Hematopoietic stem cells, hematopoiesis and disease: lessons from the zebrafish model

The zebrafish model is rapidly gaining prominence in the study of development, hematopoiesis, and disease. The zebrafish provides distinct advantages over other vertebrate models during early embryonic development by producing transparent, externally fertilized embryos. Embryonic zebrafish are easily visualized and manipulated through microinjection, chemical treatment, and mutagenesis. These procedures have contributed to large-scale chemical, suppressor, and genetic screens to identify hematopoietic gene mutations. Genomic conservation and local synteny between the human and zebrafish genomes make genome-scale and epigenetic analysis of these mutations (by microarray, chromatin immunoprecipitation sequencing, and RNA sequencing procedures) powerful methods for translational research and medical discovery. In addition, large-scale screening techniques have resulted in the identification of several small molecules capable of rescuing hematopoietic defects and inhibiting disease. Here, we discuss the contributions of the zebrafish model to the understanding of hematopoiesis, hematopoietic stem cell development, and disease-related discovery. We also highlight the recent discovery of small molecules with clinical promise, such as dimethyl prostaglandin E2, 3F8, and thiazole-carboxamide 10A.


Genome comparison
The zebrafish shows genetic similarity to other verte brates. At approximately 1.8 billion base pairs, the zebrafish genome is about twothirds the size of the human genome [10]. Although the fish genome is vastly rearranged, several areas of local synteny and some larger chromosomal regions are preserved [11]. This has greatly facilitated positional cloning projects, as chromosomal synteny can be used as a guide within the genome. Comparisons of chromosomal arrangements and indi vidual DNA sequences in the zebrafish have revealed general conservation, particularly for the Hox loci [12]. However, the zebrafish genome incurred a significant duplication that arose in teleosts about 300 million years ago. Because of the early incidence of this duplication in teleost evolution, the zebrafish genome has since under gone further alterations as subsequent deletions are believed to have removed many of the originally dupli cated genes [11]. These genomic events are demonstrated by the presence of seven Hox clusters in zebrafish com pared with only four in humans [12].
These alterations have provided unique opportunities for discovery, as they have sometimes led to a splitting of regulatory elements. For instance, the zebrafish has two independent transferrin receptor1 genes [13]. One is a general, ubiquitously expressed gene and the other is a redbloodcellspecific gene. In humans, there is a single gene for transferrin receptor1 that is expressed both highly in red blood cells and ubiquitously at a low level. Nevertheless, an independent zebrafish mutant in trans ferrin receptor1 has been isolated that lacks red blood cells. Comparative genomic analysis and study of the regulatory sequences in this mutant may prove useful.
Recent technological advances have also made zebra fish epigenetic analysis possible, as demonstrated by the use of chromatin immunoprecipitation sequencing (ChIPseq) by a number of laboratories studying specific chromatin or transcription factor binding in whole zebrafish embryos [14,15]. Furthermore, chromatin re modeling has been evaluated by analyzing specific histone modifications, such as methylation and acetyla tion. Recent studies have highlighted the specific post translational modifications H3K4me3, H3K9ac, and H4ac as activating; H3K27me3 and H3K9me3 as repressing; and H3K36me3 as being involved in transcriptional elongation [16,17]. Through these techniques, the zebra fish model has helped to clarify the relationship between epigenetics and gene function, and can be expected to further contribute to this understanding in the future.

