gridlock, an HLH Gene Required for Assembly of the Aorta in Zebrafish

See allHide authors and affiliations

Science  10 Mar 2000:
Vol. 287, Issue 5459, pp. 1820-1824
DOI: 10.1126/science.287.5459.1820


The first artery and vein of the vertebrate embryo assemble in the trunk by migration and coalescence of angioblasts to form endothelial tubes. The gridlock (grl) mutation in zebrafish selectively perturbs assembly of the artery (the aorta). Here it is shown that grl encodes a basic helix-loop-helix (bHLH) protein belonging to the Hairy/Enhancer of the split family of bHLH proteins. The grl gene is expressed in lateral plate mesoderm before vessel formation, and thereafter in the aorta and not in the vein. These results suggest that the arterial endothelial identity is established even before the onset of blood flow and implicate the grl gene in assignment of vessel-specific cell fate.

Arteries and veins are morphologically and functionally very distinct. For example, arteries deliver oxygenated blood at high pressure from the heart, whereas veins serve as capacitance vessels for blood return. Some of the morphological differences may be imposed after, and depend upon, onset of function. However, a complete vascular loop, composed of the trunk dorsal aorta and posterior cardinal vein, is needed to accommodate the output of the first heartbeat. These simple tubes of endothelium form by local aggregation of angioblasts, a process termed vasculogenesis (1). Neither mutations nor molecular markers have revealed whether there are arterial-venous distinctions between angioblast progenitors. In the mouse, ephrinB2 is selectively expressed on the arteries and EphB3 and EphB4 on the veins, but this occurs after vasculogenesis (2). Furthermore, mutation of ephrinB2 does not affect vasculogenesis, although it does disrupt later vessel formation and remodeling, a process termed angiogenesis (2).

The gridlock mutation (grl m145) was originally isolated in a large-scale chemical mutagenesis screen (3) for developmental mutations of the zebrafish,Danio rerio. Homozygous mutant embryos have no circulation to the posterior trunk and tail because of a localized block to caudal blood flow at the base of the dorsal aorta, the region where the two anterior lateral dorsal aortae merge to form the single midline dorsal aorta. Cranial vessels and other trunk vessels appear to form normally in the mutants.

To clone the grl mutation, we first established its position on the genetic map, using single-strand length polymorphism (SSLP) markers and random amplified polymorphic DNAs (4, 5), and then generated a physical map of the region using yeast artificial chromosomes (YACs) (6), bacterial artificial chromosomes (BACs), and P1 bacteriophage chromosomes (PACs) (7) (Fig. 1). Fine mapping, using recombination frequency among 2400 meioses, permitted us to refine the interval to 120 kb on overlapping BAC14 and PAC14 (Fig. 1) (8). We used two strategies to define genes within this interval. First, we assembled the sequences of BAC14 and PAC14 (9), which we analyzed by the Phred/Phrap/Consed program (9, 10), and thereby established 11 sequence contigs. Combinatory Genescan (11) and Blast Search (12) identified exons, introns, and exon-intron boundaries, with four genes (termed genes A to D). Second, we used the BAC and PAC as probes to screen embryonic cDNA libraries (13). This method identified only gene B. Subsequent fine mapping, with single-nucleotide polymorphisms, showed gene B to be outside of thegrl region. We sequenced the other three genes from wild-type and mutant embryos and examined their expression patterns. Gene C appears to be a zebrafish ortholog of human Pex7(14), and gene D appears to be novel. No consistent mutations were found in the exons and exon-intron boundaries of genes C and D in mutant embryos, and the genes were not expressed in the aorta or surrounding tissues. Gene A, in contrast, was expressed selectively in the aorta and contains a mutation in the mutant embryos (see below), leading us to conclude that it is the grl gene.

