| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
CONCISE REVIEW |
a Immunology Program, Child Health Research Institute, North Adelaide, South Australia, Australia;
b Haematology Division, Hanson Institute, Institute of Medical and Veterinary Science, Adelaide, South Australia, Australia;
c Department of Paediatrics, University of Adelaide, Adelaide, South Australia, Australia
Key Words. Hematopoietic stem cells (HSCs) • Cytokines • Hematopoietic development • Umbilical cord blood • Ex vivo expansion
Correspondence: Richard D’Andrea, Ph.D., Child Health Research Institute, 72 King William Road, North Adelaide, South Australia, 5006, Australia. Telephone: 61-8-8161-8105; Fax: 61-8-8239-0267; e-mail: Richard.dandrea{at}adelaide.edu.au
 |
ABSTRACT |
|---|
 Top Abstract IntroductionBMP4 Signaling in Generation...Clinical Applications for Ex...Potential of BMP4 in...SummaryReferences |
|---|
Although the origins of HSCs have become clearerand progress has been made in identifying genes that are critical for the formation and maintenance of HSCs, less is known about the signals that commit specific populations of mesodermal precursors to hematopoietic cell fate. Critical signals acting on these precursorcells are likely to be derived from visceral endoderm in yolk sac and from underlying stroma in the aorta-gonadmesonephros region. Here we summarize briefly the origin of yolk sac and embryonic HSCs before detailing evidence that bone morphogenic protein-4 (BMP4) has a crucial role in Xenopus and mammalian HSC development. We discuss evidence that BMP4 acts as a hematopoietic growth factor and review its potential to modulate HSC in ex vivo expansion cultures from cord blood.
|
INTRODUCTION |
|---|
|
TopAbstractIntroductionBMP4 Signaling in Generation...Clinical Applications for Ex...Potential of BMP4 in...SummaryReferences |
|---|
Ontogeny of Hematopoiesis The predominant anatomical site of hematopoiesis shifts several times during mammalian development. The first wave of hematopoiesis occurs in the yolk sac and is characterized by the production of large, nucleated primitive (or embryonic) erythroid cells. Primitive yolk sac hematopoiesis is transient, taking place between 7 and 11 days postcoitum (dpc) in the mouse before being replaced by the definitive, multilineage blood system that is sustained throughout life from multipotent HSCs [1]. Definitive hematopoiesis also progresses in a spatially and temporally ordered manner in the embryo [2]. Cells with definitive hematopoietic potential have been identified prior to circulation, as early as 7.5–8.0 dpc in intraembryonic sites such as the para-aortic splanchnopleure (P-Sp)/aorta-gonad-mesonephros (AGM) region, as well as in the yolk sac [3,4]. In mice, the fetal liver assumes the principal hematopoietic role from 10–11 dpc until birth. Late in gestation the bone marrow (BM) is established as a hematopoietic organ and becomes the major source of hematopoietic cells when fetal liver hematopoiesis declines shortly after birth. Similarly, the site of hematopoiesis shifts in humans during development. Primitive erythropoiesis is initiated in the yolk sac around week 3 of gestation and remains active until week 6. As in the mouse, the P-Sp/AGM region in humans contains definitive hematopoietic cells prior to their detection in the fetal liver [5,6]. The fetal liver then continues as a major site of hematopoiesis until week 22, by which time the fetal bones, having been colonized in weeks 8–12, become the major site of hematopoiesis [2].
It is now clear that HSCs do not originate in situ in the fetal liver [7,8] or BM [9,10], the two principal organs associated with definitive hematopoiesis. Rather, these organs are seeded by circulating HSCs that arise elsewhere in the embryo. Although it has been clearly established that BM is seeded by HSCs that develop in the fetal liver, the origins of fetal liver-colonizing HSCs have not been established unequivocally. In bird and frog species, the origin of definitive hematopoietic cells has been traced to an intraembryonic source, most notably to the dorsal lateral plate/AGM region [11,12]. Subsequently, the equivalent region in the mouse, the P-Sp (at 8.5–9.5 dpc), which develops into the AGM by 10.5–11.5 dpc of gestation, has been identified as an important site for the development of definitive hematopoietic stem cells [13]. In humans, hematopoietic cells are also present in the P-Sp/AGM region from embryonic day 19 (E19) onward [6]. The emergence of HSC at other intraembryonic sites, including the vitelline and umbilical arteries in mice [14] and humans [6,15], has also been demonstrated. Interestingly, although the AGM is an important site of HSC emergence, it does not appear to support the differentiation of these cells to specific hematopoietic lineages, thereby indicating that the AGM is exclusively involved in generating a pool of HSCs capable of seeding the fetal liver [16].
Qualitative Differences between Extraembryonic and Embryonic HSCs Whether mammalian definitive HSCs arise exclusively from the intraembryonic P-Sp/AGM region or whether they arise independently from both intra- and extraembryonic sites remains controversial (discussed in [17–19]). Several recent studies using in vitro explant cultures of dissected yolk sac and AGM tissue suggest that the ability to generate lymphoid progeny and to engraft adult mice is restricted to the AGM region in precirculation embryos [19–21]. These data have been used to argue for the in situ development of cells with definitive HSC potential exclusively in the P-Sp/AGM region.
