| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
CONCISE REVIEW |
a Department of Obstetrics and Gynaecology, School of Human Development,
b Institute of Genetics, and
c Department of Immunology, University of Nottingham, Queens Medical Centre, Nottingham, United Kingdom
Key Words. Development • Immune tolerance • Stem cell therapies • Human embryonic stem cells • Hematopoiesis • Differentiation
Correspondence: Helen Priddle, Ph.D., Department of Obstetrics and Gynaecology, School of Human Development, University of Nottingham, Queens Medical Centre, Nottingham NG7 2UH, United Kingdom. Telephone: +44 115 82 30662; Fax: +44 115 82 30651; e-mail: helen.priddle{at}nottingham.ac.uk
Received August 1, 2005;
accepted for publication November 4, 2005.
 |
ABSTRACT |
|---|
 Top Abstract IntroductionDisclosuresReferences |
|---|
|
INTRODUCTION |
|---|
|
TopAbstractIntroductionDisclosuresReferences |
|---|
Immunogenicity of hESC-Derived Transplants
The degree and mechanisms of immunogenicity of hESC derivatives are only beginning to be investigated. Most somatic cells display major histocompatibility complex (MHC) proteins on their surface. These are highly variable between individuals and are responsible for presenting foreign pathogens and transplanted donor tissues to the immune system. Most cell types display MHC class I proteins through which they present peptides derived from endogenous antigens, which are recognized by host cytotoxic T lymphocytes. However, the major task of antigen surveillance is performed by "professional" antigen presenting cells (APCs) such as macrophages and dendritic cells (DCs), using MHC class II molecules to display exogenous peptides. Bone marrow (BM)–derived interstitial DCs are present in most tissues and therefore also in donor organs. Consequent DC-mediated antigen presentation in allografts leads to proliferation of donor-reactive recipient lymphocytes and acute organ rejection. Differentiation of hESCs in vitro to form a specific tissue or cell type (other than hematopoietic lineages) is likely to exclude cells presenting donor antigens via MHC class II molecules. Thus, acute rejection via direct allorecognition may be less likely than in organ transplantation [3]. However, MHC class II expression can be induced in certain cells during inflammation [4]. Chronic rejection, by contrast, is probable because MHC class I peptides from donor cells can be taken up and indirectly presented by recipient DCs to activate helper T cells and promote indirect allorecognition [5]. Thus, the avoidance of an immune response to hESC-derived material is improbable.
Testing of the immunogenicity of hESCs and potential transplant tissues differentiated from them is under way. Drukker et al. [6] demonstrated that undifferentiated H9, H13, and HES1 hESCs did not express MHC class II and did express low levels of MHC class I. Upon differentiation in vitro and in vivo, MHC class I increased moderately and was further upregulated by interferons. Li et al. observed similar results with H1 and H9 hESCs [7]. Additionally, they observed an inflammatory response to intramuscular endotoxin injection in immune-competent mice, abrogated when undifferentiated hESC were also injected, suggesting an inhibitory effect of hESCs on inflammation. Upon further investigation, undifferentiated hESCs or the more therapeutically relevant differentiated embryoid bodies (EBs) did not elicit a lymphocyte response in a mixed lymphocyte reaction (MLR). Furthermore, hESCs prevented the proliferative response of lymphocytes to third-party antigens. Undifferentiated hESCs and their differentiated derivatives did not express co-stimulatory molecules CD40, B7.1 and B7.2, suggesting an incomplete activation of lymphocytes, giving rise to a state of "anergy." This is a demonstration of immune-privilege in hESCs, also observed in rat ESC-like cells in which allogeneic, ESC-like cells engrafted without rejection [8]. The rat cells were shown to express the pro-apoptotic FAS ligand, which may have played a role in destroying reactive lymphocytes upon contact. However, microarray data [9] suggest this mechanism does not occur with human or mouse ESCs (mESCs) because FAS and its ligand are not expressed in the H1 hESC or D3 mESC lines. Little or no MHC expression, as observed for undifferentiated hESC, usually triggers natural killer (NK) lymphocyte–mediated lysis. However, little NK-mediated lysis was observed with undifferentiated or differentiated hESCs in vitro [6], providing another example of putative immune evasion.
