Generation of red blood cells from induced pluripotent stem cells

Abstract

Purpose of Review:

Human induced pluripotent stem cells (iPSCs) are an attractive source to generate in vitro-derived blood for use as transfusable and reagent red cells. We review recent advancements in the field and the remaining limitations for clinical use.

Recent findings:

For iPSC-derived red blood cell (RBC) generation, recent work has optimized culture conditions to omit feeder cells, enhance red cell maturation, and produce cells that mimic fetal or adult-type RBCs. Genome editing provides novel strategies to improve cell yield and create designer RBCs with customized antigen phenotypes.

Summary:

Current protocols support red cell production that mimics embryonic and fetal hematopoiesis and cell yield sufficient for diagnostic RBC reagents. Ongoing challenges to generate RBCs for transfusion include recapitulating definitive erythropoiesis to produce functional adult-type cells, increasing scalability of culture conditions, and optimizing high density manufacturing capacity.

Introduction

Red blood cell (RBC) transfusions are a critical therapy for individuals with hematologic disorders as well as those in the setting of trauma, surgery, and cancer. Approximately 29,000 units of RBCs are transfused daily in the United States alone (1), where blood donations generally meet the demand. Globally, many regions lack a sufficient or safe blood supply, and thus strategies for alternative sources are needed. Alloimmunization to donor red cell antigens, which occurs more frequently in heavily transfused populations including sickle cell disease (SCD) and thalassemia, may lead to requirement for uncommon or rare RBCs. In addition to transfusion therapy, availability of reagent RBCs is reliant on blood donors, which are used for pre-transfusion antibody testing to ensure safe and compatible blood transfusions. Recent blood shortages in the setting of an international pandemic highlight the continued need for other red cell sources (2).

RBCs can be generated ex vivo starting from hematopoietic stem and/or progenitor CD34+ cells from cord blood (CB), adult bone marrow, or peripheral blood following mobilization with stimulating growth factors. RBCs can also be produced from human pluripotent stem cells (PSCs), which include embryonic stem cells (ESCs) and induced pluripotent stem cells (iPSCs) which have the capacity for unlimited self-renewal. iPSCs are reprogrammed from somatic cells and can be differentiated into any of the three germ layers and the derivative cell types (3). Although each starting source of cells have their advantages and disadvantages in generating mature transfusible RBCs, iPSCs are especially attractive for their unlimited self-renewal potential to generate in vitro-derived blood for use as transfusable and reagent RBCs. They are renewable and easily manipulated to produce RBCs with customized antigen phenotypes that would particularly benefit alloimmunized patients. Based on alloantibody combinations found in 16,486 mutiply transfused patients, only 3 human iPSC clones are needed to produce compatible red cells to match more than 99% of the alloimmunized individuals (4). Current challenges include producing sufficient cell numbers since one RBC unit contains 2 e¹² cells, reaching culture cell densities to minimize reagent costs, and recapitulating the adult developmental hematopoietic program to facilitate the final stages of erythropoiesis that include adult-type (beta) globin production, enucleation and reticulocyte maturation. These obstacles have guided the goals of recent and ongoing research in iPSC-derived RBC production (Figure 1).

Figure 1. Current goals in producing iPSC-derived RBCs.

Ongoing research in the field includes increasing cell yield and cost efficiency (A), improving erythroid maturation (B), and studying in vivo maturation and survival of in vitro-derived RBCs from iPSCs (C).

Primitive and definitive erythropoiesis

Successfully producing RBCs from iPSCs requires an understanding of both the primitive and definitive developmental stages of hematopoiesis, which have primarily been studied in mice. The site of blood production varies developmentally, beginning in the yolk sac, transitioning to the fetal liver and subsequently, the bone marrow. Although human studies are limited by availability of primary cells from 3 to 6-week old embryos, the primitive program is transient and restricted to erythroid, megakaryocyte, and macrophage lineages (5). Primitive erythropoiesis occurs in early development in the extraembryonic yolk sac where erythroblasts develop from mesoderm (6). Primitive erythroid cells are larger, express embryonic ε-, ζ-, and α- globins (7), and remain nucleated in circulation but can enucleate aided by macrophages in the placental villa (8).

