Journal list menu
Concise Review: From Greenhouse to Garden: The Changing Soil of the Hematopoietic Stem Cell Microenvironment During Development
Abstract
The hematopoietic system has been intensely studied for many decades. For this reason, it has become the best understood stem cell-derived system that serves as a paradigm for stem cell biology and has found numerous applications in the clinics. While a lot of progress has recently been made in describing the bone marrow components that maintain and control blood stem cell function in the adult, very little is currently known about the regulatory microenvironment in which the first adult-repopulating hematopoietic stem cells are formed during development. Knowledge of these processes is crucial for understanding the basic regulation of hematopoietic stem cell production and behavior and to allow their in vitro expansion and generation from embryonic stem cells or iPS cells for clinical and research purposes. This review summarizes the recent advances that have been made in defining the cellular components, as well as the soluble and physical factors, that are part of the niche involved in regulating hematopoietic stem cell generation in the embryo. The findings are compared with what is known about the adult bone marrow niche to find common pathways for stem cell regulation, but also to highlight processes uniquely required for de novo hematopoietic stem cell generation, as these are the conditions that will need to be recreated for the successful production of blood stem cells in culture. Stem Cells 2014;32:1691–1700
Introduction
The hematopoietic system is a highly dynamic system that is maintained at a rapid rate of turnover, producing millions of new cells every second to maintain homeostasis and respond to challenges such as infection, respiratory stress, or bleeding. This critical balance is maintained and supplied by a hierarchy of progenitors, stemming from a relatively small population of self-renewing, multipotent, largely quiescent cells, known as hematopoietic stem cells (HSCs). While the majority of studies to date focus on these adult HSCs, definitive HSCs arising in the embryo can provide a wealth of information to further our understanding of the production, maintenance, and cell fate decisions of HSCs. This information is particularly useful for identifying methods of generating HSCs from pluripotent stem cells. A recent review by Cao et al. 1 examines the hematopoietic microenvironment throughout development with extensive coverage of the fetal liver (FL) and beyond. Here, we provide a complementary review concentrating on the earlier origins of HSCs in the aorta-gonads-mesonephros (AGM).
Hematopoiesis in Development
Broadly speaking, hematopoiesis occurs in two waves during development. The first, known as primitive hematopoiesis, is transient and supplies the needs of the early embryo, including tissue oxygenation during rapid growth. The second wave is known as definitive hematopoiesis and culminates in the production of definitive HSCs, which are capable of long-term multilineage repopulation of irradiated adults following injection into the bloodstream (reviewed in Lensch and Daley 2).
In the mouse embryo, the first definitive HSCs appear autonomously in the AGM region at embryonic day 10.5 (E10.5) 3, 4. From E11, HSCs are also found in the yolk sac 5, 6, placenta 7, 8, and FL 4, although the AGM remains the only region of the wild-type embryo that can expand HSCs in organ culture 3, 9. While the AGM is capable of expanding HSCs at E11 10, its primary role in vivo appears to be to promote de novo HSC generation.
Unlike the AGM, the placenta hosts mitotically active hematopoietic cells 11. There is also evidence to suggest that it is capable of de novo generation of hematopoietic stem/progenitor cells (HSPCs) in addition to receiving HSCs generated elsewhere, although these results should be interpreted with caution as these experiments were performed in the confines of the Ncx1−/− mouse model, which lacks a heartbeat and is not viable beyond E10.5 12.
The yolk sac is the origin of the primitive hematopoietic wave and also of some transient definitive hematopoietic cells (reviewed in 13). Whether it can generate HSCs autonomously is still a matter of debate 3-5.
The FL is not capable of de novo HSC generation 14 and is instead colonized by HSCs originating from other sites 10. It hosts expansion of HSCs and the majority of HSCs here are considered to be in cycle 15. Unlike the AGM 16, the FL plays an important role in generating differentiated blood cells in a manner similar to adult hematopoiesis. From E17.5, HSCs start to localize to the bone marrow (BM) 8, 17, which is where they continue to be found postnatally.