An important question is whether fully differentiated hESC-derived cells, as might be used in replacement therapies, will evade or evoke an immune response. Undifferentiated hESCs and their early derivatives may have immune-privileged properties, but this might simply reflect immune evasion in early human development [10]. Although Li et al. [7] demonstrated a lack of immune response to differentiated hESCs, these cells were differentiated for only 20 days. Several hESC derivatives produced under similar conditions have fetal, rather than adult, characteristics [11–13], although it is clear that hESCs can develop to mature, functional APCs and that these are immunogenic in MLRs [14]. There is no reason to believe that the immune privilege apparent in hESC derivatives after short periods of differentiation will persist in cells differentiated, after successful engraftment, to a fully functional adult phenotype. However, it may be possible to confer immune-privileged properties on hESCs and their derivatives by exploiting mechanisms employed by immune-privileged cell types. For example, expression of transforming growth factor-ß (TGF-ß) [15] or indolamine 2,3-dioxygenase [16] can suppress local immune responses.
Strategies to Overcome the Immune Barrier in Stem Cell Therapies
Given the likelihood of rejection of hESC-derived transplants (Fig. 1A
), it is critical to circumvent the immune barrier in stem cell therapies. Strategies proposed to date are illustrated in Figure 1B–1E.
|
) from donor skin cells advanced the prospect of autologous stem cells [17] that are likely to avoid allorecognition [18]. However, in addition to ethical concerns about SCNT and the production of human embryos solely to supply cells for transplantation, SCNT will not reverse genetic defects in patient cells. Despite efficiency improvements [17], each stem cell line requires one or two healthy oocyte donors to undergo an invasive procedure. Furthermore, the high level of technical skill required and the propensity for abnormalities in SCNT embryos [19] preclude wide use of this patient-specific approach.
Alternatively, stem cell banking [20] (Fig. 1C) of non-SCNT lines could provide HLA-matched cells, adopting the process widely used for BM transplantation. However, the numbers of human blastocysts required to produce sufficient lines to cover just 70% of the population are also likely to be prohibitive to providing generic therapies [21].
Strategies to genetically modify hESCs [22, 23] to produce transplantable hESC derivatives [3] (Fig. 1D) provide another theoretical means to avoid immune rejection. At present, the complexity of the immune system and unpredictable effects of the genetic changes constitute the main limitations of this approach.
Perhaps the most promising solution to the immune barrier would be to induce a state of immune tolerance to the hESC line to be used (Fig. 1E). This requires the differentiation of hESCs to cell types able to modulate the patient’s immune system so subsequent grafts derived from the same hESC line avoid allorecognition. Several tolerogenic cell types exist in vivo. Donor BM mesenchymal stem cells are tolerogenic but reduce immune responses to third-party antigens [24], which would be undesirable. The BM hematopoietic compartment also contains tolerogenic cells. Donor dendritic cells can induce deletion of donor-reactive lymphocytes [25]. Veto activity (specific suppression of alloreactive lymphocyte proliferation) is detectable in CD34+ BM cells (potentially hematopoietic stem cells [HSCs]), CD33+ myeloid progenitors, and CD8+ cytotoxic lymphocytes [26]. Thus, hematopoietic differentiation from hESCs would provide a range of tolerogenic cells in vitro for clinical use. hESC-derived HSCs may be tolerogenic and, with their potential to populate hematopoietic niches, can provide a range of tolerogenic cells in vivo.
Hematopoietic Chimerism and Immune Tolerance
Mixed hematopoietic chimerism was first observed in a case of dizygotic twins in which 32% of lymphocytes from the male twin were XX and the female twin had 40% XY lymphocytes [27]. Such chimerism occurs in 8% of twins and 21% of triplets [28]. Development of immune tolerance for hematopoietic cells between one set of twins by mixing their hematopoietic systems in utero was confirmed when MLRs demonstrated no alloreactivity between twin lymphocytes [29]. The simplest explanation would be that mixing of cells occurs before development of the thymus and T cell selection; thus, cells from both twins are seen as "self" in each individual.
