Modeling human hematopoietic and cardiovascular diseases in zebrafish


  • Trista E. North,

    1. Division of Hematology/Oncology, Department of Medicine, Children's Hospital of Boston, Enders Research Building, Boston, Massachusetts
    Search for more papers by this author
  • Leonard I. Zon

    Corresponding author
    1. Division of Hematology/Oncology, Department of Medicine, Children's Hospital of Boston, Enders Research Building, Boston, Massachusetts
    2. Howard Hughes Medical Institute, Children's Hospital of Boston, Enders Research Building, Boston, Massachusetts
    • Howard Hughes Medical Institute, Children's Hospital of Boston, 320 Longwood Avenue, Enders Research Building, Room 761, Boston, MA 02115
    Search for more papers by this author


Zebrafish have emerged as a useful vertebrate model system in which unbiased large-scale screens have revealed hundreds of mutations affecting vertebrate development. Many zebrafish mutants closely resemble known human disorders, thus providing intriguing prospects for uncovering the genetic basis of human diseases and for the development of pharmacologic agents that inhibit or correct the progression of developmental disorders. The rapid pace of advances in genomic sequencing and map construction, in addition to morpholino targeting and transgenic techniques, have facilitated the identification and analysis of genes associated with zebrafish mutants, thus promoting the development of zebrafish as a model for human disorders. This review aims to illustrate how the zebrafish has been used to identify unknown genes, to assign function to known genes, and to delineate genetic pathways, all contributing valuable leads toward understanding human pathophysiology. Developmental Dynamics 2003. © 2003 Wiley-Liss, Inc.


The zebrafish (Danio rerio) is a powerful model organism for the investigation of vertebrate development and the perturbation of normal developmental programs in disease. Several characteristics have made the zebrafish useful for genetic studies of development and differentiation. The relatively short generation time (3 months), small size at maturation (3 to 4 cm long as an adult), and high fecundity (200 eggs per clutch) of the zebrafish facilitate the execution of large-scale genetic screens, while limiting prohibitive cost and space requirements. Additionally, external fertilization coupled with optical clarity throughout embryogenesis allow for the direct monitoring of developmental processes, and their manipulation using mechanical, chemical, or genetic techniques. Zebrafish show genetic and anatomic conservation with both mice and humans. Large segments of zebrafish chromosomes are syntenic with those of the human and mouse genome, and many genes have been demonstrated to have a high degree of sequence homology (Barbazuk et al., 2000). Furthermore, zebrafish undergo complex processes of vertebrate development such as the production of definitive hematopoietic cells and the formation of a multichambered heart (Stainier et al., 1993; Zon, 1995; Driever and Fishman, 1996; Galloway and Zon, 2003).


Many genetic techniques originally developed in other model systems such as Drosophila, Arabadopsis, Caenorhabditis elegans, and Xenopus, have been adapted for use with zebrafish. For example, mutant analysis has long been a staple for understanding the function of genes regulating embryogenesis and adult organ function. To date, zebrafish have been successfully used in several forward genetic screens. Such “phenotype-first” analyses are advantageous for the unbiased identification of new genes, as well as for the assignment of previously uncharacterized gene functions to known genes. The use of the chemical mutagen N-ethyl-N-nitrosourea (ENU) in these screens has yielded hundreds of mutations that disrupt a wide array of different early developmental processes in the zebrafish (Driever et al., 1996; Haffter et al., 1996). Many mutants isolated in the ENU screens resembled human disease states, and thus provide a useful tool for investigating the corresponding pathophysiology. Through positional and candidate cloning strategies, the genes affected in many zebrafish mutants have been identified, increasing our understanding of both normal and abnormal developmental processes in the vertebrate embryo. Continuous advances in zebrafish genomics and molecular techniques that have facilitated cloning strategies were reviewed recently (Thomas and Touchman, 2002; Rawls et al., 2003). The sequencing of the entire zebrafish genome is expected to be completed in 2004 and will help to eliminate or significantly decrease many of the time-intensive steps currently involved in cloning ENU-induced mutations.

The use of insertional mutagenesis has also been exploited in zebrafish screens (Gaiano et al., 1996; Amsterdam et al., 1999; Golling et al., 2002). The retroviral vectors used to disrupt gene sequences allow identification of the targeted genes by inverse-polymerase chain reaction (PCR) -based sequencing, thus reducing cloning time and effort. Recently, some of the genes isolated by this approach have been described (Chen et al., 2001; Sun and Hopkins, 2001; Golling et al., 2002). Several zebrafish mutants isolated in the insertion screen resemble those with ENU phenotypes and could facilitate their identification (Golling et al., 2002). Numerous genes are orthologs of those known to be critical in Drosophila, mouse, or human development; for example, caudal was identified as the gene mutated in a zebrafish embryo with a short trunk/tail (Golling et al., 2002). Other genes identified by this method are orthologous to human genes with unknown function.

Clinical or developmental processes can be investigated in detail in zebrafish through careful and creative design of zebrafish genetic screens. Many laboratories have begun to conduct “targeted” screens solely focused on a particular developmental process or specific organ (Alexander et al., 1998; Brokerhoff et al., 1998; Trede et al., 2001; Long et al., 2003). Such screens have used a variety of mutagenesis and identification strategies, including behavioral assays and gene expression analyses. (reviewed in Patton and Zon, 2001). For example, many zebrafish mutants with eye defects were uncovered through a behavioral screen for optokinetic and optomotor response (Brokerhoff et al., 1995). These focused screens will serve to saturate genetic pathways, leading to a more complete understanding of particular aspects of vertebrate development.


Zebrafish are amenable to a variety of techniques used in other model systems to evaluate genetic interactions. The placement of genes in specific pathways can be determined by traditional epistasis experiments, in which mutations are bred into the same individual and the nature of the dominant phenotype is assessed. Ectopic gene-expression, accomplished by injection of mRNA into zebrafish embryos, can also be used to delineate genetic pathways or to investigate gene function. For example, injection of scl RNA was able to partially rescue the hematopoietic defect found in cloche (clo) embryos, indicating that scl functions downstream of clo (Liao et al., 1998).

The creation of transgenic zebrafish has been used to monitor protein expression patterns during development (Stuart et al., 1988; Udvadia and Linney, 2003). Linearized plasmid DNA constructs are microinjected into zebrafish embryos at the one-cell stage and are expressed during development. A small number of embryos will have stable integration of the injected constructs and an even smaller fraction will incorporate the DNA into the germ-line and be transmitted to subsequent generations. For example, green fluorescent protein (GFP) expression under the control of the erythroid-specific promoter gata1, can be used to visualize erythroid cells in the circulation of both embryos and adults, whereas zebrafish that express fli1-GFP can be used to visualize the developing vascular system (Long et al., 1997; Lawson and Weinstein, 2002). Transgenes have also been used to examine the effect of overexpression of genes on development. Inappropriate expression of the oncogene cmyc causes leukemia (Langenau et al., 2003) and will be discussed in more detail later in this review. Finally, stable transgenic lines can be bred to one another to evaluate gene interactions, to follow the development of multiple cell types, for use in visually based screens, or to purify specific cell populations by flow cytometry.

