Keywords

1 Introduction

1.1 General

Hematopoietic stem cells (HSCs) are self-renewing multipotent cells that can generate all mature blood cells throughout the lifespan of an individual (Weissman 2000). Due to the limited lifespan of mature blood cells, HSCs must be continually available to replace cells lost in circulation. Despite this demand, most adult HSCs present in the bone marrow are quiescent and divide rarely under homeostatic conditions. This is probably to prevent the depletion of the stem cell pool or to protect HSCs from myelotoxic injury. Instead, hematopoietic progenitor cells (HPCs), which are the progeny of HSCs that have the limited or no self-renewal ability, rapidly proliferate and differentiate to satisfy the requirements for new mature blood cells (Cheshier et al. 1999; Walkley et al. 2005). HSCs in the bone marrow interact with other cell types known as niche cells such as osteoblasts, endothelial cells, and stromal cells. These niche cells tightly regulate the balance between the quiescence, self-renewal, and differentiation of HSCs through essential signaling molecules called “niche factors” (Ehninger and Trumpp 2011; Hoggatt et al. 2016; Yu and Scadden 2016; Beerman et al. 2017). A timely response to blood cell requirements is thus largely dependent on the function of the niche under normal physiological conditions and during hematopoietic crises such as loss of blood, hemolysis, or infection.

HSCs present within adult bone marrow or newborn cord blood are by far the most widely utilized stem cells in the clinic. It is always difficult, however, to find immune-matched donors or to obtain sufficient numbers of HSCs for transplantation therapies to treat blood-related disorders (e.g., anemia, leukopenia, leukemia, etc.). Generation of HSCs from pluripotent precursors, including induced pluripotent stem cells (iPSCs), is therefore a key therapeutic aim. It is of great interest to better understand the developmental mechanisms of HSCs in the embryo because elucidation of this developmental process might provide insight into the cellular and molecular cues needed to instruct and expand HSCs from pluripotent precursors (Murry and Keller 2008).

1.2 Zebrafish as a Model for the Study of HSC Development

Almost two decades ago, the zebrafish emerged as a new genetic model to analyze hematopoietic development as well as human hematopoietic diseases (Zhu and Zon 2002; Bradbury 2004). Its embryos are externally fertilized and transparent, enabling in vivo live-imaging analysis, which can capture the dynamics of HSC development, including HSC birth, migration, and cell division. The high fecundity and rapid generation time of zebrafish make phenotype-based forward genetic screening feasible, permitting the discovery of novel signaling pathways that control hematopoietic development (de Jong and Zon 2005). Moreover, the recent advent of genome-editing technologies by the CRISPR/Cas9 system will further promote the investigation of the spatial and temporal regulatory mechanisms within the specific cell types (Jao et al. 2013; Gagnon et al. 2014; Hisano et al. 2015; Shah et al. 2015; Kawahara et al. 2016). Importantly, despite the evolutional divergence, the genetic programs governing hematopoiesis are highly conserved between mammals and fish. Our understanding of the regulatory mechanisms of HSC development has been greatly improved by the zebrafish system within the last decade. This chapter discusses this conservation and highlights the current knowledge and recent advances regarding the cellular origin and molecular regulation of HSC development in the zebrafish embryo.

2 Comparison of Hematopoietic Development between the Mouse and Zebrafish

2.1 Hematopoietic Development in the Mouse

Hematopoietic development in vertebrates can be characterized by two waves of hematopoiesis: the primitive and definitive waves. In the mouse, the primitive wave is initiated in the extra-embryonic yolk sac (YS), where primitive erythrocytes are generated from unipotent hematopoietic precursors at day 7.5 post-coitum (E7.5) (Dzierzak and Medvinsky 1995; Palis et al. 1999). These blood islands consist of nucleated erythrocytes expressing embryonic globin genes (Stamatoyannopoulos 2005). Macrophages and megakaryocytes are also generated in the YS during the primitive wave (Shepard and Zon 2000; Tober et al. 2007). Primitive hematopoiesis is transient and replaced by the definitive wave during embryonic stages.

Multilineage definitive hematopoiesis is first initiated by erythro-myeloid progenitors (EMPs), which emerge in the YS and placenta at E9.5 and have the potential to differentiate into both erythroid and myeloid lineages but not the lymphoid lineage (Palis et al. 1999; Bertrand et al. 2005). Erythrocytes derived from EMPs are enucleated and express adult globin genes, while the embryonic globin gene is weakly expressed (McGrath et al. 2011). The intra-embryonic aorta-gonad-mesonephros (AGM) region gives rise to HSCs with the potential to generate all blood lineages including the lymphoid lineage (Müller et al. 1994; Cumano et al. 1996; Godin and Cumano 2002; Medvinsky and Dzierzak 1996). Lineage-tracing studies revealed that HSCs arise from a subset of specialized endothelial cells present in the ventral floor of the dorsal aorta (DA), termed hemogenic endothelial cells (HECs), via the endothelial-to-hematopoietic transition (EHT) (Zovein et al. 2008; Boisset et al. 2010). HSC formation also occurs in the umbilical and vitelline arteries, placenta, and YS (de Bruijn et al. 2000; Gekas et al. 2005; Ottersbach and Dzierzak 2005; Samokhvalov et al. 2007). Hematopoietic stem/progenitor cells (HSPCs) arising from the DA at E10.5 initially form the intra-aortic hematopoietic clusters (IAHCs), which contain cells expressing both hematopoietic (CD41 and CD45) and endothelial (CD31 and VE-cadherin) markers (Fraser et al. 2002; de Bruijn et al. 2002; North et al. 2002). HSPCs isolated at E10.5 cannot reconstitute adult hematopoiesis without ex vivo organ culture, whereas HSPCs present at E11.5 can engraft the adult niche (Taoudi et al. 2008; Rybtsov et al. 2011). Thus, developing HSCs within IAHCs (pre-HSCs) require further maturation that potentiates their engraftment and survival in future hematopoietic niches. After leaving the IAHCs, HSCs migrate to the fetal liver (FL), the main hematopoietic organ in mid-late gestation. In contrast to adult HSCs in the bone marrow, HSCs in the FL are largely cycling and frequently undergo symmetric cell divisions, in which both daughter cells retain self-renewal capacity and multipotency, resulting in a marked expansion of the HSC pool (Ema and Nakauchi 2000; Lessard et al. 2004; Khan et al. 2016). Shortly before birth (E17), HSCs finally seed the bone marrow, which becomes the major site of hematopoiesis in the adult stage.