Mixed chimerism can also occur in adulthood. In 1983, Reisner et al. [30] demonstrated that severe combined immunodeficient patients could receive and durably engraft partially mismatched BM transplants without rejection. However, mismatched transplantation was not efficient in immunocompetent patients unless large doses of HSCs were delivered [31], most likely correlating with the quantity of peripheral alloreactive lymphocytes which must be destroyed. Early studies in kidney transplantation showed that the infusion of donor-specific BM (DSBM) improved subsequent donor organ survival and was associated with a diminished reactivity of patient lymphocytes to donor lymphocytes in MLRs [32]. More recently, four HLA-mismatched kidney recipients were treated with DSBM, leading to macro-chimerism in three patients. In two patients, their lymphocytes were unreactive to donor lymphocytes in MLRs and so they began immunosupressive drug withdrawal; one of the patients was sustained rejection-free without drugs [33]. In the mouse model, use of DSBM has also led to states of mixed hematopoietic chimerism with subsequent donor-specific immune tolerance demonstrated by acceptance of mismatched allogeneic skin grafts [34, 35]. In all of these examples, conditioning has been used in addition to BM to combat peripheral lymphocyte reactivity, although in some cases without the use of irradiation [32, 35].
BM from MHC class I–deficient but not MHC class II–deficient mice supports tolerance [36], indicating a role for MHC class II–expressing cells. In mouse BM transplant/skin graft models, allograft acceptance correlates with the presence of donor MHC class II cells in the thymus and the clonal deletion of donor-reactive host cells [37]. When high level chimerism is not associated with production of donor T cells, tolerance does not occur and the presence of donor T cells correlates with clonal depletion of alloreactive cells [38]. Interestingly, both host and donor T cells can be deleted [39]. Transplantation of BM sorted for Sca1+Lin– cells allowed skin graft acceptance in mice [34], suggesting that the HSC component of BM is sufficient to induce immune tolerance. Additionally, veto activity has been demonstrated in the CD34+ fraction [40]. Thus, it is likely that the administration of HSCs adds to the effect of conditioning regimes to neutralize circulating donor-reactive cells providing "peripheral tolerance." The HSCs subsequently establish hematopoietic chimerism and differentiate producing MHC class II–expressing cells that allow clonal deletion of donor-reactive cells and develop a state of "central tolerance" within hematopoietic niches.
Tolerance can be induced by donor hematopoietic cell–induced microchimerism or even in the absence of detectable microchimerism [41]. Subsequent to peripheral lymphocyte deletion, tolerance can be maintained by indirectly activated recipient regulatory T cells that suppress alloresponses to donor tissue. These effects can occur without donor cells after peripheral tolerance induced by blockade or depletion of T cells [42]. Thus, it may be possible to induce tolerance without the need for donor hematopoietic cells. It is important to note that strategies both with and without donor hematopoietic cells have proven successful and have been tried clinically, but without donor cells, tolerance incidence was low [43].
Use of ESC-Derived Tolerogenic Cells
The proposal to use hESC-derived hematopoietic cells to tolerize a patient to transplants derived from the same line (Fig. 1E) relies on the ability to differentiate hESCs to tolerogenic cells in vitro. The recent tolerization of an MHC-mismatched mouse to an mESC line [44] demonstrated that ESC-derived hematopoietic progenitor cells (HPCs) have the potential to support mixed chimerism and induce immune tolerance. In this study, MLR showed low levels of reactivity by recipient splenocytes to donor or recipient splenocytes but effective reactivity to third-party, indicating a state of tolerance. Similarly, rat ESC-like cells injected intraportally into an MHC-mismatched rat generated mixed chimeras that engrafted donor-specific cardiac allografts in the abdominal cavity without rejection [8]. Together these data demonstrate the potential of ESC-derived hematopoietic cells to overcome the immune barrier in stem cell therapies.
What Advances Are Required in hESC Technology?
The reproducible differentiation of hESCs to specific cell types from a range of hESC lines in significant numbers is currently in development [45]. H1 and H9 hESCs were first differentiated to CD34+ HPCs in 2001 [46]. H1 hESC differentiation to MHC class II expressing functional APCs (dendritic cells) has also been demonstrated [14].
