The 'Definitive' (And 'Primitive') Guide to Zebrafish Hematopoiesis

The ‘deﬁnitive’ (and ‘primitive’) guide to zebraﬁsh hematopoiesis

Alan J Davidson1 and Leonard I Zon*,1

1Division of Hematology/Oncology, Department of Medicine, Children’s Hospital and Dana Farber Cancer Institute, Howard Hughes Medical Institute, Boston, MA 02115, USA

Progressive advances using zebrafish as a model organism to bilateral stripes of cells in the posterior meso- have provided hematologists with an additional genetic derm. These cells migrate to the trunk midline system to study blood cell formation and hematological and differentiate into ‘primitive’ proerythroblasts malignancies. Despite extensive evolutionary divergence and endothelial cells of the trunk vasculature. In between bony fish (teleosts) and mammals, the molecular cross-section, the endothelium of the axial vein pathways governing hematopoiesis have been highly encapsulates the converged mass of erythroid cells, conserved. As a result, most (if not all) of the critical thereby resembling the cellular architecture of the hematopoietic transcription factor genes identified in mammalian yolk sac blood island (Al-Adhami and mammals have orthologues in zebrafish. As in other Kunz, 1977; Willett et al., 1999). vertebrates, all of the teleost blood lineages are believed to originate from a pool of pluripotent, self-renewing hematopoietic stem cells. Here, we provide a detailed Ventral mesoderm patterning review of the timing, anatomical location, and transcrip- With the advent of modern molecular and genetic tional regulation of zebrafish ‘primitive’ and ‘definitive’ techniques, it has been possible to develop a more hematopoiesis as well as discuss a model of T-cell comprehensive understanding of the origin and forma- leukemia and recent advances in blood cell transplantation of the ICM. Fate mapping analyses have demon- tion. Given that many of the regulatory genes that control strated that the first circulating erythroid cells descend embryonic hematopoiesis have been implicated in onco- from mesoderm on the ventral side of the early gastrula genic pathways in adults, an understanding of blood cell stage embryo (Kimmel, 1990; Warga and Nusslein- ontogeny is likely to provide insights into the patho- Volhard, 1999). Following the extensive morphological physiology of human leukemias. movements of gastrulation, ventral mesoderm even- Oncogene (2004) 23, 7233–7246. doi:10.1038/sj.onc.1207943 tually comes to occupy the posterior and lateral regions of the embryo (Figure 2). In addition to blood and Keywords: zebrafish; blood; hematopoietic stem cell; vascular precursors, the posterior mesoderm also gives embryo; leukemia; hematopoiesis rise to the embryonic kidney (pronephros) that forms as bilateral stripes of cells lateral to the ICM precursors (Figure 3e). How the ventral mesoderm is patterned into different cell fates is still largely unclear. However, secreted ‘Primitive’ hematopoiesis signaling molecules belonging to the bone morphoge- netic protein (BMP), fibroblast growth factor (FGF), Successive waves of ‘primitive’ and ‘definitive’ hemato- and Wnt families have been implicated as early inducers poiesis, occurring in anatomically distinct sites, are and patterning factors of ventral mesoderm (reviewed characteristics of vertebrate embryos (Figure 1 and by Mun˜ oz-Sanjua´ n and Hemmati-Brivanlou, 2001). At reviewed by Galloway and Zon, 2003). In mammals and present, it is widely believed that a morphogenic birds, ‘primitive’ hematopoiesis occurs in the extraem- gradient of BMP signaling across the dorsal–ventral bryonic yolk sac blood islands and predominantly axis during gastrulation is primarily responsible for generates erythroid cells. The equivalent site in zebrafish patterning mesoderm into discrete cell fates. A recent is located intraembryonically and is known as the re-evaluation of the Xenopus fate map has shown that intermediate cell mass (ICM). Historically, the forma- the dorsal–ventral axis of the early gastrula embryo tion of the ICM has been studied for over 130 years more accurately reflects the arrangement of tissues in a number of diverse fish species (Oellacher, found along the anterior–posterior axis (Lane and 1872; Stockard, 1915; Al-Adhami and Kunz, 1976; Sheets, 2000; Lane and Sheets, 2002). In general, Al-Adhami and Kunz, 1977). These early light mesoderm on the dorsal side of the early gastrula gives microscopy studies traced the origin of the ICM rise to anterior structures such as the head, while ventral mesoderm generates more posterior tissues. These *Correspondence: L Zon, Children’s Hospital, KARP7, 1 Blackfan observations are also consistent with the zebrafish fate Circle, Boston, MA 02115, USA; E-mail: [email protected] map (Kimmel, 1990; Warga and Nusslein-Volhard, Zebrafish hematopoiesis AJ Davidson and LI Zon 7234 Transcriptional regulation of ICM-derived hematopoietic stem cells (HSCs)

Many of the transcription factor genes that play a critical role during mammalian hematopoiesis have been identified in zebrafish (Table 1) and have spatiotemporal expression patterns comparable to their mammalian counterparts. The stem cell leukemia (scl) gene, encoding a basic helix–loop–helix transcription factor, has been extensively characterized in both mammals and zebrafish, largely due to its early and essential role in HSC formation (Robb et al., 1995; Shivdasani et al., 1995) and its involvement in acute lymphoblastic leukemia Figure 1 Approximate time of appearance and duration of hematopoietic activity in different tissues during zebrafish embry- (Begley et al., 1989; Brown et al., 1990). It is important ogenesis (adapted from Galloway and Zon, 2003). to note, however, that scl is also an early molecular ICM ¼ intermediate cell mass; AGM ¼ aorta-gonad-mesonephros marker of vascular precursors (angioblasts). In the region equivalent (corresponding to the ventral wall of the dorsal posterior mesoderm, expression of scl is first evident aorta); hpf ¼ hours postfertilization; dpf ¼ days postfertilization around the 2-somite stage in bilateral stripes of cells flanking the paraxial mesoderm at the future level of somite six (Davidson et al., 2003 and Figure 3f). This expression pattern of scl overlaps with other transcription factor genes, including cdx4, fli1a (previously known as fli1), gata2, hhex, lmo2, spadetail/tbx16, and tif1g (Table 1 and Figure 3b, c). The coexpression of scl, lmo2,andgata2 is consistent with biochemical studies showing that these transcription factors can interact to form a multimeric complex (Valge-Archer et al., 1994; Wadman et al., 1997) and overexpression experiments have suggested crossregula- tory interactions between scl, hhex,andfli1a (Gering et al., 1998; Liao et al., 2000b; Gering et al., 2003). Whether the scl þ ICM precursors at this stage represent hemangioblasts (a common progenitor to HSCs and angioblasts) or independently derived populations of HSCs and angioblasts that simultaneously develop adjacent to each other is currently unresolved. Aside from dorsal–ventral patterning mutations, which glob- Figure 2 Schematic representation of zebrafish embryos during gastrulation and somitogenesis showing the rostral and caudal sites ally affect the patterning of ventral mesoderm, only of ‘primitive’ hematopoiesis (blue stripes). ICM precursors are the mutants cloche, spadetail/tbx16, and kugelig/cdx4 derived from a subset of ventral mesoderm (red) during gastrula- display defects at this early stage of HSC formation tion, which forms the posterior lateral mesoderm. RBI precursors (see below). are derived from more anterior lateral mesoderm. Both blood cell The expression pattern of scl in ICM precursors at the populations migrate during subsequent stages of development (indicated by white arrows). At the 20-somite stage (bottom right 3-somite stage overlaps with that of pax2.1 and pax8 embryo shown in lateral view), blood cells from the RBI are found (AJD and LIZ, unpublished results), encoding two scattered across the anterior yolk (blue spots), while blood cells homeobox genes required for pronephros formation from the ICM have converged at the trunk midline (blue stripe). (Majumdar et al., 2000; Bouchard et al., 2002). Similar For clarity, only mesoderm is shown for early and late gastrula- stage embryos. The 5-somite stage embryo is shown removed from observations have been found in the dorsolateral plate the yolk and flattened (flatmount). Black arrows indicate gastrula- mesoderm of Xenopus laevis embryos (Walmsley et al., tion movements at the marginal zone. A ¼ anterior, P ¼ posterior, 2002), suggesting that ICM precursors are derived from V ¼ ventral, D ¼ dorsal mesoderm that has initially activated a kidney gene program. The significance of this finding is unclear because by the 5-somite stage, the posterior mesoderm is clearly subdivided along the medial–lateral axis into 1999) and suggest that dorsal–ventral patterning of nonoverlapping bilateral stripes of scl þ ICM precursors the posterior mesoderm (equating to medial–lateral and pax2.1 þ kidney progenitors (Figure 3e). cloche cell fates at the end of gastrulation) occurs along mutants fail to express scl at this time and display the animal–vegetal axis. This hypothesis is still com- normal kidney stripe patterning (Figure 3f), suggesting patible with the BMP gradient model, although it that scl is not required to downregulate pax2.1 does require that the gradient is reoriented along transcription. The transient coexpression of kidney the animal–vegetal axis rather than the classical and ICM precursor genes may reflect the progressive dorsal–ventral axis. refinement in gene expression domains associated with

