Hematopoietic Stem Cells and Regeneration

Abstract

Hematopoietic stem cell (HSC) regeneration is the remarkable process by which extremely rare, normally inactive cells of the bone marrow can replace an entire organ if called to do so by injury or harnessed by transplantation. HSC research is arguably the first quantitative single-cell science and the foundation of adult stem cell biology. Bone marrow transplant is the oldest and most refined technique of regenerative medicine. Here we review the intertwined history of the discovery of HSCs and bone marrow transplant, the molecular and cellular mechanisms of HSC self-renewal, and the use of HSCs and their derivatives for cell therapy.

DISCOVERY OF HSCs AND THE BIRTH OF BONE MARROW TRANSPLANT

As a free-flowing liquid, blood is unlike any other organ in the body. So while the invention of microscopy provided histological insights into the identification of adult stem cells of many tissues based on their defined spatial relationships and morphological characteristics (such as the basal layer of epidermis or crypts of intestinal villi), the complex array of blood-forming cells in bone marrow prohibited identification of hematopoietic stem cells (HSCs) by direct visualization. Instead, discovery of HSCs in the absence of morphological or other direct descriptors required the dawn of atomic science, cell transplantation, and clonogenic assays to functionally identify multipotent cells based on their progeny (Fig. 1).

Figure 1.

Timeline of hematopoietic stem cell (HSC) and blood marrow transplant (BMT) discoveries. (HSCT) Hematopoietic stem cell transplant, (SCID) severe combined immunodeficiency, (ST-HSC) short-term HSC, (LT-HSC) long-term HSC, (iPS) induced pluripotent stem cells, (HLA) human leukocyte antigen, (CAR) chimeric antigen receptor.

The concept of a multipotent HSC in the bone marrow giving rise to all blood lineages had been hypothesized as early as the 1800s by others including Franz Neumann and Alexander Maximow. However, Owen (1945) provided the first experimental evidence for such a claim by showing that fraternal twin cattle sharing a placenta maintained lifelong presence of both siblings’ blood types. Soon afterward, the advent of nuclear radiation and lethal bone marrow failure experienced by victims of radiation toxicity inspired the first experiments showing that intravenous transplants of normal adult bone marrow cells could protect mice from otherwise lethal radiation doses by providing an alternative and sustainable source of blood cells (Jacobson et al. 1951; Lorenz et al. 1951; Ford et al. 1956). The chimeric blood composition of the mice engrafted in these experiments suggested that HSCs could be quantified by measuring their progeny. McCulloch and Till (1960) and Till and McCulloch (1961) performed the first HSC-limiting dilution assays in which defined numbers of adult mouse bone marrow donor cells were injected and linearly correlated to the number of cells required to protect recipients from radiation and the number of large spleen nodules that formed in recipients 2 weeks later. These spleen nodules represented the first colony-forming units (CFU-S) and were demonstrated to be true genetic clonal derivatives of individual HSCs (Becker et al. 1963); to contain mixtures of all blood cell types including erythroid, megakaryocytic, granulocyte/macrophage (Wu et al. 1967), and lymphoid lineages (Wu et al. 1968), as well as daughter cells capable of recapitulation of the colony in secondary transplant recipients (Siminovitch et al. 1963); and to have highly stochastically variable lineage and maturity content between HSC clones (Till et al. 1964). Together, these experiments proved that HSCs were capable of self-renewal and differentiation into all blood cell lineages and established HSC research as the first quantitative single-cell science.