2.2 Hematopoietic Development in the Zebrafish

Hematopoiesis in zebrafish also occurs in two waves, but these waves occur in distinct locations compared to mammals. The primitive wave in the zebrafish embryo is initiated in the intra-embryonic lateral plate mesoderm (LPM), which is the bilateral stripes that flank the somite during early somitogenesis (Liao et al. 1998; Thompson et al. 1998). The anterior LPM (ALPM) is the major site of primitive myelopoiesis, whereas the posterior LPM (PLPM) contains mostly erythroid precursors and a few myeloid precursors (Detrich et al. 1995; Warga et al. 2009; Galloway and Zon 2003). The PLPM also gives rise to angioblasts, a common precursor of endothelial cells and definitive HSPCs (Dooley et al. 2005; Patterson et al. 2005). At around 15 h post-fertilization (hpf), cells in the PLPM begin to migrate axially along the ventral domain of the somite and reach the midline to form the intermediate cell mass (ICM). The ICM is the major site of primitive erythropoiesis and has a similar cellular architecture to the mammalian YS blood islands (Davidson and Zon 2004; Hsia and Zon 2005; de Jong and Zon 2005). Around the same time, angioblasts also reach the midline and form the vascular cord, which subsequently develops into the lumenal DA and posterior cardinal vein (PCV) (Jin et al. 2005).

As in mammals, definitive hematopoiesis in the zebrafish embryo is initiated by EMPs (Bertrand et al. 2007). EMPs can be detected as lmo2 + gata1 + cells at around 24 hpf in the posterior blood island (PBI), which is located in the anterior-ventral tail. Shortly before the initiation of blood circulation (23 hpf), specified HECs can be observed in the ventral floor of the DA (Wilkinson et al. 2009), the AGM equivalent in zebrafish. In vivo live-imaging has directly shown a number of HSPCs to bud off from the ventral floor of the DA and enter the circulation at around 30–54 hpf (Bertrand et al. 2010; Kissa and Herbomel 2010). These nascent HSPCs express both hematopoietic (cmyb and cd41) and endothelial (kdrl) markers (Bertrand et al. 2010), similar to pre-HSCs in the mouse embryo (Fraser et al. 2002; de Bruijn et al. 2002; North et al. 2002). After arising from the DA, HSCs migrate via the circulation to the caudal hematopoietic tissue (CHT), which is a transient hematopoietic organ that is equivalent to the FL in mammals (Murayama et al. 2006; Jin et al. 2007; Blaser et al. 2017). HSPCs in the CHT actively proliferate and differentiate (Tamplin et al. 2015). Very recently, a clonal fate mapping study using drl:CreER T2; ubi:Zebrabow double-transgenic animals, which allow the labeling of individual HSCs and their progeny with a unique color, demonstrated that approximately 21 HSCs arise from the DA in the zebrafish embryo. These initial cells expand to 30 HSCs during the embryonic stage (Henninger et al. 2017). Expanded HSCs then move to the kidney marrow to sustain lifelong hematopoiesis or to the thymus, where some HSCs differentiate into T lymphocytes (Davidson and Zon 2004; de Jong and Zon 2005).

3 Molecular Regulation of HSC Specification in the Zebrafish Embryo

3.1 Transcriptional Regulation of HSC Specification

HSCs originate from a shared vascular precursor, the angioblast. Although angioblasts emerge in both the ALPM and PLPM, HSCs are specified only from endothelial cells of the trunk DA, suggesting that angioblasts in the PLPM are pre-patterned to respond to specification signals, or that HSC specification signals are spatially restricted to the trunk area of the embryo. A number of intrinsic and extrinsic signaling molecules that are required to establish HSC fate have been identified in the zebrafish embryo (Clements and Traver 2013). The transcription factor Runx1 (also known as AML1) is required for definitive but not primitive, hematopoiesis in both mammals and zebrafish (Chen et al. 2009; Kalev-Zylinska et al. 2002; Gering and Patient 2005). The expression of runx1 is detected in the ventral floor of the DA as early as 23 hpf in the zebrafish embryo (Burns et al. 2002; Wilkinson et al. 2009). Loss of runx1 in the zebrafish embryo resulted in the burst of HECs in the ventral floor of the DA and loss of definitive HSPCs (Kissa and Herbomel 2010), indicating that Runx1 plays an essential role in the initiation of EHT. Since runx1 expression is very specific to HECs/HSPCs within the DA, runx1 is widely utilized as a marker to distinguish HECs/HSPCs from other endothelial cells or differentiated blood cells in the zebrafish embryo.