Oncogene Zebraﬁsh hematopoiesis AJ Davidson and LI Zon 7235

Figure 3 Flatmounted embryos at the 5-somite stage stained by whole mount in situ hybridization for genes expressed in the rostral and caudal mesoderm. (a) Expression of scl (red) and gata1 (purple). Transcripts for gata1 are found in a subset of scl þ cells in the posterior. (b) Expression of lmo2. Note the wider expression domain of lmo2 in the posterior stripes, most likely including pronephric duct precursors. (c) Expression of gata2. In addition to blood and vascular precursors, transcripts for gata2 are also found in the ectoderm. (d) Expression of pu.1 in the RBI. (e) Expression of scl (red) and pax2.1 (purple) in mutually exclusive expression domains. (f) Expression of paraxis1 in the paraxial mesoderm (par1; purple), pax2.1 in the pronephros (purple), and scl (red) in ICM precursors of wild-type (wt) and cloche (clo) mutants. (g) Schematic representation of a flatmounted 5-somite stage embryo showing the arrangement of blood and vascular precursors in the RBI and ICM ventral mesoderm patterning, and may not necessarily erythroid cell function (Figure 4). Of the transcription represent cells with tri- or bi-potential fates. factor genes, functional requirements for ICM erythropoiesis have been demonstrated for gata1 (vlad tepes), ‘Primitive’ erythropoiesis tif1g (moonshine), and biklf (Table 1). As with other tissues in the embryo, blood cell differentiation proceeds The first unambiguous indication of hematopoietic sequentially in a rostral to caudal direction. The commitment occurs at the 4-somite stage with the anterior-most erythroid precursors begin migrating expression of the erythroid-specific transcription factor towards the trunk midline at the 10-somite stage, gata1 (Detrich et al., 1995) in a subset of scl þ cells followed progressively by more posterior cells. As the (Figure 3a and Davidson et al., 2003). Transgenic blood precursors ‘zip-up’ along the trunk, they also reporter lines carrying the green fluorescent protein undergo extensive proliferation. Thus, by the time blood (gfp) gene under the control of the gata1 promoter circulation begins between 25–26 hpf, the ICM is (Long et al., 1997) and limited fate mapping analysis comprised of at least 300 proerythroblasts (Long et al., (Lieschke et al., 2002) have confirmed that these cells 1997) that express genes required for heme and globin ultimately give rise to the first circulating blood cells. chain synthesis (a and b embryonic globins, alas2, fch, The remaining ICM precursors that fail to express gata1 urod), iron utilization (dmt1), and membrane stability are most likely angioblasts, as they differentiate into (b-spectrin, protein 4.1r; reviewed by Wingert and Zon, endothelial cells that express the vascular endothelial 2003). growth factor receptor flk1/vegfr2 (Liao et al., 1997; Over the next 4 days the circulating proerythroblasts Sumoy et al., 1997; Davidson et al., 2003). Taken synchronously mature into flattened elliptical erythro- together, these observations suggest that ICM precur- cytes (Al-Adhami and Kunz, 1977; Detrich et al., 1995; sors adopt either a HSC or a vascular fate by the 5- Weinstein et al., 1996). Surprisingly, completely anemic somite stage (Figure 3g). mutants are viable for at least 2 weeks postfertilization, Following the formation of presumptive HSCs, there suggesting that blood cell-mediated oxygen transport is a sequential activation of genes important for is not critical until relatively late stages of larval

Oncogene Zebraﬁsh hematopoiesis AJ Davidson and LI Zon 7236 Table 1 Zebraﬁsh hematopoietic transcription factor genes Gene Family Loss-of-function Reference

scl Basic helix–loop–helix Morphant Dooley et al. (in press) c/ebp1 bZIP — Lyons et al. (2001) c/ebpa bZIP — Lyons et al. (2001) c/ebpb bZIP — Lyons et al. (2001) cbfb Core binding factor — Blake et al. (2000) runx1 Core binding factor Morphant Kalev-Zylinska et al. (2002) runx3 Core binding factor Morphant Kalev-Zylinska et al. (2003) ets1 ETS — Zhu et al. (submitted) mef ETS — Zhu et al. (submitted) pu.1 ETS — Bennett et al. (2001) fli1a ETS — Brown et al. (2000) cdx4 Homeobox kugelig Davidson et al. (2003) hhex Homeobox Deletion b16 Liao et al. (2000b) ldb1 LIM domain binding — Toyama et al. (1998) lmo2 LIM domain — Thompson et al. (1998) pax5 Paired domain, homeobox — Pfeffer et al. (1998) tif1g RBCC, PhD, Bromo moonshine Ransom et al. (in press) spt/tbx16 T-box spadetail Thompson et al. (1998) stat3 STAT — Oates et al. (1999) c-myb Zinc-finger Deletion b316 Thompson et al. (1998) draculin Zinc-finger — Herbomel et al. (1999) fog1 Zinc-finger — Nishikawa et al. (2003) gata2 Zinc-finger — Detrich et al. (1995) gata1 Zinc-finger vlad tepes Detrich et al. (1995) ikaros Zinc-finger — Willett et al. (2001) klfd Zinc-finger — Oates et al. (2001) klf4 Zinc-finger Morphant Oates et al. (2001) klf12 Zinc-finger — Oates et al. (2001) nfe2 Zinc-finger — Pratt et al. (2002)

development. Although the lifespan of ‘primitive’ erythroid cells is unknown, transfusion experiments have demonstrated that labeled proerythroblasts taken from 36 hpf embryos will persist in the circulation of unlabeled host embryos for at least 10 days. However, there is a gradual decline in donor blood cell numbers beginning at 6 days postfertilization (dpf), reaching approximately 50% of their starting values at 10 dpf (Weinstein et al., 1996). The bloodless (bls) mutant, which appears to have a specific defect in ICM hematopoiesis, starts to recover circulating erythroid cells after 5 dpf (Liao et al., 2002). This has led to the suggestion that a second wave of erythropoiesis begins around this time, perhaps originating from ‘definitive’ HSCs that are resident in the pronephros (discussed below). Unlike ‘definitive’ red blood cells in mammals, teleost erythrocytes remain nucleated for the lifespan of the fish.

Myelopoiesis and the rostral blood island A second site of hematopoiesis initiating in the anterior mesoderm (herein referred to as the rostral blood island or RBI) has recently been characterized in zebraﬁsh (Figure 3). Although earlier light microscopy studies Figure 4 Temporal expression proﬁles of hematopoietic transcription factor genes expressed in ICM precursors during somitogenesis had found ‘primitive amoeboid’ cells originating from (1–30 somite stages). An asterix indicates that the exact onset of the head mesenchyme, these cells were not considered to expression has not been reported be hematopoietic in nature (Rieb, 1973; Al-Adhami and