Basic HSC research using bone marrow transplants in mice inspired clinical progress in human bone marrow transplants (or hematopoietic stem cell transplant [HSCT]) shortly after. The first unsuccessful bone marrow transplant attempted was likely a patient with aplastic anemia who received a small amount of intravenous marrow from a sibling (Osgood et al. 1939). But the technique was not formally developed until the advances in mice of the 1950s led to the first successful engraftment of bone marrow transplants from identical twin donors for patients with refractory leukemia (Thomas et al. 1957, 1959). These first bone marrow transplants followed the typical operating characteristics that would be expected from genetically identical donor grafts; they gave prompt cell line recovery after radiation (no graft failure), and did not cause the secondary wasting syndrome that had been first described previously in mice receiving bone marrow grafts from unrelated donors—graft-versus-host disease (GVHD) (Trentin 1956)—but also gave little graft-versus-leukemia (GVL) effect in that recipients relapsed a few months posttransplant. In contrast, the first transplants between genetically nonidentical but immunologically compatible siblings (allogeneic HSCT) could give durable leukemia treatment responses for years (Thomas et al. 1975, 1977, 1979) but occurred much later as they awaited the discovery and methods to type the human leukocyte antigen (HLA) system of genes on chromosome 6 governing immunologic rejection between donor and graft (van Rood et al. 1968), extensive experiments with canine sibling transplants (Storb and Thomas 1972), and the development of immunosuppression (Santos and Owens 1969; Storb et al. 1970; Kolb et al. 1973). The first successful human bone marrow transplants were major breakthroughs in hematology and immunology, and gave hope to patients with leukemias and other blood disorders.

The first bone marrow transplants used heterogenous mixtures of cells from bone marrow that included and proved the existence of HSCs in animals and humans. However, research to isolate and identify the specific HSC cell population continued. Whereas initially the cells capable of producing spleen colonies were taken to be identical to HSCs, early experiments suggested that there was greater heterogeneity in CFU-S than previously thought. It was first observed that spleen colonies appeared and disappeared rapidly after transplant and that later day 14 colonies were not formed from the first day 7 colonies (Hodgson and Bradley 1979; Magli et al. 1982). The different temporal origins of spleen colonies suggested heterogeneity in the underlying transplanted cells, and opened the question of whether cells capable of long-term hematopoiesis in the bone marrow were distinct from CFU-S and cells giving immediate radioprotection. It was later confirmed by cell centrifugation and separation experiments that HSCs with long-term repopulating ability (long-term repopulating cells [LTRCs]/LT-HSCs) could be physically separated from less primitive progenitor cells, such as CFU-S and lineage-restricted progenitors giving rise to granulocytes and macrophages (CFU-GM) (Jones et al. 1990). Physically larger CFU-S and CFU-GM were responsible for the early engraftment and hematopoiesis required to save animals from the initial aplasia of radiation injury, while long-term engraftment was achieved by LTRCs that were smaller and morphologically lymphocyte-like. These LTRCs among total adult bone marrow cells were very rare (1/10,000), as quantified in competitive repopulating unit (CRU) assays developed contemporaneously (Szilvassy et al. 1990). The field was further revolutionized by the first experiments to sort HSCs using flow cytometry and fluorescence-activated cell sorting (FACS) (Spangrude et al. 1988). The use of centrifugal cell separation and immunophenotyping marked a new era of hematopoietic research in which HSCs and other blood cell lineages were prospectively sorted using defined physical markers. LTRCs could be further defined and purified through cell surface immunophenotypes such as Lin⁻CKit⁺Sca1⁺CD135⁻CD34^Lo/− CD150⁺ in mouse or Lin⁻CD34⁺CD38⁻CD90⁺ CD45RA⁻ in human, or by Rhodamine-123 and Hoechst33342 dye efflux side population experiments as Rho⁻SP⁺ (Baum et al. 1992; Osawa et al. 1996; Kiel et al. 2005; Dykstra et al. 2007; and see Weissman and Shizuru 2008 for a history of the discovery of CD antigens to distinguish HSCs from other progenitors and differentiated blood cells).

The discovery of cell surface markers for HSCs finally gave a molecular description to the rare, morphologically indistinct cells of the bone marrow vitally responsible for long-term regeneration of blood and powerfully used in bone marrow transplants. It also gave hematologists the ability to purify HSCs and thereby study their unique molecular properties.

HSC SELF-RENEWAL: DORMANCY AND PROLIFERATION

Following the development of techniques to prospectively isolate and purify LT-HSCs, extensive molecular and cellular biology has occurred to characterize the mechanisms of HSC self-renewal. The long-term repopulating capacity of HSCs is a complex phenotype that includes both cell proliferation in response to extrinsic signaling and stress and dominant long-term quiescence to prevent exhaustion from continued cell cycling (Fig. 2). Both HSC dormancy and proliferation are exquisitely regulated by mechanisms both internal in HSCs themselves and external in cells of the bone marrow niche, and are increasingly elegantly refined by techniques with single-cell resolution. Here we review the factors affecting HSC self-renewal in both steady-state conditions as well as following injury and transplantation.