Methodologies have been optimized that allow gene “knock-down” in the zebrafish through the use of morpholino (MO) strategies. MOs are antisense nucleic acid analogs in which a morpholine moiety replaces the pentose ring (Summerton and Weller, 1997; Heasman, 2002). MOs have been useful for testing candidate genes for mutant phenotypes, as well as for the assignment of new function to known genes (Nasevicius and Ekker, 2000). MOs are targeted to bind to a specific RNA at either the translation initiation site or a splice junction. Binding will selectively inhibit translation of the targeted gene or alter protein stability causing degradation; for example, injection of a chordin antisense MO phenocopies the chordin loss-of-function mutant phenotype (Nasevicius and Ekker, 2000). Additionally, human disease states can be phenocopied by MO injection; a MO to the uroporphyrinogen decarboxylase gene results in the production of embryos with porphyria, whereas injection of a dystrophin-MO results in muscle degeneration akin to that seen in muscular dystrophy (Nasevicius and Ekker, 2000; Guyon et al., 2003). Furthermore, the phenotype observed after MO injection is directly dependent on dose, and injection of MOs targeted to both the initiation start and splice junction site of one gene will result in a more thorough elimination of function. MOs can also be targeted to two genes simultaneously to eliminate redundant/overlapping function. For example, simultaneous injection of a her1-MO and her7-MO are required to mimic the severe somitic phenotype found in the irradiation-induced zebrafish mutant b567 (Henry et al., 2002). Injection of each construct alone caused less severe phenotypes, indicating that Her1 and Her7 have partially redundant functions in somitogenesis (Henry et al., 2002).


Recent work provides a reverse genetic complement to the forward genetic screens in zebrafish. The use of routine and efficient “knock-out” strategies are still at the earliest stages of development; however, the formation and manipulation of embryonic stem (ES) cell clones in culture has been reported, indicating that targeted mutagenesis will likely be available in the future (Ma et al., 2001). Additionally, nuclear transfer from long-term cultured cells has been used to clone zebrafish, providing another method for genetic manipulation (Lee et al., 2002). A more high-throughput reverse genetic strategy is TILLING (Targeted Induced Local Lesions IN Genomes), which involves random chemical mutagenesis (ENU) of adult zebrafish followed by screening for mutations in a specific “target” gene of interest by genomic sequence analysis (Wienholds et al., 2002). Through TILLING, it is possible to create an allelic series of mutations in a gene of interest to facilitate the identification of critical functional regions or to provide information about gene function in cases where a null allele is lethal. High-throughput sequencing was used to detect ENU-based mutations in the rag1 gene from nested PCR samples of genomic DNA isolated from thousands of F1 male progeny of ENU-mutagenized zebrafish. In this case, TILLING resulted in the identification of 15 mutations ranging from silent, to those in intronic sequences, to changes in the coding sequence of rag1 that result in missense mutations or stop codons (Wienholds et al., 2002). A truncated protein produced by the premature stop was shown to be incapable to performing V(D)J recombination, suggesting that regions 3′ to the deletion are critical for this function of rag1 (Wienholds et al., 2002). The genomic DNA “library” of ENU-mutated genomes isolated from FI males can be re-screened to detect mutations in other genes. Once a mutation of interest is identified at the molecular level, mutants can be recovered by in vitro fertilization using cryopreserved sperm collected from the appropriate F1 male and reidentified in heterozygous carriers (Wienholds et al., 2002).

To date, hundreds of zebrafish mutants that resemble a diverse array of human disease states have been identified. Through synteny, candidate, positional, or insertional cloning methods the genes responsible for these disorders are being identified. A complete description of all the pertinent zebrafish models of human disease is beyond the scope of this review. Instead, a thorough description of mutations with human disease correlates in blood and heart development will be presented (Table 1).

Table 1. Phenotypic and Genetic Characterization of Zebrafish Mutants With Analogous Human Hematological and Cardiovascular Disordersa
Mutant abbreviationMutant full nameRelevant phenotypeGene productRelated human disorderReferences
  • a

    Genes marked with an asterisk are mutated in the corresponding human disorder.

 cdychardonnayHypochromic AnemiaDMT-1Microcytic anemiaRansom et al., 1996; Donovan et al., 2002
 cha/motchablis/merlotHemolytic anemiaBand 4.1*Hereditary elliptocytosisLux and Palek, 1995; Ransom et al., 1996; Shafizadeh et al., 2002
 drcdraculaPorphyriaFerrochelatase*Erythropoietic protoporphyriaWeinstein et al., 1996; Childs et al., 2000
 retretsinaHemolytic anemiaBand 3*Congenital dyserythropoietic anemia type 2Lux and Palek, 1995; Ransom et al., 1996; Paw, 2001; Paw et al., 2003
 risrieslingHemolytic anemiaβ-spectrin*Hereditary spherocytosisLux and Palek, 1995; Ransom et al., 1996; Liao et al., 2000b
 sausauternesHypochromic anemiaAlas2*Sideroblastic anemiaCotter et al., 1994; Ransom et al., 1996; Brownlie et al., 1998
 wehweissherbstHypochromic anemiaFerroportin 1*Juvenile hemochromatosisRansom et al., 1996; Donovan et al., 2000; Montosi et al., 2001
 yqeyquemPorphyriaUrod*Hepatoerythropoietic porphyriaRansom et al., 1996; Wang et al., 1998
 aceacerebellarReduced ventricleFgf8 Brand et al., 1996; Whitfield et al., 1996; Reifers et al., 1998; Reifers et al., 2000
 bonbonnie and clydeCardia bifidaMix, MixerCardia bifidaStainier et al., 1996; Kikuchi et al., 2000
 cascasanovaCardia bifidaSox-familyCardia bifidaChen et al., 1996; Alexander et al., 1999; David and Rosa, 2001; Dichmeis et al., 2001; Kikuchi et al., 2001; Reiter et al., 2001b; Aoki et al., 2002
 cyccyclopsHeart looping defect (mild)Cyclops/Ndr2Left–right laterality disordersHatta et al., 1991; Strahle et al., 1993; Baier et al., 1996; Karlstrom et al., 1996; Malicki et al., 1996a; Schier et al., 1996; Solnica-Krezel et al., 1996; Stemple et al., 1996; Halpern et al., 1997; Rebagliati et al., 1998; Sampath et al., 1998; Varga et al., 1999; Bisgrove et al., 2000; Muller et al., 2000; Karlen and Rebagliati, 2001; Dougan et al., 2003
 faufaustCardia bifidaGata-5Cardia bifidaChen et al., 1996; Reiter et al., 1999; Rodaway et al., 1999; Reiter et al., 2001a
 flhfloating headHeart looping defectsFloating head (homeobox protein)Left–right laterality disordersHalpern et al., 1995; Talbot et al., 1995; Odenthal et al., 1996; Solnica-Krezel et al., 1996; Stemple et al., 1996; van Eeden et al., 1996; Halpern et al., 1997; Masai et al., 1997; Melby et al., 1997; Sumoy et al., 1997; Amacher and Kimmel, 1998; Bisgrove et al., 2000
 fogfoggyMyocardial generation defectSpt5 Guo et al., 1999; Guo et al., 2000; Keegan et al., 2002
 grlgridlockVasculature defectHey2 Weinstein et al., 1995; Stainier et al., 1996; Zhong et al., 2000; Zhong et al., 2001
 hanhands offHeart and fin differentiation defectsHand2Heart and limb syndromeBasson et al., 1997; Li et al., 1997; Alexander et al., 1998; Yelon et al., 2000
 hasheart and soulConcentric heart chambersaPKC Malicki et al., 1996a; Schier et al., 1996; Stainier et al., 1996; Horne-Badovinac et al., 2001; Peterson et al., 2001.
 hstheartstringsHeart and fin differentiation defectsTbx5*Holt-Oram syndrome (heart and limb syndrome)Basson et al., 1997; Li et al., 1997; Garrity et al., 2002
 islisland beatDefective heartbeat, small ventricleα1C-LTCCArrhythmiaStainier et al., 1996; Rottbauer et al., 2001.
 jekjekyllAbsence of heart valvesUgdhHeart valve defectsNeuhauss et al., 1996; Stainier et al., 1996; Walsh and Stainier, 2001
 likliebeskummerCardiac hyperplasiaReptinCardiac hypertrophyRottbauer et al., 2002.
 milmiles apartCardia bifidaLysosphingolipid G-protein–coupled receptorCardia bifidaChen et al., 1996; Stainier et al., 1996; Kupperman et al., 2000
 ntlno tailHeart looping defectsNo tailLeft–right laterality disordersSchulte-Merker et al., 1994; Odenthal et al., 1996; Solnica-Krezel et al., 1996; Stemple et al., 1996; van Eeden et al., 1996; Schier et al., 1997; Sumoy et al., 1997; Feldman and Stemple, 2001; Amacher et al., 2002
 oepone-eyed pinheadCardia bifida, heart looping defectsOne-eyed pinhead, EGF-CFC family*Cardia bifida, left–right laterality disordersHammerschmidt et al., 1996; Malicki et al., 1996a; Schier et al., 1996; Solnica-Krezel et al., 1996; Stemple et al., 1996; Schier et al., 1997; Strahle et al., 1993; Zhang et al., 1998; Gritsman et al., 1999; Yan et al., 1999; Bamford et al., 2000; Bisgrove et al., 2000; Gripp et al., 2000; Feldman and Stemple, 2001; Reiter et al., 2001b; Feldman and Concha, 2002; Griffin and Kimelman, 2002
 panpandoraMyocardial generation defectSpt6 Abdelilah et al., 1996; Malicki et al., 1996a; Malicki et al., 1996b; Stainier et al., 1996; Yelon et al., 1999; Keegan et al., 2002
 pikpikwikHeart contraction defectTitin*CardiomyopathyStainier et al., 1996; Gerull et al., 2002; Xu et al., 2002
 sihsilent heartAbsence of heart beatTnnt2*CardiomyopathyThierfelder et al., 1994; Chen et al., 1996; Moolman et al., 1997; Sehnert et al., 2002
 smoslow moReduced heart rate BradycardiaStainier et al., 1996; Baker et al., 1997; Warren et al., 2001