Oncogene Zebrafish hematopoiesis AJ Davidson and LI Zon 7237 Kunz, 1977). Using Nomarski video-microscopy, fate circulation via the ducts of Cuvier, the yolk sac blood mapping, and molecular marker expression, these channels that connect the trunk vasculature to the heart, amoebocytes were identified as macrophages and their while others migrate into the head (Herbomel et al., origin traced back to the lateral mesoderm of the head 1999). Although pu.1 expression levels in these macro- at the 3-somite stage (Herbomel et al., 1999; Bennett phage precursors decline after the 18-somite stage, et al., 2001; Lieschke et al., 2002). Interestingly, vascular additional molecular markers including c/ebp1, fms, cells are also derived from a similar region of the l-plastin, and lysozyme-c become expressed at this time cephalic mesoderm, indicating that the common spatial (Herbomel et al., 1999; Bennett et al., 2001; Herbomel and temporal characteristics of blood and vascular et al., 2001; Lyons et al., 2001; Lieschke et al., 2002; Liu development are features shared by both rostral (RBI) and Wen, 2002). Functional macrophage activity in the and caudal (ICM) sites of hematopoiesis. Consistent ducts of Cuvier has been observed as early as 26 hpf, with this, transcripts for scl, lmo2, gata2, and fli1a are where macrophages can be seen ‘interrogating’ and/or found in the RBI between the 3- to 5-somite stages engulfing erythroblasts from the bloodstream (Herbo- (Figures 3 and 5; Liao et al., 1998; Thompson et al., mel et al., 1999). In addition, macrophages in embryos 1998; Brown et al., 2000). This is followed soon after by challenged with Escherichia coli or Bacillus subtilis at 30 the expression of the myeloid-specific transcription hpf are able to migrate to the site of infection and factor pu.1 (also known as spi1), in a subset of scl þ remove the bacteria by phagocytosis (Herbomel et al., RBI precursors (Bennett et al., 2001; Lieschke et al., 1999). Following the onset of blood circulation, macro- 2002). Thus, in a similar fashion to the ICM, cells in the phages expressing l-plastin can be found dispersed RBI rapidly acquire either a hematopoietic (myeloid) or throughout 28–32 hpf embryos as well as accumulating vascular cell fate (Figure 3g). Transcripts for gata1 are within, and surrounding, the ventral region of the tail not observed in the RBI, consistent with this hemato- vein (Herbomel et al., 1999; Willett et al., 1999). It is not poietic site being restricted to myelopoiesis, whereas clear if these ventral tail cells are derived from pu.1 expression is found at a low level in ICM blood circulating macrophages or instead represent a small precursors between the 10- and 19-somite stages population of ICM-derived macrophages. Interestingly, (Bennett et al., 2001; Lieschke et al., 2002). Coexpres- cells in this region of the tail rapidly take up latex beads sion of pu.1 and gata1 in presumptive erythroid or carbon particles injected into the bloodstream, progenitors in the ICM is not unexpected, given the leading to the proposal that the ventral tail region may role for pu.1 in erythroid progenitor proliferation be functionally analogous to the reticuloendothelial (Fisher et al., 2004). system of mammals (Kimmel et al., 1995; Lieschke et al., Between the 11- to 15-somite stages, pu.1 þ macro- 2001). At present, it is not known if the macrophages phage precursors migrate toward the head midline and originating from the RBI represent a transient wave form a partially converged mass of cells between the eye of ‘primitive’ macrophages that is later replaced by and the heart primordium (Herbomel et al., 1999; monocyte-derived macrophages, or if they are perma- Lieschke et al., 2002; Ward et al., 2003). These cells nently retained into adulthood. then make an abrupt 1801 change of direction and migrate laterally, scattering into single cells across the Granulopoiesis yolk sac. Some of these macrophage precursors enter Zebrafish, like other teleosts, have at least two types of granulocytes: heterophilic or neutrophilic granulocytes and eosinophilic granulocytes (also known as eo/ basophils as they exhibit characteristics of both mammalian eosinophils and basophils; Figure 6). Granulo- cytes containing characteristic neutrophilic granules can be morphologically detected in the circulatory system and in loose connective tissue at 48 hpf (Willett et al., 1999; Lieschke et al., 2001). In contrast, no cells containing the granules of eosinophil granulocytes can be detected in embryos up until 5 dpf, the latest stage that has been examined (Lieschke et al., 2001). Furthermore, no eosinophil-specific molecular markers have been isolated to date, thus the origin and developmental stage at which eosinophilic granulocytes appear during embryogenesis is unknown. Expression of the granulocyte-specific marker myeloperoxidase (mpo/mpx; encoding an enzyme abundant in the granules of mature neutrophils and eosinophils) is first detected at 18 hpf in presumptive neutrophilic Figure 5 Temporal expression profiles of hematopoietic transcription factor genes expressed in RBI precursors during somitogenesis precursors within the caudal ICM. Soon after this stage, (1–30 somite stages). An asterix indicates that the exact onset of mpo-expressing cells also appear scattered on the expression has not been reported anterior yolk sac (Bennett et al., 2001; Lieschke et al.,

Oncogene Zebraﬁsh hematopoiesis AJ Davidson and LI Zon 7238

Figure 6 ‘Definitive’ hematopoiesis in adult zebrafish kidney and morphological comparison of human and zebrafish blood cells. (a) Live image of an adult zebrafish male (2.5 cm in length). (b) Para-sagital section of an adult zebrafish stained with hematoxylin and eosin showing the location of the kidney (outlined). (c) Higher magnification view of the head kidney showing mesonephric tubules (PT) surrounded by small blue- and red-stained hematopoietic cells (white asterix). (d) Comparison of human and zebrafish mature peripheral blood cells stained with Wright Giemsa

Table 2 Zebraﬁsh models of hematological diseases in humans Mutant Gene Human disease Reference

dracula ferrochelatase Erythropoietic protoporphyria Childs et al. (2000) merlot protein 4.1r Hemolytic anemia Shaﬁzadeh et al. (2002) retsina solute carrier family 4, anion exchanger 1 Congenital dyserythropoietic anemia type II (HEMPAS) Paw et al. (2003) riesling b-spectrin Hereditary spherocytosis Liao et al. (2000a) sauternes g-aminolevulinate synthetase Congenital sideroblastic anemia Brownlie et al. (1998) vlad tepes gata1 Familial dyserythropoietic anemia and thrombocytopenia Lyons et al. (2002) weissherbst ferroportin1 Hemochromatosis type 4 Donovan et al. (2000) yquem uroporphyrinogen decarboxylase Porphyria tarda and hepatoerythropoietic porphyria Wang et al. (1998) zinfandel globin locus Hypochromic anemia Brownlie et al. (2003)

2001). At least some of these mpo þ cells coexpress pu.1, HSC mutants suggesting that they are derived from the RBI. The detection of mpo þ cells in the caudal ICM provides To date, three large-scale mutagenesis screens have been additional circumstantial evidence that the ICM gives completed in zebrafish, resulting in the isolation of rise to a small population of myeloid progenitors. In at least 26 complementation groups with defects in the contrast to the results obtained with pu.1, very few mpo þ development of the hematopoietic system (reviewed by cells coexpress l-plastin, consistent with these two Wingert and Zon, 2003). The majority of these populations representing distinct granulocyte and mutations affect relatively late stages of ICM erythro- macrophage lineages (Bennett et al., 2001). By 3 dpf, poiesis and many represent models of human disease mpo-expressing cells are scattered throughout the (Table 2). In contrast, only three mutants are thought to embryo with a prominent aggregation in the ventral be defective in the formation of HSCs: cloche (clo), tail vein region (Bennett et al., 2001; Lieschke et al., spadetail (spt), and kugelig (kgg). 2001). Histochemical staining for myeloperoxidase activity yields similar results to the gene expression cloche studies and has been used to demonstrate the rapid mobilization of granulocytes to sites of trauma induced The clo mutant, characterized by a severe deficit in by tail clipping (Lieschke et al., 2001). It is not known blood and vascular cells, was originally identified as a how long these embryonic granulocytes persist in the spontaneous mutation within a population of semiwild embryo or if they contribute to the granulocyte fish from an Indonesian fish farm (Stainier et al., 1995). population that is found in the pronephros at 7 dpf A lack of endocardium in clo hearts results in an (Willett et al., 1999). abnormally enlarged atrium, giving the heart a bell (or

