International Journal of Hematology

, Volume 91, Issue 3, pp 384–391

Hematopoiesis from pluripotent stem cell lines


    • Department of Cell Differentiation, Institute of Molecular Embryology and GeneticsKumamoto University
  • Kiyomi Tsuji-Tamura
    • Department of Cell Differentiation, Institute of Molecular Embryology and GeneticsKumamoto University
  • Minetaro Ogawa
    • Department of Cell Differentiation, Institute of Molecular Embryology and GeneticsKumamoto University
Progress in Hematology ES and iPS cells, attractive stem cells for regenerative medicine

DOI: 10.1007/s12185-010-0519-7

Cite this article as:
Sakamoto, H., Tsuji-Tamura, K. & Ogawa, M. Int J Hematol (2010) 91: 384. doi:10.1007/s12185-010-0519-7


Embryonic stem cells (ESCs) can differentiate into various types of hematopoietic cells (HPCs) when placed in an appropriate environment. Various methods for the differentiation of ESCs into specific HPC lineages have been developed using mouse ESCs. These ESC-differentiation methods have been utilized also as an in vitro model to investigate hematopoiesis in embryos and they provided critical perceptions into it. These methods have been adapted for use with human ESCs, which have the possibility of being employed in regenerative medicine; further improvement of these methods may lead to the efficient production of HPCs for use in transfusions. The generation of transplantable hematopoietic stem cells is a medical goal that is still difficult to achieve. Recently, induced pluripotent stem (iPS) cells have been established from differentiated cells. Thereby, iPS cells have expanded further possibilities of the use of pluripotent stem cell lines in clinical application. Indeed, iPS cells have been established from cells with disease genes and those which have undergone reprogramming and targeting have generated phenotypically normal HPCs. Here, we mainly summarize the recent progress in research on hematopoiesis conducted with ESCs and iPS cells.


ES celliPS cellHematopoiesisHematopoietic stem cell

1 Methods for ESC differentiation into hematopoietic progenitors and HPCs

Embryonic stem cells (ESCs) and induced pluripotent stem (iPS) cells can grow indefinitely with multipotency under appropriate culture conditions, whereas they undergo spontaneous and synchronous differentiation into all cell lineages when removed from these conditions. Methods that efficiently enable ESCs and iPS cells to differentiate are very important, because in the undifferentiated state the cells form teratomas when transplanted into the body. Numerous ESC-differentiation methods have been developed from research using mouse ESCs (mESCs). In the differentiation pathway to hematopoietic progenitors, mESCs, similar to mouse embryos, go through the mesodermal cell lineage. Three different methods have been employed to induce mESCs to become mesodermal cells: (1) the formation of embryoid bodies (EBs, three-dimensional colonies of differentiating ESCs), (2) the co-culturing of ESCs with stromal layers and (3) the culturing of ESCs on extracellular matrix proteins. The former two methods are mainly utilized to differentiate mESCs into hematopoietic progenitors (Fig. 1).
Fig. 1

A scheme representing HPC development from ESCs. ESCs are prompted to differentiate into hematopoietic progenitors by two representative methods: EB formation and co-culturing on a stromal cell layer. The progenitors express Flk1 and subsequently generate hemangioblasts and hemogenic ECs. The hemangioblasts have the potential for producing ECs and primitive HPCs, whereas the hemogenic ECs give rise to mainly definitive HPCs

EBs are formed in suspension cultures of mESCs and contain erythroid and myeloid lineages [1, 2]. Within the EBs, cell–cell interactions enhance this differentiation; however, the complexity of these structures may be disadvantageous in regulating differentiation because accessibility of external factors to cells is limited within EBs (Table 1).
Table 1

Differentiation of ESCs and iPS cells into HPCs


HPC types




Ery-P, Ery-D

[3, 6, 32, 33]

B cell

[3, 7]

T cell

[11, 39]



Meg, Platelet

[8, 36, 37]


[45, 46, 50]

 iPS cell






[34, 35]

B cell


T cell

[16, 41, 43, 44]