The basic helix-loop-helix transcription factor stem-cell leukemia (Scl, also known as Tal1) is essential for both primitive and definitive hematopoiesis as well as the vascular formation (Patterson et al. 2005; Dooley et al. 2005). In zebrafish, two distinct isoforms of Scl, Scl-α and Scl-β, are produced based on an alternative promoter site (Qian et al. 2007). Although these two isoforms are functionally redundant in primitive erythropoiesis, their roles in definitive hematopoiesis are divergent. The full-length isoform, scl-α, is expressed by primitive erythroid precursors in the ICM and in HECs after 36 hpf, whereas the N-terminal-truncated isoform, scl-β, is expressed in HECs at an earlier stage, probably before 22 hpf. Morpholino knockdown of scl-β resulted in the reduction of runx1 expression in the DA, while loss of runx1 did not affect the expression of scl-β in the DA, suggesting that Scl-β acts as upstream of Runx1 in HECs. In contrast, loss of Scl-α did not affect the expression of runx1 in the DA although the number of HSPCs in the CHT was largely reduced at 3 days post-fertilization (dpf). These data suggest that Scl-β is required for HSC specification via the regulation of runx1 expression in HECs, whereas Scl-α plays an essential role in the maintenance of HSCs after the specification process (Qian et al. 2007; Zhen et al. 2013).

Ets variant 2 (Etv2, also known as Etsrp) is a critical regulator of vascular development and primitive myelopoiesis in zebrafish (Sumanas and Lin 2006; Sumanas et al. 2008). The expression of etv2 is first detected in both the ALPM and PLPM as early as 11 hpf and later in the whole vasculature (Sumanas and Lin 2006). Morpholino knockdown of etv2 induced down-regulation of scl-α, scl-β, and fli1, an ETS domain transcription factor expressed in the LPM and vascular endothelium (Thompson et al. 1998; Lawson and Weinstein 2002), resulting in defective blood vessel formation and the complete loss of definitive HSPCs (Sumanas and Lin 2006; Ren et al. 2010). The vascular defect of etv2 morphants could be rescued by injection of fli1 mRNA. In addition, injection of scl-β mRNA along with fli1 mRNA rescued runx1 expression in the DA in etv2 morphants, suggesting that Etv2 activates the expression of both fli1 and scl-β to establish HSC fate in the angioblast (Ren et al. 2010). Surprisingly, injection of scl-α mRNA alone partially restored both the vasculature and runx1 expression in etv2 morphants (Ren et al. 2010), despite the dispensable role of Scl-α in HSC specification (Zhen et al. 2013). One possible explanation is that the specification of angioblasts, which occurs prior to HSC specification, requires a specific level of Scl-α expression, and this requirement can be partially compensated for by Scl-β. Scl-α overexpression may also be able to compensate for the loss of Scl-β, restoring the HSC fate in the angioblast. Taken together, these studies demonstrate that Etv2 plays an essential role in both the angioblast specification and HSC specification via the regulation of Scl-α, Scl-β, and Fli1.

Gata2 plays an important role in HSC development, proliferation, and survival as well as the vascular formation in mammals (Tsai et al. 1994; Tsai and Orkin 1997; Lim et al. 2012; Johnson et al. 2012; de Pater et al. 2013; Gao et al. 2013). Due to genome duplication, two paralogues of gata2 exist in the zebrafish: gata2a and gata2b. In embryonic stages, gata2a is expressed in the LPM and the trunk vasculature, whereas the expression of gata2b is more restricted to HECs/HSPCs (Detrich et al. 1995; Brown et al. 2000; Butko et al. 2015). While vascular formation was intact, loss of Gata2b led to the down-regulation of runx1 and cmyb, a downstream target gene of Runx1 that is essential for the maintenance of HSCs (Soza-Ried et al. 2010; Zhang et al. 2011). Enforced expression of runx1 in Gata2b-deficient embryos rescued the expression of cmyb in the DA (Butko et al. 2015), indicating that Gata2b is required for HSC specification through the regulation of Runx1. In contrast, deficiency in gata2a led to severe defects in vascular formation and circulation (Zhu et al. 2011; Butko et al. 2015). Although it remains unclear if zebrafish Gata2 proteins also play a role in HSC proliferation and survival as has been shown in the mouse (Tsai and Orkin 1997; de Pater et al. 2013), it is tempting to speculate that duplication of the gata2 locus in zebrafish might have led to a functional separation between the roles of Gata2 in vasculogenesis and hematopoiesis.

3.2 Regulation of HSC Specification by Notch Signaling

Notch signaling is involved in the patterning of the trunk vasculature and HECs. Canonical Notch signaling is initiated when a membrane-bound Notch ligand (Delta or Jagged) on the signal-sending cell interacts directly with a Notch receptor on the signal-receiving cell. After binding to a Notch ligand, the Notch receptor is cleaved first by ADAM TACE metalloproteases at the S2 site, then by γ-secretase at the S3 site, which releases the Notch intracellular domain (NICD) that translocates to the nucleus and modulates transcription of Notch target genes (Brou et al. 2000; Bozkulak and Weinmaster 2009; Mumm et al. 2000; Kopan and Ilagan. 2009). Mutation or deletion of Mindbomb (Mib), which encodes a ubiquitin ligase required for Notch signal transduction (Itoh et al. 2003; Chen and Casey Corliss 2004), resulted in the loss of HECs in both the mouse and zebrafish (Burns et al. 2005; Yoon et al. 2008). Chimeric mice generated from both wild-type and Notch1-deficient cells showed no contribution of Notch1-deficient cells to adult hematopoiesis, indicating that signaling through Notch1 is required for HSC specification in a cell autonomous manner (Hadland et al. 2004).