Zebrafish are particularly useful for the study of hematopoiesis. Blood circulation begins by 24 hours postfertilization (hpf) and is clearly visible under the microscope. The process of blood development and the morphology of zebrafish blood closely parallels that of mammals (for a recent review, see Wingert and Zon, 2003). There are both primitive and definitive waves of differentiation in the zebrafish, which produce primitive erythrocytes and macrophages, followed by definitive erthyrocytes, B cells, T cells, monocytes, granulocytes, and thrombocytes, respectively (Orkin and Zon, 1997; Trede and Zon, 1998; Herbomel et al., 1999; Willett et al., 1999; Bennett et al., 2001; Lieschke et al., 2001). Many zebrafish orthologs of blood-specific genes demonstrated to be important in mouse and human development have been isolated, including cmyb, gata1, gata2, globin, hhex, ikaros, lmo2, pu1, rag1, runx1, scl, and vegf (Hansen et al., 1997; Willett et al., 1997; Gering et al., 1998; Liao et al., 2000a; Liang et al., 2001; Burns et al., 2002; Lyons et al., 2002; Brownlie et al., 2003). Although the genetic program appears to be highly conserved during hematopoietic development in the vertebrate embryo, the site of hematopoiesis is not as consistent. Whereas primitive hematopoiesis in mice and humans takes place in the yolk sac, in the zebrafish, the primitive wave initiates in a region termed the intermediate cell mass (ICM; Detrich et al., 1995; Orkin and Zon, 1997). Definitive hematopoiesis appears to initiate in the dorsal aorta of the zebrafish as it does in all vertebrates studied to date; however, the site of adult hematopoiesis in the fish is the kidney, not the bone marrow as in mammals (Willett et al., 1997; Galloway and Zon, 2003). The development of the T-cell lineage occurs in the thymus in zebrafish, mice, and humans (Trede and Zon, 1998).

As part of large-scale mutagenesis screens, more than 50 mutants that affect hematopoiesis, comprising 25 complementation groups, have been isolated (Ransom et al., 1996; Weinstein et al., 1996). Fortuitously, the developing zebrafish embryo can survive in the absence of blood for several days, thus allowing for the detection of mutant larva with blood-related defects. The majority of mutations fall into five classes: those affecting mesoderm patterning, hematopoietic stem cells, committed progenitors, differentiation/proliferation, and hypochromic erythrocyte mutants. Due to the design of the screens, most of the identified blood phenotypes affect primitive hematopoiesis and the earliest stages of definitive hematopoietic development. The comprehensive spectrum of blood mutants identified to date has been reviewed recently (Wingert and Zon, 2003); this review will focus primarily on mutants in which the underlying genetic component has been identified, and those in which there is a corresponding human disease state.

Hemolytic Anemia

A significant fraction of the blood mutants isolated in the large-scale screens display an anemic blood phenotype as embryonic maturation occurs. These mutants, such as cabernet (cab), chablis (cha)/merlot (mot), grenache (gre), pale and wan (paw), pinotage (pnt), retsina (ret), riesling (ris), sticky blood (sti), and thunderbird (tbr), tend to initiate hematopoiesis normally, with wild-type numbers of cells that express normal levels of the erythroid marker gata1 (Ransom et al., 1996; Weinstein et al., 1996). After approximately 2 to 4 days of development, however, there is a dramatic decrease in the number of circulating blood cells in these mutants. This finding suggests that the mutations cause defects in hematopoietic progenitor cell proliferation or in the survival of erythroid cells. The merlot (mot) and chablis (cha) mutants (demonstrated to be allelic) have normal hematopoiesis for the first 3 days of development, followed by the onset of severe anemia by 4 days postfertilization (dpf) (Ransom et al., 1996). A close examination of erythrocytes from mot (or cha) embryos showed that they were morphologically abnormal, with spiculated membranes and bilobed nuclei. Examination of the peripheral blood of the small number of embryos that are homozygous viable demonstrated a significant reduction in red blood cell number, with an arrest at the basophilic erythroblast stage of development. Erythrocytes from homozygous mot or cha adults display membrane projections, surface pitting, and osmotic fragility. By using a combination of positional and candidate cloning approaches, the mutations were found to reside in the gene encoding the band 4.1 (also know as protein 4.1) erythrocyte-specific protein (Shafizadeh et al., 2002). Band 4.1 is a structural membrane protein found in erythrocytes, which functions to anchor the spectrin–actin cytoskeleton to the erythrocyte plasma membrane to maintain stability and red cell morphology. In humans, protein 4.1 deficiency causes hereditary elliptocytosis, a hemolytic anemia characterized by erythrocytes that are elliptical in shape rather than biconcave disks (Lux and Palek, 1995).