In the first instance, a high dose of veto cells is required to delete peripheral alloreactive lymphocytes, especially if debilitating ablative conditioning is to be avoided [34]. Thus, with typically 102 demonstrable HPCs arising from each hESC differentiation, a higher level of efficiency will be important to produce sufficient cells for effective tolerance. Even in the most productive conditions, only 7.5% of differentiated hESCs were CD34+/CD45+, with less than 0.165% of these able to differentiate to hematopoietic lineages in vitro [47]. To generate a range of tolerogenic cell types for peripheral and central tolerance and allow their in vivo persistence via long-term repopulation (LTR) of the hematopoietic niche, the production of multipotent definitive (adult) hematopoietic progenitors will be required, as opposed to restricted primitive (embryonic) hematopoiesis. Notably, the requirement for donor T cells in tolerance induction is a challenge, given the difficulties in ESC-derived T-lymphoid reconstitution. Thus, there are two major goals: one of quantitative improvements and one of qualitative improvements in hematopoietic differentiation from hESCs. To achieve these goals, there is much we can learn from developmental biology.
Lessons to Be Learned from Developmental Biology
During development, pluripotent cells are instructed by extra-cellular environmental cues that progressively alter gene expression, thereby programming specific tissues (Fig. 2). Determining the nature of these cues may allow simulation of hematopoietic developmental processes by sequential and timely administration of signals of correct magnitude. Increased quantities of tolerogenic cells might be achieved by broad application of the desired signal to all cells and/or by application of inhibitor signals for superfluous cell lineages. Perrier et al. [48], for example, achieved production of midbrain dopaminergic neurons from hESCs by application of patterning molecules important in midbrain development. They succeeded in producing quantities equivalent to an entire adult substantia nigra. However, because in utero human development is difficult to study, examination of other vertebrate models may prove most fruitful in defining the complex ontogeny of cues driving hematopoiesis from early embryonic cells. Previous strategies [47] to program hematopoiesis from hESCs have concurrently exposed cells to a range of factors that in the embryo act sequentially during mesoderm induction, HSC formation, and blood lineage differentiation. In addition, vertebrate hematopoiesis occurs in two major waves, with the earlier primitive hematopoiesis resulting in predominantly erythroid and some myeloid lineages, and LTR tolerogenic cells only arising in the later phase of definitive blood formation. Thus, the development of blood lineages from hESCs is not necessarily indicative of the required cell types for persistent immune tolerization.
|
) have been implicated in germ layer specification in a range of vertebrates [50]. In hESC differentiation, upregulation of Cripto (a component of TGF-ß/nodal signaling) and downregulation of leftyA (a TGF-ß/nodal antagonist) have been observed [51]. There is reason to believe that a "mesendodermal" layer precedes the formation of endoderm and some mesoderm and that cells of the hematopoietic system are included in this subset of mesoderm [52, 53]. A genetic regulatory network has been constructed to summarize the functions of transcription factors and embryonic signals involved in the formation of mesendoderm in the amphibian Xenopus and its progression to the mesoderm and endoderm [54]. Initially, TGF-ß signaling is induced by maternal VegT and, combined with nuclear ß-catenin, ultimately leads to the expression of the T-box transcription factor, Brachyury. As development proceeds, Brachyury expression is switched off in the mesendoderm and becomes restricted to the mesoderm, giving rise to notochord and muscle [53]. The Xenopus Brachyury promoter has been shown to be responsive to low levels of activin (a member of the TGF-ß superfamily) [55], to ß-catenin (via T cell factor-binding sites) [56], and to fibroblast growth factor [55] (which promotes mesodermal gene expression through mitogen-activated protein kinase kinase signaling [57]). In contrast, Goosecoid represses Brachyury expression in Xenopus [58], and high level expression of goosecoid in mouse ESCs leads to undetectable expression of Brachyury [59]. Also in Xenopus, high levels of activin induce expression of Mix.1, which indirectly represses Brachyury by upregulating goosecoid. Thus, the level of TGF-ß signaling determines cell fate [60], as seen during the differentiation of mESCs, in which initially Brachyury-positive cells exposed to low levels of activin supported skeletal muscle and blood development, whereas higher levels of activin supported endoderm development [61].