In zebrafish, four Notch receptor genes, notch1a, 1b, 2, and 3, have been identified. A recent study revealed that three of these receptors, Notch1a, 1b, and 3, are independently required for HSC specification (Kim et al. 2014). All of these three notch genes (notch1a, 1b, and 3) are expressed in the PLPM and somite during somitogenesis and the vascular endothelium after 24 hpf. Loss of notch1a resulted in loss of runx1 and efnb2a, a marker of the aortic endothelium, revealing the requirement for Notch1a in both HSC and aortic specification, similar to Notch1 in the mouse (Kreb et al. 2000). In contrast, loss of notch1b or notch3 resulted in loss of runx1 but did not affect efnb2a expression. These data suggest that Notch1b and Notch3 are required for HSC specification but are dispensable for aortic specification. Aortic notch1b expression is regulated in part by a zinc finger transcription factor, ecotropic viral integration site-1 (Evi1), through the phosphorylation of AKT. The Evi1–pAKT–Notch1b signaling pathway has been shown to be required cell-autonomously for HSC specification (Konantz et al. 2016). Interestingly, however, Notch3 is required non-cell autonomously for HSC specification. In Notch3-deficient embryos, enforced expression of NICD in the somite rescued the expression of runx1 in the DA, whereas enforced expression in the vascular endothelium did not. Thus, while its target gene is still unclear, Notch3-dependent signaling within the somite regulates HSC specification indirectly in the zebrafish embryo (Kim et al. 2014).

The requirement for somitic Notch ligands, Delta-C (Dlc) and Dld, in HSC specification has been shown in zebrafish (Clements et al. 2011; Kobayashi et al. 2014; Lee et al. 2014). A β-catenin-independent (non-canonical) Wnt ligand, Wnt16, is expressed within the somite during somitogenesis. Morpholino knockdown of wnt16 reduced both dlc and dld in the somite and led to a loss of definitive HSPCs, but left aortic specification was intact. Injection of both dlc and dld mRNA along with wnt16 morpholino was sufficient to rescue the expression of runx1 in the DA (Clements et al. 2011). The expression of dlc, but not dld, in the somite is mediated by fibroblast growth factor (FGF) signaling. Loss of FGF signaling or fgfr4, which is regulated by Wnt16, inhibited the formation of HSPCs in the DA, while the defect in HSPCs in these embryos was rescuable by injection of dlc mRNA (Lee et al. 2014). These data suggest that Wnt16 regulates the expression of dlc through FGF signaling (Clements et al. 2011; Lee et al. 2014). To establish an HSC fate, angioblasts directly receive Dlc and Dld signaling from the somite (Kobayashi et al. 2014). During somitogenesis, angioblasts migrate axially along the ventral domain of the somite to reach the midline. Migrating angioblasts tightly adhere to the somite based on the interaction of two different cell adhesion molecules, junctional adhesion molecule 1a (Jam1a), and Jam2a. Loss of either Jam1a or Jam2a led to a delay in angioblast migration and reduction of runx1 in the DA. Jam1a-deficiency had no effect on the expression of Notch receptor or ligand genes, but runx1 expression could be restored in the absence of Jam1a through enforced expression of dlc or dld. These data suggest that the intimate contact between the angioblast and the somite by Jam1a–Jam2a binding promotes efficient Notch signal transduction from the somite to the angioblast (Kobayashi et al. 2014). Taken together, these studies suggest that the somite regulates HSC specification through the presentation of requisite Notch ligands Dlc and Dld to the angioblasts during axial migration.

A Notch ligand Jagged1 (Jag1) expressed by the aortic endothelium also plays an important role in HSC specification in both the mouse and zebrafish (Robert-Moreno et al. 2008; Espín-Palazón et al. 2014; Monteiro et al. 2016). In mice, Jag1 is required for definitive hematopoiesis through the regulation of Gata2 expression in AGM cells, but it is not required for the establishment of arterial fate (Robert-Moreno et al. 2008). Similar to murine Jag1, zebrafish Jag1a is involved in HSC specification but not arterial specification (Espín-Palazón et al. 2014; Monteiro et al. 2016). The expression of jag1a in the DA is regulated in part by tumor necrosis factor α (TNF-α) and TNF receptor 2 (TNFR2) signaling (Espín-Palazón et al. 2014), a signaling pathway that is associated with the regulation of inflammation and immunity. Loss of TNFR2 led to a reduction in jag1a expression and loss of HSPCs in the DA (Espín-Palazón et al. 2014). In addition to TNF-α, transforming growth factor-β (TGF-β) is also involved in the regulation of aortic jag1a (Monteiro et al. 2016). TGF-β1a and TGF-β1b are produced in aortic endothelial cells, while TGF-β3 is expressed in the notochord. All these autocrine and paracrine inputs of TGF-β contribute to the expression of jag1a in the aortic endothelium through the activation of TGF-β receptor 2 (TGFβR2) (Monteiro et al. 2016). These data indicate that Notch signal transduction within the aortic endothelium via Jag1a is also necessary for HSC specification.

