De novo generation of HSCs from somatic and pluripotent stem cell sources

Bone marrow transplantation (BMT) is the most established cellular replacement therapy, dating back to 1951 when Lorenz et al first described protection from the lethal effects of X-irradiation by bone marrow (BM) injection in mice and guinea pigs.¹  Thomas et al later infused patients receiving radiation and chemotherapy with BM from fetal and adult cadavers.²  BMT remains the only curative treatment of patients suffering from a variety of hematologic disorders, including sickle cell anemia, leukemia, lymphoma, and in at least one case, HIV infection.³  The functional unit of a BM transplant is the hematopoietic stem cell (HSC), which resides at the apex of a complex cellular hierarchy and replenishes blood development throughout life.⁴  Primary BM, umbilical cord blood, or mobilized peripheral blood are the only sources of HSCs presently available. The scarcity of HLA-matched HSCs severely limits the ability to carry out transplantation, disease modeling, and drug screening. HSC expansion represents one potential source of additional transplantable units.⁵  Considerable progress has been made in defining molecular determinants that can expand HSCs in culture.^5-7  However, even the most robust current protocols achieve only a modest expansion of long-term (LT) repopulating HSCs, and the expanded stem cells often have reduced multilineage and migratory potential compared with fresh HSCs. Furthermore, for a wide range of conditions, such as BM failure syndromes, too few functional HSCs are available for autologous expansion of gene correction strategies. Thus, in parallel with the efforts to expand HSCs, many studies have aimed to generate HSCs from alternative sources. This review will consider the latest advancements in the efforts to generate HSCs either by directed differentiation from pluripotent stem cells (PSCs) or direct conversion from somatic cell types.

During mammalian embryogenesis, blood development occurs in at least 2 waves. Primitive hematopoiesis first takes place in the extraembryonic yolk sac and generates mostly myeloid cells and nucleated erythrocytes. The primitive hematopoietic system is transient and replaced by HSC-driven intraembryonic adult-type definitive hematopoiesis.⁴  HSCs then take over the blood production of the embryo and possess the capacity for self-renewal, multilineage differentiation, and homing and engraftment to hematopoietic territories, including the fetal liver and BM in the adult. Functionally, HSCs are defined by the capacity for LT reconstitution of all blood lineages following transplantation.⁸  A number of different groups have focused on developing model systems that accurately and reproducibly recapitulate in vivo hematopoiesis.^9-12  The isolation of murine and human embryonic stem cells (ESCs)^13,14  offers a novel and unique opportunity to study blood development. ESCs are distinguished by the capacity to self-renew and differentiate into all 3 germ layers. ESCs differentiated as 3-dimensional aggregates, called embryoid bodies (EBs), give rise to hematopoietic cells in the presence of mesoderm morphogens and growth factors. Access to the earliest cells in hematopoietic ontogeny and the relative ease with which genes or pathways can be manipulated enables investigation of early stages of hematopoietic development that are otherwise difficult or impossible to obtain, especially in the context of human embryogenesis. Advances in reprogramming to induced PSCs (iPSCs)¹⁵  offers an even greater advantage, ie, patient-specificity. Cells derived from patients’ own tissues can theoretically allow for autologous transplantation, disease modeling, and drug screening when primary cells from patients are often limiting or unavailable. These properties make PSCs an appealing alternative source of HSCs for research and potential clinical applications, especially for those diseases that result from the destruction and/or dysfunction of HSCs in BM failure syndromes and leukemia.