Mutagenesis
The zebrafish model has been pioneered as a genetic system for studying a variety of different diseases, includ ing hematopoietic disorders. In 1996, new mutagenesis protocols allowed the derivation of many novel blood mutants [18,19]. Male zebrafish were soaked in ethyl nitrosourea (ENU), introducing mutations to the sperm or spermatogonia. The males were then mated with females, creating progeny that carried over 100 mutations per genome. Outcrossing these mutants formed F2 families, which were subsequently crossed to analyze the autosomal recessive or dominant hematopoietic muta tions in the F3 generation. The initial experiment derived more than 50 independent blood mutants that formed 26 complementation groups when crossmated [18,19]. Most of the mutated genes have since been isolated and linked to many defects in mesoderm induction, stem, or progenitor cell formation, and erythroid or Tcell develop ment [20]. Analysis of several red blood cell mutants isolated novel genes that correlated to mutations found in several human subjects with anemia. At least three independent zebrafish mutations, and the resulting blood disorders, have contributed to the discovery of the molecular basis of human diseases (Table 2) [2123].
Other largescale approaches have been applied to the zebrafish system. One uses 'targetinginduced local lesions in genomes' (TILLING), a process by which random mutagenesis and individual exon sequencing are carried out to identify mutations within a particular gene, essentially mimicking the outcome of targeted muta genesis [24]. This approach has led to the derivation of mutants in the runx1 and gata1 genes, which encode transcription factors specific to the blood program [25,26]. Furthermore, an insertional mutagenesis screen was conducted by Nancy Hopkins and coworkers, in which a retrovirus was inserted into the genome to promote random mutagenesis [27,28]. The recovered mutations defined 315 independent mutants affecting early development. In another study, retroviral insertions were shown to be highly efficient on a genomewide scale, with nearly one in five integrations resulting in mutation [29]. The mutants discovered in these large scale screens have proved extremely useful, with some associated with cancer phenotypes and others affecting individual organs. In the zebrafish system, haploid genetics can also be used [3032]. This is extremely powerful as it reduces the extensive requirements typically associated with genetic screens. In a haploid screen, males are mutagenized (with ENU) and mated with a female. The eggs produced by the secondgeneration females are studied by in vitro fertilization with UVirradiated sperm. The UV irradia tion functionally inactivates the paternal DNA while maintaining sperm fertility, thus creating haploid animals. Zebrafish haploids typically survive for 4 days, but the reason for death is unclear. However, given that blood formation occurs within the first 36 hours of development, haploid screens can be used to study independent hematopoietic mutants. A recent variant of such a screen used early pressure to identify several mutants that affected Tcell development [20,33]. The early pressure method suppresses the second meiotic division, generates gynogenetic diploids, and thus elimi nates the additional complexity created by heterozygosity [34]. Using this particular method, the sart3 gene was found to be critically required for thymus development through regulation of the U6 small nuclear ribonucleo protein [35].

Morpholinos and gene knockdown
Morpholinos are small antisense oligonucleotides that are constructed to specifically target sequences at the transcriptional start site (ATG morpholinos) or at intron exon splice junctions (splicing morpholinos) and allow selective inhibition of a target gene [34]. The use of morpholinos has greatly expanded the versatility and importance of the zebrafish model in biomedical sciences [9,36,37].
Morpholinos have been used extensively for the knockdown of a variety of hematopoietic genes and have proven to be an important tool for genetic screens. For instance, we are currently conducting a chromatin factor screen, targeting over 480 independent factors via morpholino knockdown, to determine DNA rearrange ment requirements in hematopoiesis. This screen provides a means for determining the role of chromatin factors in the birth of hematopoietic stem cells (HSCs) in the aorta and in globin expression (HT Huang, K Kathrein, and LI Zon, unpublished).