Definitive HSCs Have Distinct Properties in the Adult and the Embryo
The earliest definitive HSCs must ultimately expand to give rise to the full complement of adult HSCs and so are highly proliferative. In contrast, adult BM HSCs are thought to divide rarely. Indeed, HSCs isolated from E14.5 FL expand at a faster rate when compared with adult HSCs 18. These fetal characteristics persist to around 3 weeks after birth, but by 4 weeks, HSCs behave similarly to adult HSCs. The HSCs found in the AGM have a number of distinct properties that suggest that they may constitute a third, functionally and phenotypically discrete population of HSCs (summarized in Table 1). This has recently been demonstrated by the expression of the cell surface marker CD41: repopulating HSCs were all found in the CD41int fraction of AGM cells from E11 embryos, whereas in the E14 FL only CD41− cells were capable of repopulating irradiated recipients 19. Further evidence for a distinction between E11.5 and adult HSCs arises from their differing expression of the transcription factor Gata3. Gata3 has been shown to be expressed in adult BM HSCs 20, while it is absent from E11.5 AGM HSCs 21. Supporting the idea that these populations are also functionally dissimilar is our finding that AGM HSCs respond differently from E14.5 FL and adult BM HSCs to the deletion of the cyclin-dependent kinase inhibitor, p57Kip2 (Cdkn1c). Loss of p57Kip2 in both E14.5 FL and adult BM HSCs leads to reduced self-renewal and maintenance, respectively 22, 23, whereas we have observed an expansion of HSCs in p57Kip2 null AGMs 24. More recently, it has been shown that despite expression of both retinoic acid receptors (RARs) α and γ on HSCs, the functionally predominant RAR transitions throughout development. HSC induction in the AGM is primarily via RAR-α, while HSC expansion in the FL occurs through both RAR-α and RAR-γ. Finally in adults, RAR-γ is required for HSC maintenance 25-27. While some of the different responses to the same molecules may be due to the expression of distinct cell-intrinsic factors, differences in the regulatory microenvironment will also make a major contribution.
| E11 AGM HSCs | E14 fetal liver HSCs | Adult bone marrow HSCs | |
|---|---|---|---|
| CD34 expression | CD34+ [85] | CD34+ [85] |
CD34− when quiescent CD34+ when activated [86] |
| Expression of SLAM markers | CD150−CD48− [87] | CD150+CD48−CD244− [88] | CD150+CD48−CD244− [38] |
| CD41 expression | Intermediate [19] | Negative [19] | CD41 marks a subset of LT-HSCs that are myeloid biased [89] |
| Behavior in vivo | HSCs emerge/mature from pre-HSCs [3] | No de novo generation, but rapid expansion of HSCs [18] | Largely quiescent unless activated [90, 91] |
| Response to p57kip2 knockout | Increased repopulation ability [24] | Reduced repopulation in tertiary recipients [22] | Conditional knockout results in reduced repopulation capacity [23] |
| Gata3 expression and effects | No expression in HSCs [21] Gata3 knockout embryos have reduced HSC emergence [21] | Fetal livers from Gata3 knockout embryos retain long term repopulating potential but have reduced self-renewal ability within the adult niche [92] |
Gata3 expression is found in LT-HSCs [20, 93] Gata3 is dispensable for HSC self-renewal [94] but is thought to play a role in HSC maintenance and quiescence [92, 95] |
| Retinoic Acid Receptors | RARα is the primary receptor mediating HSC expansion [27] | HSC expansion occurs via both RARα and RARγ in E13.5 fetal liver [27] | RARγ is essential for HSC maintenance [25, 26] |
- Abbreviations: AGM, aorta-gonads-mesonephros; HSC, hematopoietic stem cell.