Figure 1

The integrated genetic, physical, and radiation hybrid maps of the zebrafish gridlock region. Marker z536 was linked to the gridlock locus on chromosome 20 by bulked segregant analysis and used to initiate a chromosomal walk. Five and 12 recombinants were identified with z536 and z9794, respectively, among 1200 mutants. A YAC contig was constructed spanning about 2 Mb of the grl region. The YAC sizes (from 350 kb to 1.5 Mb) were determined by pulsed-field gel electrophoresis. The BAC/PAC contig was assembled from three BACs and three PACs with the T3 end of YAC133 as a starting point. G, H, I, J, K, and L correspond to YAC, BAC, and PAC ends. The number of recombinants between the markers and grl locus are shown below the Chromosome line. Both PAC14 and BAC14 were shotgun-sequenced, providing fivefold DNA sequence coverage of each. The five exons of grl were identified by sequence analysis. Some of the genetic markers were placed on a radiation hybrid (RH) map to integrate genetic, physical, and RH maps in the mutation region, as shown. The distance in centirays (cR) between these markers on the RH map is indicated below the Radiation Hybrid line. The YAC sizes are not proportional to the scale bar.

Conceptual translation of grl genomic DNA and cDNA sequences predicts a basic helix-loop-helix (bHLH) protein with an apparent human ortholog (Fig. 2A). These grlgenes are highly related to the mouse Hey2/HRT-2gene (15, 16), which belong to the family ofHairy/Enhancer-of-split related (Hesr) bHLH genes (Fig. 2B) (17). The Hesr genes are a subgroup of the Hairy-related bHLH genes, whose function is unknown (17). Like other Hairy genes (18), they are predicted to encode an NH2-terminal bHLH domain, an Orange domain (19), and a protein-protein interaction motif near the COOH terminus (Fig. 2B). This COOH-terminal motif is critical for action of Hairy proteins, but is divergent in Hesr proteins (17). Among the elements that distinguish Hesr proteins from other members of the Hairy-related family are a Pro→Gly substitution in the basic domain and a change in the COOH-terminal WRPW domain to YRPW, the latter embedded within a 13–amino acid Hesr motif (Fig. 2B) (17). Alignment of the predicted human and zebrafish Grl amino acid sequences predicts 78% overall identity, 100% identity in the bHLH domain, and 95% identity in the Orange domain (Fig. 2A). The similarity between Grl and other Hesr proteins is 85% in the bHLH domain, but only 55% in the Orange domain and 21% outside the conserved domains (Fig. 2B). The Orange domain confers specificity among members of the Hairy family (19), and the Grls and Hesrs show characteristic sequence differences in this domain (Fig. 2B). These results suggest that the Grl proteins are a distinct subgroup of the bHLH-YRPW family.

Figure 2

Sequence and domain structure of the zebrafish and human Grl proteins, sequence alignment of Grl with other Hesr proteins, and the position of the point mutation in the zebrafishgrl mutant allele. (A) Amino acid sequence alignment of zebrafish Grl and human GRL. The human GRL sequence was assembled from human expressed sequence tags (AI 727779 and AA116067). bHLH, Orange, and YRPW domains are indicated above the alignment. Black boxes, amino acid identity. Gray boxes, amino acid similarity. The GenBank accession numbers for Grl and hGRL are AF237948 and AF237949, respectively. (B) Sequence comparison of Grl with other Hesr proteins in the bHLH domain, Orange domain, and COOH-terminal motif. Black box, amino acid identity among all Grl and Hesr proteins. Dark blue boxes, amino acids that are distinct for Grl proteins and different from Hesr proteins. The Orange domain differences are shown as black squares below the sequences. Light blue boxes, amino acid similarity among all Grl and Hesr proteins. Black triangles, Pro→Gly substitution in the basic region of bHLH and Trp→Tyr substitution in the protein-protein interaction motif of the COOH-terminus. Sequences were aligned with Pileup of the GCG package (Version 10) and displayed by interface of MacBoxshade 2.15. (C)grlm145 mutation changes T to A, with a predicted effect of changing the stop codon to Gly and extending the protein by 44 amino acids at the COOH-terminus. The mutation was found in the genomic region of all eight grl mutant embryos examined and also found in reverse transcription–PCR of cDNA from a pool of 10 mutant embryos. Abbreviations for the amino acid residues are as follows: A, Ala; C, Cys; D, Asp; E, Glu; F, Phe; G, Gly; H, His; I, Ile; K, Lys; L, Leu; M, Met; N, Asn; P, Pro; Q, Gln; R, Arg; S, Ser; T, Thr; V, Val; W, Trp; and Y, Tyr.