In summary, angioblasts require at least two distinct inputs of Notch signaling to establish HSC fate in the zebrafish embryo: (1) Dlc and Dld from the somite and (2) Jag1a from the aortic endothelium. It remains unclear, however, whether these distinct Notch signaling events activate different downstream targets in HSC precursors and which receptor mediates each signaling event.

3.3 Regulatory Signaling Pathways Required for HSC Specification

The hedgehog (Hh) signaling pathway is a major regulator of cell differentiation, proliferation, and tissue polarity. Murine embryonic stem (ES) cell studies have suggested a role for Hh signaling in hematopoiesis (Maye et al. 2000; Dyer et al. 2001; Byrd et al. 2002). In zebrafish, Sonic-hedgehog (Shh) regulates the expression of vascular endothelial growth factor A (vegfa) in the somite (Lawson et al. 2002), and the Shh–Vegfa signaling axis has been implicated in the regulation of definitive but not primitive, hematopoiesis (Gering and Patient 2005). Treatment with an Hh inhibitor (cyclopamine) or mutation of Hh-related genes, such as shh or smoothened (smo), resulted in defective angioblast migration, arterial specification, and HSC specification. Similarly, inhibition of Vegf signaling also led to a reduction in efnb2a, notch3, and runx1 expression in the DA (Gering and Patient 2005). These results suggest a model in which Shh expressed by the notochord and/or floorplate induces the expression of Vegfa in the somite, which then regulates arterial and HSC programs through the Notch signaling.

In vertebrates, multiple isoforms of Vegfa are expressed in the somite. In Xenopus, a short Vegfa isoform, Vegfa-122, is responsible for the expression of arterial marker genes; an intermediate isoform, Vegfa-170, is required for HSC specification (Leung et al. 2013). A recent study showed similar divergent roles for Vegfa isoforms in zebrafish (Genthe and Clements 2017). An extracellular protein R-spondin 1 (Rspo1), which is implicated in the enhancement of Wnt/β-catenin signaling, controls HSC specification through the regulation of vegfa and wnt16 expression in the somite. Loss of rspo1 resulted in the reduction of both vegfa and wnt16 in the somite and loss of HSPCs. The effect of HSPCs in Rspo1-deficient embryos was rescued by injection of an intermediate isoform, vegfa-165, but not the short isoform, vegfa-121 (Genthe and Clements 2017). Since somitic Vegfa controls the expression of tgfb1a and tgfb1b in the DA (Monteiro et al. 2016), it is likely that Vefga-165 regulates HSC specification through the activation of the TGF-β1/Jag1a signaling pathway.

Bone morphogenetic protein (BMP) signaling is essential for patterning the ventroposterior mesoderm of the developing embryo cooperatively with the Wnt signaling. It has been shown in zebrafish that BMP4 signaling regulates HSC specification independently from the Vegfa and Notch signaling pathways (Wilkinson et al. 2009; Pouget et al. 2014). The expression of bmp4 is detected in the pronephros and the ventral mesenchyme underlying the DA at around 24 hpf (Chin et al. 1997). Loss of bmp4 led to a reduction of HECs in the DA without affecting aortic specification (Wilkinson et al. 2009). Recently, it was revealed that the expression of bmp4 in the ventral mesenchyme is negatively regulated by FGF signaling and is involved in HSC specification (Pouget et al. 2014). When FGF signaling was inhibited by heat-induction of dominant-negative (DN) Fgfr1 at around 20 hpf, the expression of bmp4 in the ventral mesenchyme was increased, resulting in the enhanced expression of runx1 in the DA. Conversely, when FGF signaling was enforced by heat-induction of constitutively active Fgfr1 at around 20 hpf, the expression of bmp4 was reduced in the ventral mesenchyme, leading to a reduction in runx1 expression in the DA. These studies also showed that FGF signaling represses bmp4 expression directly and indirectly via the induction of BMP antagonists Noggin2 and Gremlin1a expressed by the neighboring somite (Pouget et al. 2014).

Recently, proinflammatory signaling has been implicated in the regulation of HSC specification in both the mouse and zebrafish (Espín-Palazón et al. 2014; Sawamiphak et al. 2014; Li et al. 2014; He et al. 2015). As described in Sect. 3.3.2, TNF-α acts as upstream of the Notch ligand Jag1a to regulate HSC specification (Espín-Palazón et al. 2014). In contrast, the proinflammatory cytokine interferon-γ (IFN-γ) acts downstream of Notch and regulates HSC specification (Sawamiphak et al. 2014). Deficiency in IFN-γ or its receptor Crfb17 led to reduced runx1 expression in the DA, while enforced expression of IFN-γ increased the number of HSPCs. Furthermore, HSPC defects caused by the inhibition of Notch signaling could be rescued by enforced production of IFN-γ. It has also been shown that IFN-γ/Crfb17 signaling activates Stat3, an atypical transducer of IFN-γ that is required for HSC development. These studies suggest that IFN-γ, controlled by Notch signaling, regulates HSC specification through the activation of Stat3 (Sawamiphak et al. 2014).