A new era of genetic suppressor screens
Recently, we undertook a novel genetic suppressor screen in the search for recessive mutants that rescue a zebrafish mutant phenotype ( Figure 1) [2]. This screen focused on the mutant moonshine, which completely lacks blood due to a defect in the chromatin factor Tif1gamma [38]. Tif1gamma contains several motifs, including a PHD fingerbromodomain and a ring finger domain, and several laboratories have demonstrated its involvement in transforming growth factor beta (TGFbeta) signaling [39,40]. The aforementioned screen [2] sought another gene that, when mutated, would restore blood develop ment in moonshine. To this end, moonshine was rescued with a transgenic bacterial artificial chromosome (BAC) containing the wildtype copy of tif1gamma. The BAC had a ubiquitous actin promoter driving green fluores cent protein (GFP) expression, resulting in stable trans genic fish that were both green and homozygous mutant at the endogenous tif1gamma locus. These fish were mutagenized and a haploid screen was conducted. Half of the screened embryos were GFP positive and half displayed the moonshine phenotype. A suppressor was defined as a recessive mutation resulting in the rescue of blood in at least half of the mutant phenotypes. Using this zebrafish screening model, we discovered two inde pendent suppressors (sunshine and eos) [2]. We mapped the sunrise suppressor to cdc73, a gene involved in the polymeraseassociated factor (PAF) complex, which is required for transcription elongation. The PAF complex includes several other factors, which, when inactivated in the moonshine background, also resulted in rescue. This demonstrated involvement of the PAF complex in hematopoietic cell transcriptional elongation. Purifica tion of the complex bound to Tif1gamma demonstrated the transcriptional involvement of other cellspecific regulators, including Gata1 and the basic helixloophelix transcription factor Scl, and the elongation factor PTefb,  [2]. The transgene was injected into one-cell-stage embryos (right) to rescue the lethality of Tif1gamma mutant (mon) fish. (b) Schematic diagram of the suppressor screen. Stable transgenic fish are homozygous mutants for the endogenous tif1gamma locus (mon/mon) but retain viability because they are heterozygous for the transgene. The GFP marker on the transgene makes them green fluorescent. F0 males were mutagenized with ethylnitrosourea (ENU). In the F 1 generation, 25% of progeny were transgene homozygotes (Tg homo, mon/mon; Tg/Tg, bright green), 50% were transgene heterozygotes (Tg het, mon/mon; Tg/+, light green, in red circle), and 25% lacked the transgene (No Tg, mon/mon, gray). Only the progeny that were heterozygous for the transgene were raised to adults. The F 1 females were then squeezed to provide unfertilized eggs that were activated by UV-treated sperm. The UV treatment destroys the paternal DNA while still allowing fertilization. The resulting F 2 embryos were haploid and were subjected to in situ hybridization (ISH) at 22 hours post-fertilization for GFP and beta e3 globin probes. Transgenic embryos (mon;Tg) were positive for both probes, whereas non-transgenic embryos (mon) were negative for both probes. However, embryos that were negative for GFP but positive for globin indicated the presence of a genomic suppressor (sup) mutation. which is the kinase responsible for phosphorylation of RNA polymerase II and its regulator DRB sensitivity inducing factor (DSIF) [2]. This suggests a model whereby all blood gene transcription in moonshine is paused until the additional mutation in the PAF or DSIF complex promotes rescue by obstructing transcriptional inhibi tion. This novel mechanism has also been observed in other cell types, including in melanocyte cell fate regu lation [41].
In another suppressor screen we analyzed the cdx4 mutant kgg, which is defective in HSC development because of abnormal hox gene expression [42,43]. Several chemicals were found to rescue the cdx4 mutant, many of which are involved in the retinoic acid pathway. This suggests that the CdxHox pathway mediates the retinoic acid response during hematopoietic cell development. Through these types of largescale screens, the zebrafish model provides a means of defining connections between abnormal gene function and their respective pathways.

Small-molecule screens in the zebrafish
Zebrafish embryos have become a very useful tool for studying developmental responses to chemical treatment [44]. We recently conducted a chemical screen investi gating the birth of HSCs in the aorta. In this screen, individual embryos were placed into a 96well plate and chemically treated ( Figure 2) [45]. Embryos were then stained for the stem cell markers Runx1 and cMyb. The screen revealed 35 chemicals capable of enhancing HSC engraftment, the most potent of which was dmPGE2, a known small lipid mediator of inflammation that is upregulated during marrow transplantation. Following its discovery in zebrafish, we tested the efficacy of dmPGE2 in mammals using a limiteddilution competi tive repopulation assay in mouse marrow transplants, which showed a fourfold increase in HSC engraftment. This increase is sufficient for therapeutic consideration. For instance, current cord blood transplantation uses a single cord for young children, whereas adult trans plantation requires two cords. dmPGE2 increases cord blood engraftment in nonobese diabetic severe com bined immunodeficiency (NOD/SCID) animals and has been shown to be nontoxic in primate competitive trans plant models [46].
Many other smallmolecule screens have been per formed, contributing equally promising candidate chemical treatments. The discovery of 3F8, a novel inhibitor of glycogen synthase kinase 3 (Gsk3), has great potential as a candidate for therapeutic use. Gsk3 is a key member of the Wnt and hedgehog signaling pathways and has been linked to a number of human diseases, including type 2 diabetes, bipolar disorder, Alzheimer's disease, and some cancers [47]. The combination of multiple pathway involve ment and multiple disease implication makes Gsk3 a potentially important drug target. In a recent chemical screen of 4,000 compounds, 3F8 was found to phenocopy the 'noeyes' embryonic zebrafish phenotype observed in cases of Wnt overexpression, as the result of Gsk3 inhibition [48]. Subsequent analysis has shown 3F8 to be more selective and potent than the previously used GSK3 inhibitors, suggesting increased potential for research and clinical application [48].
These studies demonstrate the advantages provided by the zebrafish model as a platform for conducting large scale screens for potential molecules that target stem cell development, hematopoietic differentiation, and disease related mechanisms. Smallmolecule screens have proven invaluable to the discovery and evaluation of chemicals displaying potential for clinical research and as reagents for translational research.