The gene mutated in the riesling (ris) mutant encodes another protein required for maintaining the integrity of the erythrocyte cell membrane (Liao et al., 2000b). The ris mutant has a null mutation in the erythroid β-spectrin gene, the largest component of the erythrocyte cytoskeleton (Liao et al., 2000b). ris mutant embryos display profound anemia by 4 dpf. Adult ris fish show severe anemia, as well as expanded numbers of hematopoietic progenitors in their kidney. Erythrocytes isolated from ris adults are spherical with round nuclei and undergo rapid hemolysis. The number of microtubules in ris erythrocytes are reduced by half, suggesting a role for β-spectrin in the aggregation and maintenance of microtubules in the red cell cytoskeleton (Liao et al., 2000b). Hereditary spherocytosis, in which patients have hemolytic anemia, characterized by spherocytic (completely round) erythroid cells, is the human hematopoietic disorder analogous to the zebrafish ris mutant phenotype (Lux and Palek, 1995).

Human hemolytic anemia is also caused by mutations in several other genes shown to be critical to red cell membrane ultrastructure, including ankyrin, erythroid α-spectrin, and band 3 (Lux and Palek, 1995). Recently, mutations in band 3 (also called anion exchanger 1) were found to cause the anemia observed in the zebrafish retsina (ret) mutant (Paw, 2001; Paw et al., 2003). In ret mutant embryos, anemia appears shortly after the onset of definitive hematopoiesis and is lethal in the majority of cases. Some ret fish survive to adulthood, although their total number of erythrocytes is significantly reduced, with maturation arrest at the late erythroblast stage of development and bilobed nuclei. These features are strongly reminiscent of patients with congenital dyserythropoietic anemia type 2 (Paw, 2001; Paw et al., 2003). Further investigation of the nuclear phenotype of ret embryos demonstrated that Band 3 had a critical role in the segregation of chromosomes during anaphase and that its absence prevented cytokinesis in developing erythroblasts (Paw, 2001; Paw et al., 2003). Analysis of mice with band 3 mutations suggests that the function in cytokinesis is evolutionarily conserved (Paw, 2001; Paw et al., 2003); furthermore, it could represent the underlying cause of erythrocyte defects seen in patients. The mot, cha, ris, and ret mutants each model a different aspect of human hemolytic anemia and will allow the function of the membrane cytoskeleton during erythropoiesis to be investigated further.

Hypochromic Anemia

Several zebrafish mutants were shown to have hypochromic microcytic anemia characterized by a reduced number of primitive erythrocytes, which are pale and diminished in size. Patients that present with these blood characteristics generally have hypochromic microcytic anemia resulting from a defect in hemoglobin production. Regulation of iron availability and the de novo synthesis of heme and globin are essential for hemoglobin production. Zebrafish mutants such as chardonnay (cdy), chianti (cia), clear blood (clb), frascati (fra), sauternes (sau), weissherbst (weh), and zinfandel (zin), are likely to have mutations in the genes involved in these processes (Ransom et al., 1996; Weinstein et al., 1996). This assumption was proven correct when the sauternes (sau) mutant was genetically identified. sau was shown to have defects in aminolevulinate synthase 2 (ALAS2), an enzyme that functions at the first step of heme biosynthesis (Brownlie et al., 1998). Primitive erythrocytes differentiate abnormally in sau mutants and embryos show a decrease in blood cell number by 2 dpf. sau mutants have markedly decreased heme levels, a decrease in β-globin expression, and fail to down-regulate gata1 during differentiation. Mutations in the ALAS2 gene result in sideroblastic anemia in humans, and the sau mutant represents the first animal model of this disease (Cotter et al., 1994; Brownlie et al., 1998). Further analysis of sau and similar mutants will help to increase understanding of the intracellular link between heme levels and erythroid differentiation.

The chardonnay (cdy) mutant has a defect in the gene encoding divalent metal transporter 1 (DMT1; Donovan et al., 2002). DMT1 is required to transport iron into cells during receptor-mediated endocytosis (Donovan et al., 2002). Zebrafish embryos that lack functional DMT1 have normal numbers of circulating erythrocytes until 48 hpf; however, hemoglobin is not detectable (Ransom et al., 1996; Donovan et al., 2002). cdy mutants are viable and show normal globin expression, but differentiation of circulating erythroid cells is delayed (Ransom et al., 1996; Donovan et al., 2002). Adults have an increased number of erythroid precursors in their peripheral blood and kidneys that is consistent with an increase in erythropoiesis in response to anemia. Due to the adult survival of DMT1 mutant zebrafish, the presence of a minor alternate pathway to supply iron to developing erythrocytes has been suggested, providing a basis for further investigation (Donovan et al., 2002).

Positional cloning of the gene responsible for the hypochromic mutant weissherbst (weh) resulted in the discovery of ferroportin1 (Donovan et al., 2000). weh embryos have morphologically immature erythrocytes with large nuclei and basophilic cytoplasm, make essentially no hemoglobin, and have significantly decreased levels of iron. Injection of iron-dextran could completely rescue hemoglobin production in weh embryos, demonstrating that weh erythroid cells are capable of making hemoglobin but lack adequate iron supplies due to defective transport of iron from the yolk to the embryo (Donovan et al., 2000). In addition to its role in transfer of iron across the yolk syncytial layer and the basolateral membrane of the absorptive enterocyte, Ferroportin1 has also been identified in macrophages, where it is hypothesized to play an important role in exporting iron from stored forms to developing erythrocytes. Shortly after the identification of the ferroportin1 gene in zebrafish, patients with a severe autosomal dominant juvenile onset form of hereditary iron overload (hemochromatosis) were demonstrated to have mutations in FERROPORTIN1 (Montosi et al., 2001). Of interest, patients with FERROPORTIN1 mutations develop iron overload, particularly in the Kupffer cells (macrophages of the liver), unlike patients with HFE-related hereditary hemochromatosis (Montosi et al., 2001). This finding implicates Ferroportin1 function in iron recycling, as well as in iron absorption. Analysis of the sau, weh, and cdy mutants demonstrates the usefulness of zebrafish as a tool to uncover new genes (weh), alternate biochemical pathways (cdy), and to model human disorders (sau, weh). Additionally, the weh mutation represents the first example of a previously unknown gene that was discovered in zebrafish, and then later shown to be mutated in the corresponding human disease.