In addition to regulating many diverse functions such as cell proliferation and differentiation, adenosine signaling has recently been demonstrated to be required for HSC specification (Jing et al. 2015). The adenosine receptor A2b is expressed on endothelial cells prior to HSC emergence. Loss of adenosine signaling or A2b led to a reduction in scl-β and runx1 expression without affecting vascular formation, while elevated adenosine signaling enhanced the production of HSPCs. Adenosine signaling activates the cyclic AMP (cAMP)–protein kinase A (PKA) pathway to promote the production of Cxcl8, a chemokine expressed by endothelial cells. Injection of cxcl8 mRNA was sufficient to rescue defective HSPC formation in A2b–deficient embryos. These results suggest that adenosine signaling in endothelial cells regulates HSC specification via the production of Cxcl8 (Jing et al. 2015).

As shown in Fig. 3.1, although a number of extrinsic signaling molecules that regulate HSC specification have been identified in the zebrafish embryo, only Notch (through Dlc, Dld, and Jag1a) and BMP4 regulate HSC programs directly in angioblasts/HECs. While the possible presence of other direct signaling molecules still cannot be excluded, the precisely timed and tuned coordination of Notch and BMP4 signaling may be one of the most important factors necessary to establish the HSC fate in angioblasts.

Fig. 3.1
figure 1

The process of hematopoietic stem cell specification

Hematopoietic stem cells (HSCs) are specified from the angioblast, which arises in the posterior lateral plate mesoderm. Angioblasts expressing Etv2 and Scl-α axially migrate along the ventral domain of the somite to reach the midline. Angioblasts experience an initial Notch signaling event via the interaction with somitic cells during axial migration, leading to the expression of Gata2b. Angioblasts also begin to express Scl-β through the adenosine signaling to become the hemogenic endothelial cells (HECs). Within the aortic floor, HECs receive a second Notch signal from the aortic endothelial cells to drive the expression of Runx1. At almost the same time, BMP4 signaling from the ventral mesenchyme also induces Runx1 expression. Nascent Runx1+ HSCs then enter maturation and proliferation steps within the dorsal aorta and begin to express c-Myb. All boxed transcription factors (Etv2, Scl-α, Scl-β, Gata2b, Runx1, and c-Myb) are essential for HSC development. Among them, scl-β, gata2b, runx1, and cmyb are used as the marker for HSCs in zebrafish embryos

Full size image

4 Maintenance and Expansion of HSCs in the Zebrafish Embryo

4.1 Maintenance and Expansion of HSCs Within the Dorsal Aorta

After specification, nascent HSCs require extrinsic signals for their maturation, maintenance, and proliferation within the DA and CHT. The effects of blood flow and the shear stress on the endothelium have important roles in the maintenance of developing HSCs in the DA (North et al. 2009; Wang et al. 2011). Mutation of the cardiac troponin T2a (tnnt2a) gene, which results in a lack of heartbeat, led to a drastic reduction of HSPCs in the DA at 36 hpf although initial runx1 expression (24 hpf) in the DA was normal in these embryos (North et al. 2009; Wang et al. 2011). This indicates that blood flow is necessary for HSC maintenance/proliferation but not specification. Shear stress on blood vessel walls associated with blood flow induces the production of nitric oxide (NO) through nitric oxide synthases (NOS) (Moncada and Higgs 2006). Inhibition of NOS led to decreased numbers of HSPCs, and chemical production of NO caused a significant increase in HSPCs. In tnnt2a mutant embryos, defective HSPCs with reduced expression of a NOS gene (nos1) could be rescued by enforced production of NO. Moreover, nos1-deficient transplanted cells failed to contribute to HSPCs. These results indicate that NO signaling mediated by shear stress regulates the maintenance/proliferation of HSPCs cell-autonomously (North et al. 2009; Wang et al. 2011).

As described in Sect. 3.3.2., Notch signaling pathways are required for HSC specification cell-autonomously and non-cell autonomously. It has been reported, however, that Notch signaling must be down-regulated in HECs during the maturation of HSCs in both the mouse and zebrafish (Richard et al. 2013; Lizama et al. 2015; Zhang et al. 2015). G protein-coupled receptor 183 (Gpr183, also known as Ebi2; Epstein-Barr virus-induced gene 2) is activated by its ligand 7α-25-OHC and promotes the degradation of Notch1 through the proteasome pathway. Mutation of gpr183 in zebrafish embryos resulted in the increased expression of Notch target genes (efnb2a, her6, and her9) in the DA and loss of HSPCs, without affecting HSC specification (Zhang et al. 2015). The requirement for Notch down-regulation was further confirmed by enforced expression of heat-inducible NICD and DN-mastermind-like (MAML), which can promote and inhibit the activation of Notch target genes, respectively. Enforced expression of NICD at an early stage (20 hpf) led to an increased number of HSPCs in the DA; DN-MAML expression at the same stage resulted in the loss of HSPCs. By contrast, enforced expression of NICD at a later stage (26 hpf) led to a reduction of HSPCs, whereas DN-MAML expression increased HSPC numbers (Zhang et al. 2015). Thus, Notch signaling is not continuously required during HSC development, and down-regulation of Notch is necessary for the final maturation of HSCs.