Meg, Platelet

[14, 15, 38]

 iPS cell

Ery-D, Mye, Meg


Ery-P primitive erythrocyte, Ery-D definitive erythrocyte, Mye myelocyte, NK natural killer, Meg megakaryocyte

In the second method to obtain hematopoietic progenitors, stromal layers are employed in co-cultures with mESCs. The most used stromal cell line is the OP9 [3], which was established from an op/op mouse. op/op mice are deficient in macrophage colony-stimulating factor (M-CSF), which has some deleterious effects on the early development of hematopoietic cells (HPCs) [4]. In co-cultures of mESCs and OP9 cells, the former show high-level expression of Flk1, a marker of mesodermal cells in mouse embryos, in differentiating mESCs up to day 6 after the start of differentiation. The Flk1-expressing cells in EBs are detected up to day 4 and their number declines thereafter [5]. These findings suggest that OP9 cells provide a proper environment for the conversion of mESCs to hematopoietic progenitors by prolonging the mesodermal state during differentiation.

In addition, by supplementation with cytokines, OP9 cells can support the terminal differentiation of the hematopoietic progenitors into various HPC lineages. For instance, erythropoiesis on OP9 cells clearly shifts the primitive to definitive erythrocyte in the presence of erythropoietin (EPO) [3, 6]. OP9 cells also support in vitro B-lymphopoiesis when IL-7 and Flt3L are added to the culture medium [3, 7]. Megakaryocytes, natural killer cells and dendritic cells are also generated on OP9 cells [810]. On the other hand, OP9 cells expressing Delta-like ligand 1 (OP9-DL1), a ligand of Notch, induce the hematopoietic progenitors to differentiate into T lymphocytes [11]. OP9 and OP9-DL1 cells have been used to cause the differentiation of the progenitors derived from both nonhuman primate and human ESCs (hESCs) into HPCs [1216]. Mouse and human iPS cells also differentiate into HPCs both in EBs and on OP9 cells [1719]. However, EB formation may be a better way to obtain hematopoietic progenitors for clinical applications, because the progenitors within EBs can grow under pathogen- and animal-free conditions.

2 Hemangioblasts and hemogenic endothelial cells

The understanding of the mechanisms driving hematopoiesis has contributed to progression in clinical therapy. In embryos, proliferating hematopoietic progenitors are abundant and are useful for stem cell therapy. In research on hematopoietic progenitors, mainly mouse embryos have been used; however, these progenitors are limited in number and difficult to access in the embryos. These experimental obstacles can be removed by ESC-differentiation methods. Consequently, hematopoietic progenitors obtained by these methods, in addition to those from mouse embryos, have been broadly utilized in the research on hematopoietic progenitors. Moreover, by being applied to hESCs, ESC-differentiation methods can offer valuable knowledge of hematopoiesis in human embryos. Substantial advances in our understanding of hematopoiesis in embryos have been achieved by using ESC-differentiation methods.

In mouse embryos, a small number of cells in the lateral mesoderm, termed hemangioblasts, have the bi-potential to generate both HPCs and endothelial cells (ECs). Also, Flk1+ cells among developing mESCs have been shown to have hemangioblast activity [20]. Thus, the concept of the hemangioblast is supported by both experiments using mouse embryos and differentiating mESCs. Moreover, it is consistent with the fact that Flk1-knockout mice fail to develop both HPCs and ECs in the yolk sac [21]. However, evidence that the Flk1+ cells are not hemangioblasts, but rather multipotent progenitors, has been proven in a study using Flk1+ cells among differentiating mESCs and in embryos [20]. The Flk1+ cells generated smooth muscle cells and cardiac cells, in addition to HPCs and ECs. Similar to that for mESCs, hemangioblast activity has also been observed for hESC-derived cells [22, 23].