Figure 2.

Factors promoting hematopoietic stem cell (HSC) dormancy and proliferation. (SCF) Stem cell factor, (TPO) Thrombopoietin, (PVA) polyvinyl alcohol.

HSC Dormancy

Both initial characterization and refined modern methods have confirmed that adult HSCs of the bone marrow are largely dormant in vivo. The short-term (ST)-HSCs in initial transplants were found to be largely in a G0, nondividing cell state and therefore resistant to chemotherapy compared to malignant cells (Becker et al. 1965; Bruce et al. 1966). Multiple studies using dye incorporation, limiting dilution assays, and transgenic histone H2B-GFP pulse label experiments have shown that LTRCs as well rarely divide in vivo, as infrequently as every 145 days in mice (Bradford et al. 1997; Cheshier et al. 1999; Sudo et al. 2000; Wilson et al. 2008; Busch et al. 2015). Indeed, recent evidence suggests that murine HSCs divide only four times in adulthood with retention of LTRC activity, which is lost on the fifth cell division (Bernitz et al. 2016). Dormancy in vivo is in contrast to HSC self-renewal in the context of transplant, where they are essential for long-term repopulating activity and can expand up to 1000-fold (Ema et al. 2005). The dormancy of adult HSCs is also in contrast to fetal life, where HSCs in the fetal liver expand nearly 100-fold (Morrison et al. 1995; Ema and Nakauchi 2000), and after which the number of LTRCs remains fairly constant throughout life (Yamamoto et al. 2018; Ganuza et al. 2019), although with declining function in aging (de Haan and Lazare 2018). The dormancy of HSCs suggests that they are not the predominant cell type responsible for the massive hematopoiesis (billions of blood cells per day) required at steady state in vivo. Instead, LTRCs at the top of the hematopoietic hierarchy give rise to the ST-HSC (CFU-S) discovered earlier, which in turn differentiate into multipotent progenitors (MPPs) with no self-renewal abilities (Yang et al. 2005) that are predominantly responsible for hematopoiesis under steady-state conditions in vivo (Sun et al. 2014; McRae et al. 2019).

The limited number of HSC cell divisions is accomplished through an exquisitely regulated ground state of HSC dormancy that preserves prolonged self-renewal capacity for stress conditions (including transplantation into ablated hosts). Numerous HSC-intrinsic factors as well as interactions with cells of the bone marrow niche regulate HSC quiescence to uncontrolled division and exhaustion of LTRC capacity. The hypoxia of the bone marrow niche is a key environmental regulator of HSC quiescence to alter cellular metabolism to limit proliferation and exhaustion. Despite the dense vascularity of bone marrow, intermediate filament Nestin⁺ periarteriolar mesenchymal cells of the quiescent HSC niche are characterized by low regional perfusion (Parmar et al. 2007; Kunisaki et al. 2013) and have been shown in elegant microscale optical imaging experiments to maintain absolute local pO2 levels as low as ∼23 mm Hg for HSCs (Spencer et al. 2014) compared to 80–100 mm Hg in arterial blood. These hypoxic conditions make HSCs dependent on glycolysis for energy and give HSCs lower basal and maximal respiration rates than proliferative MPPs irrespective of mitochondrial content (Simsek et al. 2010; Romero-Moya et al. 2013; Takubo et al. 2013; Ito and Suda 2014; de Almeida et al. 2017). Use of glycolysis due to hypoxia limits energy production for proliferation as well as intracellular reactive oxygen species (ROS) production (Jang and Sharkis 2007; Mantel et al. 2012) and DNA damage (Walter et al. 2015)—but others have also suggested that dormancy is characterized by error-prone DNA repair mechanisms such as nonhomologous end joining owing to lack of HSC cell-cycle entry (Mohrin et al. 2010; Beerman et al. 2014). The need for DNA damage repair and anaerobic respiration due to the hypoxic environment of HSCs is controlled through quantitative levels of transcription factors such as HIF-1, Meis1, and Foxo3a (Miyamoto et al. 2007; Tothova et al. 2007; Simsek et al. 2010; Takubo et al. 2010). HSC dormancy also requires the antioxidant vitamins A and C (Agathocleous et al. 2017; Cabezas-Wallscheid et al. 2017). In addition to low oxygen utilization, other conservative metabolic requirements for HSC dormancy include low rates of protein synthesis and dependence on recycling organelles through autophagy and mitophagy (Signer et al. 2014; Cao et al. 2015; Ito et al. 2016; Ho et al. 2017).