Other studies suggest that cells present within the yolk sac have the potential to develop into HSCs with long-term repopulating (LTR) ability when given the appropriate signals, which may have been lacking in the above studies. For example, Matsuoka et al. demonstrated that both yolk sac and P-Sp–derived cells isolated from 8.0–8.25 dpc embryos acquire the capacity for long-term repopulation of irradiated adult recipients when the cells are cocultured for at least 4 days on the AGM-derived stromal cell line AGM-S3 [22]. Critically, using this protocol, HSCs could be generated from yolk sac cells that were isolated before circulation between the yolk sac and the embryo was established. In support of the notion that cells with HSC potential may be present at early embryonic stages but are undetectable using the adult LTR assay, Yoder et al. showed that, when they were transplanted into conditioned newborn pups, CD34+/c-kit+ cells derived from the yolk sac of 9-dpc embryos could provide long-term engraftment of these animals. In particular, BM isolated from these mice at adulthood could reconstitute secondary recipients [23–25]. These experiments suggest that cells within the yolk sac can be educated, in the appropriate environment, to adopt an adult HSC phenotype emphasizing the importance of local signals and environment in determining the characteristics of HSCs.
The seemingly independent emergence of HSCs in yolk sac and the P-Sp/AGM region raises two important questions: (a) Is there a common precursor cell from which the HSCs develop? and (b) What are the specific inductive signals that commit these two tissues to produce hematopoietic cells? Studies in mammals [26] and other vertebrates [11] indicate that hematopoietic cells are derived from mesoderm, one of the three embryonic germ layers formed during gastrulation [27]. Next we summarize the events underlying the development of HSCs within the yolk sac and intraembryonic sites and review the evidence for bone morphogenic protein-4 (BMP4) involvement in this process.
HSCs and Their Relationship to Endothelial Cells
The observation that hematopoietic cells develop in close proximity to endothelial cells in the yolk sac and AGM region has led to the hypothesis that these two lineages have a common origin (Fig. 1
). This concept has been reinforced by a number of studies demonstrating that progenitors of both lineages share the expression of a number of genes, including those encoding CD34, Flt-3 (FMS-like tyrosine kinase 3) ligand, VEGFR-1 (vascular endothelial growth factor receptor-1), VEGFR-2 (FLK1), SCL/Tal-1 (stem cell leukemia/T-cell acute leukemia), and c-kit [28–30], and gene-targeting experiments in mice demonstrating that ablation of common genes, such as Flk1 and VEGF, leads to defects in both lineages [31,32]. However, differences in the temporal order of blood cell emergence, relative to endothelial cell development in the yolk sac blood islands versus the AGM region, have led to the proposal that there are two slightly different bipotential precursors—the hemangioblast and hemogenic endothelium—at these two distinct embryonic regions [33]. At present, the exact relationship between the hemangioblast and hemogenic endothelium remains to be established definitively, with the evidence for both summarized next.
|
Hemangioblast Endothelium
Hematopoietic and endothelial cells appear to develop simultaneously from mesoderm in the yolk sac, resulting in the formation of yolk sac blood islands (Fig. 1
). The simultaneous emergence of both cell types from an apparently homogenous cord of extraembryonic mesoderm led to the proposal that these two lineages develop from a common bipotential progenitor termed the "hemangioblast," a cell type that was first proposed over 80 years ago [34]. Since the isolation of cells with hemangioblastic properties from embryos has proven difficult, the most direct evidence for its existence has come from studies using in vitro differentiation of embryonic stem (ES) cells to mimic early events in hematopoietic and endothelial development [35]. Using the embryoid body (EB) differentiation model, Kennedy et al. and Choi et al. identified a cell type termed the "blast colony-forming cell (BL-CFC)" that fulfills many of the criteria of the elusive hemangioblast [36–38]. These BL-CFCs exist only transiently in the EB between day 2.5 and 3.5 of culture, thus preceding the formation of primitive erythrocytes, which are the first identifiable hematopoietic cells. Critically, under the appropriate culture conditions, cells isolated from undifferentiated blast colonies were able to form cells of both the endothelial and hematopoietic lineages, demonstrating the bipotential nature of these cells. Although identification of a bipotential hemangioblast cell in EB is consistent with the simultaneous formation of endothelial and hematopoietic cells in the early yolk sac, most experimental data indicate that definitive HSCs arising in the AGM and other intraembryonic sites are likely to emerge from a cell with endothelial characteristics.
Hemogenic Endothelium Intraembryonic hematopoietic cells emerge after the development of endothelial cells and the vascular system. Histologically, these emerging hematopoietic cells appear as discrete clusters of rounded cells that adhere to the ventral wall of the dorsal aorta and the vitelline and umbilical arteries [30]. In mice and humans, these clusters are present from 9.5 to 11.5 dpc [30] and E27 to E40 [6], respectively, coinciding with the emergence of cells that display definitive hematopoietic characteristics at these sites. The hematopoietic nature of these cells was subsequently demonstrated by a number of criteria, including gene and cell-surface marker expression [6, 15, 39, 40] and the absence of these clusters in mice when genes critical for definitive hematopoiesis were ablated [41,42].