HPCs are known to arise from a subset of ECs, namely hemogenic ECs. In mouse embryos, budding of HPCs were observed on the ventral side of the aorta [24, 25]. Moreover, some ECs purified from mouse embryo developed into mature HPCs when grown in culture [26]. On the other hand, mESC-derived ECs also give rise to HPCs [27]. Eilken et al. [28] validated the existence of hemogenic ECs among differentiating mESCs by using time-lapse microscopy. They labeled mESC-derived ECs with low-density lipoproteins coupled to DiI (DiI–Ac–LDL) and genetically marked the ECs with the fluorescent protein Venus. Subsequently, they confirmed the formation of HPCs from the ECs. Also, they verified that counterparts of mESC-derived hemogenic ECs exist in the E7.5 mouse embryo. It has been shown that HPCs from ECs are mainly definitive. In particular, lymphopoiesis, observed only in definitive hematopoiesis, occurs in ECs from mESC [29] (Fig. 2). From these findings, hemogenic ECs have been proposed to be one of the origins of hematopoietic stem cells (HSCs).
Fig. 2

T-lymphocyte differentiation from ESC-derived ECs on OP9-DL1. ESCs were allowed to differentiate on OP9 monolayers, and then the resulting CD31+VE-cadherin+ EC cells were sorted at day 6 after differentiation. These EC cells were then cultured on OP9-DL1 monolayers in the presence of IL7, SCF and Flt3L. After 13 days, the floating cells were stained with CD8/APC and CD4/PE for identification as T cells

hESCs also give rise to bi-potential progenitors with the potential for generation of ECs and HPCs. The progenitors express CD31, Flk1 and VE-Cadherin, but not CD45 (CD45negPFV cells) [30]. It was suggested that human hematopoiesis and endothelial maturation originate exclusively from the CD45negPFV cells. By clonal analysis, it was confirmed that the CD45negPFV population contained a small number of bi-potent cells able to differentiate into ECs and HPCs. Interestingly, these cells reconstituted the hematopoietic system in immunocompromised mice on injection into the bone marrow [31], suggesting that the CD45negPFV population indeed contained HSCs.

3 Differentiation from ESCs into specific hematopoietic lineages

Early studies using mESCs showed that erythroid and myeloid lineages simultaneously develop from mESCs [2]. Control of differentiation into specific hematopoietic lineages has been achieved by optimizing the culture conditions of ESC-derived progenitors. Recently, in terms of transfusion medicine, these culture methods have been adapted to obtain hESC-derived HPCs. Application to human iPS cells will also provide further opportunities for regenerative medicine.

3.1 Erythroid cells

EBs at day 5–6 of differentiation give rise to erythroid colonies in methylcellulose cultures containing EPO and stem cell factor (SCF) [32]. mESCs co-cultured with OP9 cells efficiently differentiate into definitive erythrocytes [3, 6]. Nevertheless, this method cannot supply the large number of erythrocytes needed in the body (>1011 erythrocytes per day in humans). Carotta et al. [33] have developed an efficient method by which mESCs are directed to definitive erythroid differentiation without contamination of other HPC types. Firstly, they formed EBs and subsequently dissociated the EBs into single cells. The resultant cells expanded in number and became erythroid progenitors when grown in serum-free medium containing EPO, SCF and dexamethasone (Dex). Finally, these erythroid progenitors differentiated into mature enucleate erythrocytes by the addition of high concentrations of EPO, transferrin and insulin after the removal of SCF and Dex. These erythroid progenitors expanded 108- to 1012-fold within 9 weeks and differentiated into normal erythrocytes.

Erythrocytes from hESCs or iPS cells are in great demand from a therapeutic standpoint. Thus, universal hESCs having blood group O and suppressed expression of HLA molecules would be a good source of erythrocytes. Recently, methods for the large-scale production of erythroid cells from hESCs have been reported [34, 35]. Tsuji et al. [34] co-cultured hESCs with murine fetal liver-derived stromal cells (mFLSCs) to obtain such cells. Notably, hESC-derived progenitors were fated mostly to become definitive erythrocytes that were anucleate, underwent switching to adult-type β-globin and functioned as oxygen carriers. As much as 1 × 104 undifferentiated hESCs roughly generated 1 × 106 mature erythrocytes.