Apart from metabolic requirements, extrinsic signaling factors and internal epigenetic modulators and transcriptional programs act to maintain HSC dormancy. Extracellular cytokines have prominent roles promoting HSC dormancy by limiting excessive cell division. Signaling ligand TGF-β is produced by numerous cell types of the bone marrow niche including vascular cells, megakaryocytes, and nervous system Schwann cells (Yamazaki et al. 2011; Zhao et al. 2014); limits in vitro and in vivo HSC proliferation through its receptor TGFBR2 and transcriptional effector SMAD4 (Keller et al. 1990; Yamazaki et al. 2009, 2011); and prevents exhaustion after stress-induced HSC proliferation (Brenet et al. 2013). The ligand CXCL12 (SDF-1) through its receptor CXCR4 promotes HSC quiescence and is produced by numerous niche cells including Leptin receptor, Lepr⁺ perivascular stromal cells, CXCL12-abundant reticular cells, endothelial cells, megakaryocytes, and perivascular mesenchymal stem cells (Sugiyama et al. 2006; Nie et al. 2008; Greenbaum et al. 2013; Bruns et al. 2014). CXCL12 also inhibits HSC egressing from the bone marrow and is under the control of circadian nervous system cycles and bone marrow macrophages (Méndez-Ferrer et al. 2008; Chow et al. 2011). The ligand Netrin-1 is also produced by periarteriolar endothelial and stromal cells and promotes HSC quiescence (Renders et al. 2021). The ligand Angiopoietin-1 promotes HSC quiescence and is produced by osteoblasts in the niche (Arai et al. 2004). In addition to cytokines and growth factors above, single-cell RNA sequencing has provided new insights into novel niche factors regulating HSCs such as Angiogenin, a secreted RNase that also promotes HSC quiescence (Goncalves et al. 2016; Silberstein et al. 2016). Downstream of extracellular signaling, numerous transcription factors and epigenetic modulators also promote HSC dormancy including p21 and p53 (Cheng et al. 2000; van Os et al. 2007; Liu et al. 2009) under the regulation of E47 (Yang et al. 2011), Scl (Lacombe et al. 2010), Pbx1 (Ficara et al. 2008), Egr-1 (Min et al. 2008), Gfi-1 (Hock et al. 2004; Lee et al. 2018) under regulation of Mysm1 (Wang et al. 2013), PU.1 (Staber et al. 2013), C/EBPa (Ye et al. 2013; Hasemann et al. 2014), Znhit1 (Sun et al. 2020), Tet2 (Moran-Crusio et al. 2011), Dnmt3a (Jeong et al. 2018), Ythdf2 (Li et al. 2018b), and Tcf15 (Rodriguez-Fraticelli et al. 2020).

HSC Proliferation

Whereas LT-HSCs spend most of adult life in a dormant state, signaling from the niche and stressors from the body can awaken HSCs and stimulate cell proliferation. In addition to contributing to our understanding of hematopoietic homeostasis, the study of HSC proliferation is vital to HSC regenerative medicine in improving bone marrow transplant and ex vivo culture of HSCs for translational applications. HSCs are awakened from their dormant ground state by both physiologic stressors including hemorrhage and infection and iatrogenic stressors such as DNA damage through radiation injury or chemotherapy, and bone marrow transplant mobilizing and conditioning regimens. Infection and inflammation in response to various pathogens or pathogen-associated molecular patterns such as lipopolysaccharide that activate proinflammatory Toll-like receptor, interferon, IL-6, TNF-α, and NF-κB protein signaling networks have been shown to increase HSC cell-cycle entry and proliferation (Essers et al. 2009; Baldridge et al. 2010; Chen et al. 2010; Esplin et al. 2011; King and Goodell 2011; Megías et al. 2012; Zhao et al. 2013). However, such studies have also shown that proliferative signaling through chronic immune stimulation also injures HSC LTRC capabilities and skews differentiation potential toward myeloid lineages. The repopulating capacity of HSCs injured by interferon signaling is opposed by Irf2, a negative transcriptional regulator of interferon signaling (Sato et al. 2009). Proliferation is also stimulated by the common pre-stem-cell transplantation mobilization and conditioning methods including the cytokine G-CSF and chemotherapy (Wilson et al. 2008; Busch et al. 2015; Schoedel et al. 2016). In addition to chemotherapy, DNA damage occurring from accumulated ROS damage and aging also stimulates HSC proliferation (Morrison et al. 1996; Akunuru and Geiger 2016; Li et al. 2016). Blood loss also increases HSC proliferation (Cheshier et al. 2007).

