed in endothelial cells that span from the primary vasculature to the top of the neural tube (arrows in Fig. 6C), indicating that the endothelial cells have properly migrated to lead the formation of functional intersegmental vessels. The expression of cdh5 in functional and nonfunctional vessels demonstrates the ease of analysis of endothelial sprouting and subsequent migration with cdh5 during angiogenesis.
Expression of cdh5 Detects the Angiogenic Function of Human VEGF-A in Zebrafish Embryos
We used cdh5 as a vascular endothelial-specific expression marker to analyze the phenotype of embryos with reduced VEGF-A165 activity. In this experiment, we injected a morpholino phosphorodiamidate oligonucleotide (MPO) that targeted the 5′ untranslated region (UTR) of zebrafish VEGF-A165 to produce morphant embryos with defective blood circulation (Nasevicius et al., 2000). We observed a complete loss of intersegmental vessels throughout the tail in most MPO-injected embryos at 28 hpf (arrows in Fig. 7B, 90% ± 1), compared with uninjected controls (arrows in Fig. 7A).
We then investigated the capacity of the human VEGF-A165 homologue to restore the loss of intersegmental vessels in VEGF-A165–deficient zebrafish embryos. For this, we coinjected zebrafish embryos with VEGF-A165 MPO and a DNA mixture containing the coding sequence of zebrafish (z) or human (h) VEGF-A165 and the enhanced green fluorescent protein (EGFP) control. Expression of cdh5 revealed the complete restoration of intersegmental vessels in a majority of these embryos at 28 hpf (arrows in Fig. 7C, 82% ± 6 with zVEGF165; 89% ± 4 with hVEGF165). Most of the embryos coinjected with VEGF-A165 MPO and EGFP DNA alone did not show restored intersegmental vessels (85% ± 7) as revealed by cdh5 expression. Figure 7D represents a comprehensive summary of these injection experiments and the resulting cdh5 expression phenotypes. Our results indicate a specific effect of the VEGF-A165 knockdown, illustrated by cdh5 expression in embryos rescued with zebrafish or human VEGF-A165. Most interestingly, these studies show that the angiogenic function of human VEGF-A165 in zebrafish embryos is highly similar to that of zebrafish VEGF-A165.
This study describes the cloning and expression pattern of zebrafish cdh5 and the use of cdh5 expression as a method to examine vascular development in zebrafish. While the amino acid identity of the zebrafish Cdh5 is somewhat low, 37 to 40%, compared with that among CDH5 amino acid sequences from other vertebrates, comparisons of the zebrafish Cdh5 sequence to other cadherin family members from humans showed that the zebrafish Cdh5 is more similar to human CDH5 than any other cadherin family member (data not shown). This finding, coupled with the expression pattern of zebrafish cdh5, strongly suggests that the cdh5 gene identified in this study is an ortholog to Cdh5 in other vertebrates.
The expression of zebrafish cdh5 was specific to the vasculature, and when compared with microangiography of the functional vasculature, it is likely that cdh5 marks all of the vascular endothelial cells in the developing zebrafish embryo. This expression pattern is highly conserved in vertebrates. Furthermore, in the current study, we show that cdh5 expression is a downstream consequence of VEGF-A signaling (Fig. 7) and that VEGF-A is required for cdh5 expression in zebrafish. However, because VEGF-A is required at multiple steps during vessel formation, it is likely that the loss of cdh5 staining represents a lack of endothelial cell migration during the formation of the intersegmental vessels and not direct regulation of cdh5 by VEGF-A. In mice, Cdh5 is considered to function by forming a complex with β-catenin, phosphoinositide-3-OH kinase, and VEGF-A receptor Flk1 and is essential for the reception of VEGF-A signaling for endothelial cell survival (Carmeliet et al., 1999). Given the interaction of these molecules in mice, it is possible that VEGF-A could regulate zebrafish cdh5 during later steps of vessel formation that would not be uncovered in the current study.