3.2 Megakaryocytes and platelets

Platelets, similar to erythrocytes, are a critical component for transfusion and so highly desired to be produced from ESCs. Megakaryocytes preferentially proliferate on treatment with thrombopoietin (TPO) in co-cultures of mESCs and OP9 cells [8]. Megakaryocytes produced from mESCs by similar methods are able to bind to fibrinogen functionally [36]. Moreover, another group has also reported that mESC-derived platelets can function in aggregation, bind to fibrinogen, express P-selectin and spread on immobilized fibrinogen [37].

Similarly, hESCs also generate functional megakaryocytes and platelets. Hematopoietic progenitors, arising from hESCs in co-culture with the murine bone marrow cell line S17 or the yolk sac endothelial cell line C166 in semisolid medium, form hematopoietic colonies containing megakaryocytes [38]. Megakaryocytes from OP9 co-cultures also show DNA polyploidy and expression of cytoplasmic and cell-surface proteins similar to those generated from peripheral or cord blood CD34+ cells [14]. Moreover, Takayama et al. [15] have estimated that mature megakaryocytes that arise from hESCs by day 24 in the presence of TPO generate relatively large numbers of mature megakaryocytes (2–5 × 105 platelet-producing megakaryocytes per 105 hESCs) that produce functional platelets. However, since these megakaryocytes yield fewer platelets than megakaryocytes do in vivo, further improvement will be required for clinical application.

3.3 T and B lymphocytes

Nakano et al. [3] reported the generation of B220+ B cells from mESCs co-cultured on OP9 cells with IL-7. Most of these B cells were IgM cells that had completed immunoglobulin DJ gene rearrangement. Subsequently, a small part of HPC clusters from mESCs gave rise to IgM+ cells that expressed mRNA for the complete μ chain. Furthermore, Zuniga-Pflucker et al. [7] added an Flt3 ligand to the above-mentioned OP9 co-cultures. B cells obtained by this method became mature Ig-secreting cells after LPS treatment. On the other hand, T lymphocyte differentiation from mESCs was obtained by using fetal thymic organ cultures (FTOCs), which provide complex thymic stromal interactions. Zuniga-Pflucker et al. [39] allowed mESCs to differentiate on OP9 cells and then reseeded the resulting Flk1+CD45 cells onto FTOCs, thereby observing normal CD4 and CD8 thymic subsets. Moreover, they found that OP9-DL1 cells supported the differentiation of mESC-derived progenitors into T lymphocytes [11]. These T lymphocytes had undergone the normal program of T-cell development, including the formation of γδ and αβ TCR-bearing T cells, and expressed a high level of TCR. In addition, CD8+ T cells among them produced IFN-γ.

Human ESC-derived CD34+ cells give rise to a small percentage of CD19+ B cells when cultured on MS-5 stromal cells, which support B-lymphoid hematopoiesis, in the presence of SCF, Flt3-L, IL-7 and IL-3 [40]. In addition, mRNAs for VpreB and Igα (CD79a or mb-1) components of pre-B-cell receptor complex are also detected. There are contradictory results from hESCs in T-cell development. Galic et al. [41] developed progenitors from hESC on OP9 monolayers and then injected them into mice with severe combined immunodeficiency that had small pieces of human thymus and fetal liver implanted beneath their renal capsule (SCID-hu model). In the conjoint liver and thymus of this mouse model, the human fetal liver and the thymus provide stromal elements required for HSCs and T cells, respectively [42]. T-cell development was observed in the conjoint organ at 3–5 weeks after transplantation. Recently, this group employed EB formation of hESCs to generate T progenitor cells and the resultant CD34+ cells were injected into SCID-hu model mice [43]. This experiment showed that CD3/CD28 activation increased CD25 expression on hESC-derived T cells. Moreover, Timmermans et al. [16] have also represented that CD34high CD43low cells from hESCs generated T cells on OP-DL1 monolayers. These hESC-derived T cells proliferated by stimulation with PHA. These findings suggest that hESCs can give rise to functional T lymphocytes. Nevertheless, CD34+ CD45+ cells from hESCs did not produce T cells, not only on OP9-DL1 monolayers but also in FTOC cultures; although umbilical cord blood cells generated T cells under both co-culture conditions [44]. Additional experiments will be required to clarify the difference between hESC lines or culture conditions.