Definitive erythropoiesis involves the production of RBCs that are small, anucleate, and express fetal and adult globins. The first definitive RBCs arise in the yolk sac and mature in the fetal liver and bone marrow (9). The burst-forming unit erythroid (BFU-E) is the most immature cell that is committed to the definitive erythroid lineage (10–12). These progenitor cells develop in the yolk sac and then travel to seed the fetal liver. The BFU-E wave is followed by a colony-forming unit erythroid (CFU-E) wave which is 4-5 cell divisions upstream of mature RBCs (9, 13). These progenitors also travel from the yolk sac to the fetal liver (10). After the progenitor stage, erythroblast precursors mature from the proerythroblast to the orthochromatic erythroblast in the fetal liver and bone marrow, generating reticulocytes and mature RBCs (9). Fetal γ- and adult β-globin genes are expressed in human definitive RBCs.

Initial liquid culture systems developed for growing human erythroid precursors allowed not only studying cytokine and hormone requirements for expansion and differentiation of erythroblasts, but also provided insight into defined culture conditions that paved the path for large-scale cultures for generating differentiated RBCs expressing adult globin for clinical applications (14, 15).

Methods for producing iPSC-derived red blood cells

The development of protocols to produce RBCs from iPSCs is based on work focused on the expansion and enucleation of erythroid cells from CD34+ hematopoietic stem cells (HSCs) from CB, adult bone marrow, or peripheral blood. CB-derived CD34+ cells can be differentiated into erythroid progenitors with up to 200,000-fold expansion, achieve 4% enucleation by day 17 of culture, and can continue to amplify in vivo after transfusion into NOD/SCID mice (16). Co-culture of CD34+ cells with murine MS-5 stromal cells supports red cell generation with 90-100% enucleation after 15-18 days and terminal maturation when reticulocytes are transfused into NOD/SCID mice (17). These findings suggest stromal co-culture may produce an ex vivo microenvironment that facilitates erythroid cell growth and differentiation.

While CD34+ HSCs provide a viable source of in vitro-derived RBCs, the final cell yield is limited by the number of progenitor cells collected from the donor. Human ESCs first provided an alternative, renewable cell source. Nearly two decades ago, Olivier et al developed a 5-step protocol to produce large numbers of erythroid cells from ESCs by co-culture with human fetal liver clone B (FH-B-hTERT) cells, a telomerase immortalized cell line, to produce CD34+ hematopoietic cells (18). ESCs were induced to differentiate, erythroid progenitors stimulated with specific cytokine cocktails, and then co-culture with murine MS-5 cells led to terminal differentiation; resultant RBCs were nucleated and expressed both embryonic and fetal globins (18). Lu et al expanded this work by developing a system to generate RBCs from ESC-derived hemangioblasts that is serum-free, thus decreasing the variability in efficiency and reproducibility that serum can contribute to cultures (19).

iPSCs provide similar advantages to ESCs and erythroid differentiation from either PSC source showed no significant difference in the erythroblasts produced (20). iPSCs cultured with FH-B-hTERT cells were shown to undergo the same globin switches as ESCs regardless of origin from embryonic, fetal or adult sources (21). Most iPSC hematopoietic differentiation methods involve induction to mesoderm and hemogenic endothelium and subsequently to hematopoietic progenitor cells (HPCs), and primarily recapitulate embryonic or early fetal blood development. This is achieved either through formation of embryoid bodies (EBs), which are suspension aggregates of iPSCs, or a monolayer differentiation with or without co-culture with feeder cell types to generate HPCs (20–24). Subsequent HPC culture with specific cytokines facilitates the proliferation and differentiation to mature RBCs. These cells most closely resemble orthochromatic erythroblasts, which contain nuclei and are larger than adult RBCs. Flow cytometric analysis of cell surface markers can be used to assess the erythroid stages and maturation during in vitro differentiation and include CD36, CD71 (transferrin receptor), and CD235a (glycophorin A). CD36 increases in erythroid progenitors but subsequently decreases as erythroid cells terminally mature (16). CD71 and CD235 expression increase with erythroid differentiation but only CD235 is found on mature erythrocytes. The hemoglobin produced is largely fetal (HbF) (20–22, 25), and/or embryonic (23, 24, 26, 27).