Different methods exist to evaluate the development of zebrafish vasculature. Transgenic embryos that express fluorescent proteins employ promoters that are active in vascular tissues offer in vivo analysis of the living vasculature. Examples include transgenic lines with the following promoter:fluorescent gene combinations: TG(fli1:EGFP) (Lawson and Weinstein, 2002) and TG(VEGFR2:G-RCFP) (Cross et al., 2003). This method, however, involves the propagation and maintenance of one or more transgenic strains. Furthermore, the data that transgenic embryos provide are difficult to preserve for future analysis. For high-throughput studies, an analytical tool that visualizes the vasculature in fixed tissue would complement these live analyses. One example is the detection of endogenous alkaline phosphatase activity (AP staining) to visualize blood vessels in the trunk of fixed embryos (Serbedzija et al., 1999). However, AP activity fails to offer a clear and reproducible readout of the vasculature, because it is present in other tissues in addition to the vasculature and AP has a late onset of expression in the vascular endothelium.
We have demonstrated that in situ hybridization with cdh5 provides an alternative method for investigating the vascular network. The comprehensive expression of this gene is useful for three-dimensional marking of vascular endothelial cells in whole zebrafish embryos. The clarity of cdh5 expression allows for detailed visual inspection of the circulatory system as it develops and matures as apposed to the use of late vascular markers such as AP. Fixed signals from cdh5 in situ hybridizations are advantageous for analyzing multiple experimental samples. Additionally, cdh5 as a probe is useful in MPO experiments. A potential side effect of MPO injections is transient neural degeneration (Ekker and Larson, 2001). Under these conditions, the pattern of cdh5 is unaltered (data not shown). In combination with rapid developing zebrafish embryos, cdh5 expression is a measurable and reproducible strategy for studying the vasculature.
We established the use of cdh5 in situ hybridization to identify vascular defects in zebrafish embryos as a model for the analysis of functional conservation between zebrafish and human genes required for vascular development. In the current example, human VEGF-A was able to restore proper endothelial cell development in zebrafish embryos that lack complete VEGF-A activity as a result of the morpholino-mediated knockdown of VEGF-A. The utility of this cost-effective, high-throughput model is valuable for further research in pro- or antiangiogenic evaluations of human genes and can assist in large-scale drug screening and target validation of corresponding therapeutics.
Bioinformatics and Cloning of Zebrafish cdh5
A single genomic trace sequence was identified (zfish44906-2739a12.p1k) with a best-hit match to human CDH5 by BLASTX (Altschul et al., 1990). This sequence was used to identify the appropriate genomic supercontig sequence (z06s043902) by BLASTN (Altschul et al., 1990). These genomic sequence data were produced by the Zebrafish Sequencing Group at the Sanger Institute and can be obtained from ftp://ftp.sanger.ac.uk/pub/zebrafish. GenScan (Burge and Karlin, 1997) analysis (http://genes.mit.edu/GENSCAN.html) coupled with TBLASTN (Altschul et al., 1990) analysis of the human CDH5 protein to this supercontig using bl2seq (Tatusova and Madden, 1999) yielded the putative coding sequence for the zebrafish cdh5 ortholog. Amino acid alignment analysis was carried out by using the ClustalW (Thompson et al., 1994).
To verify the coding sequence for this potential ortholog, RT-PCR was performed on RNA isolated from 24 hpf embryos and reverse transcribed with Superscript II (Invitrogen). Initially, a 1,395-bp segment of the cDNA was amplified with Taq DNA polymerase (Bioline) by using the following primers: 5′-TTCATGGTTGCTGTCCTGCTCACAT-3′ and 5′-GCTCCCATTGGTTAGTTCTGGTGCAT-3′. This segment was cloned into pCR4-TOPO by using the TopoTA cloning kit (Invitrogen) and subsequently used as a probe for in situ hybridization experiments.
The 5′-rapid amplification of cDNA ends (RACE) PCR (Clontech) was also carried out to define the 5′-end of the cdh5 cDNA. The 2.5 μg of RNA were reverse transcribed as above with the following primers: 5′-CCAAGTTCGTACAAAAAAGCAGGCTCGGGG-3′ and 5′-ACCACTTCCTACAACAAAGCTGGGTTTTTTTTTTTTTTTTTTTTTT-3′. Primary 5′-RACE PCR reactions were carried out with a cdh5 specific primer, a long 5′-RACE primer, and a short 5′ RACE primer: 5′-CGTGTTTCCTCATACGCATACAGTTTATCC-3′, 5′-AATGGCTTCCGAGACCTGCTGTCCAAGTTCGTACAAAAAAGAGGC-3′, and 5′-AATGGCTTCCGAGACCTGCTGT-3′, respectively. The primary reaction was diluted and used as a template for secondary amplification with a nested cdh5-specific primer and a nested 5′-RACE primer: 5′- CCAATCTCTTTTATGTCTTTGAAGTGCTGC-3′ and 5′-GAGACCTGCTGTCCAAGTTCGT-3′, respectively. The 5′-RACE products were cloned into pCR4-TOPO as above.