Erythropoietin Porphyria

Photosensitive blood mutants include dracula (dra), desmodius (dsm), friexenet (frx), and yquem (yqe; Ransom et al., 1996; Weinstein et al., 1996). These mutants produce normal numbers of erythrocytes; however, the cells rapidly lyse upon exposure to light, emitting autoflouresence in the process. Human erythropoietic porphyria disorders also display these characteristics and are caused by defects in enzymes involved in the heme biosynthesis pathway. Perturbations in the structure or function of these enzymes lead to the accumulation of porphyrin intermediates that, when broken down upon exposure to light, release free radicals and cause cell lysis. Genes corresponding to two zebrafish photosensitive mutants have been cloned to date. The yquem (yqe) and dracula (dra) mutants encode the heme enzymes uroporphyrinogen decarboxylase (UROD) and ferrochelatase, respectively (Wang et al., 1998; Childs et al., 2000). The yqe mutant represents an accurate model of human hepatoerythropoietic porphyria (HEP), in which patients maintain a deficiency in UROD (Wang et al., 1998). The dra zebrafish provide a model for patients with erythropoietic protoporphyria (EPP; Childs et al., 2000). Patients with EPP have photosensitive, autofluorescent red cells and eventually develop liver disease, thought to be caused by the accumulation of toxic substances released from lysed erythrocytes (Childs et al., 2000). The yqe and dra mutants can be used to examine protoporphyrin-induced organ toxicity in controlled light conditions. In addition, the mutant zebrafish could be used in chemical screens to discover agents that could prevent or alleviate the symptoms of HEP and EEP in patients.


The transcription factor RUNX1 is often mutated in human leukemias; one leukemic fusion resulting from a translocation of Runx1 in humans, RUNX1-CBF2T1, was transiently expressed in zebrafish to examine its role in leukemogenesis (Kalev-Zylinska et al., 2002). Similar to expression of RUNX1-CBF2T1 in mice, transgene expression severely perturbed normal hematopoiesis and resulted in embryonic lethality (Kalev-Zylinska et al., 2002). Recently, transgenic zebrafish that express the oncogene cmyc fused to GFP under the control of the rag2 promoter were created (Langenau et al., 2003). Adult rag2-cmyc-GFP fish develop T-cell leukemia that is highly reminiscent of that seen in human patients (Langenau et al., 2003). The leukemic cells originated in the thymus and subsequently spread throughout the fish to infiltrate both hematopoietic and non-hematopoietic organs (Langenau et al., 2003). Leukemic cells that were transplanted into irradiated wild-type adult fish could efficiently home to the thymus and produce a pervasive leukemia (Langenau et al., 2003). Due to the speed of leukemia progression and death, it was impossible to propagate the rag2-cmyc-GFP line through mating. Experiments are under way to produce an inducible system in which the onset of leukemia can be controlled, thereby facilitating the maintenance of the rag2-cmyc-GFP fish line. After the creation of stable inducible lines of leukemia-prone fish, screens can be conducted to discover “second-hit” mutations that influence the development of the leukemic phenotype as well as provide a resource for the discovery of compensating mutations or chemicals that would prevent leukemia development.


In the vertebrate embryo, the heart is the first organ to develop and function (Stainier et al., 1993; Lee et al., 1994). The embryonic heart of the zebrafish has a close anatomic resemblance to that of a human heart at 3 weeks of gestation. The zebrafish heart consists of both an atrial and a ventricular chamber, composed of two concentric layers: an inner endothelial layer called the endocardium and an outer muscular layer termed the myocardium (Stainier et al., 1993; Glickman and Yelon, 2002). The formation of the zebrafish heart begins at approximately 16 hpf when zebrafish cardiac precursors begin to migrate to the midline from bilateral regions of the anterior lateral plate, marked by Nkx2.5 expression, termed the cardiac field (Stainier et al., 1993; Lee et al., 1994; Chen and Fishman, 1996; Yelon et al., 1999; Glickman and Yelon, 2002). Each bilateral cardiac sheet contains lateral atrial precursors that express Cardiac myosin light chain 2 (Cmlc2), and medial ventricle precursors that express both Cmlc2 and Ventricular myosin heavy chain (Vmhc) (Yelon et al., 1999); the ventricular precursor cells coalesce to the midline in advance of the atrial precursors (Yelon et al., 1999; Glickman and Yelon, 2002). As the cardiomyocytes reach the midline around 18 hpf, posterior ventricular cells contact and fuse, followed by posterior atrial cells (Yelon et al., 1999; Glickman and Yelon, 2002). This process continues along the anterior portion of both populations to create the cardiac cone, with ventricular cells at the apex and atrial cells at the base (Yelon et al., 1999; Glickman and Yelon, 2002). The cone tilts and extends to produce the heart tube, which begins to beat as a peristaltic wave at 22 to 24 hpf (Stainier et al., 1993; Glickman and Yelon, 2002). Heart formation is completed through a series of remodeling steps that properly position the ventricle and atrium, and produce cardiac valves to ensure unidirectional blood flow through the heart (Stainier et al., 1993; Glickman and Yelon, 2002). At 36 hpf, coordinated contractions of atrium and ventricle provide circulation to the head and trunk (Stainier et al., 1993; Glickman and Yelon, 2002).

The zebrafish is an ideal organism for the study of heart development and for the detection of perturbations in cardiogenesis because heart formation and function can be assessed visually in the embryo. Additionally, because the fish is not dependent on blood circulation for survival during embryogenesis (Pelster and Burggren, 1996), defects in heart development and/or function are more likely to be detected and recovered. Many zebrafish mutants with defects in cardiac development and function were identified in the large-scale mutagenesis screens and are described in detail elsewhere (Chen et al., 1996; Stainier et al., 1996). The cardiac defects identified in both large- and small-scale zebrafish screens cover numerous aspects of heart development and function including defects in cardiac precursor migration/fusion, formation/remodeling of heart structures, heart size, and in contractility/rhythmicity.

Cardia Bifida

In zebrafish and humans with cardia bifida, two “hearts” form due to a failure in the migration and fusion of cardiac primordia during embryogenesis. The underlying genetic bases of several zebrafish mutants that develop cardia bifida have been identified. Four mutations were found to disrupt genes required in Nodal signaling in the endoderm: a mix-family protein (bonnie and clyde, bon), a sox-related gene (casanova, cas), gata5 (faust, fau), and an EGF-CFC-related gene (one-eyed pinhead, oep; Chen et al., 1996; Ransom et al., 1996; Stainier et al., 1996; Zhang et al., 1998; Alexander et al., 1999; Gritsman et al., 1999; Reiter et al., 1999; Rodaway et al., 1999; Kikuchi et al., 2000, 2001; Dichmeis et al., 2001). A genetic pathway has begun to emerge in which oep and bmp2 (bone morphogenic protein 2/swirl; Kishimoto et al., 1997; Lee et al., 1998) regulate gata5/fau expression in cardiac precursors to influence cardiac migration and myocardial differentiation (Reiter et al., 2001b), while fau and bon work in parallel pathways upstream of cas to regulate endodermal cell fate (Alexander et al., 1999; Reiter et al., 2001a; Aoki et al., 2002). Implantation of wild-type endoderm into cas embryos rescues cardia bifida and further demonstrates that endodermal Nodal signaling is required for normal heart formation from mesodermal tissue (David and Rosa, 2001).

In miles apart (mil) embryos, cardia bifida results from a mutation in a lysosphingolipid G-protein–coupled receptor (Chen et al., 1996; Kupperman et al., 2000). This mutation produces a non-cell autonomous signaling defect that effects myocardial migration to the midline (Kupperman et al., 2000). Sphingosine-1-phosphate was found to be a functional ligand for Mil in cell culture; thus, mil embryos demonstrate that bioactive lipids have a role in precursor cell migration during heart formation (Kupperman et al., 2000).