The first direct evidence that intra-aortic hematopoietic clusters are derived from endothelial cells resulted from lineage-tracing studies performed in the chick where the endothelial tree was marked from within the vessel lumen [43,44]. As labeling of endothelial cells occurred before hematopoietic cells emerged in the floor of the aorta, the subsequent detection of labeled hematopoietic clusters strongly suggested that these cells originated from vascular endothelial cells. In mice, cell-marking studies have also suggested that HSCs arise from the endothelial layer of dorsal aorta [45]. Similarly, candidate endothelial cells isolated by flow cytometry from both mouse [21] and human [46] yolk sac and AGM regions have been shown to have hematopoietic potential in vitro. Nishikawa and colleagues, using a variation of the ES cell in vitro differentiation system and flow cytometry, have also shown that at least a fraction of hematopoietic cells go through a vascular endothelial (VE)–cadherin+ intermediate [47,48]. Lastly, Sugiyama et al. have recently demonstrated the generation of circulating hematopoietic cells from DiI-conjugated acetylated low-density lipoprotein (Ac-LDL-DiI)–labeled endothelial cells in the 10-dpc mouse embryo, consistent with the direct derivation of hematopoietic cells from pre-existing endothelial cells [49].
|
BMP4 SIGNALING IN GENERATION OF HSCS |
|---|
|
TopAbstractIntroductionBMP4 Signaling in Generation...Clinical Applications for Ex...Potential of BMP4 in...SummaryReferences |
|---|
BMP4 and other members of the transforming growth factor beta (TGF-ß) superfamily of growth factors transmit intracellular signals by binding to and activating a family of heteromeric cell-surface receptors consisting of type I and type II serine/threonine protein kinase subunits (reviewed in [56–59]). BMP dimers are able to bind to both type I and type II receptor subunits leading to the formation of an active complex. Once a BMP type I/type II receptor complex is formed, the constitutively active kinase domain of the type II receptor transphosphorylates a conserved 30 amino acid stretch (GS domain) on the type I receptor. This results in the activation of the type I receptor kinase domain, which then phosphorylates members of the Smad family of proteins. In mammals, eight members of the Smad protein family have been identified, and these have been grouped into three different classes based on structure and function [60].
One class of Smads consists of the receptor-regulated Smads (R-Smads) that act as latent transcription factors. R-Smads transiently associate with the type I receptor and are phosphorylated on their C-terminus by the type I receptor kinase domain. Phosphorylated R-Smads then interact with the co-Smad, Smad4, resulting in translocation of the heteromeric complex to the nucleus where it interacts with cell-specific transcription factors to induce target gene expression [61]. Binding of different TGF-ß family members to their respective receptors results in the activation of different R-Smads, with R-Smads-1, -5, and -8 phosphorylated and translocated to the nucleus upon BMP stimulation, while R-Smads-2 and -3 are associated with TGF-ß and activin signaling [62]. The last class of Smads consists of the inhibitory Smads-6 and -7, which function as negative regulators of signaling. Smad6 primarily regulates BMP signaling by competing with Smad4 for binding to activated Smad1, resulting in the formation of an inactive Smad6/Smad1 complex [63]. Smad7, which lacks the conserved C-terminal phosphorylation site of R-Smads, is unable to translocate to the nucleus and downregulates BMP, activin, and TGF-ß signaling by inhibiting the phosphorylation of R-Smads through competition with R-Smads for type I receptor binding [64].
The Smad complexes do not bind DNA strongly, so activation of specific target genes is achieved by the ability of different R-Smad/Smad4 complexes to interact with other transcription factors that are also present at the various promoters. For example, the activin/TGF-ß–responsive R-Smads, Smad2 and Smad3, have been shown to bind to the mouse goosecoid promoter in association with the winged-helix transcription factor FAST1/2 (FoxH1) [65,66], while the BMP-responsive R-Smad, Smad1, has been shown to require a transcription factor—either Olf-1/EBF–associated zinc finger (OAZ), PEBBP2
A, early hematopoietic zinc finger protein (EHZF), or Xvent-2 [67–70]—to activate the Xenopus Xvent2B promoter in response to BMP2. A detailed list of proteins that interact with the various Smad proteins can be found in Miyazono et al. [71]. In addition, different R-Smads bind several DNA elements with different affinities. Smad3 and Smad4 preferentially bind to the Smad-binding element (SBE) containing the core consensus sequence CAGAC, while Smad1 and Smad4 have been shown to bind to SBEs and GC-rich motifs, depending on the promoter [68, 72–75].
Role for BMP4 in the Induction of Hematopoiesis in Xenopus
Mesoderm Induction In Xenopus and zebrafish, a number of techniques—including tissue recombination, cell transplantation, cell fate mapping, and overexpression studies—have been used to investigate the molecular mechanisms of mesoderm induction, patterning, and blood formation [76]. In Xenopus, TGF-ß family members Xnr1, Xnr2, Xnr-4 to -6, and derriere released from the vegetal blastomeres cause a broad band of cells in the region midway between the animal and vegetal poles (the equatorial marginal zone) of the late blastula embryo to adopt a mesodermal fate. Production of these zygotic mesoderm-inducing signals is thought to be controlled by the maternally derived and vegetally localized VegT transcription factor [77–80].