Oncogene Zebrafish hematopoiesis AJ Davidson and LI Zon 7239 ‘cloche’ in French) shape. clo mutants have been found in ventral mesoderm during gastrulation and extensively characterized using a wide range of HSC later in the bilateral ICM precursors, are absent in spt and angioblast genes including scl (Liao et al., 1998), mutants from the 3-somite stage onwards (AJD and lmo2 (Thompson et al., 1998), gata2 (Thompson et al., LIZ, unpublished results). Hence, in addition to 1998), runx1 (Kalev-Zylinska et al., 2002), fli1a (Brown regulating proper cell migration during gastrulation, et al., 2000), and flk1 (Liao et al., 1997). These analyses spt may also be required for the commitment of ventral have shown that the clo defect results in an almost mesoderm to an HSC fate. Interestingly, overexpression complete absence of HSCs and angioblasts in the RBI of scl in spt mutants can rescue erythroid development and ICM. The clo defect appears specific to the (Dooley et al., 2004, in press), further suggesting that spt hematopoietic and vascular lineages, as adjacent tissues, acts upstream of scl during ICM-HSC formation. such as the pronephros and somites, appear to form normally in the mutants (Figure 3). A small number kugelig (10–20) of presumptive angioblasts and blood cells can be detected around the tailbud from the 5-somite stage The kgg mutant was identified in the first large-scale onwards, indicating that the clo mutation does not mutagenesis screen performed in Tu¨ bingen (Haffter completely prevent hematovascular development. While et al., 1996; Hammerschmidt et al., 1996) and is transplant experiments have revealed a cell autonomous characterized by a shortened tail and reduced yolk tube requirement for clo during vascular development, there extension (the constricted portion of the yolk sac that appears to be both cell autonomous and noncell forms beneath the trunk). As a result, the mutant autonomous requirements for clo during blood cell remains relatively spherical (or ‘kugelig’ in German) in development (Parker and Stainier, 1999). Rescue experi- appearance. In addition to the tail defects, kgg embryos ments have shown that the clo defect can be overcome are also severely anemic. Similar to spt mutants, by overexpressing scl (Liao et al., 1998; Kalev-Zylinska angioblasts and RBI-derived myeloid cells form nor- et al., 2002) but not bmp4 (Liao et al., 2002), suggesting mally in kgg mutants (Davidson et al., 2003). The cdx4 that clo acts downstream of the BMP pathway but gene, belonging to the caudal-related homeobox tran- upstream of scl. Similarly, ectopic expression of runx1 scription factor family (comprising cdx1, cdx2, and cdx4 in clo embryos can expand blood cell numbers in vertebrates), was identified as the defective locus in (Kalev-Zylinska et al., 2002). At present, the mechanism kgg mutants. In vertebrates such as mouse, frog, and by which clo regulates HSC and angioblast formation is chick, members of the cdx family have been implicated unclear. Positional cloning attempts to identify the gene in AP patterning of the embryonic body axis by acting are ongoing but have been hampered by the telomeric as ‘master regulators’ of the Hox genes, another family location of the gene (Liao et al., 2000b). of homeobox transcription factors (Lohnes, 2003). The Hox genes, which in mammals comprise 39 genes spadetail grouped together in four distinct clusters (HoxA, HoxB, HoxC, and HoxD), are expressed in overlapping The spt mutation is caused by a defect in tbx16, expression domains along the AP axis (reviewed by encoding a T-box transcription factor (Griffith et al., Favier and Dolle, 1997). The nested expression patterns 1998; Ruvinsky et al., 1998). Altered cell movements of the hox genes are believed to be important for during gastrulation in spt embryos causes cells that directing where particular tissues form along the AP would normally contribute to the trunk somites to axis. While the molecular mechanism of this patterning instead accumulate in the tailbud (Kimmel et al., 1989). is unknown, it is thought that unique combinations In addition, spt embryos fail to make erythroid cells of Hox genes (the so-called ‘Hox code’) can confer even though they form relatively normal numbers of positional information to segments of the embryo angioblasts, RBI-derived myeloid cells, and pronephric (reviewed by Krumlauf, 1994). duct precursors (Thompson et al., 1998; Amacher et al., Consistent with the role of the cdx family as master 2002; Lieschke et al., 2002). Thus, the hematopoietic hox gene regulators, at least nine hox genes coming from defect in spt mutants appears relatively specific to the the four different hox clusters display perturbed expres- ICM, as opposed to a general absence of posterior tissue sion patterns in kgg mutants (Davidson et al., 2003 and types. Based on the expression of early hematopoietic unpublished results). In general, hox genes expressed in markers such as scl, lmo2, and gata2 during mid- the anterior trunk show expanded expression domains somitogenesis, it was initially assumed that spt mutants towards the posterior, while more caudally expressed form ICM-HSCs, but that these cells are blocked in their hox genes show reduced or absent expression. Con- ability to further differentiate (Thompson et al., 1998; comitant with these perturbations in hox expression Oates et al., 1999). However, given that molecular domains, there is a severe reduction in ICM-HSCs but markers previously considered HSC-specific are also not in the adjacent angioblast population. Part of this expressed by angioblasts, it is equally likely that spt hematopoietic defect may be explained by a ‘homeotic mutants lack ICM-HSCs altogether. Consistent with transformation’ of cell fates (i.e. an expansion of rostral this, extensive coexpression of scl and flk at the 10- cell fates at the expense of more caudal tissue types). somite stage suggests that most, if not all, scl þ cells in However, while the bilateral stripes of blood precursors spt embryos are angioblasts (AJD and LIZ, unpublished in kgg mutants are shorter, they are also considerably results). Furthermore, transcripts for draculin, which are thinner due to a reduced number of scl þ /gata1 þ cells.

Oncogene Zebrafish hematopoiesis AJ Davidson and LI Zon 7240 This is not the result of a general posterior patterning adaptive immune system including T cells, B cells, defect, as the adjacent stripe of pronephric duct antigen-presenting cells, and natural killer-like cells precursors has a thickness comparable to that of wild- (reviewed by Traver et al., 2003a). Most studies, type embryos. These observations suggest an additional however, have been performed in adult fish and less and more specific requirement for hox genes in the is known about the embryonic development of the specification of ICM-HSC fate. Consistent with this, lymphoid system. In zebrafish embryos, a combination hox rescue injections, in which genes such as hoxb7 of histological and gene expression analyses have and hoxa9a are overexpressed in kgg mutants, restore identified the thymus, pancreas, and pronephros as sites erythroid development without correcting the morpho- of lymphopoiesis. logical defects of kgg embryos. Furthermore, over- Similar to other fish, the bilateral zebrafish thymi expression of scl, which can induce ectopic blood in form as small outgrowths between the third and fourth wild-type embryos, fails to induce blood in kgg embryos. pharyngeal arches just ventral to the developing ear (otic While a complete understanding of the role that cdx4 vesicle). By ultrastructural analysis, the thymi first plays during HSC formation awaits more study, the become populated by immature lymphoblasts around current data suggest that the cdx4-hox pathway acts 65 hpf (Willett et al., 1997, 1999). Between 3–4 dpf, these upstream of scl, and perhaps other essential scl cells can be detected by examining the expression of cofactors, to specify ICM-HSC fate (Davidson et al., genes such as ikaros, gata3, lck, and recombination 2003). activation genes 1 (rag1) and rag2, which encode the recombinases required for rearrangement of the T-cell receptor and immunoglobulin genes (Willett et al., 1997; Trede et al., 2001; Willett et al., 2001). The expression of ‘Definitive’ hematopoiesis the T-cell-specific markers gata3 and tcra (encoding a T-cell receptor subunit), and an absence of transcripts Formation of ‘definitive’ HSCs for pax5 (a gene important for B-cell differentiation), In mice, ‘definitive’ HSCs, as defined by their ability to suggest that the lymphoid cells in the thymus at this reconstitute all of the hematopoietic lineages in a stage are restricted to the T-cell lineage (Trede et al., lethally irradiated adult, are first detected in the aorta- 2001). The thymi continue to increase in cellularity gonad-mesonephros (AGM) region and umbilical ves- during larval development and reach their adult size by sels at E10-11 (reviewed by Yoder and Palis, 2001). Sites 1 month postfertilization (Willett et al., 1997; Lam et al., analogous to the AGM have been identified in other 2002). Although initially thought to lack the compart- vertebrates, including the para-aortic splanchnopleura mentalization of other vertebrates, it is now clear that in humans (Tavian et al., 1996) and chickens (reviewed distinct cortical and medullar regions exist in the adult by Dieterlen-Lievre et al., 2001), and the dorsolateral zebrafish thymus (Lam et al., 2002; Schorpp et al., plate in Xenopus (Chen and Turpen, 1995). Morpho- 2002). As in mammals, mature T cells expressing tcra logically, presumptive HSCs appear as clusters on the are found in the medulla, whereas immature T cells floor of the dorsal aorta and are thought to form by expressing rag1 and rag2 localize to the cortex (Lam budding from a ‘hemogenic endothelium’. In mammals, et al., 2002; Langenau et al., 2004). these HSCs are believed to enter circulation and seed The origin of the lymphoid progenitors that first the fetal liver, the first site of definitive hematopoiesis. populate the thymi is not known. Given that these Gene targeting studies in mice have shown that the cells appear in the thymus several days before kidney runx1 transcription factor gene is essential for AGM- hematopoiesis begins, it is unlikely that HSCs residing in derived HSC formation (Okuda et al., 1996; Wang et al., the pronephros are the source for these lymphocytes. 1996). Transcripts for the zebrafish runx1 orthologue are Indeed, lymphopoiesis occurs relatively late in the found in the ventral wall of the dorsal aorta between kidney and is not detected until 3 weeks postfertilization 24–48 hpf (Burns et al., 2002; Kalev-Zylinska et al., (Willett et al., 1999). In the mouse, the splanchnopleural 2002), suggesting that this site is analogous to the mesoderm (from which the AGM region is derived), AGM region. In support of this, other hematopoietic rather than the yolk sac blood islands, gives rise to T-cell transcription factor genes are also expressed in the precursors (Cumano et al., 1996). Thus, it has been dorsal aorta during this time, including c-myb (Thomp- proposed that in zebrafish, the lymphoid progenitors son et al., 1998), ikaros (Willett et al., 2001), scl (Jenna that seed the thymus are descended from AGM-HSCs Galloway and Liz, manuscript in preparation), and lmo2 generated between 24 and 48 hpf (Willett et al., 1999; (Hao Zhu and Liz, manuscript in preparation). Con- Trede et al., 2001). While this hypothesis has yet to be firming the HSC identity of these AGM cells and tested, it is interesting to note that ikaros-expressing cells whether they are responsible for seeding ‘definitive’ can be detected both in the AGM region as well as hematopoiesis in the kidney awaits lineage labeling or throughout the pharyngeal arches at 48 hpf (Willett cell transplantation experiments. et al., 2001). Relative to T lymphopoiesis, B-cell formation initiates Lymphopoiesis later in zebrafish development. By polymerase chain reaction (PCR), immunoglobulin (Ig) gene rearrange- From immunological studies of catfish and trout, it has ments are first detected at 4 dpf, whereas transcripts for long been known that teleosts possess a fully functioning the membrane form of IgM appear at 7 dpf (Danilova