A chemical genetic screen in zebrafish embryos has identified the role of prostaglandin in the HSC proliferation. Prostaglandin E2 (PGE2) is the main effector of prostanoid and is regulated by both cyclooxygenase 1 (Cox1, also known as Ptgs1) and Cox2 (also known as Ptgs2a). While treatment of zebrafish embryos with PGE2 increased the number of HSPCs, Cox inhibition decreased HSPC numbers (North et al. 2007). PGE2 regulates the Wnt signaling pathway through the stabilization of β-catenin (Goessling et al. 2009). β-catenin-dependent (canonical) Wnt signaling is crucial for the development, maintenance, and proliferation of HSCs (Ruiz-Herguido et al. 2012; Luis et al. 2009). Wnt signaling is induced when a Wnt ligand binds to a Frizzled family receptor. Upon activation of the receptor, β-catenin becomes stabilized and enters the nucleus, where it interacts with a transcription factor T-cell factor (TCF, also known as LEF; lymphoid enhancer binding factor) to drive the expression of Wnt target genes. In zebrafish embryos, induction of a membrane-level Wnt antagonist, dickkopf1 (dkk1), led to the reduction of HSPCs in the DA, an effect that could be rescued by treatment with PGE2. In contrast, induction of axin1, which encodes a protein that promotes the destruction of β-catenin, also led to a reduction of HSPCs, but this effect could not be rescued by PGE2 treatment. These data suggest that PGE2 interacts with Wnt signaling pathway to block the degradation of β-catenin (Goessling et al. 2009). Recently, the requirement for a canonical Wnt ligand, Wnt9a, in HSC proliferation has been shown in zebrafish (Grainger et al. 2016). Wnt9a is expressed by the somite and acts as a paracrine signal to induce the expression of the cell cycle regulator c-Myc in angioblasts during angioblast migration. When Wnt signaling was inhibited by enforced expression of DN-tcf or injection of wnt9a morpholino, the expression of cmyb in the DA was normal until 30 hpf but reduced after 31 hpf. This effect was rescued by enforced expression of cmyc, suggesting that the Wnt9a/c-Myc signaling pathway is not necessary for initial HSC specification but is required for the proliferation of HSPCs in the DA (Grainger et al. 2016). Together, canonical Wnt signaling promotes the proliferation of HSCs in the DA through the regulation of c-Myc, and this signaling pathway is stabilized by PGE2 during somitogenesis.

Hypoxic stress during the developmental stage is also important factor for proliferation of HSPCs (Kwan et al. 2016). Hypoxic stress activates the hypothalamic-pituitary-adrenal/interrenal (HPA/I) stress response axis, which is regulated by the neurotransmitter serotonin within the central nerve system (CNS). Exposure to serotonin or a hypoxia mimetic cobalt chloride (CoCl2) induced the production of cortisol and increased HSPC numbers in the DA. Inhibition of neuronal tryptophan hydroxlyase (Tph), a synthesizer of serotonin, led to reduced HSPC numbers. Mutation of the nr3c1 gene, which encodes a receptor that can bind to cortisol (glucocorticoid receptor, GR), also resulted in the reduction of HSPCs. Selective induction of nr3c1 under the runx1 promoter rescued the number of HSPCs in nr3cl mutant embryos, indicating that hypoxia-induced stress signaling directly stimulates the proliferation of HSPCs within the DA (Kwan et al. 2016).

Vitamin D is also a positive regulator of HSC proliferation during the developmental stage (Cortes et al. 2016). Vitamin D synthesis begins with the transformation of 7-dehydrocholesterol to the non-active vitamin D precursor cholecalciferol (D3) by UV radiation in the skin. D3 is then modified by the cytochrome P450 enzymes 2R1 (Cyp2r1) and 27B1 (Cyp27b1) to generate 1, 25-hydroxy vitamin D (1,25(OH)D3), the active form of vitamin D. Treatment of zebrafish embryos with 1,25(OH)D3 resulted in an increase in HSPCs in the DA. In contrast, loss of vitamin D receptor a (vdra) or cyp27b1 caused a reduction in HSPCs. Increased numbers of HSPCs were also observed when embryos were treated with calcipotriol, a vitamin D3 analog that has been shown to be 100-fold less calcemic than 1,25(OH)D3, indicating that vitamin D regulates HSC proliferation independently of Ca2+ regulation. Taken together, vitamin D3 acts directly on HSPCs, independent of calcium regulation, to increase proliferation in the DA (Cortes et al. 2016).

The extracellular matrix (ECM) is an important component of the hematopoietic microenvironment that can sequester cytokines and regulate signal transduction (Davis and Senger 2005). The ECM structure is dynamic and actively remodeled by degrading proteases known as matrix metalloproteinases (MMPs) (Heissig et al. 2003). In zebrafish embryos, Mmp2 is expressed in the trunk vasculature and mesenchyme and is involved in the egress of HSPCs from the DA. Inhibition of Mmp2 led to the accumulation of an ECM component fibronectin, resulting in the retention of HSPCs and formation of an abnormal pattern of hematopoietic clusters in the DA. Inhibition of Mmp2 also led to the delay of HSPC colonization in the CHT. These data suggest that Mmp2 facilitates the process of EHT by remodeling ECM in the DA (Theodore et al. 2017).