4 Development of hematopoietic stem cell from ESCs

Advancing techniques of ESC differentiation have facilitated the generation of specific HPCs from ESCs. However, despite extensive efforts, efficient differentiation of transplantable HSCs from ESCs has still been difficult. Till now, two approaches to generate HSCs from mESCs have been reported: (1) a method that modifies culture conditions for ESC differentiation and (2) a method that employs genetically modified ESCs.

Burt et al. [45] reported on mESCs without genetic modification that allowed formation of EBs for 7–10 days in the presence of SCF, IL-3 and IL-6, after which the CD45+c-Kit+ cells in the EBs were transplanted into irradiated recipient animals. Surprisingly, despite transplantation into allogeneic recipients, these cells generated extensive hematopoietic chimerism and contributed to both the myeloid and lymphoid lineages. However, this approach does not lead to routine isolation of HSCs from ESCs. Since the FCS they used may have contained factors that promoted differentiation into HSCs, establishment of serum-free conditions would be needed to enable other investigators to reproduce this result.

An alternative approach to obtain HSCs from mESCs is to force the expression of genes that promote hematopoiesis and enhance HSC functions. Kyba et al. [46] succeeded in transplanting mESC-derived progenitors with forced expression of the homeobox transcription factor HoxB4. Donor cells were present in the transplanted mice up to 12 weeks after repopulation in primary recipients and as long as 20 weeks in secondary recipients. In addition, expression of HoxB4 led primitive hematopoietic progenitors of the yolk sac to become definitive HSCs [46]. Nevertheless, HoxB4-knockout mice had only a mild proliferation defect in their hematopoiesis [47], suggesting that HoxB4 is not required for the generation of HSCs or the maintenance of steady-state hematopoiesis in the mouse. Thus, the self-renewal effects of ESC-derived HSCs are likely to be attributed to the ectopic expression of HoxB4. In Kyba’s report [46], the ESC-derived HPCs with HoxB4 overexpression were predominantly Gr-1+ cells in the recipient mice, which is consistent with previous reports that HoxB4 blocks lymphopoiesis [48, 49]. However, at least some of the mESC-derived T cells with ectopic expression of HoxB4 were immunologically functional [48]. Similarly, mouse iPS cell-derived progenitors with the introduction of HoxB4 also reconstituted the hematopoietic system in irradiated mice [17]. On the other hand, Wang et al. [50] transplanted hematopoietic progenitors from mESCs expressing HoxB4 alone or both HoxB4 and Cdx4. Mice bearing engrafted hematopoietic progenitors treated with the two transcription factors showed a high level of donor chimerism and especially a high degree of lymphoid reconstitution, compared with mice having transplanted cells treated with HoxB4 alone. These findings suggest that the introduction of HoxB4 with Cdx4 could be the best way to generate HSCs from mESCs. In contrast to the relevance of HoxB4 in mice, ectopic expression of HoxB4 in hESCs has no effect on the repopulating capacity of their HSCs in immuno-compromised mice [31]; whereas HoxB4-transfected hESCs show an increased appearance of HPCs in vitro development [31, 51].