Hematopoiesis in the zebrafish and mammals
Zebrafish hematopoietic development occurs in two waves, an embryonic and a definitive wave, and seems to be highly conserved in mammals ( Figure 3) [49,50]. The zebrafish embryonic wave initiates at the 13 somite stage when hemangioblasts develop. This process is com parable to mammalian primitive hematopoiesis, which takes place in the yolk sac mesodermal cells [51]. The cells arising from these tissues are the early progenitors of endothelial and hematopoietic cells. The further differentiation of these tissues occurs early in develop ment (about 15 hours postfertilization (hpf )) in zebrafish and about 19 days postfertilization (dpf ) in humans) [51]. In zebrafish, this differentiation is characterized by two stripes of lateral mesoderm that converge toward the midline before fusing to form the blood island [51]. The blood island serves as the functional equivalent of the mammalian yolk sac and is the developmental site of primitive erythrocytes and some myeloid components [52]. At 36 hpf, HSCs are formed in the ventral wall of the dorsal aorta in a similar manner to that seen in other vertebrates, a process that occurs at day 27 in human development [49]. This HSC formation, in the aorta gonad mesonephros (AGM) region of each organism, marks the beginning of the definitive wave of hemato poiesis, with the majority of these cells functioning as progenitors and a few others acquiring selfrenewal ability. The zebrafish definitive wave continues in the caudal hematopoietic tissue (CHT; about 3 dpf ) before seeding the kidney (about 4 dpf ), whereas in humans the definitive wave continues in the fetal liver and placenta (about 35 dpf ) before seeding the spleen, thymus, and bone marrow [53,54]. The ability to study primitive and definitive hematopoiesis in an externally fertilized, and thus more accessible, vertebrate species has facilitated the dissection of several signaling pathways regulating hematopoiesis.

Hematopoietic stem cell development and emergence
The ontogeny of HSCs has been a major focus of research in the blood research community. Use of the cd41GFP zebrafish transgenic line has shown that HSCs are first derived in the AGM region and are marked by CD41 positivity [55,56]. Further analysis using the cd41GFP line has led to the observation that CD41positive cells exist in two distinct populations, which are manifested as GFP (hi) or GFP (lo) cells in this system [56]. After sorting by flow cytometry, each CD41 population was evaluated for longterm engraftment and multilineage reconstitution in sublethally irradiated zebrafish. The resulting data indicate that cd41GFP (lo) cells represent true HSCs, as these cells are capable of both engraftment and long term sustainment of the hematopoietic program [56].
The origin of HSCs has long been an important topic in the hematopoietic field. However, recent advances in zebrafish live imaging technology have provided new insights into HSC emergence from the AGM region. Transgenic zebrafish with redlabeled endothelial cells and greenlabeled blood cells have been used to directly visualize the budding process of HSCs from aorta endo thelial cells [5759]. Using the kdrGFP transgenic zebra fish line, which drives GFP expression under the control of the kdrl gene promoter in vasculature starting at 18 hpf, timelapse fluorescence confocal microscopy revealed endothelial cells emerging from the aortic floor and entering the subaortic space starting at 30 hpf, a process that has been termed endothelial hematopoietic transition (EHT) [59]. The emergent kdrGFP + cells are morphologically consistent with hematopoietic progenitor cells and are shown to seed the CHT (35 hpf ) and thymus (3 dpf ). runx1 morpholino knockdown in the kdrGFP line has also demonstrated that the EHT event is a Runx1dependent process, as the budding process does not occur in the absence of Runx1 [58,59].  [45] to identify 82 compounds that influence hematopoietic stem cell differentiation, the most prominent of which was dimethyl prostaglandin E2 (dmPGE2). Modified with permission, from [45].