Oncogene Zebrafish hematopoiesis AJ Davidson and LI Zon 7241 and Steiner, 2002). By whole mount in situ hybridiza- surround the blood vessels between the renal tubules tion, transcripts for rag1 are found in the region of and glomeruli (Al-Adhami and Kunz, 1977; Willett the pancreas primordium of 4 dpf larva, suggesting that et al., 1999). By 7 dpf, erythroblasts, myeloblasts, and this organ is the initial site of B lymphopoiesis. In neutrophilic granulocytes are readily observed within support of this result, expression of cm, encoding the the pronephros. Although the amount of hematopoeitic heavy chain subunit of IgM, can be found in cells tissue progressively expands during the next 2 weeks localized within the exocrine compartment of the of larval development, it remains restricted to the pancreas at 10 dpf (Danilova and Steiner, 2002). myeloerythroid cell lineages (Willett et al., 1999). However, lymphocytes labeled with GFP are not Lymphoblasts are not found in the pronephros until detected in the pancreas of larval rag2gfp or lckgfp approximately 19 dpf, where they can be identified both transgenic zebrafish (Langenau et al., 2004). While the morphologically and by the expression of rag1 and cm reason for this discrepancy is unknown, it is clear that transcripts (Willett et al., 1999; Danilova and Steiner, the pronephros becomes an active and major site for B 2002). At this stage of development, the amount of lymphopoiesis starting at 19 dpf, consistent with its role hematopoietic tissue in the kidney has enlarged, as the mammalian bone marrow equivalent (Danilova concomitant with the formation of additional renal and Steiner, 2002). tubules comprising the mesonephros (the second and final form of the kidney found in teleosts). A more Thrombopoiesis detailed time course study is needed to determine the appearance and expansion of the thrombocytic and By light microscopy, zebrafish thrombocytes appear as eosinophilic lineages within the kidney. small round nucleated cells that are morphologically In adults, the entire kidney is hematopoietically active similar to T lymphocytes, albeit smaller (Figure 6; and all blood cell lineages and their precursors are found Jagadeeswaran et al., 1999). Based on their aggregation as a heterogeneous population between the renal tubules and adhesion response to platelet agonists such as ADP, and the blood vessels (Figure 6). Thus, the kidney collagen, thrombin, arachidonic acid, and ristocetin, it marrow displays a cellular complexity comparable to appears that teleost thrombocytes are functional homo- that of the mammalian bone marrow. The kidney logues of mammalian platelets (Lloyd-Evans et al., marrow can be separated by flow cytometry into four 1994; Hill and Rowley, 1996; Jagadeeswaran et al., 1999; relatively distinct populations based on forward scatter Gregory and Jagadeeswaran, 2002). The functional (cell size) and side scatter (granularity). By cell equivalence between thrombocytes and platelets sug- morphology and transgenic reporter gene expression, gests that the factors involved in the formation of these these four populations correspond to (1) mature cell types may also be conserved. Accordingly, genes erythroid cells, (2) myelomonocytic cells (neutrophils, encoding transcription factors important for megakar- monocytes, macrophages, and eosinophils), (3) lym- yocyte development in mammals, such as fli1, fog1, phoid cells, polychromatophilic erythroblasts, and gata1, nfe2, and runx1, have also been identified in mature thrombocytes, and (4) immature precursors of zebrafish (Table 1). In addition, the thrombocytic/ all lineages (Traver et al., 2003b). The ease with which megakaryocytic marker genes itga2b, encoding the light scatter characteristics can be used to assess the platelet glycoprotein GPIIb (also known as CD41 and relative percentages of the major blood lineages has integrinaIIb), and c-mpl, encoding the thrombopoietin permitted subtle hematopoietic defects in adult fish receptor, have also been isolated in zebrafish (Lin et al., to be examined. For example, zebrafish that are 2000; Lin et al., 2001). heterozygous carriers for mutations in the genes solute Little is known about the origin and ontogeny of the carrier family 4, anion exchanger 1 (Paw et al., 2003), or first thrombocytes in zebrafish. Recent experiments have erythrocytic b-spectrin (Liao et al., 2000a) display a mild taken advantage of the fluorescent lipophilic dye DiI- anemia and an increase in the percentage of cells within C18, which selectively labels thrombocytes, to evaluate the ‘precursor’ fraction (Traver et al., 2003b). the developmental stage at which thrombocytes enter In some fish, such as trout (Salmo trutta) and perch circulation (Gregory and Jagadeeswaran, 2002). Micro- (Perca fluviatilis), the spleen is an additional or the injection of DiI-C18 into embryos at 25, 36, and 48 hpf sole site of adult hematopoiesis, respectively (Catton, revealed that a small number (5–20) of presumptive 1951). In zebrafish, the spleen contains predominantly thrombocytes could be labeled at 36 hpf, with greater mature erythrocytes with minor populations of mature numbers observed at 48 hpf. These results agree with lymphocytes, myelomonocytes, but an absence of RT–PCR analysis for itga2b transcripts, which can be immature precursors (Lieschke et al., 2001; Danilova detected in embryos starting at 48 hpf (Lin et al., 2000). and Steiner, 2002; Traver et al., 2003b). Thus, with the Further investigation is needed to determine the source exception of T lymphocytes, which need to be educated of these embryonic thrombocytes, their lifespan, and in the thymus, all the mature blood cell lineages in later sites of thrombopoiesis. zebrafish are derived from the adult kidney. As a corollary, it is generally accepted that HSCs reside Kidney hematopoiesis within the kidney marrow, most likely existing in specialized niches together with reticular stromal cells The first hematopoietic cells appear in the pronephros (Catton, 1951; Zapata, 1979; Meseguer et al., 1991; at 4 dpf and become organized into cellular cords that Willett et al., 1999).