According to the prevailing view [53,54], mesoderm induction and patterning in Xenopus occurs by a three-signal process (the three-signal model) where mesoderm is established initially in a binary state consisting of a small region termed the "Spemann organizer," specified around the site of initial blastopore lip formation, with the remaining mesoderm termed the "ventral-lateral marginal zone." Subsequently, molecules secreted from the Spemann organizer are proposed to establish the dorsal-ventral axis of the mesoderm. This dorsal-ventral patterning of the mesoderm occurs along an axis running from the Spemann organizer (dorsal-most tissues) across the equatorial marginal zone to the ventral-most tissue directly opposite the organizer and is proposed to result from antagonism between growth factors that promote ventral fates and molecules secreted by the Spemann organizer, which inhibit the activity of these ventralizing factors [53, 54, 81]. In this model, BMP family members, and BMP4 in particular, are critical components of the ventralizing signal. Numerous studies in which BMP activity is artificially increased or decreased have been interpreted to indicate that BMP4 acts as a morphogen, with a gradient of BMP4 activity specifying the different mesodermal territories along the marginal zone of the Xenopus gastrula. In particular, high levels of BMP4 activity specify ventral mesoderm fate such as blood, intermediate BMP levels specify more lateral mesoderm derivatives, and the lowest level of BMP activity (in the region closest to the organizer) specifies dorsal fates. In keeping with the three-signal model, the gradient of BMP activity is largely established by a post-translational mechanism involving the binding and inhibition of BMP4 by organizer-secreted molecules such as chordin, noggin, and follistatin (reviewed in [53,82]).
New cell fate mapping and gene-expression data have recently challenged the model just described. For example, mesodermal cells fated to form ventral blood islands, and thus proposed to represent the ventral-most mesoderm, are not restricted to the tissue directly opposite the Spemann organizer. Rather, cells that contribute to the ventral blood islands appear to be derived from a band of mesodermal cells located around the circumference of the vegetal-most portion of the equatorial marginal zone, known as the leading-edge mesoderm (Fig. 2C). Critically, this includes cells located next to the Spemann organizer, which are therefore exposed to very low BMP4 levels [12, 83–86].
|
). These authors reinterpret the evidence for the Spemann organizer–induced BMP4-gradient playing a role in mesoderm patterning as a combination of changes to rostral-caudal patterning and alterations in the developmental timing of mesoderm differentiation, which normally proceeds in a rostral-to-caudal manner [76,87]. At present it is unclear whether such a fundamental shift in ideas regarding mesoderm patterning and axis orientation in the amphibian embryo will be accepted; however, this proposal clearly warrants further investigation. To avoid confusion with many earlier articles, in subsequent sections here we have used the terms dorsal marginal zone (DMZ) to describe the marginal zone on the side where the blastopore lip first forms and ventral marginal zone(VMZ) to describe the region located horizontally opposite the DMZ.
Hematopoiesis-Inducing Activity of BMP4 in Xenopus Whichever model of mesoderm patterning proves to be correct, it must incorporate the substantial experimental evidence demonstrating a role for BMP4 in specifying blood formation during gastrulation, while also accounting for the finding that some blood progenitor cells arise from the mesoderm that is initially adjacent to the Spemann organizer and therefore exposed to low BMP4 levels.
Several transplantation studies and cell fate maps in Xenopus indicate that adult and primitive blood have distinct origins [12, 83, 88, 89], analogous to the separate origins of avian and mammalian primitive and definitive hematopoietic systems. In the lineage tracing studies of Ciau-Uitz et al. and Walmsley et al., the yolk sac ventral blood islands (VBI) and the dorsal lateral plate (DLP; the amphibian equivalent of the mammalian P-Sp/AGM region), associated with primitive and definitive blood formation, respectively, were shown to be derived from distinct blastomeres of the 32-cell embryo (Fig. 2B) [12,83]. Furthermore, the cells that contribute to the posterior and anterior VBI also appear to be derived from distinct blastomeres, with progeny of the D4 blastomere contributing to the posterior VBI, while C1 and D1 blastomere-derived cells contribute to the anterior VBI [12,83]. Although other lineage-tracing experiments have since shown that all C-tier blastomeres (C1 to C4) contribute to primitive blood (Fig. 2C) [85,86], all of these studies consistently show that mesodermal cells adjacent to the Spemann organizer (and therefore thought to be exposed to low levels of BMP) also contribute to primitive blood. While VBI formation in response to high levels of BMP4 is well established [84, 90–94], the demonstration that blood is also derived from other mesodermal sources (DMZ and DLP) exposed to lower levels of BMP4 raised the question of whether BMP4 signaling is essential for all embryonic blood formation. However, the finding that primitive and definitive blood formation is inhibited by expression of a dominant negative BMP type I receptor in either the DMZ and VMZ (anterior and posterior VBI) or the DLP, respectively, strongly suggests that BMP signaling is necessary for all stages of embryonic blood formation [83].
This apparent anomaly of cells of the anterior VBI requiring BMP4 to develop into blood while initially coming from a region of low BMP4 (DMZ) appears to be explained by several recent experiments indicating that the major points at which BMP4 is required for blood specification occur during late gastrulation and postgastrulation rather than in the patterning of mesoderm [84, 95–97]. Cells destined to form VBIs are derived from leading-edge mesoderm [98], which involutes in the early gastrula embryo and begins to migrate upward across the roof of the blastocoel, eventually arriving at the site of VBI formation near the ventral midline and underneath the epidermal progeny of the A3 and A4 blastomeres. This involution and migration begins with C1- and D1-derived portions of the leading-edge mesoderm and then proceeds laterally across the embryo, with C4- and D4-derived leading-edge mesoderm involuting last.