Many directed differentiation protocols from PSCs have been established but these protocols invariably produce short-lived progenitors without bona fide HSC functionality (Table 1). Chadwick et al showed that hematopoietic growth factors and BMP-4, a ventral mesoderm inducer, promoted hematopoietic development in the context of EBs.¹⁶  The isolated CD45⁺ hematopoietic progenitors were capable of normal lineage maturation when plated into colony-forming assays. Other groups have developed coculture methods with BM stromal cells, such as OP9, and reported the emergence of CD34⁺ cells.^17,18  These protocols demonstrate that hematopoietic cells with clonogenic capacity can be generated from human PSCs (hPSCs), although these hematopoietic progenitors lack robust lymphoid differentiation and engraftment potential. An in-depth kinetic analysis of blood differentiation from ESCs by Keller et al revealed that blood development in EBs largely recapitulated primitive hematopoiesis.¹⁹  Like yolk sac progenitors, EB-derived hematopoietic cells lacked robust and durable repopulating capability in lethally irradiated adult recipients, hindering their therapeutic potential. This deficiency is believed to reflect an immature developmental program.²⁰  In support of this belief, when yolk sac progenitors are transplanted into neonates,²¹  or cultured on stroma derived from the blood-forming aorta-gonad-mesonephros region,²²  they engraft in the adult. Although these data suggested that primitive embryonic blood progenitors can acquire definitive potential, it is unclear whether the embryonic program can only be matured with proper microenvironmental cues or if the definitive program can be induced with cell-intrinsic regulators.

Although the molecular mechanisms that distinguish primitive yolk sac and definitive hematopoiesis remain largely unknown, investigation of the homeobox (Hox) genes in primitive and definitive hematopoietic populations has revealed that several of the Hox A and B clusters are highly expressed in definitive, but not yolk sac, cells.^23,24  Building on these studies, Kyba et al demonstrated remarkably that ectopic expression of HoxB4 endowed both yolk sac and mouse ESC-derived hematopoietic cells with multilineage engraftment potential.²⁵  However, the donor grafts exhibited marked myeloid skewing, indicating that overexpression of HoxB4 alone is insufficient to fully convert primitive ESC-derived progenitors into bona fide HSCs. Nonetheless, these seminal experiments with HoxB4 set the precedent for the induction of HSCs using defined transcriptional regulators of ontogeny. However, induction of HOXB4 in hPSCs did not generate engraftable cells, suggesting species-specific differences.^26,27  Molecular analysis of human ESC (hESC)-derived hematopoietic cells compared with somatic HSCs revealed that these cells express high levels of HOX B genes but only low levels of the HOX A cluster. Ramos-Mejía et al demonstrated that overexpression of HOXA9 enhanced hematopoietic commitment from hESCs but was not sufficient to confer engraftment.²⁸  Taken together, these data suggest that HOX A genes may be more important in augmenting definitive hematopoietic potential in human cells than HOX B genes.

The marked deficiency of stem cell properties in hPSC-derived progenitors has prompted the need to develop refined hematopoietic differentiation protocols. Achieving this goal requires a detailed understanding of the cellular and molecular pathways that lead to blood formation, specifically, the pathways that distinguish primitive and definitive hematopoiesis. During embryogenesis, endothelial and hematopoietic cells develop in parallel, and the close developmental association has led to the hypothesis that they arise from a common precursor, ie, the hemangioblast.^29,30  In the embryo, hemangioblast development is followed by a transient wave of primitive hematopoiesis that gives rise to primitive erythroid, macrophage, and megakaryocyte (MK) lineages, which is then replaced by definitive hematopoiesis bearing true HSCs with lymphoid potential.⁴  During hESC differentiation, Wang et al used cell surface markers associated with endothelial cells capable of early hematopoietic potential in the human embryo to identify a subpopulation of endothelial-like cells that possess hemangioblastic properties.^31,32  Until recently, it was unclear whether a parallel definitive program existed in PSC differentiation cultures.