Heart and Limb Syndromes

The hands off (han) mutant was found to have defects in myocardial and ventricular differentiation, as well as in pectoral fin patterning (Alexander et al., 1998). The genetic defect resides in the locus encoding the basic helix-loop-helix (bHLH) transcription factor hand2. In the fin, disruption of Hand2 function results in the production of a small bud of undifferentiated fin mesenchyme that fails to initiate chondrogenic condensation or to elongate, whereas in the heart, hand2 mutation decreases the number of cardiac precursors and inhibits cardiac migration (Yelon et al., 2000). The myocardial tissue that does form in han embryos is not patterned properly and fails to maintain expression of the critical cardiac gene tbx5; similarly tbx5 expression is initiated but not maintained in the fin bud of han embryos (Yelon et al., 2000). Mutations in TBX5, a member of the T-box family of transcription factors, result in Holt-Oram syndrome, a human genetic disorder characterized by heart and upper limb defects (Basson et al., 1997; Li et al., 1997; Yelon et al., 2000). In a specific screen for mutations affecting zebrafish cardiac function, the mutant heartstrings (hst) was isolated and later shown to encode the zebrafish ortholog of human TBX5 (Garrity et al., 2002). hst embryos do not develop pectoral fins (analogous to mammalian limbs) and have severe heart defects, including reduced heart rate, failure of heart looping, and cardiac deterioration/stretching (Garrity et al., 2002). Haploinsuffiency of TBX5 in humans causes a variety of either bilateral or asymmetric upper limb defects, ranging from minor thumb deformities to deletion of the entire arm (Basson et al., 1997; Garrity et al., 2002). Low-level morpholino inhibition of tbx5 also results in a variety of bilateral and asymmetric fin (limb) deformities, suggesting an additional role for Tbx5 in fin/limb outgrowth. The mild bradycardia detected in hst embryos at the onset of heart function, may be present in human patients with Holt-Oram syndrome, but has not been described (Garrity et al., 2002). Expression of Hand2 in the lateral plate mesoderm of hst embryos appears normal; however, by 32 hpf, expression of Hand2 cannot be detected in the heart or fins (Garrity et al., 2002). The han and hst mutant zebrafish underscore the presence of a conserved parallel genetic pathway between heart and limb formation that requires a critical threshold of Tbx5 for the maintenance of Hand2 expression.

Atrium and/or Ventricle Defects

Defects in the processes of cardiac migration and fusion can result in the improper assignment of atrial and ventricular cell fates, as well as flaws in the maturation of each heart chamber. The heart and soul (has) mutant has a defect in cardiac fusion that influences polarity of the developing heart tube, as well as other organs (Schier et al., 1996; Stainier et al., 1996). An atypical protein kinase C (aPKCλ) that functions in epithelial polarity and tight junction integrity is encoded by the has locus (Horne-Badovinac et al., 2001; Peterson et al., 2001). The hearts of has mutant embryos fail to develop distinctive chambers, with ventricular muscle appearing inside of the atrium (Horne-Badovinac et al., 2001; Peterson et al., 2001). In a recent screen for chemical agents that affect heart formation, the chemical concentramide was found to phenocopy the heart defect in has mutants, with the ventricle forming inside the atrium in treated embryos (Peterson et al., 2001). Unlike has embryos, concentramide does not affect cell polarity, but appears to shift the converging heart field (Peterson et al., 2001). Addition and subtraction of concentramide at precise time-points during zebrafish development demonstrated that the chemical had to be present before the 14-somite stage to produce the concentric chamber phenotype (Peterson et al., 2001). Examination of the heart tissue during that critical time period in has and concentramide-treated embryos showed a defect in the generation of the midline cone (Peterson et al., 2001). In has and concentramide-treated embryos, the fusion of the posterior ends is delayed, occurring well after that of the anterior end; this delay appears to affect later processes of tilting and looping resulting in atrial cells bending around and enclosing the ventricular population (Yelon et al., 1999; Peterson et al., 2001). Analysis of has embryos demonstrates that proper fusion of the heart tube is critical for appropriate atrium and ventricle chamber separation and development.

In addition to has, mutants such as pandora (pan), foggy (fog), and acerebellar (ace) affect heart tube formation and atrioventricular development (Brand et al., 1996; Chen et al., 1996; Driever et al., 1996; Jiang et al., 1996; Stainier et al., 1996; Guo et al., 1999, 2000; Yelon et al., 1999). ace embryos, originally identified on the basis of their brain phenotype, have a defect in fibroblast growth factor 8 (fgf8) that causes a severe reduction in the size of the ventricle (Brand et al., 1996; Reifers et al., 1998, 2000). Fgf8 is expressed in cardiac precursors and in the ventricle, and Fgf8 signaling was shown to be critical for myocardial induction in ace mutants (Reifers et al., 2000). The function of Fgf proteins in vertebrate heart development has been difficult to study due to functional redundancies between family members and early embryonic lethality. Thus, the ace mutant provides important evidence that specific Fgf proteins are not only expressed but required during heart formation (Reifers et al., 2000). Morpholino knock-down of fgf8 mimics the ace phenotype, suggesting that the use of morpholinos to simultaneously inhibit other members of the zebrafish Fgf family alone or simultaneously could be used to investigate their roles in vertebrate heart development further (Araki and Brand, 2001). pan and fog mutants have a defect in the generation of myocardial cells that results from mutations in the genes encoding the transcription elongation factors Spt6 and Spt5, respectively (Guo et al., 2000; Keegan et al., 2002). Spt6 and Spt5 appear to work cooperatively to increase transcriptional efficiency during development; the defects seen in heart formation in pan and fog embryos likely arise due to the cumulative effect of inadequate transcription of several critical heart-specific genes (Keegan et al., 2002).

Heart-Looping and Laterality Defects

Shortly after the heart begins to function, it is remodeled through the processes of cardiac jogging to place the heart to the left of the dorsal midline and cardiac looping to place the ventricle to the right of the atrium. Defects in the assignment of left–right asymmetry in patients, situs inversus or situs ambiguous, interrupt the process of cardiac looping; additionally, many congenital human heart diseases are associated with flaws in the rightward looping of the heart tube (Bisgrove et al., 2000; Bisgrove and Yost, 2001). Several zebrafish “looping” mutants have been isolated and many of these mutants have alterations in the positioning of other organs as well as the heart; mutants that have aberrations in the establishment of midline structures, such as the notochord in no tail (ntl), floating head (flh), and momo (mom), or the floor plate as in cyclops (cyc/ndr2) display randomized heart looping (Odenthal et al., 1996; Halpern et al., 1997; Rebagliati et al., 1998; Sampath et al., 1998; Bisgrove et al., 2000; Amacher et al., 2002). The development of the notochord and the floor plate are hypothesized to act at the midline as molecular barriers that prevent the dispersion of asymmetrical signals (Bisgrove and Yost, 2001). In humans and zebrafish, left-sided Nodal signaling is thought to be critical for determining the left–right axis. Mutations in EGF-CFC genes, oep in zebrafish, and CFC1 in humans, block Nodal signaling and cause random organ positioning and incorrect heart placement (Zhang et al., 1998; Bamford et al., 2000; Feldman and Concha, 2002). Recently, another Nodal-related gene, southpaw (spaw), was identified in zebrafish and demonstrated by means of morpholino knock-down to be crucial for the expression of left-specific genes and the processes of cardiac jogging and looping (Long et al., 2003).