This early migration of C1- and D1-derived leading-edge mesoderm away from the DMZ (thought to be rich in BMP antagonist) to a region high in BMP4 activity has been suggested as an explanation for these cells forming the anterior VBI in a BMP4-dependent manner [83,85]. Subsequently, ectoderm-derived BMPs have been shown to have an important role in primitive blood formation by experiments demonstrating that increasing or decreasing BMP signaling within a localized area of the ectoderm leads to, respectively, expansion or suppression of globin expression in the adjacent mesodermal layer [84, 96, 97].
Another recent study in which BMP signaling was blocked in a controlled manner in Xenopus embryos and explants suggests that BMP signaling is required after gastrulation within the ventral mesoderm itself for primitive erythropoiesis to occur [95].
Although these experiments do not rule out the possibility that BMP signaling may also be required at earlier gastrula stages for marking mesoderm as competent to form blood, they clearly demonstrate that BMPs play a crucial part in the final stages of blood formation, independent of any effects on mesoderm patterning.
BMP4-Induced Gene Expression Critical to our understanding of the role of BMP4 is the identification of direct BMP4 target genes and genetic programs induced by BMP in the mesoderm, hematopoietic primordia, and HSCs. BMPs are known to induce different genes in different cell types by the specific recruitment of BMP-activated Smads to target promoters via their interaction with cell-specific transcription factors [68].
In Xenopus embryos, BMP signaling results in the induction of a number of genes involved in either mesoderm patterning or hematopoiesis (or both), including the homeobox genes msx-1 [99], the Vent family members Xvent-2 (and the related genes Xom, Vox, Xbr-1, and Xvent-2B) [70,100] and Xvent-1 (and the related genes PV.1 and Xvent-1B) [101,102], the Mix family member mix-1 [103] and Xhox3 [104], as well as hematopoietic-specific genes such as GATA-1 [105], GATA-2 [106], SCL [107], and LMO-2 [108], the ETS family member Xfli1 [83], the Wnt-signaling molecule Xwnt8 [109], and, most recently, a novel long-terminal-repeat-retrotransposon Xretpos [110]. To date, Xvent-2, GATA-2, and Xretpos genes have been identified as direct targets of BMP signaling [68, 111, 112], while Xvent-1 is downstream of Xvent-2 and GATA-2 [112].
Role for Xvent Proteins in Hematopoiesis
Based on the three-signal model of mesoderm formation, the Vent class of homeobox genes have been proposed to have a pivotal role in BMP-induced ventralization of Xenopus and zebrafish embryos by mediating a positive autoregulatory loop for BMP4 and suppression of dorsal genes such as XFD-1' [113], chordin [114], and goosecoid [115], thus allowing cells to acquire a ventral fate.
The Vent family of homeobox proteins has been divided into two subfamilies, Vent-1 and Vent-2, on the basis of amino acid sequence [116,117]. Apart from the multiple family members present in Xenopus, one member of each subfamily has been described in zebrafish (vox/vega1 and vent/vega2) [114,118], while it appears that only one Vent protein is present in the invertebrate Amphioxus (AmphiVent) [119] and in humans (VentX2) [120]. In Xenopus, the expression domains and temporal order of expression of Vent-1 and Vent-2 family members are consistent with each playing different regulatory roles. Xvent-2 appears to act as either a context-specific transcriptional repressor or an activator, depending on the target gene [70,112], whereas Xvent-1 has been shown to function only as a repressor [121]. Amajor function proposed for Xvent-2 is the establishment of a positive autoregulatory loop involved in the zygotic activation and subsequent maintenance of BMP4 expression during the late blastula and early gastrula stage of Xenopus embryonic development [50,54]. Xvent-2 maintains this positive autoregulatory feedback loop by binding to and activating its own promoter [70], as well as the promoter of the BMP4 gene [122]. Moreover, the Xvent-2 promoter contains a functional BMP responsive element, and therefore increased BMP4 expression is presumed to further upregulate Xvent-2 transcription [69,100]. The autoregulatory activation of BMP4 and Xvent-2 expression, combined with downregulation of BMP inhibitors such as chordin by Vent family members, may be critical for generating the threshold levels of BMP signaling that are required to induce a blood cell fate.
However, as discussed, the role that the BMP4 gradient induced by the Spemann organizer has in specifying ventral mesoderm and blood formation has been questioned. Indeed, overexpression of either Xvent-1 or Xvent-2 in embryos leads to a decrease in globin expression rather than to the expansion of blood, as might be expected from the three-signal model of mesoderm induction [84,123]. Kumano et al. propose that the effects of overexpression of Vent family members is consistent with the caudalization of the embryos and a resultant differentiation delay at the caudal end of the embryo, since globin levels increase at later developmental stages [84].