4.2 Proliferation of HSCs Within the Caudal Hematopoietic Tissue

The CHT contains a sinusoidal structure of the vascular endothelium and is equivalent to the FL in mammals (Murayama et al. 2006; Jin et al. 2007). Live-imaging analysis of HSPCs in the CHT revealed the specific interaction of HSPCs with vascular endothelial cells and mesenchymal stromal cells. When HSPCs reach the CHT, they are enwrapped in endothelial cells (described as“endothelial cuddling”). This interaction induces and maintains contact between an HSC and a single mesenchymal stromal cell, which can induce and orient the cell division of the HSC (Tamplin et al. 2015). Chemical inhibition of Cxcl12/Cxcr4 signaling, which is known to play a role in the homing and mobilization of HSCs (Lapidot et al. 2005; Sugiyama et al. 2006), led to a reduction of HSPCs in the CHT. In contrast, inhibition of TGF-β expanded the HSPC population, which is consistent with mouse studies that showed that TGF-β signaling negatively regulates HSPC proliferation (Soma et al. 1996; Yamazaki et al. 2011). These data suggest that the expansion of HSCs is largely dependent on vascular and/or perivascular niches in the CHT (Tamplin et al. 2015).

The Cxcl8/Cxcr1 signaling regulates HSPC colonization and proliferation in the CHT (Blaser et al. 2017). Heat-shock induction of Cxcl8 or Cxcr1 after 36 hpf increased HSPC numbers in the CHT (Blaser et al. 2017). In contrast, loss of Cxcl8 led to the reduction of HSPCs in both the DA and CHT, due to defects in HSPC specification and probably also due to impaired HSPC colonization of the CHT (Jing et al. 2015; Blaser et al. 2017). Interestingly, neither cxcl8 nor cxcr1 was expressed in HSPCs at 72 hpf when HSPCs are proliferating in the CHT, indicating that Cxcl8/Cxcr1 signaling regulates HSPC proliferation non-cell autonomously. Enhanced Cxcl8/Cxcr1 signaling increased CHT volume and the expression level of cxcl12a within the endothelium, suggesting that Cxcl8/Cxcr1 signaling promotes the proliferation of HSPCs by remodeling of environmental niche cells in the CHT (Blaser et al. 2017).

Endothelial cells within the CHT produce an essential cytokine to promote HSPC proliferation. A basic helix–loop–helix transcription factor Tfec (transcription factor EC) is expressed by endothelial cells present within the CHT. Enforced expression of tfec led to increased numbers of HSPCs, while tfec mutants exhibited reduced definitive hematopoiesis (Mahony et al. 2016). Tfec mediates the expression of kit ligand b (kitlgb, also known as scf; stem cell factor), which encodes a cytokine that can bind to c-Kit and plays an important role in HSC proliferation and maintenance (Ding et al. 2012). Injection of kitlgb mRNA in tfec mutant embryos rescued HSPC numbers, suggesting that Tfec expressed by endothelial cells regulates HSPC proliferation via the production of Kitlgb in the CHT (Mahony et al. 2016). Thus, endothelial cells within the CHT play an essential role in the expansion of HSPCs.

Mmp9 is an ECM remodeling protein that is expressed by primitive neutrophils present within the CHT. Inhibition of Mmp9 increased the number of HSPCs in the CHT but decreased them in the thymus. Interestingly, the total number of HSPCs in the whole embryo was unchanged in these embryos, suggesting that Mmp9 is involved in the mobilization of HSPCs from the CHT. Similar phenotypes were also observed when cxcl12b was overexpressed in the endothelium. In addition, the accumulation of HSPCs by Mmp9 inhibition was ameliorated by knockdown of cxcl12a, raising the possibility that Mmp9 expressed by primitive neutrophils controls HSPC mobilization through the repression of Cxcl12/Cxcr4 signaling (Theodore et al. 2017).

Taken together, the maintenance and proliferation of HSCs are very dependent on environmental factors, such as blood flow, hypoxia, and niche cell functions. This suggests that although early hematopoiesis occurs in a transient hematopoietic organ, the signaling cascades that can respond to blood cell requirements are mostly established during embryonic stages. Signaling molecules involved in the maturation, maintenance, and proliferation of HSCs are summarized in Fig. 3.2.

Fig. 3.2
figure 2

Maintenance and proliferation of hematopoietic stem cells

The maintenance of hematopoietic stem cells (HSCs) within the ventral floor of the dorsal aorta (DA) requires shear stress associated with the blood flow, leading to the production of nitric oxide through nitric oxide synthases (NOS). While Notch signaling is up-regulated in the specification process, it is, in turn, down-regulated via the Gpr183 signaling pathway in the maturation process of HSCs. Wnt9a signaling from the somite induces the expression of c-Myc, which promotes the proliferation of HSCs in the DA. In addition, cortisol and vitamin D3 also directly influence the HSC proliferation in the DA. After budding from the DA, HSCs migrate via the circulation to the caudal hematopoietic tissue (CHT), where environmental niche cells (endothelial cells, mesenchymal stromal cells, etc.) produce cytokine signals that accelerate HSC proliferation. Cxcl8-Cxcr1 signaling and Kitlgb positively regulate and TGF-β signaling negatively regulates the expansion of the HSC pool. Cxcl12-Cxcr4 signaling is involved in the homing and mobilization of HSCs

Full size image

5 Conclusion

Our understanding of the molecular mechanisms of vertebrate HSC development has been largely improved by the zebrafish system over the past decade. This contribution could not be possible without the unique and wide variety of experimental approaches available to the zebrafish model. These include innovative approaches as well as well-established tools and methodologies, including transgenic/mutant animals, live-imaging, microinjection, transplantation, cell culture systems, chemical and forward genetic screens, and genome editing systems. In summary, the zebrafish represents an ideal system to dissect the fundamental mechanisms of HSC development and will continue to be leveraged for its unique attributes in order to make new discoveries regarding the molecular cues needed to instruct and expand HSCs from pluripotent precursors.