Cardiac Valve Defects

Another critical remodeling step after heart formation, and the onset of function, is the production of heart valves. In the jekyll (jek) mutant, blood cells “toggle” between the two chambers of the heart due to the absence of valves (Stainier et al., 1996). jek mutants fail to form cardiac cushions, which are the precursors of cardiac valves (Walsh and Stainier, 2001). Of interest, jek embryos also do not show distinct gene expression (bmp4, br146, and notch 1) patterns in the valve forming regions of the heart (Walsh and Stainier, 2001). This finding suggests that the jek mutation is required for proper gene expression to delineate the atrioventricular boundary and that this designation is needed for cushion (and later valve) formation. Mutations in the uridine-diphosphate (UDP)-glucose dehydrogenase (UGDH) locus cause the jek phenotype (Walsh and Stainier, 2001). UGDH is required starting at gastrulation to produce heparan sulfate proteoglycans; however, jek embryos appear to bypass this earlier requirement through maternal contribution of UGDH (Walsh and Stainier, 2001). Jek/UGDH is proposed to act during cardiac valve development as a critical component in a previously unidentified signaling cascade (Walsh and Stainier, 2001).


In silent heart (sih) mutants, the process of cardiac migration, fusion, valve formation, and looping all proceed normally, however, the heart fails to contract (Chen et al., 1996). The mutations causing sih are found in noncoding regions of the sacromere component cardiac troponin T (tnnt2; Sehnert et al., 2002). The identity of sih as tnnt2 phenotype was confirmed by morpholino and DNA rescue experiments (Sehnert et al., 2002). In the absence of Tnnt2, cardiac sarcomeres cannot be assembled as a result of defects in thin filament stability, and heart muscle is rendered nonfunctional (Sehnert et al., 2002). In humans, mutations in TNNT2 cause familial hypertrophic cardiomyopathy (Thierfelder et al., 1994; Moolman et al., 1997; Sehnert et al., 2002). The sarcomere loss and myocyte disarray associated with TNNT2 mutation can result in heart failure and sudden death (Thierfelder et al., 1994; Moolman et al., 1997; Sehnert et al., 2002). The sih mutant represents the first animal model of Tnnt2 deficiency.

Sarcomeres are also the affected target in the mutation that causes the pickwick (pik) phenotype (Xu et al., 2002). Little if any blood is ejected from the heart in pik mutants due to a severe reduction in systolic pressure (Stainier et al., 1996; Xu et al., 2002). pik embryos form thin cardiac myocytes that can contract but fail to generate higher order sarcomeres (Xu et al., 2002). A mutation in cardiac-specific exons of titin (ttn) was shown to cause pik (Xu et al., 2002). Antisense inhibition or deletion of the TTN cardiac scaffold protein was previously shown to cause defective sarcomere assembly in cardiac cell lines (Person et al., 2000; van der Ven et al., 2000). The phenotype associated with pik resembles the defects seen in human familial dilated cardiomyopathy, a disease with a strong genetic contribution in which the genes responsible for the defect have been identified only in several rare cases (Xu et al., 2002). A companion study has shown that mutations in TTN cause human familial dilated cardiomyopathy (Gerull et al., 2002).

Cardiac Hypertrophy

Prolonged cardiac hypertrophy in humans can lead to dilation, poor contractility, and eventually heart failure and death. Hypertrophy initially results in response to load demands on the heart, and how it progresses to cardiac failure is not understood at this time. The zebrafish mutant liebeskummer (lik) shows embryonic cardiac hyperplasia, with underdevelopment of endoderm-derived organs (gut hypoproliferation) (Rottbauer et al., 2002). By positional cloning, lik was shown to result from an activating mutation in a conserved domain of reptin (Rottbauer et al., 2002). Reptin functions as part of a multimeric ATPase complex in chromatin remodeling and transcription in all eukaryotes. Pontin, often found in complexes with Reptin, has an antagonistic affect on Reptin during development (Bauer et al., 2000). Reduction in Pontin was also found to cause cardiac hyperplasia in zebrafish (Rottbauer et al., 2002). Thus, the Reptin/Pontin ratio plays a critical role in the regulation of heart growth in the zebrafish. The lik mutation causes a dramatic increase in the thickness of the cardiac wall due to an increase in the number of cardiomyocytes after the heart tube has formed (Rottbauer et al., 2002). The load demands on the lik mutant heart eventually cause poor contractility and failure reminiscent of that seen in patients with prolonged hypertrophy, suggesting that Reptin or associated proteins could have a similar role in the development of hypertrophy and heart failure in human hearts (Rottbauer et al., 2002).

Arrhythmia and Bradycardia

The zebrafish mutant island beat (isl) has a defect in heartbeat and cardiac morphogenesis (Stainier et al., 1996). Positional cloning was used to identify the genetic defect in isl embryos as a mutation in the α1C L-type calcium channel subunit (C-LTCC; Rottbauer et al., 2001). The C-LTCC protein is the primary ion-conducting pore-forming subunit of the L-type calcium channel in cardiac tissue (Rottbauer et al., 2001). Calcium plays a critical role in heart formation and function, regulating elements of the heartbeat that include pacemaking, contractions, and atrio-ventricular conduction (Rottbauer et al., 2001). Calcium is also thought to influence cell growth (Rottbauer et al., 2001). In isl embryos, which lack functional C-LTCC, the atrium is relatively normal in size, but contracts in a disordered manner that is not propagated to the ventricle (Rottbauer et al., 2001). The ventricle of isl embryos is small and does not beat (Rottbauer et al., 2001). The growth defect in the ventricle results from a decrease in number of cardiomyocytes (Rottbauer et al., 2001). Transplantation analysis demonstrated that the isl mutation has distinctive effects on atrial and ventricular cells; in the atrium, the contraction defect was shown to be cell-autonomous, while in the ventricle there was a non-cell autonomous requirement for C-LTCC function (Rottbauer et al., 2001).

The slow mo (smo) mutation also affects heartbeat. smo fish have a dramatic reduction in their “pacemaker current” (Ih), resembling the human heart condition bradycardia (Stainier et al., 1996; Baker et al., 1997; Warren et al., 2001). Although exhibiting bradycardia, a portion of smo mutants can survive to adulthood, and have hearts that are indistinguishable morphologically from wild-type hearts. Ih was shown to be smaller in smo atrial cells than in wild-type cardiomyocytes (Warren et al., 2001). The genetic component for smo has not been identified to date; however, the isolation of slow mo provides the first genetic evidence that cardiac pacemaking is regulated by Ih in the adult zebrafish. Congestive heart failure in humans can result from a variety of mutations that affect cardiac output/function, and the identification of smo, as well as other heart mutants that impact the heartbeat, may be beneficial.