As in dormancy, HSC proliferation is stimulated by extracellular cytokines and growth factors. Two classic HSC proliferative cytokines include c-kit ligand (stem cell factor [SCF], steel factor) and Thrombopoietin (TPO). c-kit ligand is secreted by perivascular and endothelial cells of the niche and increases survival and proliferation of HSCs and downstream progenitors both in vivo and in vitro in combination with other cytokines such as IL-3, -6, -11, -12, -27, and/or TPO (Geissler et al. 1981; Carow et al. 1991; Metcalf and Nicola 1991; Ogawa et al. 1991; Li and Johnson 1994; Sharma et al. 2007; Ding et al. 2012; Dahlin et al. 2018). The fact that HSCs rely on c-kit and possess the c-kit receptor can be used by coating HSCs with anti-c-kit antibodies for destruction by antibody-dependent cell-mediated cytotoxicity and may one day allow for bone marrow transplants with novel biological-based preconditioning regimens (rather than radiation or chemotherapy that has been traditionally used, as discussed below) (Chhabra et al. 2016). TPO produced by megakaryocytes and osteoblasts is another critical cytokine responsible for expansion of HSCs in vivo and in vitro in combination with SCF and IL-3 (Sitnicka et al. 1996; Kimura et al. 1998; Ema et al. 2000; Fox et al. 2002), but also promotes quiescence and ensures lack of exhaustion of HSCs (Qian et al. 2007; Yoshihara et al. 2007). TPO is under the negative regulation of the Lnk adaptor protein in HSCs, with deletion of Lnk leading to large increases in HSC self-renewal (Ema et al. 2005; Seita et al. 2007). A recent study highlighted the importance of TPO and c-kit ligand in showing that synergistic dosing of TPO and c-kit with no other cytokines provided long-term ex vivo expansion of murine HSCs in vitro to a scale suitable for engraftment without preconditioning (Wilkinson et al. 2019). However, in this study, a fibronectin-coated substrate and replacement of fetal bovine serum with polyvinyl alcohol (PVA) to serve as a bioactive cytokine and lipid carrier were also used, which highlights the importance of insoluble and noncytokine factors in optimizing ex vivo HSC culture.

Apart from c-kit and TPO, Wnt and Notch extracellular signaling factors have also been shown to stimulate HSC expansion. Following injury, hematopoietic and stromal cells of the bone marrow up-regulate Wnt ligands, which trigger β-catenin-mediated signaling to increase proliferation and repopulation, and decrease intracellular ROS, thereby promoting HSC regeneration (Reya et al. 2003; Duncan et al. 2005; Nemeth et al. 2007; Congdon et al. 2008; Lento et al. 2014). Notch ligands such as DLL4 and DLL1 primarily expressed through vascular endothelial cells of the niche and signaling through the Notch1 intracellular domain and transcription factor Hes1 have been shown to promote HSC proliferation in vivo and in vitro and prevent myeloid skewing or predifferentiation of HSCs (Varnum-Finney et al. 2000; Stier et al. 2002; Kunisato et al. 2003; Tikhonova et al. 2019). The ligand Adiponectin from adipocytes in bone marrow has also been shown to increase HSC proliferation in vitro and in vivo through the p38 MAPK pathway without sacrificing long-term repopulating activity (DiMascio et al. 2007). Flt3 ligand is another cytokine that stimulates HSC proliferation, largely in potentiation of other growth factors (Tsapogas et al. 2017). Prostaglandin E2 is a nonpeptide eicosanoid factor that has also been shown to increase HSC proliferation and self-renewal (Hoggatt et al. 2009).