Patterning of Mesoderm
Mesoderm is subdivided along three different axes: dorsal-ventral, anterior-posterior, and medial-lateral. In Xenopus, the primitive and definitive blood arise from different populations of cells, with the primitive blood developing from the ventral blood island mesoderm and definitive blood from the dorsal lateral plate (DLP) mesoderm [62]. Human definitive hematopoiesis arises from the splanchnoplural mesoderm of the lateral plate [49]. Taken together, it is likely that, to promote definitive hematopoiesis, we will need to understand signals that promote DLP mesoderm formation.
Lateral mesoderm that gives rise to adult HPCs is derived from caudal/posterior regions of the primitive streak in the mouse [63, 64] and chick [65]. In mouse, cells failing to express nodal (Fig. 2) segregate to the anterior of the epiblast, and when nodal expression is inactivated in the epiblast there is no mesoderm induction [66]. Furthermore, BMP4 (Fig. 2) expression (required for mesoderm induction, as shown by knockout mouse experiments [67]) is not maintained in the absence of nodal signaling. However, experiments in Xenopus suggest that the role of bone morphogenetic protein is in the patterning of, rather than induction of, mesoderm [68]. BMP4 has been shown to dramatically increase the numbers of HPCs formed from rhesus monkey ESCs [69] and to increase the number formed from cytokine-treated human EBs by 30% [47]. Both BMP-2/-4 and Wnt-8 signaling in Xenopus are required to inhibit more dorsoanterior gene expression and promote ventroposterior lateral mesoderm formation [70]. The effect of BMP-4 on zebrafish mesoderm patterning is dose-dependent, with injection of BMP-4 ventroposteriorizing the embryo at the expense of the dorsoanterior mesoderm. Dominant negative BMP-4 receptor or the BMP-4 antagonist, noggin, had the converse effect [71].
The cells of the Xenopus animal cap have been shown to be pluripotent when isolated in vitro and can be programmed to form different mesodermal subpopulations, in much the same way as we seek to differentiate hESCs (reviewed in Okabayashi and Asashima [72]). Low concentrations (<1 ng/ml) of activin (Fig. 2) induced ventral mesodermal tissues (including blood), higher concentrations (5–10 ng/ml) induced dorsal mesodermal tissues, whereas at 10–100 ng/ml notochord, heart, and endodermal tissue induction occurred. Similar attempts have been made to mimic mesoderm patterning during in vitro differentiation of mESCs: lower levels of activin were successful in inducing skeletal muscle and blood [61]. Fluorescence-activated cell sorting of differentiating cells determined that lower levels of Brachyury expression were associated with a mesodermal population that gave rise to blood progenitors [73].
Specification of Hematopoietic Progenitors
Hematopoietic potential (intra- and extra-embryonic) is detected as early as day 19 of human development [49]. Precursors in the para-aortic splanchnopleura and dorsal aortae give the first demonstrable lympho-myeloid potential at approximately day 27, when the dorsal aortae fuse to a single dorsal aorta where the first clusters of CD34+ cells are found adhering to the ventral wall (although some endothelial cells are also CD34+). The dorsal aorta continues to contain cells with lympho-myeloid potential until day 40, just before hematopoietic progenitors are detectable in the adjacent fetal liver [49]. Consistent with this, expression of the endothelial and hematopoietic progenitor marker, flk-1, is first observed in the human embryo at the beginning of week 4 in cells that later co-express CD34 [74]. The earliest hematopoietic precursors have endothelial potential and are known as hemangioblasts [75]. Such bipotential progenitors are present in the aorta-gonad-mesonephros (AGM) region of the mouse fetus [76], adult mouse BM [77], human cord blood, and adult human BM [78]. They have also been identified during in vitro differentiation of human ESCs, arising from cells with an immature endothelial phenotype [79].