In recent years, there has been a focus on improving methods to optimize erythroid proliferation, increase maturation and enucleation, and improve cost efficiency. Earlier studies used co-culture of iPSCs with stromal cells to induce hematopoietic differentiation that was followed by erythroid specific culture; cell yield was relatively limited and low enucleation rates (2 – 10%) were attained (24). To improve enucleation, Bernecker et al developed a method with an intermediate hematopoietic cell forming complex (HCFC); the HCFC is made from adherent EBs and surrounding stromal cells that release hematopoietic stem/progenitor cells from which CD43⁺ hematopoietic cells could be collected repeatedly over a period of up to 6 weeks for further erythroid differentiation (25). These cells expanded 100- to 1000-fold and enucleation rates improved to 40-60%, suggesting the HCFC may provide a niche that mimics the in vivo microenvironment. To decrease culture costs, robust erythroid differentiation (RED) protocols were developed with chemically defined culture medias that lack albumin or animal components and significantly less transferrin (26). In this 31-day culture system, 1 e⁶ iPSCs expand to 1 e¹² erythrocytes with up to 50% enucleation. Table 1 summarizes the culture length, erythroid cell yield, hemoglobin production, and enucleation rates between highlighted protocols. It is noteworthy that the current cell yield from these methods is not yet sufficient for large-scale production of iPSC-derived RBCs.

Table 1.

Comparison of selected methods to generate iPSC-derived red cells

Year	Author (et al)	Feeder cells vs. embryoid body formation	Length of culture	Yield	Hemoglobin/globin made	Enucleation
2010	Lapillone	EB	20 days for EB formation + 26 days in erythroid culture	1x106 iPSC → 4.4x108 erythroid cells	Predominantly fetal hemoglobin	4%-10%
2011	Dias	OP9 cells MS5 cells	7-8 days in co-culture + 5-20 days in erythroid culture Co-culture for 20-25 days to facilitate enucleation	1x106 iPSC → 2x1011 erythroid cells	Predominantly fetal and embryonic hemoglobin	2- 10%
2012	Kobari	EB	27 days for EB formation + 25 days in erythroid culture	1 x 106 iPSC → 15-28.3x108 erythroid cells	Predominantly fetal chains	20-26%
2015	Dorn	EB	21 days for EB formation + 25 days in erythroid culture	106 seeded cells → 33x106 erythroid cells	Fetal and embryonic hemoglobin	21-29%
2019	Olivier	EB	10 days + 21 days in erythroid culture	1x106 iPSC → 1x1012 erythroid cells	Fetal and embryonic hemoglobin	<5%
2019	Bernecker	EB	5 days for EB formation + 14 days for HCFC formation + 18 days in erythroid culture	0.07 x 10^6 iPSC → 1.2x1010	Predominantly fetal hemoglobin	37 – 60%

EB = embryoid body, FhB-hTERT= human fetal liver-derived cell line, OP9 = macrophage-derived embryonic stem cell line, MS5 = murine stromal cell line, HCFC = hematopoietic cell forming complex formation, HPC = hematopoietic progenitor cell

Although the majority of current methods to produce iPSC-derived RBCs recapitulate primitive hematopoiesis, manipulating signaling pathways to produce definitive hematopoietic progenitors may allow generation of RBCs with improved maturation capacity and expression of adult-type globins. Blocking activin-nodal signaling early in human PSC differentiation can enrich production of definitive hematopoietic progenitors (28). Specification of definitive hematopoiesis has also been shown to require Wnt-beta-catenin signaling, in part due to strong CDX4 expression (29, 30). RBCs generated from these definitive iPSC-derived hematopoietic progenitors closely mimic primary fetal liver-derived erythroblasts, as measured by transcriptomic profiling, morphology, and red cell antigen expression (31).