Although the function that Vent family members have in specifying ventral mesoderm may be in doubt, the finding that Xvent-2 interacts with transcription factors known to play a part in hematopoieis—such as GATA-2 [112], Smad1 [70], and ETS family members fli-1 and Ets-related gene (ERG) [124]—suggests that the Vent proteins may have a later role in blood formation. However, the function of Vent family members in mammals is still unclear, given that only a single family member appears present in humans [120] and no homologues of Xvent-1 and Xvent-2 have yet been identified in mouse.
BMP4 May Specify Blood by Activating LMO-2, SCL, and GATA Factors Consistent with BMPs having a crucial role in the specification of blood from mesoderm, a number of important hematopoietic transcription factors such as GATA-1, GATA-2, LMO-2, and SCL have been identified as BMP-inducible genes in Xenopus. Of these, GATA-2 has been shown to be a direct BMP4 target gene [112]. Whether these other transcription factors are also directly regulated by BMP4 signaling or whether they are induced indirectly has not yet been addressed. GATA-2 is expressed in both endothelial and hematopoietic cells and their precursors in the yolk sac and in the AGM region of mouse, zebrafish, and Xenopus embryos [125]. In Xenopus, GATA-2 expression precedes that of SCL in the ventral mesoderm [125], consistent with recent data indicating that a complex consisting of GATA-2 and the ETS transcription factors Fli-1 and ELF-1 activates the SCL gene in mouse and Xenopus hemangioblasts, endothelial cells, and hematopoietic cells, including HSCs [126,127]. Similarly, activation of GATA-1 and LMO-2 genes in the ventral mesoderm by ectodermally derived BMP4 is proposed to lead to the formation of a complex consisting of GATA-1, LMO-2, and SCL, which functions synergistically to drive blood formation in the developing blood islands and AGM [108].
Expression of the Xenopus Runx1 (AML1) orthologue, Xaml, may also require BMP4 [128], as Xaml expression is lost when BMP4 signaling is inhibited in Xenopus embryos [83]. While Xaml expression occurs early in primitive blood islands, its expression in the progeny of the dorsal lateral plate mesoderm is restricted to the developing hematopoietic clusters and the ventral floor and underlying mesenchyme of the dorsal aorta [12,83]. This restricted pattern of Xaml expression in the AGM region is reminiscent of the polarized expression of BMP4 that is observed in the dorsal aorta of mammals (see the following discussion), which suggests that BMP4 has a role in inducing Runx1 expression. In addition, Runx1 can interact with R-Smads [129,130], raising the possibility that the Runx1 and Smad proteins may jointly regulate hematopoietic gene expression and lineage commitment [131].
Thus, irrespective of the manner in which mesoderm is patterned, localized high concentrations of ectoderm-derived BMP4—through its ability to directly or indirectly active critical transcription factors such as GATA-2, SCL, LMO-2, and Runx1—can be visualized as playing an important part in the initiation of the hematopoietic program in adjacent mesoderm (Fig. 3).
|
There is now accumulating evidence from both in vivo and in vitro experiments indicating that BMPs, and BMP4 in particular, can act at multiple points in mammalian hematopoiesis. During embryonic development, BMP4 has been implicated in the commitment of yolk sac mesoderm to hematopoietic and endothelial fates. In this setting, it is proposed to act as a mediator of the effects of the hedgehog family member, Indian hedgehog (Ihh) [134–136]. Consistent with this, high levels of BMP4 surround the blood islands in the human yolk sac [137]. In the AGM region, BMP4 expression occurs in a highly polarized pattern in the mesenchyme underlying the hematopoietic clusters that emerge from the ventral walls of the dorsal aorta, which is highly suggestive of BMP4 involvement in HSC formation [137].
An Endoderm-Derived Hematopoietic Signal in Yolk Sac May Activate BMP4 Expression Hematopoietic cells normally develop from splanchnopleuric (endoderm-associated) mesoderm, and several lines of evidence suggest that endoderm-derived signals are important for their development [138,139]. In mammals, the contribution and identity of this endoderm-derived signal has been extensively examined in the yolk sac. In the mouse, only nascent extraembryonic mesoderm that contacts visceral endoderm goes on to form blood islands [136]. Using an explant culture system in which blood formation in the yolk sac can be followed in pre- or early gastrulation–stage embryos, Baron and colleagues demonstrated that the visceral endoderm secretes a diffusible signal that is required for induction of primitive hematopoiesis [136]. Surprisingly, this endoderm-derived signal was also able to respecify anterior epiblast cells, normally fated to give rise to neurectoderm, to a more posterior fate, resulting in the induction of hematopoietic and endothelial cells.
Subsequently, Ihh [140] was shown to be capable of substituting for visceral endoderm in this assay [134]. Ihh is expressed in the visceral endoderm, whereas components of the hedgehog signaling pathway—including patched (Ptch), smoothened (Smo), and members of the Gli family of transcription factors—are expressed in the epiblast and extraembryonic mesoderm, but not the visceral endoderm, of gastrulation-stage embryos [141]. In other systems, BMP molecules have been shown to be downstream targets of the hedgehog-signaling cascade [142], and BMP4 expression is induced in anterior epiblast cells cocultured with visceral endoderm or recombinant Ihh [134]. Lastly, it has been reported that BMP4 can substitute for visceral endoderm or Ihh in these respecification assays [136], raising the possibility that some of the effects of Ihh signaling are mediated by BMP4 in vivo.