Environmental Regulation of HSCs
Stem cell regulation is achieved through the integration of intrinsic and extrinsic factors. The concept of a stem cell “niche” was first put forward in the field of hematopoiesis in 1978 40. The presence of this local tissue microenvironment where stem cells reside is now thought to be an important factor in regulating stem cell behavior in a diverse range of stem cell systems 41. The niche is the convergence of metabolic, humoral, paracrine, neural, structural, and physical signals that act to regulate stem cells 42. For HSCs, it is predominantly comprised of stromal cells, which provide cell-cell contact for HSC anchorage, signaling, and trafficking; and an extracellular matrix, which contains soluble factors secreted by these cells. These factors can either act directly on HSCs, or indirectly via the microenvironment. Other cell types in the vicinity may also provide signals that affect HSC behavior.
Given the innate ability of HSCs to behave differently according to physiological need, and the known role of the niche in influencing HSC activity, it is unsurprising that while the HSC niches in the embryo and adults share a number of characteristics, they must be custom-made to support the features of stem cell activity that predominate at the time. This is most obvious in the embryo, where HSC niches in different anatomical locations support different functions. Meanwhile, the adult BM must both support highly quiescent HSCs and also manage their differentiation and mobilization into the blood, seemingly via different compartments. Understanding specific environmental cues that affect HSC birth and behavior is critical for recreating culture conditions conducive to generating and expanding HSCs from both hematopoietic tissues and pluripotent stem cells in vitro.
Here, we compare different soluble, cellular, and mechanical components of the embryonic and adult HSC microenvironments, focusing mainly on the AGM in the embryo, as this is the site where definitive HSCs first emerge. While the commonalities may be informative about the elements that are essential for HSC maintenance, differences may highlight factors that support the de novo generation of HSCs.
Soluble Factors
Soluble factors can affect various facets of hematopoiesis and, interestingly, may behave differently in adult and embryonic microenvironments, suggesting the presence of different downstream targets in either the HSCs or their niches. Identifying sources of soluble factors that influence hematopoiesis also provides clues as to which cell types make up the hematopoietic microenvironment. Table 2 summarizes the roles of a number of important soluble factors in the AGM, FL, and adult microenvironment. While adult and developmental HSC niches show some overlap in the soluble factors involved in HSC regulation, this overlap may be underrepresented as roles for these factors have not yet been fully investigated in both systems. However, some factors, such as IL-3, appear to have opposing effects on the more proliferative embryonic HSCs compared with the more quiescent adult HSCs.
| Signal | Cell of origin | Effect on hematopoiesis | Reference |
|---|---|---|---|
| Bone morphogenic protein 4 (BMP4) | Ventral mesenchyme | Preserves HSPC potential | [96] |
|
Stromal cells Hepatocytes |
HSC maintenance Stress Erythropoiesis However, conditional loss of Smad1 and Smad5 (members of BMP signalling pathway) in FL haematopoietic cells did not affect HSCs |
[97–99] | |
| Osteoblasts Endothelial cells Megakaryocyes | Reduced LSK cells Defective microenvironment Reduced support of wildtype HSCs | [100] | |
| Catecholamines | Developing SNS | Increased HSC production | [21] |
| Not locally produced | Ottersbach (unpublished) | ||
| SNS | Enhances HSC mobilization in response to G-CSF Increased HSC proliferation Suppresses osteoblasts and reduces bone CXCL12 | [58, 59] | |
| Delta-like 1 (Dlk1) | Developing SNS Smooth muscle cells Ventral mesenchyme | Reduced HSPC production - but this effect may be limited to the membrane bound form | [70] |
| Hepatic stem/progenitor cells Stromal cells | HSC maintenance and expansion | [68, 69] | |
