This article is a US Government work and, as such, is in the public domain in the United States of America.
We identified four mutants in two distinct loci exhibiting similar trunk vascular patterning defects in an F3 genetic screen for zebrafish vascular mutants. Initial vasculogenesis is not affected in these mutants, with proper specification and differentiation of endothelial cells. However, all four display severe defects in the growth and patterning of angiogenic vessels in the trunk, with ectopic branching and disoriented migration of intersegmental vessels. The four mutants are allelic to previously characterized mutants at the fused-somites (fss) and beamter (bea) loci, and they exhibit comparable defects in trunk somite boundary formation. The fss locus has been shown to correspond to tbx24; we show here that bea mutants are defective in the zebrafish dlC gene. Somitic expression of known vascular guidance factors efnb2a, sema3a1, and sema3a2 is aberrantly patterned in fss and bea mutants, suggesting that the vascular phenotype is due to loss of proper guidance cues provided by these factors. Developmental Dynamics 235:1753–1760, 2006. Published 2006 Wiley-Liss, Inc.
The trunk vasculature provides a useful model for studying the regulation of blood vessel assembly and patterning during early development. Trunk vessels have a simple anatomy and are easily imaged, and secondary trunk vessels form in metameric repeating units along the trunk together with the somites, facilitating controlled experimental analysis. The first vessels to assemble in the trunk are the axial vessels (dorsal aorta and cardinal vein), which form by vasculogenesis, or coalescence of angioblasts (endothelial progenitor cells) derived from the lateral mesoderm (Risau and Flamme, 1995). After the establishment of these vessels, additional trunk vessels form at later stages by angiogenesis, or sprouting and growth of new vessels from preexisting vessels. These later trunk vessels include intersegmental vessels, parachordal vessels, and vertebral arteries, all of which are common across vertebrate phyla. Using multiphoton time-lapse imaging of TG(fli1:egfp)y1 zebrafish (Lawson and Weinstein, 2002), we recently described a two-step mechanism for formation of the initial angiogenic vessels in the trunk (Isogai et al., 2003). Primary intersegmental vessel sprouts emerge bilaterally from the dorsal aorta close to the intersegmental boundaries between the somites. They continue to grow dorsally, continuing to track very closely along the intersegmental boundaries as they grow. Studies from our laboratory and others showed that semaphorins expressed in the central portion of the somites act as repulsive guidance cues to help restrict the growth of the primary intersegmental vessels to intersegmental borders, both in zebrafish and in mice (Gitler et al., 2004; Torres-Vazquez et al., 2004; Gu et al., 2005).
To further probe the requirements for trunk vessel patterning, we undertook an F3 N-ethyl-N-nitroso-urea (ENU) mutagenesis screen in the TG(fli1:egfp)y1 transgenic background to search for new mutants that disrupted proper patterning of trunk angiogenic blood vessels. We uncovered four mutants with similar intersegmental vessel mispatterning phenotypes. Despite their aberrant vascular patterning, endothelial specification and differentiation is normal in these mutants. Further characterization revealed that these mutants were allelic to fused somites (fss) and beamter (bea) mutants, caused by defects in the zebrafish tbx24 and deltaC (dlC) genes, respectively. Our findings highlight the link between somite patterning and guidance and patterning of the trunk vasculature.
Isolation and Characterization of Mutants With Trunk Angiogenic Vessel Defects
During an F3 ENU mutagenesis screen in the TG(fli1:egfp)y1 background for recessive mutants with defects in vascular patterning, we identified four mutants with very similar vascular phenotypes falling into two complementation groups, y56/y70 and y55/y66. No defects are noted in formation of the vasculogenic primary axial vessels (dorsal aorta and cardinal vein) or early head vessels (primordial hindbrain channel, basilar artery, and so on) in either y56/y70 or y55/y66 mutants (see Isogai et al., 2001, for information on zebrafish vascular anatomy). However, all four mutants display defects in later-developing trunk angiogenic vessels (Fig. 1). Primary intersegmental sprouts normally emerge from the dorsal aorta close to the intersegmental boundaries at approximately 20 hpf, and extend along the intersegmental boundaries without branching except at their dorsal-most extent where they branch rostrally and caudally, interconnecting to form the dorsal longitudinal anastomotic vessels (Isogai et al., 2003). Initial emergence of primary intersegmental sprouts is delayed in y56/y70 or y55/y66 mutants. At 22 hours postfertilization (hpf), only 15% of y56 and 13% of y66 mutant embryos have primary intersegmental vessel (SeV) sprouts extending to the horizontal myoseptum, compared with over 50% of their wild-type counterparts (Fig. 1A, top). Mutant SeV sprouts do eventually catch up with their wild-type counterparts, however. By 30 hpf comparable numbers of SeV sprouts have extended dorsal to the horizontal myoseptum in y56 and y66 mutants and their wild-type siblings (Fig. 1A, bottom). The delay in intersegmental vessel sprouting was not a result of general developmental delay, as the onset of pigmentation and gross morphological staging landmarks other than somite formation (see below) appeared at a normal rate (data not shown).