It is now widely accepted that definitive HSCs arise from HE rather than a hemangioblast.³³  During mammalian embryogenesis, definitive HSCs originate in the vascular microenvironment of the aorta-gonad-mesonephros region³⁴  and arterial endothelium in other anatomical sites. Emergent HSCs have been observed to bud directly from HE cells lining the dorsal aorta.^35,36  Accordingly, hPSC differentiation protocols must include a stepwise addition of morphogens to specify a definitive HE. Choi et al identified a VE-cadherin⁺CD73⁻CD235a/CD43⁻ HE population distinct from hemangioblasts that were capable of forming endothelium and primitive blood cells.³⁷  In an attempt to define the signaling pathways that distinguish primitive and definitive hematopoietic development from hPSC differentiation cultures, Kennedy et al defined a CD34⁺CD43⁻ HE population that gave rise to T lymphoid, myeloid, and erythroid lineages. Primitive hematopoiesis was found to be temporally dependent on activin/nodal signaling, a pathway previously used to predict hPSC lines prone to hematopoietic specification.³⁸  However, the definitive wave, marked by T lymphopoiesis, was not dependent on this signaling pathway.³⁹  The same group sought to separate primitive and definitive progenitors during early hPSC differentiation, particularly during the stage of activin/nodal dependence. Sturgeon et al found that T-cell potential was restricted to the KDR⁺CD235a⁻ mesoderm-derived HE population. Furthermore, inhibition of Wnt/β-catenin signaling during the mesoderm specification stage negatively affected the definitive potential of KDR⁺CD235⁻ HE, whereas stimulation with a Wnt agonist restored T-lymphoid potential.⁴⁰  Collectively, these studies established a system where primitive hematopoiesis is specified by the combination of activin/nodal signaling and Wnt inhibition, whereas activation of Wnt/β-catenin and activin/nodal inhibition promotes definitive hematopoiesis from hPSC. However, a major limitation of these studies is that T-cell potential alone may not accurately define definitive potential. Böiers et al recently detected restricted lymphomyeloid progenitors that emerged before definitive HSCs during murine development.⁴¹  Since HSC-independent lymphopoiesis can arise in the developing embryo, these results imply that T-lymphoid potential might not unequivocally indicate conditions permissive to the emergence of HSCs. True HSCs are defined by multilineage potential and engraftment capability, therefore, a more accurate in vitro read-out would encompass multilymphoid (B- and T-cell) potential. Nonetheless, these experiments have provided valuable insights into the signaling pathways that control the emergence of primitive or definitive populations. Although no verifiable engraftment potential has yet been reported from hPSC-derived HE, these protocols have provided a unique opportunity to better understand the complexities of hematopoietic ontogeny.

Two recent reports suggest that HSCs with LT, multilineage repopulating potential can be derived from hPSCs under experimental conditions. Amabile et al reported that direct injection of human iPSCs (hiPSCs) into NOD-scid IL2rγ^null mice produced human CD45⁺ cells capable of mobilization and engraftment via teratoma formation. The flow-sorted CD34⁺CD45⁺ teratoma-derived cells were capable of multilineage engraftment and could engraft secondary recipients though at very limited levels.⁴²  In a similar report, Suzuki et al co-injected either mouse- or human-derived iPSCs with OP9, and subcutaneously implanted a micro-osmotic pump to supply hematopoietic cytokines in vivo. They reported that iPSC-derived HSCs migrated from teratomas to the mouse BM. Transplantation of murine teratoma-derived cells into irradiated recipients resulted in LT multilineage reconstitution in serial transfers. Isolation of teratoma-derived human CD34⁺CD45^dull cells showed multilineage differentiation in colony assays.⁴³  Extensive molecular analysis of these teratoma-derived cells was not performed, so it remains unclear how they compare with primary HSCs. Tumor formation from injecting PSCs is an obvious limitation of this differentiation system. Moreover, generation of HSCs by way of teratoma formation is inherently stochastic and variable. Still, these studies suggest that human hematopoietic cells with LT multilineage repopulating activity can in principle be derived from hPSCs. The in vitro system is devoid of supportive cues from the physiological microenvironment, but without precise knowledge of the cell intrinsic and extrinsic factors that contribute to the efficient production of HSCs, it remains a significant challenge to generate HSCs from hPSCs for any therapeutic application.

The concept of direct conversion, or directly switching cellular fate without reverting to a pluripotent state, was first demonstrated with myocytes. Taylor and Jones provided the initial evidence that somatic cells exhibit cell fate plasticity by treating fibroblasts with 5-azacytidine, an inhibitor of DNA methylation, and observing spontaneous differentiation into muscle and fat cells.⁴⁴  This finding suggested that DNA methylation restricted the expression of genes that regulate differentiation into alternative lineages. Davis et al later showed that the transcription factor (TF) MyoD was sufficient to convert fibroblasts into contracting myocytes.⁴⁵  This seminal finding demonstrated that trans-acting factors have the powerful capacity to regulate the epigenome and reprogram cell identity. In the decades that followed, many direct conversions between somatic cells have been reported, including many therapeutically relevant cell types, such as induced motor neurons,⁴⁶  pancreatic islet cells,^47-50  β-cells,⁵¹  and cardiomyocytes.⁵² 