Oncogene Zebrafish hematopoiesis AJ Davidson and LI Zon 7242 Hematopoietic cell transplantation et al., 1998), lymphocytes (rag2gfp; Jessen et al., 2001), lckgfp; Langenau et al., 2004), myeloid cells (pu.1gfp;Ward Recent methodological advances have now made it et al., 2003), thrombocytes (itga2bgfp; Lin et al., 2001), possible to perform hematopoietic cell transplantation and leukocytes but not erythrocyes (bÀactingfp, Traver experiments in zebrafish (Langenau et al., 2003; Traver et al., 2003b). et al., 2003b; Langenau et al., 2004). These techniques provide a means to evaluate the cell autonomous T-cell homing and engraftment requirement of mutant genes, to assay leukemic transformation (Langenau et al., 2003), to test for The homing properties of zebrafish lymphocytes have HSC activity, and to examine cell migration and recently been examined using the lckgfp transgenic fish. homing. In the adult kidney, expression of the lckgfp transgene is restricted to mature T cells (Langenau et al., 2004). Thymocytes isolated from adult lckgfp fish that are Embryonic transplantation injected into the circulation of 48 hpf embryos will home Although still in the early stages of refinement, to the thymus and ventral tail region within 24 h hematopoietic transplantation has successfully been (Langenau et al., 2004). These experiments indicate that used to rescue or delay the lethal anemia associated the embryonic thymus expresses receptors and/or with the vlad tepes/gata1 and moonshine/tif1g mutations, chemoattractants that facilitate the migration of mature respectively (Ransom et al., 1996; Lyons et al., 2002; T cells to the thymus. Traver et al., 2003b; Ransom et al., 2004). In the case Like mammals, the hematopoietic system of adult of vlad tepes, transplantation of wild-type kidney zebrafish is sensitive to g-rays, and the lethality of marrow cells (carrying a gata1gfp transgenic reporter) is irradiation can be rescued by intraperitoneal or intra- sufficient to rescue a small fraction of mutants (B20%) cardiac injections of nonirradiated donor kidney mar- for at least 8 months. The relatively low percentage row (Traver et al., 2004). Sublethal doses of g radiation of long-term transplant survival may be related to can be used to transiently ablate the hematopoietic the rarity of HSCs in the kidney marrow coupled system and thereby provide a suitable environment for with the maximum number of cells that can be donor cell engraftment. The development of this transplanted into an embryo (Traver et al., 2003b). transplantation methodology in zebrafish has enabled Transplanting into anemic mutant embryos, rather the location of long-term thymic repopulating cells to than conditioned adults, prior to the onset of lympho- be examined. Injection of lckgfp kidney marrow cells poiesis tolerizes the host and prevents graft rejection into sublethally irradiated adults results in long-term associated with histocompatibility differences. This is an (46 months post-transplantation) maintenance of important issue as it is not yet possible to derive GFP-positive T cells in the thymus. In contrast, only genetically identical inbred strains of zebrafish and weak and transient engraftment of the thymus is therefore allogeneic transplant rejection is a significant observed when lckgfp thymic cells are used. These concern. Interestingly, there are also no overt signs of experiments suggest that HSCs residing in the kidney graft-vs-host disease in the transplanted embryos, give rise to T-cell progenitors that then migrate to perhaps due to low immunogenicity of embryonic the thymus. These observations are similar to findings tissues and/or small numbers of mature T lymphocytes in mice and highlight the conservation of lympho- in the transplants. poiesis between teleost and mammals (Brecher et al., When different kidney marrow ‘scatter populations’ 1993). No engraftment is seen with either kidney are tested, only the ‘lymphoid fraction’ (containing marrow or thymus cells when the host is not irradiated, lymphoid cells, polychromatophilic erythroblasts, and suggesting that donor cells are rejected in nonirradiated mature thrombocytes) is capable of long-term erythro- recipients and/or that repopulation is dependent upon cyte generation (B6 months), suggesting that this the opening of niche space within the kidney (Langenau population also harbors HSCs (Traver et al., 2003b). et al., 2004). Confirming HSC activity in the ‘lymphoid’ fraction will depend on the development of additional reagents and methodologies such as monoclonal antibodies Zebrafish as a model for leukemia selective for HSCs, higher engraftment efficiency, multilineage readouts, and serial transplantation. Pre- For at least the last 90 years, numerous species of liminary transplant studies in the bls mutant suggest teleosts (including zebrafish) have been used to study that this host may provide a more amenable environ- spontaneous or carcinogen-induced cancers including ment to assay HSC function, as it is relatively free of melanoma, neuroblastoma, pancreatic cancer, and competing host cells for the first few days of develop- hepatocellular carcinoma (Stanton, 1965; Hawkins ment (Traver et al., 2003b). Further advancements are et al., 1985; Harshbarger and Slatick, 2001; Wellbrock likely to come from the continued use of lineage-specific et al., 2002). The neoplasms of teleosts are histologically transgenes that express fluorescent reporter proteins in similar to human tumors and the orthologues of discrete blood cell populations. Currently, there are mammalian genes involved in cell cycle regulation, transgenic zebrafish that have fluorescent erythrocytes apoptosis, as well as known oncogenes and tumor (gatagfp; Long et al., 1997), eosinophils (gata2gfp; Jessen supressors, are conserved in zebrafish (reviewed by Stern

Oncogene Zebrafish hematopoiesis AJ Davidson and LI Zon 7243 and Zon, 2003). Given the advances in zebrafish genetic the embryonic cdx-hox pathway may be directly relevant manipulation and genomic infrastructure, it is not to human leukemias. In support of this, ectopic surprising that interest has refocused on zebrafish as expression of CDX2 was found in a patient with acute model to study oncogenesis. myeloid leukemia (AML), and overexpression of Cdx2 The utility of zebrafish to induce and study leukemia in the bone marrow of mice induces fatal and has been demonstrated recently using transgenic tech- transplantable AML (Chase et al., 1999; Rawat et al., nology (Langenau et al., 2003). The zebrafish rag2 2004). promoter was used to drive lymphoid expression of Zebrafish models of human malignancies, such as murine c-Myc, an oncogene known to participate in the the Myc-induced T-cell leukemia model, are likely to pathogenesis of mammalian lymphoid malignancies. become more prevalent and will provide valuable Approximately 5% of the F0 founder fish developed resources for understanding and treating cancer. Per- lymphoblastic leukemia with aggressive infiltrations haps one of the greatest promises that the zebrafish in the intestine, gills, skeletal muscle, and fins within system holds for the future is the ability to do 30–131 days. A molecular analysis of the transformed suppressor/enhancer screens to identify modifier genes cells suggested that the Myc-induced leukemia repre- that prevent or exacerbate a phenotype. While ‘modifier’ sented a clonal expansion of transformed T-lymphocyte screens are challenging to do genetically, preliminary precursors. Injection of the rag2-myc transgene into reports suggest that they can be successfully performed rag2gfp transgenic embryos permitted the lymphoblasts using small molecule libraries (Stern and Zon, 2003). of the subsequent leukemic fish to be purified by GFP These so-called ‘chemical suppressor’ screens involve fluorescence. Injection of these GFP-positive lympho- testing compounds for their ability to prevent or delay blasts into sublethally irradiated wild-type adults the onset of a mutant phenotype or disease. The small demonstrated the transplantability of the disease, thus size of zebrafish embryos, combined with the specific verifying the malignant nature of the T-cell neoplasm. labeling of cells with fluorescent reporters, makes them Furthermore, due to the translucent nature of the fish, ideally suited for large-scale chemical screening and the dissemination and homing of the leukemic cells paves the way for the discovery of novel hematopoietic could be visualized in vivo (Langenau et al., 2003). This regulators, oncogenes, tumor supressors, and therapeu- experimental approach, making use of transgenic tic agents. technology, flurorescent reporters, and cell transplanta- The high fecundity of zebrafish facilitates functional tion, provides a template for the generation of future genomic approaches such as high-throughput in situ cancer models. hybridization screening. Efforts are currently underway to examine the expression patterns of all spatially restricted, developmentally active genes (see ‘Gene Conclusions and future directions expression’ at http://zfin.org). In the near future, developing organs will be defined by stage-specific ‘gene In the last 25 years, zebrafish have evolved from a sets’ and this information will be a valuable resource common aquarium pet to an established genetic work- for determining genetic hierarchies and interactions horse in the laboratory. Their genetic versatility and the during organogenesis. Such data will also be useful for high degree of conservation in gene function between making sense of microarray results in which whole teleosts and mammals make them well suited for the embryo datasets are compared, thereby allowing re- study of normal and malignant hematopoiesis. The search efforts to be focused only on genes expressed in translucent nature of zebrafish embryos is advantageous the tissue of interest. The functional importance of for the analysis of hematopoietic cell transplantation these genes can be rapidly tested in zebrafish embryos experiments, particularly when the donor cells express using morpholino-mediated ‘knockdown’. In addition, fluorescent reporters. Furthermore, the optical clarity of genetic lesions in selected genes can be isolated using the embryo permits the spatial and temporal character- the reverse genetics approach of ‘targeting induced istics of gene expression to be studied at the single cell local lesions in genomes’ (TILLING). This latter resolution. Such a detailed level of analysis will lay the approach has been successfully used to obtain rag1 foundation for understanding the genetic pathways that mutants that are defective in V(D)J recombination control the formation of HSCs from mesoderm, HSC (Wienholds et al., 2002). lineage decisions, and progenitor differentiation. To- Looking ahead, it is clear that the zebrafish system, wards this goal, forward genetic screens in zebrafish with its extensive array of genetic and developmental have identified a number of new, functionally important, tools, will continue making significant contributions hematopoietic transcription factor genes including cdx4, to our understanding of hematopoiesis and leukemo- spt/tbx16, and tif1g. Understanding the role of these genesis. genes during normal embryonic development will provide new insights into human hematopoietic malig- Acknowledgements nancies in which the genetic pathways controlling cell We thank Paula Fraenkel and Howard Stern for con- growth and differentiation have gone awry. For tributing images to Figure 6 and Noe¨ lle Paffett-Lugassy, instance, deregulation of hox gene expression is im- Rebecca Wingert, Nikolaus Trede, and Barry Paw for plicated in leukemic transformation (reviewed by Owens critical reading of this manuscript and their insightful and Hawley, 2002), thereby raising the possibility that comments.