Following stimulation by stress or extracellular factors as described above, HSCs shift metabolism toward increased oxidative phosphorylation and nucleic acid metabolism necessary for cell proliferation (Ito et al. 2012; Karigane et al. 2016; Karigane and Takubo 2017). Calcium and vitamin D are also extrinsic metabolic requirements for HSC self-renewal and proliferation (Cortes et al. 2016; Umemoto et al. 2018). In addition to niche cell signaling and metabolic environmental cues, a strong role for intrinsic factors in HSC proliferation is also evident from transplantation experiments showing greater than 500-fold clonal variations in proliferation between individual LT-HSCs (Sieburg et al. 2011; Benz et al. 2012). Furthermore, the preexisting epigenetic state of HSCs dictates the diverse individual clonal responses to stress signaling (Yu et al. 2017). Transcription factors and epigenetic modulators increasing HSC proliferation include Evi-1 (Goyama et al. 2008), Mef (Elf4) (Lacorazza et al. 2006), Bmi1 (Park et al. 2003; Iwama et al. 2004), MLLT3 (Calvanese et al. 2019), and HOXB4 (Antonchuk et al. 2002). RNA-binding factors also contribute to HSC maintenance and proliferation. For example, Musashi-2 is an RNA-binding protein that alters translation of target gene RNA and promotes HSC maintenance and proliferation by repression at the protein level of the Notch-inhibiting membrane protein Numb (Ito et al. 2010; Rentas et al. 2016). Dppa5 is an RNA-binding protein required for endoplasmic reticulum (ER) stress regulation whose overexpression leads to HSC expansion (Miharada et al. 2014). U2af1 is a splicing factor required for promoting HSC proliferative gene expression (Dutta et al. 2021).

The discovery of factors promoting the various components of HSC self-renewal—HSC dormancy, maintenance of LTRC activity, and HSC proliferation—has enriched our understanding of HSC contributions to hematopoiesis and allowed for the development of ex vivo HSC culture conditions. These methods to grow HSCs are also a requirement for novel translational research areas seeking to create synthetic HSCs from various sources and expand HSCs and their derivatives for therapeutic applications.

CELL THERAPY BY HSCs AND DERIVATIVES

Discoveries on HSC self-renewal and their utility in regenerative medicine for HSCT have developed in parallel since the beginning of HSC research. For over 60 years, HSCT has been deployed for both solid and hematologic malignancies and a wide range of nonmalignant diseases such as bone marrow failure and congenital immunodeficiency syndromes. Whereas in principle HSCT can cure any hematologic disorder, in practice HSCT is limited by donor availability, patient eligibility, treatment-related morbidity and mortality (including infection, organ failure, and GVHD), cost (Khera et al. 2014), and relapse of the primary disease. The therapeutic GVL effect of allogeneic HSCT has also been further refined in chimeric antigen receptor (CAR) cell therapy using HSC derivatives (e.g., CAR T cells). Here we review the sources of donor HSCs for HSCT, including the translational potential of iPS cells as new sources for HSCs, the common therapeutic indications for HSCT (Fig. 3), and the recent expansion of anticancer cell therapies by HSC derivatives.

Figure 3.

Current blood marrow transplant (BMT) indications separated by source of donor hematopoietic stem cells (HSCs) (autologous vs. allogeneic). (MM) Multiple myeloma, (NHL) non-Hodgkin lymphoma, (AML) acute myeloid leukemia, (MDS) myelodysplastic syndromes, (ALL) acute lymphocytic leukemia, (HL) Hodgkin's lymphoma, (CML) chronic myeloid lymphoma. Benign disorders include bone marrow failure syndromes such as aplastic anemia, congenital diseases, and other nonmalignant conditions. (Figure adapted from Phelan et al. 2020. Full complement of slides related to this figure are available at www.cibmtr.org.)