In addition to delays in primary intersegmental sprout emergence, y56/y70 mutants and y55/y66 mutants also exhibit defects in intersegmental vessel patterning (Fig. 1B–I). In both y56/y70 (Fig. 1C) and y55/y66 (Fig. 1D) mutants, the regularly spaced arrangement of intersegmental vessels apparent in wild-type animals (Fig. 1B,E) is not observed, and intersegmental vessel sprouts are not mirrored across the left–right axis (Fig. 1B,C,I). Ectopic branching of intersegmental vessels is also noted in mutants. Intersegmental vessels with either two ventral roots (Fig. 1F) or premature dorsal branches (Fig. 1G,H) are noted with approximately equal frequency. Although the pattern of intersegmental vessels is disrupted in mutants, their number is only slightly increased, with an average of 1.5 and 1.8 additional sprouts over the first 16 somites in y56 and y66 classes of embryos, respectively.
The only significant difference between the vascular phenotypes of y56/y70 vs. y55/y66 mutants is a lack of intersegmental vessel defects in the first four somites of y56 and y70 mutants. In y55 and y66 mutants, the defects noted above are observed throughout the entire rostral–caudal extent of the trunk. In y56 and y70 mutants, however, the growth and patterning of intersegmental vessels appears normal in the first four trunk segments, with appropriate growth along intersegmental boundaries, proper left–right mirroring of vessels, and lack of ectopic branching.
Endothelial Differentiation Is Not Affected in y56/y70 and y55/y66 Mutants
To assess whether vascular differentiation was affected in y56/y70 or y55/y66 mutants, we examined expression of a variety of molecular markers of endothelial differentiation by whole-mount in situ hybridization. Markers of hemangiogenic mesoderm such as fli1 and early pan-endothelial markers such as flk1 appear normal in mutants (data not shown), suggesting that angioblast specification and formation of endothelium occurs normally. The endothelial-specific semaphorin receptor plxnD1, which we have shown previously is required for proper guidance and patterning of trunk intersegmental vessels (Torres-Vazquez et al., 2004), is also normally expressed in both y56-class and y66-class mutant embryos (Fig. 2A–C). Specification of arterial and venous endothelial cell fates also occurs normally in y56 and y66 mutants, as shown by in situ hybridization using the arterial marker efnb2 and venous marker flt4 (Fig. 2D–I). Taken together with the observation that the vasculature of both mutants is functional (adult homozygous mutant animals are viable and fertile), these data suggest that differentiation of the vasculature is unperturbed in both classes of mutants.
Molecular Characterization of y56/y70 and y55/y66 Mutants
In addition to vascular patterning defects, we noted defects in somite boundary formation in both classes of mutant embryos (Fig. 3A,B and data not shown). Only the first four somites were formed properly in y56 and y70 mutants; none of the somites were correctly patterned in y55 and y66 mutants. The somite phenotypes we observed in y56/y70 and y55/y66 mutants bore a striking similarity to those previously described for the somite patterning mutants beamter (bea) and fused somites (fss) mutants, respectively. Complementation testing was performed between y56 and y66 and known alleles of fss and bea. These tests revealed that y56 fails to complement bea, whereas y70 fails to complement fss (see the Experimental Procedures section). The unique phenotype of loss of somite patterning in all but the first four somites and failure to complement an existing bea allele indicates that y56 and y70 are bea alleles, which we now designate beay56 and beay70. Mutations at the fss locus have been shown previously to correspond defects in the T-box transcription factor tbx24 (Nikaido et al., 2002), so we performed bulk segregant analysis and fine genetic mapping to determine their chromosomal locations. The y55 and y66 mutants both map within 2 cM of z21911, a marker tightly linked to fss/tbx24 on LG12 (Fig. 3C), confirming that these are indeed alleles of fss, which we now designate fssy55 and fssy66.