Oncogene Zebraﬁsh hematopoiesis AJ Davidson and LI Zon 7244 References

Al-Adhami MA and Kunz YW. (1976). Willhelm Roux’s Gering M, Rodaway ARF, Go¨ ttgens B, Patient RK and Green Arch.Dev. Biol., 179, 393–401. AR. (1998). EMBO J., 17, 4029–4045. Al-Adhami MA and Kunz YW. (1977). Dev. Growth Differ., Gering M, Yamada Y, Rabbitts TH and Patient RK. (2003). 19, 171–179. Dev., 130, 6187–6199. Amacher SL, Draper BW, Summers BR and Kimmel CB. Gregory M and Jagadeeswaran P. (2002). Blood Cells Mol. (2002). Development, 129, 3311–3323. Dis., 28, 418–427. Begley CG, Aplan PD, Davey MP, Nakahara K, Tchorz K, Griffith KJP, Amacher SL, Kimmel CB and Kimelman D. Kurtzberg J, Hershfield MS, Haynes BF, Cohen DI, (1998). Development, 125, 3379–3388. Waldmann TA and Kirsch IR. (1989). Proc. Natl. Acad. Haffter P, Granato M, Brand M, Mullins MC, Hammersch- Sci. USA, 86, 2031–2035. midt M, Kane DA, Odenthal J, van Eeden FJ, Jiang YJ, Bennett CM, Kanki JP, Rhodes J, Liu TX, Paw BH, Kieran Heisenberg CP, Kelsh RN, Furutani-Seiki M, Vogelsang E, MW, Langenau DM, Delahaye-Brown A, Zon LI, Fleming Beuchle D, Schach U, Fabian C and Nusslein-Volhard C. MD and Look AT. (2001). Blood, 98, 643–651. (1996). Development, 123, 1–36. Bouchard M, Souabni A, Mandler M, Neubuser A and Hammerschmidt M, Pelegri F, Mullins MC, Kane DA, Brand Busslinger M. (2002). Genes Dev., 16, 2958–2970. M, van Eeden FJ, Furutani-Seiki M, Granato M, Haffter P, Brecher G, Bookstein N, Redfearn W, Necas E, Pallavicini Heisenberg CP, Jiang YJ, Kelsh RN, Odenthal J, Warga MG and Cronkite EP. (1993). Proc. Natl. Acad. Sci. USA, RM and Nusslein-Volhard C. (1996). Development, 123, 90, 6028–6031. 143–151. Brown L, Cheng JT, Chen Q, Siciliano MJ, Crist W, Buchanan Harshbarger JC and Slatick MS. (2001). Mar. Biotechnol., 3 G and Baer R. (1990). EMBO J., 9, 3343–3351. (Suppl 1), S115–S129. Brown LA, Rodaway AR, Schilling TF, Jowett T, Ingham Hawkins WE, Overstreet RM, Fournie JW and Walker WW. PW, Patient RK and Sharrocks AD. (2000). Mech. Dev., 90, (1985). J. Appl. Toxicol., 5, 261–264. 237–252. Herbomel P, Thisse B and Thisse C. (1999). Development, 126, Brownlie A, Donovan A, Pratt SJ, Paw BH, Oates AC, 3735–3745. Brugnara C, Witkowska HE, Sassa S and Zon LI. (1998). Herbomel P, Thisse B and Thisse C. (2001). Dev. Biol., 238, Nat. Genet., 20, 244–250. 274–288. Brownlie A, Hersey C, Oates AC, Paw BH, Falick AM, Hill DJ and Rowley AF. (1996). Br. J. Haematol., 92, Witkowska HE, Flint J, Higgs D, Jessen J, Bahary N, Zhu 200–211. H, Lin S and Zon L. (2003). Dev. Biol., 255, 48–61. Jagadeeswaran P, Sheehan JP, Craig FE and Troyer D. (1999). Burns CE, DeBlasio T, Zhou Y, Zhang J, Zon L and Nimer Br. J. Haematol., 107, 731–738. SD. (2002). Exp. Hematol., 30, 1381–1389. Jessen JR, Jessen TN, Vogel SS and Lin S. (2001). Genesis, 29, Catton WT. (1951). Blood, 6, 39–60. 156–162. Chase A, Reiter A, Burci L, Cazzaniga G, Biondi A, Pickard J, Jessen JR, Meng A, McFarlane RJ, Paw BH, Zon LI, Roberts IA, Goldman JM and Cross NC. (1999). Blood, 93, Smith GR and Lin S. (1998). Proc. Natl. Acad. Sci. USA, 1025–1031. 95, 5121–5126. Chen XD and Turpen JB. (1995). J. Immunol., 154, 2557–2567. Kalev-Zylinska ML, Horsfield JA, Flores MV, Postlethwait Childs S, Weinstein BM, Mohideen MA, Donohue S, JH, Chan JY, Cattin PM, Vitas MR, Crosier PS and Crosier Bonkovsky H and Fishman MC. (2000). Curr. Biol., 10, KE. (2003). Dev. Dyn., 228, 323–336. 1001–1004. Kalev-Zylinska ML, Horsfield JA, Flores MV, Postlethwait Cumano A, Dieterlen-Lievre F and Godin I. (1996). Cell, 86, JH, Vitas MR, Baas AM, Crosier PS and Crosier KE. 907–916. (2002). Development, 129, 2015–2030. Danilova N and Steiner LA. (2002). Proc. Natl. Acad. Sci. Kimmel CB. (1990). Development, 108, 581–594. USA, 99, 13711–13716. Kimmel CB, Ballard WW, Kimmel SR, Ullmann B and Davidson AJ, Ernst P, Wang Y, Dekens MP, Kingsley PD, Schilling TF. (1995). Dev. Dyn., 203, 253–310. Palis J, Korsmeyer SJ, Daley GQ and Zon LI. (2003). Kimmel CB, Kane DA, Walker C, Warga RM and Rothman Nature, 425, 300–306. MB. (1989). Nature, 337, 358–362. Detrich HW, Kieran III MW, Chan FY, Barone LM, Yee K, Krumlauf R. (1994). Cell, 78, 191–201. Rundstadler JA, Pratt S, Ransom D and Zon LI. (1995). Lam SH, Chua HL, Gong Z, Wen Z, Lam TJ and Sin YM. Proc. Natl. Acad. Sci. USA, 92, 10713–10717. (2002). Dev. Dyn., 225, 87–94. Dieterlen-Lievre F, Pardanaud L, Caprioli A and Lane MC and Sheets MD. (2000). Dev. Biol., 225, 37–58. Jaffredo T. (2001). Hematopoiesis: A Developmental Lane MC and Sheets MD. (2002). Dev. Dyn., 225, Approach Zon, LI (ed). Oxford University Press: New York, 434–437. New York. Langenau DM, Ferrando AA, Traver D, Kutok JL, Hezel JP, Donovan A, Brownlie A, Zhou Y, Shepard J, Pratt SJ, Kanki JP, Zon LI, Look AT and Trede NS. (2004). Proc. Moynihan J, Paw BH, Drejer A, Barut B, Zapata A, Law Natl. Acad. Sci. USA, 101, 7369–7374. TC, Brugnara C, Lux SE, Pinkus GS, Pinkus JL, Kingsley Langenau DM, Traver D, Ferrando AA, Kutok JL, Aster JC, PD, Palis J, Fleming MD, Andrews NC and Zon LI. (2000). Kanki JP, Lin S, Prochownik E, Trede NS, Zon LI and Nature, 403, 776–781. Look AT.. (2003). Science, 299, 887–890. Dooley KA, Davidson AJ and Zon LI. (2004). Dev. Biol.,in Liao EC, Paw BH, Oates AC, Pratt SJ, Postlethwait JH and press. Zon LI. (1998). Genes Dev., 12, 621–626. Favier B and Dolle P. (1997). Mol. Hum. Reprod., 3, 115–131. Liao EC, Paw BH, Peters LL, Zapata A, Pratt SJ, Do CP, Fisher RC, Slayton WB, Chien C, Guthrie SM, Bray C and Lieschke G and Zon LI. (2000a). Development, 127, Scott EW. (2004). Leuk. Res., 28, 83–89. 5123–5132. Galloway JL and Zon LI. (2003). Curr. Top. Dev. Biol., 53, Liao EC, Trede NS, Ransom D, Zapata A, Kieran M and Zon 139–158. LI. (2002). Development, 129, 649–659.