Maturation of iPSC-derived RBCs following transfusion in murine models

Since the microenvironment likely facilitates the final stages of erythroid maturation, several groups have assessed the effects of transfusion of iPSC-derived RBCs into mice. Using erythroblasts differentiated from iPSCs created from patients with and without SCD, transfusion into NOD/SCID mice resulted in mature RBCs with nucleus expulsion and adult hemoglobin synthesis (22). Similarly, transfusion of iPSC-derived RBCs into immunodeficient NOD/SCID-γ mice resulted in only 0.15% of circulating cells at 1 hour after transfusion, which decreased to 0.05% at 24 hours, but then increased after day 15 to 1.2% of circulating cells on day 24 and were 95-98% enucleated (32). Moreover, iPSC-RBCs found in the peripheral blood had cell surface marker profiles more comparable to human adult RBCs than the iPSC-derived RBCs just prior to transfusion (32). The authors hypothesize that the late increase in circulating iPSC-derived RBCs in murine blood was due to initial homing of cells to the bone marrow, which then differentiated and seeded the peripheral blood.

Generation of designer RBCs from iPSCs

Red cell alloimmunization remains a challenge for chronically transfused populations, patients with hemoglobinopathies, and multiparous women. Identification of the exact red cell antibody formed by a patient can be hampered by a lack of reagent RBCs with uncommon or rare antigen phenotypes; unavailability of reagent cells that either express, or conversely, lack a specific red cell antigen can delay or prevent adequate pre-transfusion testing that is necessary to provide safe transfusion. RBC generation from human iPSCs offer an attractive alternative to produce diagnostic red cell reagents since iPSCs are amenable to genome editing, allowing for customized antigen phenotypes. Since alloimmunization can lead to difficulty identifying sufficient compatible donor units, another source of RBCs such as iPSC-derived may afford a life-saving transfusion.

Type O, Rh null RBCs are particularly attractive as a universal donor cell as they completely lack the most immunogenic membrane proteins, ABO and Rh, rendering them superior to current “universal” donor type O- cells that lack RhD but express other Rh family proteins. CRISPR/Cas9 gene editing of blood group system genes RHAG, GYPB, and XK in iPSCs was used to generate erythroblasts that lacked Rh, glycophorin B, or Kell antigens (33). These cells can facilitate antibody identification against high prevalence antigens that >99% of the population express. In addition to Rh null iPSCs, our laboratory engineered a panel of iPSCs with rare or uncommon Rh antigen phenotypes to allow precise identification of complex Rh antibody targets, including variant Rh antigens prevalent in Black patients (27). We designed Rh null iPSCs that lack both RHD and RHCE genes and used these iPSCs to express a specific Rh antigen such as RHD alone (the very rare D- - phenotype), or RhD variants (e.g. DAK or Go^a). The RBCs derived from these iPSCs are compatible with common diagnostic assays used in the blood bank and can quickly identify rare alloantibodies in patients requiring immediate transfusion (27). Notably, RBCs generated from methods to produce definitive hematopoietic cells from iPSCs demonstrate blood group antigen expression that more closely mimics that of adult erythroid cells (31).

iPSC-derived type O Rh null RBCs and those with other rare antigen phenotypes can be an alternative source of blood for alloimmunized patients. As proof of principle, iPSCs derived from a rare Rh null individual with blood type A were converted to the universal donor type O using a targeted gene editing strategy that reproduced the c.261delG polymorphism present in the inactive ABO*O.01 allele (34). This approach can be applied to other iPSCs reprogrammed from somatic cells of individuals with rare red cell antigen phenotypes who are not blood type O.