Coarctation of the Aorta

The zebrafish mutant gridlock (grl) closely resembles the human congenital disorder coarctation of the aorta (Weinstein et al., 1995; Stainier et al., 1996; Zhong et al., 2000). grl embryos display an absence of blood circulation to the truck and tail, despite a functioning heart and the presence of anterior circulation. Further analysis of grl embryos indicated a block to caudal blood flow at the base of the aorta, resulting from a defect in the fusion of the two lateral dorsal aortae into a single midline structure (Weinstein et al., 1995). The genetic defect in grl embryos was identified by positional cloning as a mutation in a gene encoding a bHLH protein in the Hairy/Enhancer of Split family that is related to Hey2/HRT-2 (Zhong et al., 2000). By in situ hybridization, grl was found to be expressed prior to vessel formation in bilateral stripes of angioblasts that converge toward the midline to form the aorta, and expression persists at high levels in dorsal aorta (Zhong et al., 2000). grl is not expressed in venous tissue and was found to play a critical role in artery/vein specification in embryonic development (Lawson et al., 2001; Zhong et al., 2001). The grl mutant phenotype suggests that Grl function continues to be required for fusion of the paired dorsal aortae.


Large-scale and targeted mutagenesis screens in the zebrafish are increasing the identification of unknown genes responsible for the formation of particular cells and tissues during vertebrate development. Additionally, several zebrafish mutants have revealed unknown genetic interactions and functional requirements of genes previously identified in other organisms. It is clear that many more genetic pathways influencing hematopoietic and cardiovascular development and differentiation will be delineated as mutants isolated in the large-scale screens continue to be cloned. Many of the zebrafish blood and heart mutants identified to date evince similar phenotypic pathology as human disease states and thus can be used to model disease onset and progression.

While the large-scale ENU and targeted mutagenesis screens successfully identified hundreds of mutants affecting known and previously unidentified genes, they did not reach genetic saturation. For example, there are several genes known to be critical for hematopoiesis in mice and humans, such as lmo2, runx1, or scl, for which no zebrafish ENU mutants were isolated. A second large-scale ENU mutagenesis screen was completed in Tubingen, Germany in 2000 in an attempt to further saturate genetic pathways used in zebrafish development (Habeck et al., 2002; Knaut et al., 2003). This screen yielded several new blood mutants that are currently under investigation in the Zon lab. Additionally, mutants that possess deficiencies in conserved genes controlling blood and vasculature development such as Flk1 (VEGF-R2) were also identified in the second large-scale screen (Habeck et al., 2002).

Currently, small-scale “targeted” screens that are focused on the identification of genes in particular developmental pathways are ongoing in many laboratories. One such screen being conducted in the Zon laboratory is examining the process of thymopoiesis (the production of T cells) through the identification of mutants that lack expression of the critical lymphoid gene rag1 (Trede et al., 2001). Eight mutants were isolated on the basis of an altered in situ expression pattern and are in the process of being genetically identified (Trede et al., 2001). By using both in situ expression patterns of artery-specific markers and a transgenic zebrafish line that expresses GFP under the control of the vascular-specific fli1 promoter, several mutants in the VEGF signaling pathway required for artery development were recently isolated (Lawson and Weinstein, 2002; Lawson et al., 2003). One mutant with artery-specific defects, y10, was found to encode the zebrafish homolog of phospholipase C gamma-1 (plcg1); through microinjection experiments, Plcg1 was shown to be required for Vegf function during arterial development (Lawson et al., 2003).

Furthermore, once interesting mutations are identified or created in zebrafish, modifier screens can be conducted to identify genetic interactions that can alter or prevent the development of a particular phenotype, or to change the spatiotemporal expression of the mutated gene as visualized by in situ pattern or GFP (Patton and Zon, 2001; Amatruda et al., 2002). For example, ENU-treated males can be bred to females that carry the mutation of interest and the progeny would be examined for a modified phenotype. If the secondary mutation is a “suppressor” the original phenotype would be absent, decreased, or delayed, whereas if the mutation is an “enhancer,” the phenotypic defect would show greater severity or earlier onset. Identification of the “modifying” mutations could highlight other genes that function in that particular pathway, or alternatively, could identify parallel pathways in organogenesis.

Large-scale in situ-based screens are being used to systematically identify the spatio-temporal expression pattern of hundreds of genes during embryogenesis and will also assist in elucidating all the members of a particular genetic pathway. In the Thisse laboratory, one such screen examined the expression pattern of random cDNA probes isolated from a normalized, embryonic cDNA library; this screen can quickly uncover new genes expressed in specific cell types, as well as confirm or modify the expression domain of known genes (Herbomel et al., 1999; Thisse et al., 2000). More than 60 cDNAs, representing both known and unknown genes, expressed specifically in the blood and vasculature were identified by this in situ expression screen; thus providing new information about genes that may function in hematopoietic development as well as indicating promising candidate genes for the ENU mutants (Thisse and Zon, 2002). The identification of the cdy mutant, described earlier in this review, was accomplished by this approach; DMT1 was discovered among the 41 random cDNAs expressed in hematopoietic cells and shown by means of candidate cloning to encode the gene mutated in cdy zebrafish (Donovan et al., 2002).

Following the genetic identification of zebrafish mutants with relevance to human disease, small molecule screens, in which mutant embryos are exposed to a library of chemical compounds, can also be used to discover therapeutic agents that alleviate or suppress the phenotypic “symptoms” of the mutation (Langheinrich et al., 2002; Stern and Zon, 2003). For example in the Zon laboratory, a small molecule screen was used effectively to ascertain one chemical that could suppress the phenotype associated with an embryonic cell cycle mutant (Stern and Zon, 2003). Chemicals identified in the small molecule screens can be further analyzed to determine whether they have pharmacologic potential as agents to effectively treat or to prevent the onset of a particular disease state. The validity of using zebrafish to accurately model human physiological responses to known and unknown drug compounds was recently examined. Zebrafish were used to screen for drugs that affect heart rate; the screen for bradycardia successfully identified known QT-elongation agents and mimicked the effects of known drug–drug interactions (Milan et al., 2003). This finding demonstrated that zebrafish could be used to effectively test the effects of new pharmacologic drugs on heartbeat (Milan et al., 2003).

In addition to the hematopoietic and cardiovascular disease models presented here, an increasing number of zebrafish mutants have been identified in other organ systems such as bone, brain, cartilage, ear, eye, fins, kidney, gut, muscle, and vasculature (Driever et al., 1996; Haffter et al., 1996; Golling et al., 2002). The identification of some of these mutants will likewise be beneficial for exploring the etiology of relevant human disease states. For example, the violet beauregarde (vbg) mutant, which has endothelial cell defects caused by a mutation in activin receptor-like kinase 1 (acvrl1/alk1), provides a model for human hereditary hemorrhagic telangiectasia type 2, characterized by vessel malformations that can cause hemorrhage or stroke (Roman et al., 2002). The impending completion of the zebrafish genome will facilitate the genetic cloning of the remainder of the mutations uncovered in both large- and small-scale screens, providing new information about the development of human disease correlates in many of these organ systems. The use of zebrafish as a model for human hematopoietic and cardiovascular disease has contributed valuable leads toward the current understanding human pathophysiology in these organ systems, and is poised to continue to add to our knowledge of the genetic bases of human disease states.


We thank Alan Davidson, Kimberly Dooley, Paula Fraenkel, Noëlle Paffett-Lugassy, and Rebecca Wingert for critical review of the manuscript. T.E.N. is supported by Postdoctoral Fellowship Grant PF-03-221-01-LIB from the American Cancer Society. L.I.Z. is an investigator of the Howard Hughes Medical Institute.