Because the defective gene in bea mutants (van Eeden et al., 1996) had not been reported, we proceeded with molecular cloning of the y56 mutation. We mapped both beay56 and beay70 to within 0.5 cM of deltaC (dlC) on LG 15 (Fig. 3D). Previous studies implicated dlC function in somite patterning (Holley et al., 2002), so we cloned and sequenced the dlC gene from y56 mutants and their phenotypically wild-type siblings. The cDNA from beay56 mutant embryos contained a deletion of the guanine at nucleotide 54 of the wild-type coding sequence, resulting in a missense mutation with a subsequent stop codon at amino acid number 26 and early truncation of the predicted dlC peptide. To further confirm that a mutation in dlC was responsible for the bea phenotype, we injected both ATG and 5′ utranslated region (UTR) morpholinos at specified doses (see the Experimental Procedures section) and examined somite and blood vessel phenotypes (Fig. 3E–J). Loss of somite borders has been demonstrated previously when dlC levels are lowered using morpholino (MO) injection (Holley et al., 2002). We saw a reproducible phenocopy of bea mutants upon dlC MO injection, with loss of distinct intersegmental boundaries in all but the first four segments (Fig. 3H,I,K,L) and similar trunk intersegmental vessel patterning defects (Fig. 3J). These phenotypes were observed with as little as 2.5 ng of either the MO directed against the 5′ UTR or the ATG. We observed a dose-dependent effect, with a high percentage of embryos exhibiting a loss of posterior somite borders at higher doses of MO injected (Table 1). Together, these results show that the bea mutant phenotype is indeed due to defects in dlC. As previously reported, we find that delC mRNA is expressed in the somites and dorsal aorta during somitogenesis (Fig. 3M), with no difference in expression apparent between wild-type and bea mutants.
Table 1. ATG and UTR Morpholinos Against dlC Both Result in Dose-Dependent Phenocopy of the beamter Mutationa
Amount of MO (ng)
MO, morpholino; UTR, utranslated region.
Although beay56 and beay70 mutants are deficient in markers of somite maturation such as myoD (data not shown), they do have some weak somite borders. However, the somites no longer retain their characteristic chevron shape and the remaining borders are irregularly spaced along the trunk. To determine whether residual somite borders in mutant embryos function as sites of intersegmental vessel sprouting similar to wild-type embryos, mutant embryos were fixed and stained with rhodamine phalloidin, a marker of actin filaments to visualize somite boundaries. Green fluorescent protein (GFP) expression in the blood vessels withstands both fixation and immunohistochemical staining protocols, enabling simultaneous visualization of somites and ISVs using confocal microscopy. We quantified the distance between the location of sprout initiation and somite boundaries in wild-type or existing residual somite boundaries in beay56 and beay70 embryos. In both wild-type and mutant animals intersegmental vessel sprouts were initiated within one cell diameter of a visualized somite boundary; there was no significant difference between any of the genotypes (data not shown). These results support the idea that intersegmental vessel defects result from lack of proper somite boundary formation and that vessels do sprout appropriately where residual boundaries are present.