Oncogene Zebraﬁsh hematopoiesis AJ Davidson and LI Zon 7245 Liao W, Bisgrove BW, Sawyer H, Hug B, Bell B, Peters K, Rawat VP, Cusan M, Deshpande A, Hiddemann W, Grunwald DJ and Stainier DY. (1997). Development, 124, Quintanilla-Martinez L, Humphries RK, Bohlander SK, 381–389. Feuring-Buske M and Buske C. (2004). Proc. Natl. Acad. Liao W, Ho CY, Yan YL, Postlethwait J and Stainier DY. Sci. USA, 101, 817–822. (2000b). Development, 127, 4303–4313. Rieb J-P. (1973). Ann. Embryol. Morphol., 6, 43–54. Lieschke GJ, Oates AC, Crowhurst MO, Ward AC and Robb L, Lyons I, Li R, Hartley L, Kontgen F, Harvey RP, Layton JE. (2001). Blood, 98, 3087–3096. Metcalf D and Begley CG. (1995). Proc. Natl. Acad. Sci. Lieschke GJ, Oates AC, Paw BH, Thompson MA, Hall NE, USA, 92, 7075–7079. Ward AC, Ho RK, Zon LI and Layton JE. (2002). Dev. Ruvinsky I, Silver LM and Ho RK. (1998). Genes Evol., 298, Biol., 246, 274–295. 94–99. Lin HF, Freedman M, Zhou Y, Zon LI and Handin RI. Schorpp M, Leicht M, Nold E, Hammerschmidt M, Haas- (2000). Blood, 96, 144 (abstract). Assenbaum A, Wiest W and Boehm T. (2002). Mech. Dev., Lin HF, Paw BH, Zon LI, Gregory M, Jagadeeswaran P and 118, 179–185. Handin RI. (2001). Blood, 98, 2147 (abstract). Shaﬁzadeh E, Paw BH, Foott H, Liao EC, Barut BA, Cope JJ, Liu F and Wen Z. (2002). Mech. Dev., 113, 69–72. Zon LI and Lin S. (2002). Development, 129, 4359–4370. Lloyd-Evans P, Barrow SE, Hill DJ, Bowden LA, Rainger Shivdasani RA, Mayer EL and Orkin SH. (1995). Nature, 373, GE, Knight J and Rowley AF. (1994). Biochim. Biophys. 432–434. Acta, 1215, 291–299. Stainier DY, Weinstein BM, Detrich HW, Zon III LI and Lohnes D. (2003). BioEssays, 25, 971–980. Fishman MC. (1995). Development, 121, 3141–3150. Long Q, Meng A, Wang H, Jessen JR, Farrell MJ and Lin S. Stanton MF. (1965). J. Natl. Cancer Inst., 34, 117–130. (1997). Development, 124, 4105–4111. Stern HM and Zon LI. (2003). Nat. Rev. Cancer, 3, 533–539. Lyons SE, Lawson ND, Lei L, Bennett PE, Weinstein BM Stockard CR. (1915). Am. J. Anat., 18, 227–327. and Liu PP. (2002). Proc. Natl. Acad. Sci. USA, 99, Sumoy L, Keasey JB, Dittman TD and Kimelman D. (1997). 5454–5459. Mech. Dev., 63, 15–27. Lyons SE, Shue BC, Oates AC, Zon LI and Liu PP. (2001). Tavian M, Coulombel L, Luton D, Clemente HS, Dieterlen- Blood, 97, 2611–2617. Lievre F and Peault B. (1996). Blood, 87, 67–72. Majumdar A, Lun K, Brand M and Drummond IA. (2000). Thompson MA, Ransom DG, Pratt SJ, MacLennan H, Development, 127, 2089–2098. Kieran MW, Detrich III HW, Vail B, Huber TL, Paw B, Meseguer J, Esteban MA and Agulleiro B. (1991). Arch. Brownlie AJ, Oates AC, Fritz A, Gates MA, Amores A, Histol. Cytol., 54, 299–309. Bahary N, Talbot WS, Her H, Beier DR, Postlethwait JH Mun˜ oz-Sanjua´ n I and Hemmati-Brivanlou A. (2001). Dev. and Zon LI. (1998). Dev. Biol., 197, 248–269. Biol., 237, 1–17. Toyama R, Kobayashi M, Tomita T and Dawid IB. (1998). Nishikawa K, Kobayashi M, Masumi A, Lyons SE, Weinstein Mech. Dev., 71, 197–200. BM, Liu PP and Yamamoto M. (2003). Mol. Cell. Biol., 23, Traver D, Herbomel P, Patton EE, Murphey RD, Yoder JA, 8295–8305. Litman GWCA, Amemiya CT, Zon LI and Trede NS. Oates AC, Brownlie A, Pratt SJ, Irvine DV, Liao EC, Paw BH, (2003a). Adv. Immunol., 81, 253–330. Dorian KJ, Johnson SL, Postlethwait JH, Zon LI and Wilks Traver D, Paw BH, Poss KD, Penberthy WT, Lin S and Zon AF. (1999). Blood, 94, 2622–2636. LI. (2003b). Nat. Immunol., 4, 1238–1244. Oates AC, Pratt SJ, Vail B, Yan YI, Ho RK, Johnson SL, Traver D, Winzeler A, Stern HM, Mayhall EA, Langenau Postlethwait JH and Zon LI. (2001). Blood, 98, 1792–1801. DM, Kutok JL, Look AT and Zon LI. (2004). Blood (in Oellacher J. (1872). Z. Wiss. Zool., 23, 373–421. press). Okuda T, van Deursen J, Hiebert SW, Grosveld G and Trede NS, Zapata A and Zon LI. (2001). Trends Immunol., 22, Downing JR. (1996). Cell, 84, 321–330. 302–307. Owens BM and Hawley RG. (2002). Stem Cells, 20, 364–379. Valge-Archer VE, Osada H, Warren AJ, Forster A, Li J, Baer Parker L and Stainier DYR. (1999). Development, 126, R and Rabbitts TH. (1994). Proc. Natl. Acad. Sci. USA, 91, 2643–2651. 8617–8621. Paw BH, Davidson AJ, Zhou Y, Li R, Pratt SJ, Lee C, Trede Wadman IA, Osada H, Grutz GG, Agulnick AD, Westphal H, NS, Brownlie A, Donovan A, Liao EC, Ziai JM, Drejer AH, Forster A and Rabbitts TH. (1997). EMBO J., 16, Guo W, Kim CH, Gwynn B, Peters LL, Chernova MN, 3145–3157. Alper SL, Zapata A, Wickramasinghe SN, Lee MJ, Lux SE, Walmsley M, Ciau-Uitz A and Patient R. (2002). Dev., 129, Fritz A, Postlethwait JH and Zon LI. (2003). Nat. Genet., 34, 5683–5695. 59–64. Wang H, Long Q, Marty SD, Sassa S and Lin S. (1998). Nat. Pfeffer PL, Gerster T, Lun K, Brand M and Busslinger M. Genet., 20, 239–243. (1998). Development, 125, 3063–3074. Wang Q, Stacy T, Binder M, Marin-Padilla M, Sharpe AH Pratt SJ, Drejer A, Foott H, Barut B, Brownlie A, Postlethwait and Speck NA. (1996). Proc. Natl. Acad. Sci. USA, 93, J, Kato Y, Yamamoto M and Zon LI. (2002). Physiol. 3444–3449. Genomics, 11, 91–98. Ward AC, McPhee DO, Condron MM, Varma S, Cody SH, Ransom DG, Bahary N, Niss K, Traver D, Burns C, Paffett- Onnebo SM, Paw BH, Zon LI and Lieschke GJ. (2003). Lugassy N, Saganic WJ, Lim CA, Hersey C, Zhou Y, Barut Blood, 102, 3238–3240. BA, Lin S, Kingsley PD, Palis J, Orkin SH and Zon LI. Warga RM and Nusslein-Volhard C. (1999). Development, (2004). PloS. Biol., in press. 126, 827–838. Ransom DG, Haffter P, Odenthal J, Brownlie A, Vogelsang E, Weinstein BM, Schier AF, Abdelilah S, Malicki J, Solnica- Kelsh RN, Brand M, van Eeden FJ, Furutani-Seiki M, Krezel L, Stemple DL, Stainier DY, Zwartkruis F, Driever Granato M, Hammerschmidt M, Heisenberg CP, Jiang YJ, W and Fishman MC. (1996). Development, 123, 303–309. Kane DA, Mullins MC and Nusslein-Volhard C. (1996). Wellbrock C, Gomez A and Schartl M. (2002). Microsc. Res. Development, 123, 311–319. Tech., 58, 456–463.

Oncogene Zebraﬁsh hematopoiesis AJ Davidson and LI Zon 7246 Wienholds E, Schulte-Merker S, Walderich B and Plasterk Wingert RA and Zon LI. (2003). Hematopoietic Stem RH. (2002). Science, 297, 99–102. Cells. Godin I and Cumano A (eds). Landes Bioscience: Willett CE, Cherry JJ and Steiner LA. (1997). Immunogenetics, Georgetown, TX. 45, 394–404. Yoder MC and Palis J. (2001). Hematopoiesis: A Willett CE, Cortes A, Zuasti A and Zapata AG. (1999). Dev. Developmental Approach. Zon LI (ed). Oxford University Dyn., 214, 323–336. Press: New York, New York. Willett CE, Kawasaki H, Amemiya CT, Lin S and Steiner LA. Zapata A. (1979). Dev. Comp. Immunol., 3, 55–65. (2001). Dev. Dyn., 222, 694–698.

Oncogene