| Stromal cells | Inhibits growth of Lin- cells in specific cell contexts Inhibits colony formation by CSF in the presence of SCF | [69, 101] | |
| Hedgehog | Developing gut | Increased HSC emergence Possible increased HSC expansion | [66] |
| Stromal cells | HSPC proliferation | [102] | |
| Stromal cells | Exogenous Sonic hedgehog treatment promotes HSC expansion, but conditional deletion of Smoothened had no apparent effect on hematopoiesis Role in adult hematopoiesis is still controversial | [103, 104] | |
| Insulin-like growth factor 2 (Igf-2) | Intra-aortic clusters Endothelial cells Smooth muscle cells | Preferential promotion of more primitive progenitors and of erythroid progenitors | [24] |
| Stromal cells | Stimulates HSC expansion | [68, 105] | |
| Non-HSC stromal cells | Expands HSCs ex vivo | [105] | |
| Interleukin-3 (IL-3) |
Stomach Aortic lumen Cardinal veins Vitelline artery |
Increased HSC survival Increased HSC proliferation Increased repopulation capability |
[9, 14] |
| No expression detected | Ottersbach (unpublished) | ||
| T-cells | Contradictory findings showing increased proliferation and differentiation or increased self-renewal and in some cases reduced repopulation capability | [106–108] | |
| Retinoic acid (RA) | Unknown | Promotes HSC development from hemogenic endothelium and pre-HSCs via RARα and βcatenin-dependent Wnt signalling | [27] |
| Unknown | Promotes erythropoiesis via RARα and erythropoietin from E9.5 to E11.5 | [109] | |
| Unknown |
Enhances maintenance and self-renewal RARγ but not RARα is critical for balance between self-renewal and differentiation |
[25, 110] | |
| Nitric oxide | Endothelium - in response to shear stress | Increased HSPCs | [75, 76] |
| Stromal cells | Expression of nNOS in FL stromal cell lines correlates with their ability to support HSCs | [77] | |
| Stromal cells | Extends replating ability of HSCs As in FL, expression of nNOS in BM stromal cell lines correlates with their ability to support HSCs | [77] | |
| Thrombopoietin (Thpo) | Expression of Thpo RNA detected in E10.5 & E11.5 AGMs | Delayed production of HSCs in Mpl−/− embryos(Mpl is the receptor for Thpo) | [111] |
| SCF+/DLK1+ Fetal hepatic stem/progenitor cells | HSC expansion | [68] | |
| Osteoblasts Stromal cells |
Increased survival Marginal increase in expansion Thpo ko mice have 4× reduced HSCS and 40× reduced repopulation capacity |
[112, 113] | |
| Angiopoietin (Ang-1) | Unknown | Tie1 and Tie2 receptors dispensable for hematopoiesis | [114] |
| Not reported | Not reported | ||
|
Osteoblasts Endothelial cells Mesenchymal progenitors |
Enhanced repopulation ability and quiescence Tie1 and Tie2 receptors required for hematopoiesis | [114–116] |
- Abbreviations: CSF, colony-stimulating factor; HSC, hematopoietic stem cell; HSPC, hematopoietic stem/progenitor cell; SNS, sympathetic nervous system.
Niche Cell Types
Cells in the HSC niche can use physical and chemical signals to influence hematopoiesis. These signals can originate not only from the stromal or endothelial cells immediately surrounding them, but also other nearby cells. Gene expression profiling of the AGM around the time of HSC emergence has revealed that upregulated genes are not limited to those of the developing hematopoietic system, but also include the developing vascular, muscular, skeletal, and nervous systems 24. Given the short distances between the developing organs in the AGM, it is likely that signals from one organ system affect cells in another, while the vascularity and nerve supply of the BM likewise expose it to alternative extrinsic signals. Figure 1 depicts some of the main cell types and signals emanating from them in the AGM.
Osteogenic Cells
In the adult BM HSC niche, as early as 1975, a supportive role for osteogenic cells was inferred by the finding that progenitors and transplanted HSCs are concentrated in the endosteal region of the BM 73, 74. Increasing the frequency of osteoblasts in the BM results in a parallel increase in HSCs, which supports a specific role for osteoblasts as niche cells; however, other studies have shown that depletion of the osteoblast lineage does not necessarily lead to reduced HSC function 75-77.