Direct conversions between hematopoietic cell types have also been described, owing to the fact that lineage-specific TFs are relatively well-defined in hematopoietic lineages.⁴  GATA1, a critical erythroid TF, directs granulocyte-macrophage progenitors and lymphoid progenitors to erythroid and megakaryocytic cell fates.⁵³  B and T lymphocytes can be reprogrammed to macrophages through overexpression of C/EBPα.^54,55  GATA3, an essential TF for early T-cell development, can drive T-lineage precursors to acquire mast cell identity.⁵⁶  These studies demonstrate that cells of the hematopoietic system are amenable to reprogramming and highlight the instructive role of TFs in fate conversions.

Successes in TF-mediated reprogramming has encouraged several groups to adopt the strategy of directly converting somatic cells to HSCs (Figure 1). Szabo et al reported that ectopic expression of OCT4 combined with hematopoietic promoting cytokines could induce restricted myeloerythroid progenitors with limited self-renewal capacity from human fibroblasts.⁵⁷  The conversion likely relies on partial reprogramming, and a recent follow-up study with molecular analysis revealed that OCT4-transduced fibroblasts display elevated levels of developmentally related genes associated with multiple cell types but not those associated with pluripotency.⁵⁸  More recently, Pereira et al employed an H2BGFP fibroblast reporter line to identify a combination of TFs capable of inducing a hemogenic program from murine fibroblasts. They found that the hematopoietic TFs Gata2, Gfi1b, cFos, and Etv6 were sufficient to convert murine fibroblasts to hematopoietic progenitors. Extensive molecular analysis of factor-induced intermediates showed sequential downregulation of a fibroblast gene expression signature, followed by induction of an endothelial program, and finally, the activation of a hematopoietic gene expression profile.⁵⁹  The TF-induced fibroblasts transitioned through an HE-like intermediate, mimicking the in vivo progression of blood development. Although emergent hematopoietic cells possessed a LT-HSC gene expression profile and surface antigen phenotype, their functional potential remains to be fully interrogated. In a similar report, Batta et al converted embryonic and adult murine fibroblasts into multilineage hematopoietic progenitors with short-term engraftment potential by ectopic expression of Erg, Gata2, Lmo2, Runx1c, and Scl.⁶⁰  Interestingly, p53-deficient fibroblasts were more efficiently reprogrammed to hematopoietic progenitors with erythroid, myeloid, megakaryocytic, and lymphoid potential than wild-type fibroblasts, suggesting that p53 expression may be a barrier for reprogramming to blood.

The close ontogenetic relationship between endothelium and hematopoietic cells makes endothelial cells an attractive starting cell type for reprogramming. In an attempt to recreate a vascular microenvironment, Sandler et al cocultured transduced human umbilical vein endothelial cells and adult dermal microvascular endothelial cells with E4ECs, engineered endothelium-derived stromal cells that support HSC maintenance by producing physiological levels of paracrine factors. Using this coculture system, the introduction of FOSB, GFI1, RUNX1, and SPI1 into primary vascular endothelial cells induced multilineage hematopoietic progenitors with durable repopulating potential.⁶¹  FOSB, GFI1, RUNX1, and SPI1-reprogrammed cells expressed HSC markers and importantly, maintained serial engraftment in NOD-scid IL2rγ^null mice. However, the engrafted cells had limited lymphoid contribution. Sustained expression of some of the transgenes could interfere with lymphoid differentiation. The authors addressed this possibility using an inducible SPI1 transgene, which modestly augmented T-cell potential. Further optimization of TF combinations is required to achieve a balanced multilineage output. In addition, without a full molecular comparison of reprogrammed cells to primary HSCs and progenitors to establish cell identity relationships in an unbiased fashion, it is difficult to ascertain the true fidelity of fate conversion. Nonetheless, these experiments have established a novel approach to HSC reprogramming underscoring the important role of the vascular niche in providing instructive and supportive signals for reprogramming.