Vascular Guidance Molecules Are Misexpressed in Somite Boundary Mutants
In addition to giving rise to tendon, muscle, and cartilage (Brent and Tabin, 2002), somites serve as signaling centers for adjacent tissues, including developing blood vessels. Vascular endothelial growth factor (VEGF), a critical endothelial cell mitogen, is expressed in ventral–medial somitic tissue (Ferrara et al., 1991; Neufeld et al., 1994; Liang et al., 1998). The somites also express sema3a1 and sema3a2, which serve as vascular guidance molecules. Endothelial cells expressing the semaphorin receptor plexinD1 are repelled by the somite-derived signal, restricting the migration of intersegmental vessel sprouts to semaphorin-free intersegmental boundaries. Loss of either semaphorin or plexinD1 results in a highly disrupted pattern of intersegmental vessels (Gitler et al., 2004; Torres-Vazquez et al., 2004; Gu et al., 2005). The morphologic defects in somitogenesis in fused-somite–like mutants have been well characterized (Holley and Nusslein-Volhard, 2000; Brent and Tabin, 2002). Multiple studies have noted defects in proper timing and expression patterning of multiple genetic regulators of the somitogenesis clock in these mutants (Holley et al., 2000, 2002). We suspected that the disruptions in timing and differentiation of the somites might play a role in the disruption of vascular patterning through defects in expression of vascular guidance molecules. To examine this possibility, we compared the expression of known somite-expressed vascular guidance molecules efnb2a, sema3a1, and sema3a2 in wild-type and mutant embryos.
The Eph/Ephrin signaling pathway plays a well-established role in somite boundary formation (Durbin et al., 1998). Ephrin-B2 is also involved in arterial–venous differentiation, and its expression in blood vessels becomes restricted specifically to arteries in all vertebrates. Although the arterial vascular expression of zebrafish efnb2a is not affected in either beay56 or fssy66 mutants (Fig. 2H,I), somitic efnb2a expression is highly irregular during somitogenesis. In beay56 mutants, somitic efnb2a expression is mottled and no longer appears in distinct blocks as it does in wild-type animals (Fig. 4A,B). Similar defects are noted in the somitic expression of the sema3a1 and sema3a2. Both semaphorins are expressed in the central region of the somites, where they provide guidance cues restricting the growth of plexinD1-expressing vascular endothelial cells along the intersegmental boundaries. In wild-type animals, sema3a1 and sema3a2 are expressed during early somitogenesis and their expression slowly decreases as the anterior somites undergo differentiation. Expression of plxnD1 is normal in the axial vessels of 24 hpf beay56 mutants (Fig. 2), but sema3a1 (Fig. 4C,D) and sema3a2 (Fig. 4E,F) are expressed in a stippled and irregular pattern rather than in more discrete blocks or stripes and the expression of both genes wanes precociously (data not shown). Similar defects in sema3a1 and sema3a2 expression are noted in the trunks of fssy66 and beay70 mutants (data not shown). Together, these data, coupled with the fact that other markers of somite maturation and regionalization are disordered and misexpressed (Fig. 4G,H; data not shown), suggest that proper somitogenesis and somite boundary stabilization are required for accurate vascular guidance to proceed.
We have identified four new ENU-induced mutants, y55, y56, y66, and y70, each of which displays a similar mispatterning of trunk angiogenic blood vessels. The mutations fall into two complementation groups, y55/y66 and y56/y70, and within each complementation group, the two mutants display identical phenotypes. The only significant phenotypic difference between the two complementation groups is a lack of defects in the first four somites of y55/y66 mutants. In addition to their vascular defects, the y55/y66 and y56/y70 groups of mutants exhibit somite mispatterning phenotypes identical to those previously described for beamter (bea) and fused somites (fss) mutants, respectively. Complementation testing confirms that the four new mutants are allelic to these previously identified mutants. We have, therefore, designated the new mutants beay56, beay70, fssy55, and fssy66. Genetic fine mapping confirmed that fssy55 and fssy66 are both tightly linked to tbx24. The molecular identity of the bea locus had not been reported, so we mapped and cloned beay56, showing that the defective gene in these mutants is deltaC (dlC), a gene implicated in the establishment of somite pattern (Holley et al., 2002). The identification of dlC as the defective gene in bea mutants has also been very recently reported by another group (Julich et al., 2005). Julich et al. have shown that bea mutants have neurogenic phenotypes and function in the Notch signaling pathway. They also show that bea/dlC has an important role in regulating the somitogenesis clock and somite patterning.