Sources for transplanted HSCs either come from the patient themselves or a healthy donor. Autologous HSCT involves withdrawal and preservation of a patient's own HSCs for reintroduction after administration of an otherwise lethal dose of radiation or chemotherapy (conditioning regimen). Autologous HSCT is the most common type of bone marrow transplant at over 14,000 performed yearly in the United States with multiple myeloma and non-Hodgkin lymphoma being the dominant indications (Fig. 3; D'Souza et al. 2020). Allogeneic HSCT involves conditioning followed by the delivery of donor HSCs from another individual—in order of frequency, an HLA-matched unrelated donor, HLA-matched sibling, partially HLA-matched (haploidentical) family member, followed by other sources at lower frequency—and is most commonly used in acute leukemias (acute myeloid leukemia [AML] and acute lymphocytic leukemia [ALL]) and myelodysplastic syndromes (MDS) (Fig. 3; D'Souza et al. 2020). Typically, four or five HLA genes (8–10 alleles) on chromosome 6-A, -B, -C, -DRB1, and -DQB1 are assessed to determine matching for allogeneic HSCT. Thus, a fully HLA-matched sibling or an ideal matched unrelated donor is referred to as an 8/8 or 10/10 match and a haploidentical family member is a 4/8 or 5/10 match (Howard et al. 2015). The greater degree of HLA mismatch increases the risk of both host-mediated graft rejection (Ciurea et al. 2009) and GVHD, but not necessarily greater GVL (Arora et al. 2009; Ringdén et al. 2009; Pidala et al. 2014; Sweeney and Vyas 2019). GVHD can also be minimized by high-dose cyclophosphamide in the immediate posttransplant induction phase of HSCT (Luznik et al. 2012) and use of a bone marrow rather than peripheral blood source for HSCs (Anasetti et al. 2012), making this the preferred source for patients with nonmalignant conditions (for whom there is no theoretical advantage to greater GVL from a peripheral blood source) withstanding donor availability for the more invasive procedure. Peripheral blood, with harvesting of HSCs by apheresis after mobilization by G-CSF, chemotherapy, and, more recently, CXCR4 inhibition (DiPersio et al. 2009), is nonetheless the most common source accounting for 80% of transplants in adults (D'Souza et al. 2020). Ethnic representation in the National Marrow Donor Program (NMDP) limits donor sources and therefore certain disease indications for allogeneic transplants: matched unrelated donors can be found for 79% of Caucasian patients but just 33% of African-Americans (Switzer et al. 2013). This limits allogeneic HSCT for these patients in malignant and nonmalignant diseases and is especially relevant to sickle cell anemia, for which allogeneic HSCT can be an effective treatment option (Hsieh et al. 2009; Leonard et al. 2020).

Given issues of donor availability and GVHD from traditional sources of HSCs, there has been intense preclinical scientific interest in creating autologous HSC sources from patient-derived human-induced pluripotent stem cells, iPS (Takahashi and Yamanaka 2006; Takahashi et al. 2007; Yu et al. 2007). In this process, a patient's somatic cells are reprogrammed into iPS cells (classically by virally inserted transcription factors, but more recently through a variety of potentially safer vector-free or small molecule techniques [Yu et al. 2009; Hu et al. 2011; Kim et al. 2020]) and then would be differentiated into HSCs for delivery back to a patient (Fig. 4). The patient's iPS cells could also be genetically manipulated to correct a congenital condition. Hanna et al. (2007) showed an impressive proof-of-principle for this approach using mouse iPS cells in which mouse skin fibroblasts were taken from a sickle cell anemia mouse model, reprogrammed to iPS cells, genetically corrected to replace hemoglobin genes with non–sickle cell alleles, then differentiated to hematopoietic progenitors in vitro and finally transplanted back into the mice. The mice then recovered from their sickle cell phenotype. Despite this impressive demonstration, the derivation of bona fide LT-HSCs from human iPS cells with long-term production of all blood lineages has since been a major technological hurdle. Reports of deriving HSCs from embryonic cell populations or induced pluripotent stem cells have used specialized feeder cells, cell sorting, transcription factor expression, and cytokines to show short-term myeloid-biased engraftment but low lymphocyte production and lack of long-term engraftment (Kyba et al. 2002; Doulatov et al. 2013; Guo et al. 2020; Shan et al. 2020). These reports of pluripotent cell-derived HSC immaturity are in line with the relative immaturity of pluripotent stem cell derivatives from other tissues, such as pancreatic β cells, neurons, and cardiomyocytes (Pagliuca et al. 2014; Weick 2016; Biermann et al. 2019). A major question therefore is whether HSC immaturity is a reflection of the synthetic limitations of in vitro culture, or intrinsic to iPS cells as sources. But arguing against the latter, it has been shown by different groups that iPS cells engrafted into immunocompromised animals as teratomas in vivo (xenografts classically created to assess the differentiation potential of pluripotent iPS cells in that they contain tissues from ectoderm, mesoderm, and endodermal germ layers) contain serially transplantable HSCs with multilineage differentiation and long-term engraftment (Amabile et al. 2013; Suzuki et al. 2013; Tsukada et al. 2017). Whereas use of teratomas would not be suitable for translational development, these studies suggest that more faithful recapitulation of the in vivo development of HSCs may aid maturation. Most recently, Sugimura et al. (2017) differentiated iPS cells first to hemogenic endothelial cells, the origin site of HSCs during embryonic development, and then expressed a larger number of transcription factors (ERG, HOXA5, HOXA9, HOXA10, LCOR, RUNX1, SPI1) to specify HSCs. These iPS-HSCs showed impressive, up to 16-week, engraftment in primary and secondary recipients often with multilineage differentiation potential. Thus, while technical limitations remain to generating LT-HSCs from human iPS cells, iPS-HSCs may provide novel sources of genetically correctable HSCs to widen the range of indications for autologous HSCT.