How do the correct mesodermal subpopulations become committed to a hematopoietic fate? Flk1 (otherwise known as vascular endothelial growth factor receptor or kinase insert domain receptor) is required for hematopoietic and endothelial lineages in mice [80] and is expressed in embryonic and fetal hematopoietic progenitors prior to the HSC markers, CD34 and Sca-1 [81]. During mESC differentiation, germ layer segregation occurred giving rise to Brachyury-positive cells as mesoderm was induced [73]. These cells became Flk1-positive as mesoderm was specified to a hemangioblast/angioblast fate. In similarly in vitro–differentiated mESCs [80], the absence of Flk1 prevented expression of the transcription factor stem cell leukemia (SCL) (Tal1), whereas partial restoration of SCL expression rescued hematopoietic potential. SCL was also demonstrated to inhibit a smooth muscle fate. However, SCL expression was not able to restore hematopoietic and endothelial development in vivo, so it seems clear that while SCL is downstream of Flk1 in promoting mouse hematopoietic development, both are required. In a serum-free, in vitro differentiation model, mESCs were used to elucidate the factors involved in the stepwise generation of Flk1+ and SCL+ cells [82]. BMP4 (Fig. 2) was required to give rise to a Flk1+ population via Smad1/5 signaling. VEGF (Fig. 2) then increased the number of SCL+ cells, using Flk1 as a receptor and signaling via the MAP kinase pathway.
In contrast to the situation in mice, zebrafish Flk1 expression is dependent on SCL [83]. Furthermore, SCL expression, at least in primitive blood, does not require VEGF signaling [84]. Although it is likely that hESCs will behave more like mESCs or mouse embryos, it will be important to take cognizance of the situation in other model systems.
In mESCs, activin A further augments the SCL expressing population, whereas TGF-ß has an inhibitory affect [82]. The ectopic expression of SCL in zebrafish embryos led to an increase in blood and endothelial progenitors, suggesting that SCL has a role in specifying the hemangioblast from mesoderm [85]. Hemangioblast induction occurred only in tissues where the SCL binding partner, Lmo2, was additionally expressed. When both Lmo2 and SCL were expressed ectopically, the range of mesodermal tissues that could be induced to form hemangioblasts was extended [86]. When SCL–/– mESCs were differentiated in vitro, the expression of ventral mesodermal genes was normal, as was the expression of genes coexpressed in progenitors for endothelial and hematopoietic lineages, but hematopoietic-specific genes, such as Pu.1 and GATA-1, were not expressed [87]. This is consistent with the hypothesis that SCL is critical for the specification of prehematopoietic mesoderm to a hematopoietic fate. Conditional knockout studies have demonstrated that SCL is required only for a narrow window of time, ending after the onset of Tie2 expression, and is not required for HSC function [88].
In mice, visceral endoderm is in close proximity to the mesoderm that will form hematopoietic precursors. Indian hedgehog is expressed in visceral endoderm and the adjacent cells of the epiblast express receptors and transducers of that signal. When early streak mouse epiblast is cultured with recombinant Indian hedgehog, HPCs and endothelial cells are induced [89]. Studies with Smo–/– (hedgehog signal transducer) mESCs and mice show that whereas mice are able to undergo hematopoiesis, differentiated mESCs are not, suggesting that hedgehog signals are more important in ESC hematopoiesis than in vivo, where the developing blood island is exposed to a greater range of neighboring tissues [90]. In support of the intact mouse data, when the hedgehog pathway is blocked in zebrafish, either by mutagenesis or inhibitors, primitive hematopoiesis proceeds normally [84]. Importantly, however, definitive hematopoiesis does not proceed. Thus, as also seen for VEGF in the zebrafish, the signal requirements for HSC formation differ from those for primitive blood.
Thrombopoietin (TPO; Fig. 2) is able to augment serum-induced differentiation of mESC to hemangioblast-like colony–forming cells. This occurs independently of and supplementary to VEGF [91]. TPO treatment of HPCs stimulates Hoxb4 gene expression [92] and HOXA9 nuclear transport [93] through the MAP kinase pathway. Mice deficient in TPO or its receptor (c-Mpl) do not show a defect in the production of HPC but do show decreased numbers [94]. Because the effect of TPO is supplementary to VEGF, it is difficult to determine whether TPO exerts an effect on generation or expansion of HPCs. Resultant downstream expression of Hoxb4 may act as in mESCs, where it promotes hematopoietic differentiation capable of LTR of erythroid and myeloid lineages in irradiated mice, although with limited lymphopoiesis [95].