Although master regulators such as GATA1 and C/EBPα are capable of inducing conversions between mature hematopoietic cell types, no TF combination had been reported to interconvert mature cells to HSC-like cells until a recent study by Riddell et al.⁶²  The authors identified 36 factors specific to HSCs and systematically investigated their capacity to convert B-cell or myeloid progenitors into LT engrafting stem cells. Using immunoglobulin rearrangement as a genetic marker to track the clonal origin of donor-derived cells, they found that pro-B cells transduced with the 36 TFs could engraft recipients. Retrospective analysis of transgene integration reduced the 36 factor cocktail to just 6 necessary factors: Hlf, Lmo2, Pbx1, Prdm5, Runx1t1, and Zfp37. The addition of Meis1 and Mycn, and delivery of factors in polycistronic cassettes markedly enhanced reprogramming efficiency. The transduction of these 8 TFs as polycistronic constructs imparted LT multilineage differentiation capacity with serial transplantation capability. At the single-cell level, these induced HSCs clustered closely to bona fide HSCs based on quantitative analysis of 151 genes. This was the first study to report the de novo generation of functional murine HSCs; however, translation to human cells is complicated by several factors. First, the same set of TFs may not function similarly to reprogram human cells. Second, it is unclear whether the in vivo microenvironment is essential for reprogramming or merely to sustain HSC potential once the stem cell program has been established by the TFs. If the in vivo microenvironment is indeed necessary, the critical issue is how the same experimental procedure could be applied in a more translational setting. Since there is little control over the reprogramming process once the cells are transplanted in vivo, it is difficult to predict efficiency of conversion and success rate of multilineage reconstitution, if this procedure were attempted directly in patients. A better understanding of the cell intrinsic and extrinsic reprogramming factors that can establish an HSC transcriptional network will provide cues necessary for the de novo generation of human HSCs.

Direct conversions between somatic and hematopoietic cells have been shown to be a promising alternative to classical directed differentiation for HSC generation, but primary cells ultimately lack the scalability that pluripotent cells offer. Elcheva et al recently identified 2 transcriptional combinations capable of directly converting hPSCs to endothelium, which subsequently gave rise to blood cells with restricted pan-myeloid (ETV2 and GATA2) or erythro-megakaryocytic (GATA2 and TAL1) potential.⁶³  The authors did not detect meaningful engraftment of these cells, suggesting that the TF-induced cells have similar properties to hematopoietic progenitors differentiated from hPSCs using more conventional protocols. A more detailed molecular comparison between TF-induced cells and hematopoietic progenitors differentiated from hPSCs, as well as primary human HSC and progenitor populations, may be useful to identify potential regulators capable of inducing a definitive program directly from hPSCs.

Using a hybrid approach between directed differentiation and reprogramming, Doulatov et al differentiated hPSCs to CD34⁺CD45⁺ myeloid precursor cells and screened for a combination of TFs to reprogram to an HSC-like state, a process termed “respecification.”⁶⁴  The rationale for this approach is that first differentiating hPSCs to CD34⁺CD45⁺ precursor cells would minimize the “epigenetic distance” to a true HSC. Although the endothelial-to-hematopoietic transition in humans remains poorly understood, the molecular differences between primary human HSCs and progenitors have been well-characterized by gene expression profiling, providing a rational approach to introduce stem cell genes back into progenitors.^65,66  Previous attempts involving overexpression of single TFs, such as HOXB4, SCL, and RUNX1 in hPSCs have failed to yield durably engrafting cells.^26,67,68  A combination of HOXA9, ERG, and RORA endowed myeloid precursors with self-renewal capacity. The addition of SOX4 and MYB to this cocktail of TFs conferred engraftment potential with myeloid and erythroid lineages. Notably, the engrafted human erythrocytes underwent definitive maturation and hemoglobin switching to express β-hemoglobin in vivo. Although lymphoid lineages were absent in vivo, this is the first report of robust TF-mediated hematopoietic engraftment from hPSCs. However, reported engraftment was transient and restricted to myeloid and erythroid lineages. Further optimization of TF combinations will be necessary to generate HSC-like cells.