Regulators of osteogenesis were also found to be upregulated at the time of HSC emergence 24; however, the importance of this has currently not been investigated further.
Endothelial Cells
There is now a substantial amount of evidence to suggest that AGM HSCs are derived from specialized endothelial cells, known as hemogenic endothelium, which can produce blood cells via a process termed endothelial-hematopoietic transition. This is thought to result in the appearance of intra-aortic clusters of cells that express endothelial as well as hematopoietic markers. These clusters are particularly abundant in the middle region of the dorsal aorta around the junction with the vitelline artery, which is also the region where HSCs are concentrated 24, 78, and they are absent in embryos deficient in HSC production 79. Whether endothelial cells and cells within the intra-aortic clusters also act as mininiches that support emerging HSCs is currently unknown.
In the BM, more than 90% of CD150+CD48−Lin− HSCs (SLAM HSCs) can be found within five cell diameters of sinusoidal endothelial cells, whereas only 8%–21% are within five cell diameters of the endosteum 32. These sinusoidal capillaries allow the transition of HSCs between the BM and the peripheral blood system. The cells that comprise these sinusoids are distinct from other vascular endothelial cells and express the VEGFR3 receptor. Endothelial cells have been shown to be capable of supporting HSCs in vitro 80-82, and mouse models targeting HSC supportive signaling in endothelial cells show a defect in HSC frequency or function 83, 84. Furthermore, transfusion of endothelial cells following irradiation leads to improved recovery 85.
Mesenchymal Stem/Stromal Cells
Stromal cells are derived from mesenchymal stem/stromal cells (MSCs): multipotent cells that can differentiate along connective tissue lineage pathways 86. During development, mesenchymal stem/progenitor cells have been identified in hematopoietic organs and circulating blood at times when they harbor HSC activity 87. Although clonality has not been examined, only the AGM region is able to give rise to osteogenic, chondrogenic, and adipogenic cells at E11. Of hematopoietic regions in development, the same property is found in E12–14 circulating blood, E14 FL, neonatal, and adult BM, but not E12 yolk sac 87. MSCs have also been isolated from human placenta 88. Crisan et al. have prospectively identified MSCs in human organs including the placenta based on CD146, Neurogenin 2 (NG2), and platelet-derived growth factor receptor β (PDGFRβ) expression in the context of absent hematopoietic, endothelial, and myogenic markers 89. These multipotential cells are principally pericytes and can also be found in nonhematopoietic tissues including pancreas and skeletal muscle, although the hematopoietic-supportive ability of MSCs derived from such areas is unknown. We have also detected expression of each of the above markers (CD146, NG2, and PDGFRβ) in perivascular cells in the E11 AGM (Fig. 2).
The coincident timing and location of HSCs and MSCs suggest that the production and support of these two cell populations may be linked, although the presence of MSCs in Runx1-deficient embryos, which lack HSCs, indicates that they arise independently of definitive hematopoiesis 87. Apart from having a similar pattern of distribution, MSCs in development are also similar to HSCs in that they appear in waves. The first wave of MSCs in mice arises from E9.5 Sox1+ neuroepithelial cells, partly through a neural crest intermediate stage 90. By E14.5, there is a second population of MSC progenitors in the embryo, which is derived from an alternative, but as yet unidentified, source. While the initial MSC progenitor population is present after birth, it rapidly decreases in the neonate, where MSCs are found within the BM. The difference between these two populations, both in general and in relation to hematopoiesis, remains to be investigated.