In the chick, anterior endoderm produces Wnt antagonists that signal to anterior mesoderm to promote heart development over blood development. When Wnt inhibitors are expressed in posterior lateral plate mesoderm, ectopic heart development is induced, and when Wnt-3a or Wnt-8c (Fig. 2) are expressed in precardiac anterior mesoderm, ectopic primitive erythrocyte markers are activated [96]. Wnt-3a has been demonstrated to have an effect on the self-renewal of HSCs, activating Hoxb4 and Notch1 through ß-catenin [97]. However, when BM HSCs lack ß-catenin, they can still self-renew and differentiate [98], suggesting that Wnt signaling may be redundant in this role.
Induction of Definitive Hematopoiesis
It is not yet clear for mammals whether definitive hemangioblasts develop by maturation of primitive hemangioblasts or arise from a separate mesodermal cell population or from waves of both [99]. Multipotent LTR HSCs were first demonstrated in the mouse AGM at day 10, at a time when the yolk sac and fetal liver are devoid of this activity [100]. However, day-9 yolk sac cells transplanted into neonatal mice [101] or day-8 yolk sac cells co-cultured with AGM [102] are capable of LTR activity, suggesting that the yolk sac contains HPCs capable of homing and/or maturing to a definitive phenotype in immature hematopoietic niches. Alternatively, the emergence of definitive HSCs could be linked with development of germ cells. In the chick, primordial germ cells (PGCs) colocalize with HPCs in the ventral wall of the dorsal aorta and these co-migrate into the splanchnic mesoderm [103]. Additionally, mouse PGCs have the capacity to differentiate into hematopoietic lineages [104]. Whether this reflects natural developmental progression, trans-differentiation, or the totipotency of PGCs is unclear. Xenopus definitive hematopoiesis is separate from primitive hematopoiesis with distinct origins clearly identified [105]. Both populations require BMP for formation, but their resultant blood and endothelial progenitors are differentially programmed by BMP (Fig. 2). Adult progenitors require BMP for initial activation of an early regulatory gene (Fli-1), whereas the embryonic progenitors do not [106]. Interestingly, BMP-4 is specifically expressed on the ventral side of the dorsal aorta, proximal to the hematopoietic clusters as they emerge, but this polarity is lost after the intra-aortic clusters disappear [107].
Yolk sac mesoderm from Notch1–/– mice can give rise to hematopoiesis, whereas mesoderm from para-aortic splanchnopleura cannot, suggesting that Notch1 plays a regulatory role in definitive, but not primitive, hematopoiesis [108]. Recent studies in zebrafish have also demonstrated a requirement for Notch, as well as hedgehog and VEGF (Fig. 2), in zebrafish definitive (but not primitive) hematopoiesis and the formation of the dorsal aorta [84]. This may reflect a requirement for the progenitors of definitive hematopoiesis to transit through a Hh and Notch dependent, arterial endothelial state prior to Notch dependent emergence of HSCs.
Can Definitive HSCs Be Produced from hESCs?
It is clear that nonhuman primate [69, 109, 110] and human [13, 14, 46, 47, 79, 111, 112] ESCs can be differentiated in vitro to give rise to hematopoietic precursors. Methodologies include differentiation of ESCs as EBs or on supportive cell layers and/or supplemented at various times with hematopoietic cytokine cocktails, VEGF, or BMPs. It is not so clear, however, whether such protocols can give rise to multilineage, self-renewing, definitive hematopoietic precursors capable of generating hematopoietic chimerism in a healthy adult human. Microarray comparison of the gene expression profiles of CD34+ CD38– cells derived from differentiation of hESCs with the CD34+CD38– cells of human adult BM reveals that hESCs derived HPCs are embryonic in character, with very low levels of flt-3, MHC, and adult globin gene expression [13]. However, Umeda et al. [110] reported the production of definitive hematopoiesis from cynomolgus monkey ESCs demonstrating the production of definitive-like enucleated erythrocytes and sequential induction of expression of embryonic (
and 
), fetal (
) then adult (
and ß) globins, although enucleation of erythrocytes is not unique to definitive hematopoiesis [113]. Differentiation of hESCs to NK- and B-lymphocyte lineages has recently been reported [112], suggesting that definitive hematopoiesis can be achieved. However, their ability to long-term repopulate irradiated mouse models and, ultimately, patients remains to be demonstrated.