The involvement of the Notch pathway in arterial–venous differentiation has been well-described (reviewed in Torres-Vazquez et al., 2003). Studies in zebrafish and mice have shown that notch signaling is part of a molecular pathway required for arterial differentiation (Lawson et al., 2001, 2002). Delta family members are Notch ligands and recent work has shown that the Dll4 gene is required for vascular development and arterial differentiation in the mouse (Duarte et al., 2004; Gale et al., 2004; Krebs et al., 2004). Because dlC is expressed in both somitic tissue and in blood vessels (Smithers et al., 2000), the vascular phenotype in bea mutant embryos could reflect loss of dlC function in vascular endothelium, or in somites, or in both tissues. Several lines of evidence argue that the vascular phenotype is an indirect result of the somitogenesis defects. First, fss and bea mutants have strikingly similar phenotypes in both the affected trunk somites and in the adjacent vasculature, despite that the affected gene in fss mutants, tbx24, is not expressed in the vasculature (Nikaido et al., 2002). Second, bea and fss homozygous mutant animals are viable and fertile with a functional vasculature, unlike zebrafish and mouse mutants with known defects in vascular notch signaling. Third, markers of endothelial specification and arterial–venous differentiation are unaffected in either fss/tbx24 or bea/dlC mutants, unlike previously described notch gain- and loss-of-function in the zebrafish vasculature (Lawson et al., 2001, 2002). Fourth, loss of dlC in bea mutant embryos does not affect the first four somites. The intersegmental vessels adjacent to these first four somites have no defects in launching, branching, or timing of sprout initiation, despite that dlC is absent from these and all other vessels. Taken together, these observations strongly support the conclusion that somite patterning disruptions, and not endothelial cell-autonomous defects, are the fundamental basis for the vascular patterning defects in both fss and bea mutants.
The reason for a lack of Notch signaling or arterial differentiation defects in the vasculature of bea/dlC mutants remains unclear. The beay56 mutation is predicted to cause an early truncation of the dlC protein, so the phenotype we observe almost certainly represents loss-of-function for this gene. However, other Notch ligands are known to be present in the zebrafish vasculature, including dll4, which as noted above has been shown to be required for arterial differentiation in the mouse vasculature. A mutation in zebrafish dll4 was isolated recently as a suppressor of the longfin mutant (Steve Johnson, U. Washington, personal communication). The mutation causes defects in formation and/or differentiation or fin arteries in adult zebrafish but, like bea mutants, has little or no effect on arterial differentiation of at least the major vessels in embryonic and larval zebrafish. These findings suggest that Notch ligand function may be either substantially redundant within the vascular endothelium in the zebrafish, or alternatively the most critical ligand(s) for Notch signaling in blood vessels has yet to be identified in this species. In the mouse, the more dramatic loss-of-function phenotype for Dll4 is likely due to the fact that it is the primary Notch ligand whose expression in vessels makes it capable of responding to both Notch1 and Notch4 (Krebs et al., 2000; Shutter et al., 2000).
Although our studies do not provide evidence for an essential role for dlC in Notch signaling and/or arterial differentiation within zebrafish vascular endothelium, they do support the idea that the establishment of somitic boundaries is critical for the proper spatial expression of trunk vascular guidance molecules. Published work from our laboratory (Torres-Vazquez et al., 2004) and other laboratories (Gitler et al., 2004; Gu et al., 2005) has shown that semaphorin signals from the somites provide a repulsive guidance cue for plxnD1-expressing intersegmental vessels. Indeed, expression of sema3a1 and sema3a2 is mispatterned in fss and bea mutant somites as patches instead of distinct stripes or blocks running down the middle of each somite. Somites also express other guidance cues implicated in intersegmental vessel patterning, such as the ephrin-B ligands. In Xenopus, misexpression of ephrin-B ligands or disruption of their signaling by expression of dominant-negative ephB4 receptors results in abnormal intersegmental vessel growth into adjacent somitic tissue (Helbling et al., 2000). Like sema3a1 and sema3a2, somitic expression of efnb2a is mispatterned in fss and bea mutants, with expression in small patches instead of stripes, although it does continue to be expressed appropriately in an arterial-specific manner in the vasculature.