In the adult mouse, a number of candidate MSC populations have been put forward as HSC niche cells. These include a population of subendothelial MSCs, equivalent to the CD146+ adventitial reticular cells found in human BM 91, known as CXCL12 abundant reticular cells in mice 92, 93. Mouse BM MSCs can also be prospectively enriched by the cell surface phenotype PDGFRα+Sca-1+CD45−Ter119−, termed PαS cells 94. A further subset of MSCs proposed to play a role in supporting BM HSCs are those that express the filament protein Nestin. These Nestin+ cells are closely associated with catecholaminergic nerve fibers and have been strongly implicated as candidate components of the perivascular HSC niche. More recently, a PDGFRα+CD51+ subset of CD146+ cells has been implicated as MSCs, which overlap with Nestin+ cells 95. A recent study proposes that MSCs can be subdivided based on marker expression (CD105, Thy1, and 6C3) and that these subpopulations collaborate to control different aspects of HSC behavior 96.
Studies assessing the effects of CXCL12 deletion in various candidate niche populations support the idea that there are at least two distinct BM niches: an osterix-expressing stromal cell niche that primarily supports lymphoid progenitors and a perivascular niche that harbors and supports HSCs 83, 97. Aside from endothelial cells, this perivascular niche is thought to comprise a combination of cell types, including MSCs and cells of the nervous system. It has since been proposed that rare NG2+ pericyte cells associated with arterioles, which are found preferentially in endosteal regions, may be key niche components for maintaining HSC quiescence 98.
Sympathetic Nervous System
Investigation of the role of the transcription factor Gata3 has revealed that signals from the sympathetic nervous system regulate HSC emergence in the embryo 21. Gata3−/− embryos suffer from impaired sympathoadrenal differentiation and, consequently, lack of catecholamine production. They have reduced functional and phenotypic HSCs compared with wild-type embryos. However, HSC numbers in these embryos can be rescued by the addition of catecholamines in vitro or in vivo. AGMs from Th−/− embryos, which cannot synthesize catecholamines, and wild-type AGMs cultured as explants in the presence of a tyrosine hydroxylase inhibitor, also have reduced repopulating ability after transplantation. Moreover, nascent HSCs express beta2-adrenoceptors that are stimulated by catecholamines. Therefore, release of catecholamines from the developing sympathetic nervous system promotes HSC production in the AGM.
In the adult HSC niche, the sympathetic nervous system has also been shown to play a role in HSC regulation 48, 49, 99. Pharmacological or genetic ablation of catecholamine signaling causes osteoblast suppression, CXCL12 downregulation, and inhibited HSPC mobilization following the administration of granulocyte colony-stimulating factor 48. In addition, denervation to remove Schwann cells around BM nerves compromises HSC quiescence 99. While the mechanism proposed was via activation of TGFβ, the effect of removing the nerve cells cannot be discounted, especially since a direct effect of catecholamines on HSC proliferation has also been reported 49.
Signals from Cells in the Gut
Hematopoietic activity has been associated with ventral structures in several organisms 100-103. The ventral domain of the dorsal aorta contains most of the HSC activity at E11.5 and it is the only domain that is able to initiate HSCs in E10.5 explant cultures and expand HSCs in E11.5 explant cultures 104. This suggests that the ventral region of the dorsal aorta contains exclusive signals that are capable of inducing and expanding definitive HSCs. One ventral signaling pathway, which predominantly acts at early E10 and arises from the developing gut, is the Hedgehog pathway 54. AGM:gut explants from early E10 embryos contain HSCs after 3 days in culture. This is not achieved with explants of AGM alone or when AGM and gut explants are cultured individually and transplanted together. Moreover, lineage tracing using a chimeric AGM reaggregate culture system shows that signals from the gut are inducing HSCs from within the AGM. This effect is blocked if anti-hedgehog antibody is added to the culture medium, and is likewise precipitated in a concentration-dependent manner in early E10 AGM explants cultured with exogenous Ihh or Shh. Expression of Gli1, one of the intracellular effectors of hedgehog signaling, is most prominent in subaortic mesenchymal cells, suggesting that the positive effect of hedgehog signaling derived from the gut acts via mesenchymal cells. Indeed, the role of the subaortic mesenchyme in inducing Runx1 expression and intra-aortic cluster formation was recently demonstrated in grafting experiments in the avian system 105.