Fate mapping in the zebrafish
One of the greatest attributes of the zebrafish model is the ability to trace hematopoietic cell fate as differen tiation occurs in the embryo. Caged fluorescein dye, which changes color in response to a laser pulse, can be injected into embryos [6163]. Laser activation of single cells, or groups of cells, allows the tracking of individual cell derivation over time. This technique has been particularly useful in the study of HSC development within the aorta. 'Uncaged' HSCs were followed as they colonized the CHT. The cells arising from the CHT then seeded the thymus and the kidney [64]. In zebrafish, the kidney serves as the primary site of larval and adult hematopoiesis [50]. Analysis of fluorescently labeled, mutant, or morphant (morpholino knockdown) cells has enabled the investigation of cell migration and develop ment. This has led to the discovery of chemokine receptors that are responsible for thymus colonization in the zebrafish [65]. In addition, fate mapping can now make use of transgenic zebrafish containing a CreErt2 (mutated estrogen receptor) construct that, when initiated, switches the expression of an integrated construct from the green label GFP to the red label DsRed in specific cells or tissues [66]. The progeny of these switched cells maintain DsRed expression and are easily traced through development. These studies have enabled visualization of the hematopoietic system at significant resolution and have been extremely useful for defining the sites of zebrafish hematopoiesis.
In the zebrafish, blastula transplantation provides a model for examining cell autonomy in many cell types, including HSCs [67]. Mutant or morphant cells are injected with a fluorescent dye and then transplanted into a wildtype embryo or vice versa [68]. The implanted cells are tracked using their fluorescence. Transplantation of a fluorescent mutant cell that results in the lack of

HSC induction
fluorescent blood indicates that the gene acted in a cell autonomous manner. More recently, this technique has been improved to allow transplantation of blastula cells from a mybGFP donor. This transgenic line contains a BAC expressing GFP under the control of a myb promoter, which marks donor cells as they form HSCs in the dorsal aorta [45]. These cells are then injected into a recipient containing a red fluorescent protein (RFP) construct that labels the vasculature red. The derivation of green cells adjacent to the red endothelial cells indi cates autonomous effects of stem cell production. These techniques allow the tracking of individual cells, which is very informative in the study of such a dynamic system.

Blood diseases in zebrafish
The zebrafish model has been used in the discovery of many new compounds with potential for clinical and thera peutic applications (Table 2), including several zebra fish cancer models that have been introduced over the past few years. These models are generally easy to mani pu late and study while showing high genetic similarity to human cancer lines [69]. One such model uses a conditional Cre/loxregulated system under the control of a heat shock promoter that drives rag2 expression in developing T cells [70,71]. Several recent publications have investigated this system in the study of TALL and cancer biology. A recent TALL study found that high levels of the apoptosis regulator Bcl2, the Gcoupled protein receptor S1p1, and the cell adhesion protein Icam1 blocked tumor cell intravasation, an important initial step in metastasis [72]. In addition, results obtained using the zebrafish model have allowed the differences between human Tcell lymphoblastic lymphoma (TLBL) and human TALL to be defined according to their cellular and molecular components. Currently, human T LBL and TALL are treated with the same regimens; however, these data have demonstrated key molecular differences that could allow more targeted treatments in the future [72].
The characterization of the ferroportin gene by zebra fish gene cloning is a prime example of the relevance of the zebrafish model for the discovery of diseaserelated genes [21]. Ferroportin was mutated in the weissherbst mutant and, using this model, was found to be the iron transporter responsible for delivering maternally derived iron from the yolk to the embryo. Human placental cells have since been found to express ferroportin [73]. Thus, maternal iron delivery to the fetus by ferroportin has been evolutionarily conserved for 300 million years. Furthermore, anemia of chronic disease has been linked to this gene through the ligand hepcidin, which binds ferroportin and promotes its internalization. Dysregula tion of this pathway can lead to hemochromatosis, an iron imbalance disorder [74]. Ferroportin mutations have been found in several patients with hemochromatosis, and this illustrates how studies of a zebrafish mutant have contributed to the definition of a human disease. More recently, mitoferrin and glutaredoxin 5 have also been linked to iron defects.
Since its discovery as an enhancer of HSC development in zebrafish, dmPGE2 is advancing towards clinical use. A clinical trial is currently analyzing dmPGE2 and its potential for enhancing engraftment in cord stem cell transplants. In that trial, leukemia or lymphoma patients are recruited and treated with highdose chemotherapy before being transplanted with two independent cord blood samples. One of the cords is pretreated with dmPGE2, and following transplantation the level of chimerism is evaluated to determine which is the dominant cord. Thus, the trial will investigate whether dmPGE2stimulated cells might display better engraft ment capability over time, a result that could greatly increase the efficacy of cord blood and bone marrow transplantation in humans.