BMP4 May Be a Stromal Signal Mediating HSC Formation in AGM Signals originating from the underlying stroma of the AGM region are also likely to be important for hematopoietic development in the AGM region. Tissue-grafting experiments performed in the chick indicate that two subsets of mesoderm produce endothelial precursors: paraxial mesoderm, which produces cells displaying only endothelial potential (angioblasts), and splanchnopleural mesoderm, which forms progenitors with both hematopoietic and endothelial potential [143]. These experiments demonstrate that the microenvironment may provide signals important for restricting the differentiation potential of mesoderm in the AGM region. In particular, endoderm-derived signals were found to be important for the homing of splanchnopleural mesoderm progeny to the floor of the dorsal aorta and the acquisition of hematopoietic potential by these cells.
Although other factors appear to be involved in the homing of these cells to the floor of the dorsal aorta [138,139], the observation that BMP4 is expressed at high levels in the stromal region beneath the ventral wall of the dorsal aorta in human embryos suggests that BMP4 may have a role in the acquisition of hematopoietic potential by these cells [137]. This gradient of BMP4 is first observed in 28-day-old human embryos, when hematopoietic clusters are just forming. At this developmental stage, BMP4 is concentrated in the stromal layer of the aortic floor underlying the sites of intra-aortic hematopoietic clusters. By day 34, a gradient of BMP4 has been established across the dorsal-ventral axis of the AGM region, with high levels present in the ventral mesoderm and low levels in the dorsal mesoderm, although BMP4 expression is still concentrated in the area underlying the intra-aortic hematopoietic clusters. However, by day 38, the hematopoietic clusters have disappeared, as has the polarized expression of BMP4 to the ventral floor. Uniform expression of BMP4 is now detected around the entire aorta. The higher concentration of BMP4 detected in the ventral floor of the dorsal aorta may therefore explain why hematopoietic induction is restricted to this region and not detected around the roof and sides of the vessel.
In Vitro Differentiation of ES Cells Suggests a Role for BMP4 in Hematopoiesis A role for BMP4 in hematopoiesis has also been demonstrated using mouse ES cell in vitro differentiation systems. BMP4 has been shown to be essential for the generation of multiple hematopoietic lineages, including erythroid, myeloid, and lymphoid (B and natural killer) cells, from ES cells differentiated in chemically defined, serum-free media [52, 144, 145]. However, the point in the differentiation pathway at which BMP4 was required appeared to vary between studies, making it difficult to determine the exact function of BMP4. Data from two groups are consistent with BMP4 having an early role such as mediating the formation of lateral and extraembryonic-like mesoderm [51, 52, 146]. Given the role of BMP4 in mesoderm formation in vivo [132], it was surprising that Adelman et al. reported that BMP4 appeared to be dispensable for mesoderm formation and its transition to early hematopoietic progenitors in EB cultured in a serum-free semisolid media. In this system, BMP4 was essential for formation of committed erythroid cells [145]. In particular, expression of the transcription factors GATA-1 and EKLF was significantly reduced in EB cultured in the absence of BMP4 [145]. Presumably, differences in ES cell lines or differentiation conditions account for the discrepancies between studies.
Of potential significance in the development of protocols for the directed differentiation of human stem cells for use in cell-replacement therapies is the inclusion of BMP4 to a cocktail of hematopoietic growth factors. This has recently been shown to increase hematopoietic cell formation during in vitro differentiation of rhesus monkey and human ES cells [147,148] and to increase the induction of hematopoietic progenitor cells from isolated neural tissue [149]. Most interestingly, BMP4, independently of hematopoietic cytokines, appeared to support the maintenance of a primitive progenitor cell type with enhanced self-renewal capacity in differentiating human ES cells [148]. Self-renewal in this assay was measured by the ability of single cells derived from primary hematopoietic clonogenic colonies to retain colony-forming unit (CFU) capacity upon secondary replating. In particular, BMP4 increased both the number of primary CFU able to generate secondary CFU and the frequency of secondary CFU per individual primary CFU. Interestingly, there did not appear to be a continuous requirement for BMP4 as cells were only maintained in BMP4 during ES cell differentiation and it was not included in the CFU assays. Whether this increase in self-renewal capacity represented an expansion of a primitive progenitor population with self-renewal capacity or the maintenance of progenitor cells in a more primitive state in the EB is unclear from this study. The benefit of such an approach for generating cell types with in vivo repopulating ability from human ES cells has yet to be proven, but based on work with mouse ES cells, the generation of hematopoietic cells with repopulating ability from human ES cells may require their genetic modification [150].
Recently, several signaling molecules and nuclear factors involved in the self-renewal of HSCs have been identified. These include the Wnt [151], Notch [152], and sonic hedgehog (Shh) [153] signaling pathways; the stromal-membrane factor mKirre [154]; the Polycomb group member Bmi-1 [155]; and the homeobox protein HOXB4 [156]. However, apart from the demonstration that the Shh effects on HSCs are at least in part mediated by its induction of BMP4 [153], it is not known whether BMP4 regulates the expression of any of these factors or whether it lies downstream of any of these pathways.