Dlk1-Expressing Cells
Dlk1 is a paternally expressed imprinted gene encoding an atypical Notch ligand, Dlk1, which exists in both soluble and membrane-bound forms. The gene is upregulated in the HSC-rich middle third of the dorsal aorta compared with the rostral and caudal thirds 24. It has been shown to promote FL hematopoiesis 51, 52 and examination of the AGM at E11 shows that it is expressed in sympathoadrenal cells, where it is downstream of Gata3, as well as the smooth muscle layer of the dorsal aorta and the ventral mesenchyme, where it is downstream of Runx1 50. Paradoxically, functional studies support its role as a negative regulator of AGM hematopoiesis: AGMs from Dlk1 transgenic embryos that overexpress Dlk1 have reduced HSC repopulating ability compared with wild-type embryos, while AGMs from Dlk1 knockout embryos give rise to increased numbers of progenitors. These results can be replicated using in vitro stromal cell coculture systems, which use stromal cells that express varying amounts of Dlk1 to support HSPCs. However, addition of soluble Dlk1 to these cocultures has no effect, suggesting that it is the membrane-bound form of Dlk1 acting via the stromal cell microenvironment that inhibits HSCs. The findings in this study are strikingly different from those observed by Moore et al. in the FL, where both soluble and membrane-bound Dlk1 appear to promote hematopoiesis 52. Incidentally, Chou et al. found that long-term expansion of HSCs also required physical contact with Dlk1-expressing fetal hepatic progenitor cells 51. The observation that Dlk1 expression is induced by Runx1 hints at the possible existence of a negative feedback loop that limits HSC expansion in the AGM. The HSC pool greatly increases in size after colonization of the FL, where Dlk1 is expressed in a different cell type and seems to perform a very different function.
Other Hematopoietic Cells
The dynamic and responsive nature of the hematopoietic system implies that the behavior of HSCs is influenced by their differentiated progeny in order to adequately regulate supply and demand of specific blood cell types both under homeostatic as well as under stress situations. This occurs most commonly through the release of cytokines that act on HSCs. For example, in mutant mice with defective megakaryopoiesis, Thpo levels are altered, resulting in impaired HSC function 106. HSC regulation by mature blood cells can also occur indirectly via the niche as macrophages have been demonstrated to interact with Nestin+ cells to prevent egress of HSCs from the niche 107. Finally, a direct contact between HSPCs and macrophages has also been observed that seems to preserve HSC function under stress 108. It is unknown whether emerging HSCs are also influenced by hematopoietic progenitors that are already present at that point in development. Intra-aortic hematopoietic clusters are a mixture of cells possibly at different stages of maturation. Whether there exists crosstalk between them remains to be shown.
Physical Factors
There is evidence that the circulation in the embryo stimulates AGM hematopoiesis. Nitric oxide production from endothelial cells is triggered by the sheer stress induced by pulsatile flow that occurs as a consequence of the heart beating 109. Progenitor numbers from para-aortic splanchnopleura-derived cells (the precursor tissue of the AGM) from Ncx1−/− embryos, which lack a heartbeat, can be increased by exposure to shear stress that is equivalent to the hemodynamic shear stress experienced by the embryonic aorta at E10.5 64. Reduction of nitric oxide synthase activity, and therefore nitric oxide, results in reduced progenitors compared with wild-type E10.5 AGMs 64 and reduced phenotypic and functional HSCs from E11.5 AGMs 65. The positive effect of shear stress on HSCs may explain why HSCs emerge in the middle third of the dorsal aorta, in particular at its junction with the vitelline artery 24. Interestingly, nitric oxide synthase 1 (nNOS) production by stromal cells in the FL is associated with negative regulation of hematopoiesis 66, revealing another difference between the AGM and FL niches.