Despite the recent progress, fully functional human HSCs are yet to be successfully generated by direct conversion or differentiation. Existing derivation protocols and TF combinations must be refined and the converted cells further characterized. However, they are nonetheless invaluable model systems for hematologic diseases, since primary patient samples are often scarce, such as in BM failure syndromes. Easily obtainable tissues, such as fibroblasts^57,59,60  and endothelial cells,⁶¹  can be converted to engraftable hematopoietic progenitors to model diseases in vitro and in vivo. Patient-derived iPSCs can help elucidate the molecular pathways involved in disease through the establishment of “disease in a dish.”⁶⁹  To date, numerous pluripotent lines have been generated from patients with hematologic diseases, including Fanconi anemia,⁷⁰  sickle cell anemia,⁷¹  Diamond Blackfan anemia,⁷²  Shwachman-Diamond syndrome,⁷³  chronic myelogenous leukemia,⁷⁴  JAK2^V617F myelo-proliferative disorder,⁷⁵  dyskeratosis congenita,⁷⁶  Pearson syndrome,⁷⁷  and others. Hematopoietic differentiation of these iPSCs have recapitulated some aspects of these disorders,^72,77  and uncovered mechanistic links between multiple tissue deficits.⁷³  However, hematopoietic differentiation of hPSCs yield short-lived progenitors and mature cells^16,78  and without the possibility of generating large numbers of hematopoietic stem and progenitor cells, these protocols limit the scope of experiments that can be performed in vitro and preclude disease modeling in vivo.

The various transgene-driven systems outlined above present new platforms for chemical screening and in vivo modeling of hematologic disease. For example, in iPSC respecification, TF-mediated expansion of CD34⁺CD38⁻ cells followed by lineage-inductive culture conditions provides a large number of target cells available for chemical screening. A similar drug screening platform has been applied to iPSC-derived motor neurons to identify therapeutic compounds for amyotrophic lateral sclerosis.⁷⁹  The near-unlimited capacity to produce erythroid precursors may be suitable for transfusion-related applications,⁸⁰  disease modeling of red blood cell (RBC) disorders, and chemical screening in vitro. Importantly, the disease phenotypes can now be recapitulated in vivo, permitting greater insight into the etiology and a platform for drug testing. For instance, in vivo maturation of erythrocytes may offer a unique opportunity to perform screens for new therapies for the β-globin disorders.⁸¹ 

There is an enormous demand for transfusable blood products. Chronically transfused patients often develop erythrocyte alloimmunization⁸²  or suffer from iron overload complications. The in vitro generation of RBCs of a defined antigen profile from PSCs represents an alternative to classic transfusion products. Hiroyama et al established methods for generating RBCs from murine ESCs and demonstrated that transfusion of in vitro differentiated RBCs protected mice from lethal hemolytic anemia.⁸³  Several groups have reported protocols for differentiating RBCs from hESCs^84,85 ; however, blood group polymorphisms of different hESC lines may not be suitable for generating universal donor RBCs. On the other hand, iPSCs in principle could represent an inexhaustible supply of autologous RBCs for transfusion. Chang et al⁸⁶  and Papapetrou et al⁸⁷  have described methods for generating RBCs from hiPSCs, but these RBCs express embryonic and fetal hemoglobin. Interestingly, patients who retain expression of fetal hemoglobin do not manifest any RBC abnormalities in adult life,⁸⁸  so the fetal phenotype of iPSC-derived RBCs does not necessarily present a barrier to transfusion, especially in patients with hemoglobinopathies. However, similar to primitive yolk sac erythrocytes, iPSC-derived RBCs do not enucleate efficiently, and ongoing efforts are aimed at promoting terminal erythroid maturation.