BMP4 Acts as an HGF In Vitro
BMP4 has been found to have pleiotropic effects in vitro on hematopoietic progenitors isolated from hematopoietic tissues. Highly purified human primitive hematopoietic progenitor populations (CD34+CD8–Lin– cells) express mRNA for BMP4; the BMP receptor subunits ALK-3 and ALK-6; and the intracellular components of the BMP signaling pathway, smad1, smad4, and smad5, as well as the BMP4 inhibitor noggin [157]. This suggests that these cells are capable of responding to BMP4. Subsequently, BMP4 has been found to act on cell populations enriched for HSCs [157], as well as on more mature hematopoietic progenitors (CD34+ cells) in vitro [158].
Varying effects of BMP4 on the differentiation of CD34+ cells into the different blood lineages have been reported. In one study, for example, BMP4 significantly increased the number of burst-forming unitserythroid (BFU-E) but not CFU-granulocyte-macrophage (CFU-GM) in cord blood (CB)–derived CD34+ cells cultured in serum-free conditions [158]. However, a second study demonstrated that BMP4 functions synergistically with stem cell factor (SCF) and granulocyte-macrophage colony-stimulating factor to increase both CFU-erythroid (CFU-E) and CFU-GM in cultures of BM-derived CD34+ cells [159]. Athird study found no effect of BMP4 on the number of BFU-E or CFU-E formed in vitro by BM CD34+ cells [160]. Whether variations in the amount of serum, hematopoietic cytokines, time in culture, or BMP4 treatment regimen caused these different outcomes is unclear.
|
CLINICAL APPLICATIONS FOR EX VIVO EXPANSION OF CB HSCS |
|---|
|
TopAbstractIntroductionBMP4 Signaling in Generation...Clinical Applications for Ex...Potential of BMP4 in...SummaryReferences |
|---|
Traditionally, the BM has constituted the source of HSCs used in transplantation [162]. Umbilical CB is an attractive, alternative source of HSCs for BMT and has been the target for ex vivo expansion protocols [163]. CB contains significantly higher frequency of HSCs, functionally defined as nonobese (NOD)/severe combined immunodeficiency (SCID) mice–repopulating cells (SRCs), and is much less immunogenic due to the naïveté of the immature T-cell population compared with adult BM or mobilized peripheral blood [164]. In addition, CB is a readily available resource, and a number of large CB banks have been established in major centers [165]. However, a significant factor limiting the success of CB transplantation is the finite cell harvest, resulting in a small HSC dose, which produces a characteristic delay in engraftment, and generally restricting CB transplantation to the treatment of small children [166]. This delay in engraftment appears to reflect the low cell number in CB grafts, although it cannot be ruled out that the immaturity of CB cells or the lack of facilitating cells in the graft also contributes to this delay. However, the current evidence suggests that approaches such as ex vivo expansion to increase HSC numbers from CB would represent a major step toward the widespread use of CB as a source of HSCs. For example, a recent report demonstrating that even a small increase in the number of nucleated CB cells transplanted contributes to an improved disease-free survival rate of adult recipients [167]. This suggests that only minimal ex vivo expansion of CB may be required to reach a threshold level of cells necessary for the successful engraftment of larger individuals.
|
POTENTIAL OF BMP4 IN EX VIVO EXPANSION OF HSCS |
|---|
|
TopAbstractIntroductionBMP4 Signaling in Generation...Clinical Applications for Ex...Potential of BMP4 in...SummaryReferences |
|---|
The HGF activity of BMP4 and its role in the development of HSCs suggest that it may also have an ability to modulate the properties of HSCs in ex vivo cytokine expansion cultures. Indeed, recent findings suggest that HSCs may respond to BMP4 ex vivo. For example, treatment of CB cell populations highly enriched for human HSCs (CD34+CD38–Lin–) with a high dose of BMP4 (25 ng/ml) results in a modest increase in the length of time SRC can be maintained in serum-free media containing the hematopoietic cytokines SCF, fetal liver tyrosine kinase ligand-3 (Flt-3), granulocyte colony-stimulating factor (G-CSF), interleukin (IL)-3, and IL-6 [157]. However, the BMP4 effects were highly dependent on concentration because a lower dose of BMP4 in the culture resulted in complete differentiation of the starting CD34+CD38–Lin– population to CD34+CD38+ cells by day 4, with resultant decreases in cell expansion, colony-forming cells, and a significant reduction in day-4 SRC. Other BMPs (BMP2 and BMP7) were also tested in these assays and found to have concentration-dependent effects (Table 1). While low doses of these BMPs did not affect day-4 SRC, higher doses significantly reduced the ability of the cytokine cocktail to maintain SRC to day 4. No net expansion of SRC was detected in the cultures with BMP4 at 25 ng/ml, suggesting that BMP4 may provide a survival signal for HSCs in this culture system.
|
|
SUMMARY |
|---|
|
TopAbstractIntroductionBMP4 Signaling in Generation...Clinical Applications for Ex...Potential of BMP4 in...SummaryReferences |
|---|
|
ACKNOWLEDGMENTS |
|---|
|
TopAbstractIntroductionBMP4 Signaling in Generation...Clinical Applications for Ex...Potential of BMP4 in...SummaryReferences |
|---|
and 2. The two senior authors (Lewis and D’Andrea) contributed equally to this article. |
REFERENCES |
|---|
|
TopAbstractIntroductionBMP4 Signaling in Generation...Clinical Applications for Ex...Potential of BMP4 in...SummaryReferences |
|---|