The adult BM contains poorly perfused hypoxic areas, particularly around the endosteum, which were found to harbor all cells with long-term culture-initiating cell potential and the majority of the repopulating cells, including the more dormant LSK SLAM cells 110, 111. It was also shown that HSCs stained highly for the hypoxia marker, pimonidazole. Together with the finding that HSCs maintain their function best under hypoxic conditions in vitro, it seems likely that hypoxic conditions are a key BM niche component that can influence the behavior of these cells. In the developing embryo, the AGM including intra-aortic clusters, the FL, and the placenta are hypoxic as seen by pimonidazole staining. In addition, there is a reduction in intra-aortic clusters in the aorta and repopulation capacity of the AGM and placenta when the hypoxia inducible factor 1 alpha transcription factor, a key component of hypoxia responsive signaling, is conditionally knocked out in endothelial cells 112.
A very recent development in the study of stem cell niches is the concept of substrate elasticity and thus the mechanical properties of the niche. There is evidence to suggest that flattening of osteoblasts which increases their stiffness leads to enhanced HSC mobilization 113. Whether such physical changes in the niche cells contribute to the migration of HSCs during development remains to be determined.
Conclusions
Hematopoiesis occurs sequentially at multiple sites in the developing embryo. Each site harbors a different balance of HSC emergence, maintenance, proliferation, differentiation, and death, and must therefore comprise a niche that functions accordingly. It is important to understand the AGM in hematopoiesis because it is a conserved location of HSC emergence from zebrafish to humans, presumably due to an evolutionary advantage for HSCs to arise in the aorta 65. Some of the regulatory signals, such as hedgehog, Bmp, nitric oxide, prostaglandins, and metabolic factors, are also conserved 65, 114-116. One difficulty with identifying factors that affect the AGM is that only perturbation of those which play a critical role in hematopoiesis appears to result in phenotypic effects in the newborn/adults. This suggests that compensatory mechanisms exist throughout development to ensure production of adequate numbers of hematopoietic cells, and means that assays to identify regulatory signals should specifically measure HSC numbers within the AGM.
Niche factors can be soluble or membrane-bound, and be under physical, temporal, or spatial regulation. Moreover, as exemplified by Dlk1 and nitric oxide, the same signal may have opposing effects in different tissues, further highlighting the importance of the microenvironment. Due to the small size of the embryo and the presence of a heartbeat from E8.5, signals influencing hematopoiesis can arise locally, from the surrounding stromal cells, from within the dorsal aorta, or from other nearby developing organs such as the gut or the sympathetic nervous system. The latter may include signals intentionally regulating more than one developing organ system, or alternatively may subject the AGM to bystander effects. The presence of negative regulators of hematopoiesis not only in the AGM but also in the BM and FL hematopoietic microenvironment suggests that they are required and that a negative effect cannot be assumed to be an unwanted signal. The reason for the limited number of HSCs within a niche is unknown; is it simply a matter of niche capacity, or does it have more far-reaching consequences for the organism? Plausible drawbacks of a system purely under positive control include reduced HSC quality and an increased risk of malignancy due to uncontrolled proliferation. Future work will ultimately aim to derive HSCs from embryonic stem cells and induced pluripotent stem cells and to culture HSCs independently of stromal cells. In addition to optimization of physical culture conditions, this requires the identification of further signals, both positive and negative. In light of the findings discussed above, the search for the source of these signals should not be limited to the immediate vicinity of HSCs in vivo. Instead, the net needs to be cast far and wide in order to gain a true understanding of the breadth of signals which affect developing HSCs.
Author Contributions
B.M.-S., S.R.F. and K.O.: wrote this review together.
Disclosure of Potential Conflicts of Interest
The authors indicate no potential conflicts of interest.