Lessons from the zebrafish model
Through mass mating procedures, the zebrafish can be used in various highthroughput genomic techniques that have not been possible with other vertebrate models. The advantages provided by the zebrafish in visualization, fate mapping, and early embryonic development contri bute greatly to cell biological studies, particularly as they pertain to early hematopoietic development and HSCs. In addition, mutagenesis, chemical, and other largescale screens are important methods for the discovery of novel pathways and potential therapeutics targeting hematopoiesis.
As mentioned, transplantation assays have also been developed in the zebrafish [75,76]. The first marrow transplants were performed using GFPpositive whole kidney marrow transplanted into irradiated adults. GFP positive blood cells can be seen in the host up to 6 months after transplantation. Serial transplantation has also demonstrated effectiveness, as recipients retain GFPpositive blood for months after transplant. More recently, competitive repopulation studies between red and green fluorescently tagged marrow cells have been performed in the Casper line, in which marrow cells are pretreated with a chemical and evaluated for competitive advantage [60]. Through the use of this technology, chemicals can be screened to assess their ability to enhance transplantation, and thus to enhance the robustness of HSC development, engraftment, and retention.

Implications for translational stem cell research
Recent advances in epigenetic and sequencing technolo gies, particularly the development of ChIPseq and RNA seq, have allowed the investigation of molecular interactions on a genomewide scale [77]. Recently, the genomewide binding sites of the essential hematopoietic transcription factors Gata1, Gata2, Runx1, Fli1, and Scl in human megakaryocytes were identified [78]. Analysis revealed 144 regions representing 151 candidate genes that showed simultaneous binding of all five factors. Of these genes, 18 had known functions in hematopoiesis, and the zebrafish model was then used to further investigate these genes. Eight genes were chosen at random and targeted for knockdown using morpholinos. In each case, morpholino injection caused a significant reduction in erythrocyte, thrombocyte, and/or HSC number. This study demonstrates the effectiveness of the zebrafish model in validating results found in other organisms using a highthroughput in vivo system [78].
The use of ChIPseq analysis has also led to resolution of the molecular interplay among external signaling transcription factors and cellspecific regulators during hematopoietic regeneration. In a recent study using a combination of zebrafish, murine, and human inputs, the BMP and Wnt signaling pathways were shown to be essential for hematopoietic regeneration following acute hematopoietic injury [79]. In this study, ChIPseq analysis demonstrated that Smad1 and Tcf7l2 cooccupy sites with cellspecific master regulators in a dynamic manner throughout differentiation. These data suggest that the hematopoietic program is coordinated by a finely tuned collaboration between master regulators and external signaling factors, in which master regulators direct the binding profiles of the signaling transcription factors.
In addition to serving as an effective chemical screening platform, the zebrafish model has shown promise as an efficient means of prescreening small molecules for drug candidacy. A recent study evaluated the specificity of three molecules that are known to inhibit pololike kinase 1 (Plk1) in vitro, a protein that is overexpressed in many tumors and thus is considered a potentially important target for cancer therapy [80]. Analysis of Plk1 has revealed high conservation between the zebrafish and human homologs, including a nearly identical active site composition [81]. The study investigated the Plk1 inhibitors LFMA13, ON01910, and thiazolecarboxa mide 10A to determine which molecule provided the most specific and effective inhibition in vivo. The embry onic phenotypes resulting from each chemical treatment were compared with the phenotype resulting from direct morpholino knockdown of Plk1. The results indicated that whereas each inhibitor showed promise in vitro, only one, thiazolecarboxamide 10A, selectively inhibited Plk1 in vivo. This result highlights the difficulty associated with the discovery of drug candidates through in vitro methods, as well as the significant advantage provided by using the zebrafish model to prescreen potential therapeutics in vivo [80].

Conclusions and future directions
The zebrafish model provides a tremendous balance between scale and applicability. The ease of mutagenesis, high fecundity, and visualization techniques, in conjunc tion with the largely conserved hematopoietic system that the zebrafish provides, allow largescale genomic analysis while maintaining relevance in higher organisms. The definition of genes involved in TALL and hypo chromic anemia, and the discovery and assessment of dmPGE2, thiazolecarboxamide 10A, and 3F8 have demonstrated the relevance of the zebrafish model for clinical and therapeutic research. This model will con tinue to help define genetic and epigenetic mechanisms in blood cells using the highthroughput procedures ChIPseq, RNAseq, and morpholino screening. Further studies of HSC development, selfrenewal, and differen tiation using the zebrafish model have great potential to contribute to advances in the treatment and management of numerous blood diseases and cancers.

Competing interests
LIZ is a founder and stockholder of Fate, Inc., and a scientific advisor for Stemgent.