The 3′-end of the coding sequence was verified by amplification of a 1,075-bp band using the following primers: 5′-CATTGAAGACGCAAACTGTCCTGTAGATG-3′ and 5′-TCAGTAGGAGCTATCCGAATCAGAGCC-3′. Furthermore, the 3′-end of the cdh5 cDNA was verified using 3′-RACE PCR (Clontech). The same reverse transcribed RNA was used for 3′-RACE as for 5′-RACE PCR. The primary PCR reaction contained a gene-specific primer, a long 3′-RACE primer, and a short 3′-RACE primer: 5′-GATGGCATTCCCTATGACACATTACACATC-3′, 5′-CGACAGAACACGCTCCAGCATTGACCACTTCCTACAACAAAGCTGGGT-3′, and 5′-CGACAGAACACGCTCCAGCATTG-3′, respectively.
After amplification, the primary PCR reactions are diluted 1:50, and 2 μl was used as the template for the secondary PCR with the nested gene-specific primer and 3′-RACE primer: 5′-CTCGTCGTCCAGTGGGTCTAACCTG-3′ and 5′-CGCTCCAGCATTGACCACTTCCTAC-3′, respectively.
Secondary PCR products were cloned into the pCR4-TOPO vector (Invitrogen) as above and sequenced. Subsequent sequence analysis identified a 3′-EST sequence with overlap to the 3′-end of the coding sequence confirmed above (GenBank accession no. CA472457). Pfam domains (Bateman et al., 2002) were determined based on the human sequence using the HMM search tool at the Sanger Center Web site (http://www.sanger.ac.uk/Software/Pfam/search.shtml).
In Situ Hybridization
For antisense, digoxigenin-labeled RNA probe synthesis, the plasmid containing the initial 1,395-bp cdh5 DNA segment was linearized with NotI and used as a template for in vitro transcription using T3 polymerase (Roche Biochemicals). Hybridizations were performed with the nonradioactive, whole-mount zebrafish in situ hybridization kit from R&D Systems, Incorporated (Minneapolis, MN; catalogue no. ISH001).
To knockdown VEGF-A activity, the VEGF-A165 MPO (Gene Tools, LLC) targeting the 5′-UTR of zebrafish VEGF-A165 with sequence: 5′-TAAGAAAGCGAAGCTGCTGGGTATG-3′ was injected as described (Nasevicius and Ekker, 2000).
The open reading frame (ORF) of human VEGF-A165 (accession no. NM_003376) was isolated from a human VEGF-A165 ORF clone (kindly provided by R&D Systems) with the following primers: 5′-GGAATTCATGAACTTTCTGCTGTCTTG-3′ and 5′-GGAATTCTCACCGCCTCGGCTTGTC-3′. An EcoRI site was introduced into the 5′ end of each primer, and the PCR fragment was subcloned into the EcoRI site of the FRM vector (Gibbs and Schmale, 2000). The ORF of zebrafish VEGF-A165 was subcloned into the FRM vector as described (Chen et al., 2004). The FRM EGFP expression construct (Gibbs and Schmale, 2000) was injected as a DNA distribution control. Mixtures of zebrafish or human VEGF-A165 with EGFP DNA were injected into the yolk–blastomere interface at the one-cell stage. Embryos positive for GFP expression were chosen for in situ hybridization analysis.
Fluorescein isothiocyanate (FITC) –dextran was prepared and injected into the common cardinal vein of anesthetized embryos as described (Nasevicius et al., 2000). Fluorescence was visualized with a Zeiss Axioskop 2 microscope fitted with an FITC filter.
Microscopy was performed on a Zeiss Axioskop 2 microscope fitted with differential interference contrast microscopy optics. Digital images were captured with a Nikon Coolpix 995. Negative images were inverted from their originals with Photoshop 7.0.
We thank Dr. Pat Gibbs for the use of the FRM vector. We also thank Patty Brooks, Angie Grant, and Dr. Alexander Kalyuzhny of R&D Systems, Inc., for in situ kit development. E.C. and S.C.E. were funded by the National Institutes of Health.