Enhancing proliferation of in vitro-derived RBCs through manipulation of gene expression

Gene editing of iPSCs may be harnessed to optimize red cell yield by manipulating genes that regulate their proliferation. Since rare loss-of-function mutations in the gene SH2B3 are associated with increased red cell count, CRISPR/Cas9-mediated genome editing was used to create homozygous SH2B3 deletions in ESCs that resulted in 3-fold increased erythroid cell production (35). While manipulation of BMI1 or KIT has focused on expanding or immortalizing erythroid precursors (erythroblasts) from donor peripheral blood mononuclear cells (PBMCs), these genes can be targeted in iPSCs to potentially enhance red cell production. Human PBMC-derived erythroblasts with ectopic expression of the BMI1 transgene exhibited continued proliferation for 60 days with >10¹⁴ fold expansion (36). Introduction of a constitutively activating mutation of the stem cell factor (SCF) receptor gene KIT into the immortalized erythroblast cell line HUDEP-2 induced SCF-independent expansion. Importantly, transduced HUDEP-2 cells retained the ability to terminally mature, and overall, this strategy significantly decreased culture costs by eliminating the need for SCF (37). Thus, genetic manipulation of BMI1 and KIT genes in iPSCs may provide novel strategies to enhance iPSC-derived erythroid proliferation.

Use of bioreactor and agitation systems for iPSC-derived red cell expansion

To reach sufficient quantities of in vitro-derived RBCs for clinical use, scalability and increased cell culture densities using bioreactors or agitation systems to increase nutrient and oxygen availability are needed. The use of bioreactors to scale up cultured RBCs has primarily been studied using PBMC and CB CD34+ cell sources. Culturing RBCs from PBMCs in a 1L bioreactor has achieved a 3 e⁶-fold expansion (38). A bottle turning device culture system produced 200 million RBCs from one human CD34+ CB cell, estimating that one CB unit could produce 500 units of transfusable blood (39). PBMC-derived RBCs in stirred tank reactors up to a 3-liter capacity achieved 196-fold growth with comparable cell viability and maturation compared to static conditions (40). Oxygen requirements were shown to be highest during the expansion phases and decreased at later stages of culture as cells underwent terminal differentiation (40).

A suspension agitation culture platform for differentiating iPSC-microcarrier aggregates to erythroid cells achieved a cell density of up to 1.7 e⁷ cells/ml (41). The iPSC-derived RBCs were similar to adult-derived RBCs in expression of adult and fetal hemoglobins, oxygen binding and membrane integrity. However, generating a unit of RBCs containing 1 e¹² cells in a maximum bioreactor volume of 20 L would require achieving cell densities higher than 1 e⁸ cells/ml, which would necessitate media replenishment and perfusion strategies to prevent accumulation of toxic metabolites such as lactate and ammonia (41). High-density culture of nearly 35 million iPSC-derived erythroblasts/ml in a perfusion bioreactor system provides promise for scalable production of iPSC-derived RBCs (42).

Conclusion

The past decade has brought about significant advancements towards the generation of iPSC-derived RBCs for clinical use. However, large-scale generation of universal RBCs requires overcoming three major limitations: 1) the ability to mimic adult RBCs in hemoglobin expression, enucleation, and deformability, 2) scalability to generate 2 trillion RBCs equivalent to one donor unit, and 3) high density manufacturing capacity to minimize the cost of growth factors and media. Ongoing research in the field includes investigating methods to increase in vitro-derived RBC yield via gene manipulation and optimization of culture conditions, the use of perfusion bioreactor systems to scale up production and generating functional adult-like red cells by recapitulating definitive erythropoiesis. As the goal is to translate in vitro-derived red cells to a clinical context, we require further understanding of how these cells mature and survive when transfused into animal models and ultimately into humans (Figure 1). Current cell yield and maturation is sufficient for iPSC-derived reagent RBCs for diagnostics, but the ability to produce sufficient functional RBCs for transfusion remains a significant goal for the field.

Key Points.

• A number of protocols are available for production of red blood cells from iPSCs but cell yield and functionality remain a limitation for translating these cells for transfusion.

• Genome editing of iPSCs provides strategies to increase cell yield and produce customized RBCs with rare antigen phenotypes.

• Available technologies produce sufficient cell quantities for diagnostic red cell reagents, but use of agitation and perfusion bioreactor strategies are needed to reach production scale for transfusion.