The breakpoint cluster region/Abelson murine leukemia viral oncogene oncogene allows self-renewal of the HSC and was retrovirally transfected into mouse EBs to allow expansion of any HSCs formed, combating rapid differentiation to lineages [114]. In vitro observations suggest that only the earliest ESC-derived hematopoietic progenitors have T-lymphoid potential and that these cells lack CD49d required for migration to the thymus [115]. This may help explain the lack of in vivo T-cell reconstitution from mESC-derived HPCs. Multilineage hematopoietic repopulation of irradiated mice by mESC-derived HPCs has been achieved [116]. This success may have hinged upon selection of HPCs that expressed CD44, which is known to have a role in homing and proliferation of HPCs [117].
The difficulties in hematopoietic repopulation from mESC-derived material may be due to a failure to initiate definitive hematopoiesis [95]. Expression of HoxB4 enables yolk sac precursors and in vitro differentiated mESCs to repopulate (with poor lymphoid reconstitution) irradiated mice, suggesting a role for HoxB4 in establishing definitive hematopoiesis. However, it is still possible to hypothesize that rare definitive HSCs underwent expansion in the presence of HoxB4. Loss of HoxB4 in null mice allows development of all hematopoietic lineages, albeit with reduced proliferation of HSCs [118], suggesting that HoxB4 either does not have a role in induction of definitive hematopoiesis or is compensated for by other factors.
The Future
It is likely that we can generate immune tolerance to hESC lines if we can efficiently derive definitive HSCs from them. Vodyanik et al. [112] have succeeded in making definitive blood lineages from hESCs, demonstrating that this feat is achievable. However, the co-culture strategy used provided few clues about the factors required. Our review of Xenopus, zebrafish, chick, mouse, mESC, and human developmental biology has highlighted factors important for definitive hematopoiesis (Fig. 2). Induction of mesoderm requires activin and ß-catenin, and resulting cells are patterned by nodal, BMP, and possibly continued activin treatment. Specification of definitive hemangioblasts involves BMP, VEGF, TPO, and Wnt. These hemangioblasts then differentiate either directly or via an arterial endothelial intermediate to HSCs, involving VEGF, Notch, and BMP. The task now is to establish whether these factors are useful for hematopoiesis from hESCs and to define the necessary timing and magnitude of each signal. Consequent efficient production of definitive HSCs (potentially also their differentiation to immature DCs) will provide veto cells to induce peripheral tolerance. Selection of CD44+ HSCs will also enhance the ability of HSCs to reconstitute hematopoietic niches in the recipient and provide multiple cell types in vivo for central tolerance.
With such improvements, it becomes possible to imagine a future in which both transplant tissues and tolerogenic cells are continuously produced in parallel from the same few hESC lines within clinical laboratories, ready for immediate, off-the-shelf treatments. A patient would receive initial conditioning followed by simultaneous delivery of the tolerogenic and therapeutic cell types. Initial immune suppression could be used until immune tolerance has developed, after which time it could be withdrawn. Thus, a wide spectrum of patients, of all races and means, could expect drug-free durable engraftment of tissues, affording them a quality of life otherwise denied.
|
ACKNOWLEDGMENTS |
|---|
|
TopAbstractIntroductionDisclosuresReferences |
|---|
|
DISCLOSURES |
|---|
|
TopAbstractIntroductionDisclosuresReferences |
|---|
|
FOOTNOTES |
|---|
|
TopAbstractIntroductionDisclosuresReferences |
|---|
|
REFERENCES |
|---|
|
TopAbstractIntroductionDisclosuresReferences |
|---|