The genomic sequence of grl in mutant embryos contains a T-to-A transversion (Fig. 2C) that would change the stop codon to Gly and extend the protein by 44 amino acids. To confirm that this mutation is responsible for the mutant phenotype, we attempted to rescue the mutant embryos by injecting them with wild-type grl RNA.grlm145 is a fully penetrant recessive mutation, and the most robust phenotype is absence of demonstrable circulation to the trunk at 48 hours post fertilization (hpf), as shown by microangiogram (Fig. 3, A and B). We injected one- to- four-cell-stage embryos from grlm145 /+ in-crosses with in vitro–transcribed grl RNA (20). At 48 hours, each injected embryo was examined and genotyped by using a closely linked SSLP marker. Wild-typegrl RNA (grl wt) rescued 26% of genotypically mutant embryos (of 39 genotypically mutant embryos injected, 10 were phenotypically wild type) (21). We confirmed that four of these phenotypically wild-type embryos were genetically mutant by sequence analysis (Fig. 3C). Mutant grlRNA (grl mut) rescued only 5% (of 37 injected genotypically mutant embryos, 2 were phenotypically wild type) (21). This is compatible with the mutation being hypomorphic rather than null (i.e., reducing rather than abolishing protein function).

Figure 3

Phenotypic complementation of the zebrafish grl mutation. (A) Phenotype of wild-type zebrafish embryos by microangiogram, revealing robust blood flow to the trunk through the dorsal aorta (long arrow), with return through the axial vein (short arrow). (B) Microangiogram of a grl mutant showing absence of trunk circulation, because of the blockage at aortic junction of the paired dorsal aortae (arrows point to position where the vessels would be in the wild type). Fluorescent dextran accumulates over the yolk because of the circulatory blockade. (C) After injection, some genotypically mutant embryos become phenotypically wild type. The microangiogram shows relatively complete restoration of trunk circulation through the dorsal aorta (long arrow) and axial vein (short arrow). The COOH-terminal sequence of the grl gene is shown to the right in four wild-type, four mutant, and four rescued embryos. (These were selected for sequence confirmation, but had already been genotyped by a closely linked marker.) The grlmutant embryos and the rescued grl mutant embryos have a T-to-A point mutation, compared with the wild-type sequence. Cappedgrl RNA was injected at the one- to four-cell stage into wild-type and mutant embryos. Phenotypic analysis was carried out by phase contrast, and in a few cases, as shown, by angiographic analysis (3), for the presence or absence of trunk circulation at 48 hpf. Bar, 100 μm.

In Hairy-related proteins, the COOH-terminus is the site of interaction with the transcriptional corepressor Groucho (22). It is conceivable that the mutation-induced protein extension interferes with the function of this region. Consistent with this hypothesis,grl mRNA lacking the YRPW domain (grl del) was unable to rescue the mutant phenotype when injected into genotypically mutant embryos (0 of 38 embryos rescued) (21).

In gene expression analyses, the grl transcript is detected as bilateral mesodermal stripes as early as the 10-somite stage in zebrafish embryos, before vessel formation (Fig. 4A) (23). Thereafter, some ofgrl-expressing cells appear to converge toward the midline to form the primordium of the dorsal aorta at the 24-somite stage (Fig. 4D). By the 30-somite stage, when blood flow begins, grl is expressed strongly throughout the dorsal aorta, including the trunk region (Fig. 4, B, C, F, and G) and anterior bifurcation (Fig. 4, E and J). No grl-expressing cells are observed in the region of the axial vein, ventral to the aorta (Fig. 4, D, F, and G). In contrast, fli (24), an early endothelial marker, is expressed both in arteries and veins (Fig. 4, H and I). In addition, grl is transiently expressed in the heart, aortic arches, and lateral somites, a pattern similar to that described for the mouse Hey and HRT genes (15,16).