Platelets are a central component of the blood-clotting cascade. As with RBCs, current human donor-derived platelets are limited in supply and have a short shelf life. The in vitro hematopoietic differentiation of PSCs to MKs, the precursors to platelets, represents a potentially unlimited supply of platelets for transfusion. Previous efforts to increase MK yield from hESCs have had limited success.^89,90  In a novel strategy, Nakumura et al generated immortalized self-replicating MK progenitor cell lines from hiPSCs using BCL-XL, c-MYC, and BMI1. These immortalized lines could be matured into CD42b⁺ platelets with in vivo functionality in a thrombocytopenic mouse model upon transgene silencing.⁹¹  The clinical application of this system could provide a plentiful supply of platelets from screened and selected immortalized megakaryocytic clones.

Adoptive T-cell transfer for cancer immunotherapy is often limited by the number of antigen-specific T cells that can be expanded from individual donors. Previous studies have demonstrated the feasibility of generating T lymphocytes from hPSCs, although the yield has been low.^39,92  Themeli et al combined iPSC and CAR technologies to produce antigen-specific T cells. CARs are engineered antibody-based receptors comprised of an extracellular domain of chosen specificity fused with cytosolic domains with signal transduction and co-stimulatory signaling capabilities.⁹³  Themeli et al generated iPSC lines from donor T cells and engineered them to express CAR targeting CD19 showing that the engineered T cells elicited strong antitumor responses in a xenogeneic Burkitt lymphoma mouse model.⁹⁴  CAR-engineered iPSCs are amenable to further genetic modification to remove potentially unwanted recognition from the native T-cell receptor locus. These studies highlight the therapeutic value of combined hematopoietic differentiation from iPSCs and CAR technologies to yield clinically useful cell types for cancer immunotherapy.

De novo generation of human HSCs remains a “Holy Grail” of regenerative stem cell biology. For many years, the concept of direct conversion has been an attractive alternative approach to hematopoietic differentiation from PSCs for the de novo generation of HSCs. In the past year, a number of studies have demonstrated the proof of principle that hematopoietic cells with varying degrees of stem cell potential can be induced from somatic cells. The critical issue with all these cell fate conversions, however, is the relationship of derived hematopoietic cells to their native counterparts. HSCs are functionally defined by their ability to reconstitute the entire hematopoietic system in transplant recipients, and this metric remains the gold standard for HSC engineering. However, functional interrogation must be combined with extensive molecular analysis of the transcriptome and epigenome to assess the relatedness of derived HSC-like cells to their native counterparts. This may be accomplished using advanced bioinformatics algorithms such as CellNet, a network biology platform for quantifying how closely engineered cell populations resemble their target cell type.^95,96  We would argue that all of the direct conversion and differentiation strategies aimed at generating HSCs remain to be more fully characterized and improved before any consideration of clinical application. However, the goal of obtaining de novo HSCs now seems attainable. In addition, differentiation protocols have produced other useful blood cell types, including RBCs, platelets, and T cells with the potential to transform the practice of blood transfusion and immunotherapy. The efforts devoted to HSC derivation have significantly advanced our understanding of early human hematopoietic commitment, endothelial-to-hematopoietic transition, critical HSC gene networks, and hematologic disease mechanisms. Moving forward, network biology must be combined with rigorous functional interrogation to provide a comprehensive strategy to reconstruct gene regulatory networks that control HSC identity, and ultimately enable conversion of heterologous cells into clinical grade HSCs.

The authors thank Sergei Doulatov for helpful discussion, and apologize to the many authors whose work they were unable to cite because of space constraints.

G.Q.D. is supported by grants from the National Institutes of Health, National Institute of Diabetes and Digestive and Kidney Diseases (R24-DK092760) and the National Heart, Lung, and Blood Institute Progenitor Cell Biology Consortium (UO1-HL100001), Alex’s Lemonade Stand, and the Doris Duke Medical Foundation. L.T.V. is supported by the National Science Foundation Graduate Research Fellowship. G.Q.D. is an investigator of the Howard Hughes Medical Institute and the Manton Center for Orphan Disease Research.

Contribution: L.T.V. reviewed the literature; L.T.V. and G.Q.D. wrote the paper.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: George Q. Daley, Boston Children’s Hospital, Harvard Medical School, Karp Family Research Building, 7th Floor, 300 Longwood Ave, Boston, MA 02115; e-mail: george.daley@childrens.harvard.edu.