Though HoxB4 is a valuable gene to generate functional HSCs from hESCs, the next step in generating clinically useful HSCs will be to induce differentiation into HSCs without any genetic modification, because of the possibility that any exogenous gene might have a negative influence on normal development. Nevertheless, since exogenous genes essential for establishing iPS cells are excluded by Cre recombinase [52], this technique could be applied to eliminate HoxB4 and/or Cdx4. Intriguingly, the injection of mESC-derived progenitors into the bone marrow leads to faster and more effective reconstitution of hematopoiesis than that by the intravenous route [45]. Similarly, hematopoietic progenitors from hESC are also engrafted successfully when directly injected into the intra-bone marrow, though they fail to reconstitute by injection into a tail vein [31]. These findings suggest a deficiency in the potential of ESC-derived progenitors for homing and migration to, and interaction with, the bone marrow niche, similar to that observed in progenitors in the early mouse embryos [53]. Given that the fetal liver is a site of maturation into HSCs with high homing capacity to the bone marrow, one important approach to produce transplantable HSCs could be to provide the ESC-derived progenitors with proper conditions such as the fetal liver.

5 Application of iPS cells for regenerative medicine

iPS cell technology generates pluripotent cells from readily available differentiated cells, thereby obviating ethical concerns surrounding hESCs, as no human embryos are harmed. Human iPS cells after direct reprogramming and targeting can be used for therapeutic purposes. Indeed, Hanna et al. [17] employed a humanized knock-in mouse model of sickle cell anemia to investigate the therapeutic potential of iPS cells. They established iPS cells from tail tip-derived fibroblasts of the mice and subsequently corrected sickle globin genes by homologous recombination with a targeting construct containing the human wild-type globin gene. When hematopoietic progenitors that arose from these iPS cells after introduction of HoxB4 were transplanted into irradiated mice with sickle cell anemia, all hematological and systemic parameters of sickle cell anemia improved substantially and were comparable to those in the control mice. Xu et al. [54] cured a hemophilia mouse model by transplantation of ECs that were generated from murine wild-type iPS cells. These murine experiments suggest that human iPS cells can be utilized for regenerative and therapeutic applications.

Most recently, patient-specific iPS cells have been established. Raya et al. [18] reprogrammed dermal fibroblasts and/or epidermal keratinocytes of Fanconi anemia (FA) patients to generate iPS cells, which were followed by genetic correction with lentiviral vectors encoding FANCA or FANCD2, by which the FA phenotype was corrected. These iPS cells have not shown any detectable differences in the differentiation ability compared to that of either hESCs or normal iPS cells and, most importantly, the corrected FA-specific iPS cells have generated phenotypically normal HPCs. This study has opened up new opportunities for therapy applications of iPS cell technology (i.e., the generation of disease-corrected, patient-specific iPS cells).

6 Perspective

It has been widely recognized that cell transplantation is useful for regenerative medicine. In fact, bone marrow transplantation has cured malignant or non-malignant hematopoietic diseases for over decades. Therefore, derivatives of hESCs and iPS cells are exclusively expected to be employed as a cell source for regenerative medicine. However, certain issues have to be settled before clinical practice can begin, e.g., augmenting the number of cells produced from ESCs and iPS cells. In particular, red blood cells and platelets are required in enormous amounts for transfusion medicine. As red blood cells and platelets have been efficiently generated from hESCs by a great deal of effort [14, 15, 34, 35, 38], it could be feasible to use HPCs generated from hESCs and iPS cells by advancing other techniques, such as bioengineering. On the other hand, basic science concerning hematopoiesis can also contribute to resolving these issues. For instance, HPCs develop in the context of a complex and 3-dimensional microenvironment composed of paracrine factors, stromal contacts and physical forces that vary during development. Understanding the roles of this microenvironment in hematopoiesis may provide a key to more efficiently reconstitute hematopoiesis in vitro by using ESCs. Moreover, knowledge of hematopoietic regions could help to develop HSCs from ESCs and iPS cells that home to the bone marrow via the circulation without genetic modification. Overcoming the hurdles facing the use of pluripotent stem cell lines will open a new era of regenerative medicine.


Our studies described in this manuscript were partially supported by a grant (No.21591208) from the program, Grants-in-Aid for Scientific Research (C), 2009, of the Ministry of Education, Science, Sports and Culture. Our work was also supported in part by the Global COE Program (Cell Fate Regulation Research and Education Unit), MEXT, Japan.

Copyright information

© The Japanese Society of Hematology 2010