In conclusion, our results show that the fused-somites–like class of somite morphogenesis mutants in zebrafish display defects in trunk angiogenic vessel patterning. Concordance between improper somite border stabilization and misguided intersegmental vessel growth is observed; bea mutants only exhibit vascular defects after the fourth somite, whereas all intersegmental vessels are affected in fss mutants. Somite patterning defects are accompanied by mispatterned somitic expression of vascular guidance cues sema3a1, sema3a2, and efnb2a. Our results suggest that the failure to properly express repulsive vascular guidance cues in the somites of bea and fss mutants is responsible for the lack of proper patterning of angiogenic trunk intersegmental vessels.
Fish and Handling
Zebrafish were bred and maintained as described (Westerfield, 1993). The TG(fli1:egfp)y1 transgenic line has been described previously (Lawson and Weinstein, 2002). Complementation testing was performed to beab633 and fssae114. Three separate crosses were performed for each test, with comparable results for each cross (either complementation or failure to complement).
To facilitate genetic mapping, the TG(fli1:egfp)y1; beay56/y70, and TG(fli1:egfp)y1; fssy55/y66 founder pairs were crossed to the wild-type TL line. Identified mutant carriers derived from this map cross were incrossed, and embryos scored based on the presence or absence of somite boundaries and vascular defects. Bulk segregant analysis was performed using a 192-marker panel of available CA markers. The list of markers in this panel is available upon request. Additional SSLP markers were found using the data from the ZFIN Web site. Genomic DNA isolation and polymerase chain reaction (PCR) protocols were described elsewhere (Roman et al., 2002). Oligonucleotide sequences for the markers noted in Figure 3C,D are available at http://zebrafish.mgh.harvard.edu/mapping/ssr_map_index.html. Oligonucleotide primers for PCR to identify additional polymorphisms were designed on the basis of available zebrafish genomic sequence obtained through Blastn searches of trace data from the Sanger Institute (http://trace.ensembl.org/perl/ssahaview?server=danio_rerio) using zebrafish dlC coding sequence. Resultant PCR fragments were sequenced directly. Polymorphism analyses and sequence comparisons were performed using SeqMan alignment software (DNASTAR, Inc.). Primers for polymorphic dinucleotide repeats flanking the dlC gene were designed from Zv4_scaffold1303 (1303aFor-caaagcacccactgactg and 1303aRev-cacactggcgtttgaccta, and 1303bFor-gcacggcacataggttg and 1303bRev-ggcttgc- cagacattcag).
Specified amounts of morpholinos (Gene-Tools, http://www.gene-tools.com) were injected at the one-cell stage into the single cell (dlCatg, 5′-gcacgttaataaaacacgagccatc-3′; dlCutr, 5′-ttgccttcttgtctgctactgaacg-3′) in 2× injection buffer (0.1% phenol red, 0.4 M KCl).
Imaging of blood vessels in all embryos was performed using a multiphoton laser or confocal laser microscope as described previously (Lawson et al., 2002). Transmitted light images were obtained with a Leica MZ12 or Zeiss Axiophot2 microscope equipped with a Pro-gRes mF digital camera. For measurements of vessel launching with respect to somitic boundaries, embryos were fixed in 3.8% paraformaldehyde in phosphate-buffered saline (PBS), rinsed in PBS, and stained for 1 hr with AlexaFluor 568–phalloidin (Molecular Probes); images were taken and measurements were made using MetaMorph software. All quantitation of ISV guidance defects was performed using confocal microscopy of the first 16 or the 5th through 16th ISVs in fss or bea embryos, respectively. A minimum of 30 mutant and 45 wild-type embryos were scored for each set of analysis.
Whole-Mount In Situ Hybridization
Antisense mRNA probes for efnB2a, fli1, dlC, plxnD1, sema3a1, and sema3a2 were prepared as described previously (Smithers et al., 2000; Lawson et al., 2001; Torres-Vazquez et al., 2004). The her1 antisense probe was made by digesting her1-pBluescriptKS with SalI and using T7 polymerase (her1 was a gift of I. Dawid and M. Tsang). Whole-mount in situ hybridization was performed as described elsewhere (Hauptmann and Gerster, 1994).
The authors thank Igor Dawid for critical reading of this manuscript. We also thank Steve DeVoto and Sharon Amacher for providing mutant fish lines. K.M.S. was supported by an American Cancer Society Post-Doctoral Fellowship. This research was supported (in part) by the Intramural Research Program of the National Institute for Child Health and Human Development (NICHD).