Figure 4

Expression of grl mRNA in zebrafish. (A) grl is expressed as bilateral stripes at lateral plate mesoderm (arrows) at the 10-somite stage (dorsal view). (B and C) Lateral views of the anterior trunk, posterior trunk, and tail, showing grl transcripts in the dorsal aorta (arrows). (D) Sagittal section of the posterior trunk showing grl expression in the primordium of dorsal aorta (arrow), but not in the axial vein (arrow) at the 24-somite stage. (E) Transverse section of the anterior trunk, showing grl expression in the paired dorsal aortae (arrows) ventral to the notochord, but not in the bilateral cardinal veins [see arrowheads in (I)]. (F and G) Transverse section of posterior trunk, showing grl expression in the dorsal aorta (arrowhead) beneath the notochord, and extending into what is likely a sprout (intersomatic artery) (short arrow), but not in the axial vein (arrow). (H) fli expression highlights both the axial vein (arrow) and the dorsal aorta (arrowhead) in the posterior trunk. (I) Transverse section through the anterior trunk at the level of the first somite. fli is expressed in the bilateral common cardinal veins (arrowheads) and the paired dorsal aortae (arrows). (J) Longitudinal section of the anterior trunk showing a dorsal view of grl expression in the aortic bifurcation. Dorsal is up in (B) to (I), anterior is to the left in (B), (C), and (J). The in situ hybridization was carried out at the 30-somite (24 hours) stage (B, C, E to I, J). N, notochord; da, dorsal aorta; v, vein. Bars, 100 μm (A, B, C, D, and J), 50 μm (E and I), and 50 μm (F to H).

It is not known how or when angioblasts assume a particular vascular fate in the early embryo. Nor is it known which genes drive vessels to assemble into tubes of a particular branching form. The finding of grl as a member of a bHLH transcriptional regulatory gene family, and as a gene needed for proper formation of the aorta, speaks to issues of both angioblast cell fate and vascular morphogenesis.

Grl is a member of the Hairy-related family of bHLH proteins, which are important for cell fate determination in other cell types (25). In the Drosophila nervous system, for example, members of the Hairy family act downstream of Notch as transcriptional repressors and help to “single out” neuronal precursor cells within “equivalence groups” (25). We speculate that Grl plays a similar role specifically for aortic angioblasts and that other bHLH proteins may be required for specification of vein angioblasts. Different bHLH proteins, including stem-cell leukemia/T-cell acute leukemia 1, Tfeb, hypoxia-inducible HIF-1α and HIF-2α, and the dominant inhibitory HLH factors Id1 and Id3, are involved in a variety of endothelial functions (26), including angiogenesis, but no bHLH proteins have been clearly linked to vasculogenesis. Vascular endothelial growth factor (Vegf) and its receptor Flk are essential for early vasculogenesis (1, 27), and it will be important to determine whether Grl functions in the Vegf/Flk pathway.

Although grl is expressed throughout the entire vessel, the most anterior region of the aorta, the bifurcation, is particularly affected in the mutant. This may be due to the fact that the mutation reduces but does not eliminate Grl function. Perhaps other genes provide redundant function in the other regions of the aorta. Indeed, three related Hey and HRT genes are expressed in the mammalian aorta (15, 16). The aortic bifurcation is particularly susceptible to congenital dysmorphogenesis in humans, suggesting that it may be more sensitive than the rest of the aorta to perturbation. Some of these clinical disorders, such as coarctation, show a high sibling recurrence (28), and it will be of interest to examine grl as a candidate gene for these diseases.

Zebrafish mutations are proving to be especially informative about vertebrate-specific processes, such as organogenesis. The hope of the zebrafish genetic screens was that they would not only inform about the logic of development, and the role of known genes, but also lead to discovery of novel genes. As exemplified here by grl, and byoep (29) and weh (30), the positional cloning infrastructure for zebrafish now is sufficiently robust to make this a reality.

  • * These authors contributed equally to this work.

  • Present address: Aventis Pharma, 26 Landsdowne Street, Cambridge, MA 02139, USA.

  • Present address: Jake Gittlen Cancer Research Institute, Pennsylvania State University College of Medicine, 500 University Drive, Hershey, PA 17033, USA.

  • § Present address: Unit on Vertebrate Organogenesis, Laboratory of Molecular Genetics, National Institute of Child Health and Human Development, Building 6B, 6 Center Drive, Bethesda, MD 20892, USA.

  • || To whom correspondence should be addressed. E-mail: fishman{at}


View Abstract

Navigate This Article