Figure 4.

Process of the creation of patient-specific iPS stem cells, genetic correction, and differentiation to hematopoietic stem cells (HSCs) for the treatment of hematological genetic diseases (such as sickle cell anemia). (BMT) Bone marrow transplant, (iPS) induced pluripotent stem cells.

Therapeutic indications for HSCT have changed over time. For example, autologous HSCT was once widely used for metastatic breast cancer, but eventually randomized controlled trials argued against a benefit to the procedure (Vogl and Stadtmauer 2006). Similarly, chronic myelogenous leukemia (CML) was often treated with allogeneic HSCT before the development of the tyrosine kinase inhibitor imatinib, and allogeneic HSCT for CML is now mainly used for disease refractory to targeted therapy or in leukemic transformation (Benyamini and Rowe 2013). Today, therefore, autologous HSCT is most commonly performed for multiple myeloma and non-Hodgkin lymphoma, and allogeneic HSCT for acute leukemias and MDS.

Multiple myeloma is the most lethal plasma cell dyscrasia and involves clonal plasma cells causing renal failure, bone destruction, and anemia. Autologous HSCT for consolidation after primary induction therapy has been standard of care for patients healthy enough to receive it after the first randomized trials showed deepened remissions (Attal et al. 1996). Whereas primary therapy for myeloma has advanced substantially since then—specifically, the addition of proteasome inhibitors (such as bortezomib and carfilzomib) and immunomodulatory imide drugs (lenalidomide, pomalidomide) to create triple therapy regimens in combination with steroids—progression-free survival benefits on the order of years continue to favor the addition of autologous HSCT for consolidation therapy (Palumbo et al. 2014; Cavo et al. 2020). Non-Hodgkin lymphoma, the most common histologic subtype being diffuse large B-cell lymphoma (DLBCL), is the second leading indication for autologous HSCT. DLBCL is an aggressive malignancy of clonal B cells in the lymph nodes and extranodal sites of the body that is usually responsive to combination chemotherapy regimens with rituximab, anti-CD20 (Coiffier et al. 2010). However, in the setting of disease that is relapsed or refractory to initial treatment, salvage chemotherapy followed by autologous HSCT in eligible patients provides a large benefit on response rates and overall survival compared to salvage therapy alone (Philip et al. 1995; Oliansky et al. 2011).

In both myeloma and DLBCL, the goal of autologous HSCT is to use patients’ own HSCs as rescue to deliver a myeloablative dose of chemotherapy. However, in the postremission management of acute leukemias (AML and ALL), allogeneic HSCT is added for its GVL effect. AML and ALL are aggressive proliferations of malignant clones of myeloid or lymphoid progenitors and blasts in the blood and bone marrow. AML mostly affects adults, while ALL is the most common cancer of childhood. Despite these different patient populations, allogeneic HSCT confers relapse prevention and survival benefits compared to consolidative chemotherapy alone in eligible patients with high-risk disease based on molecular and cellular features (Yanada et al. 2005; Schrauder et al. 2006; Cornelissen et al. 2007; Seibel et al. 2008; Koreth et al. 2009; Schrappe et al. 2012). The clinical existence of GVL as the mechanism of benefit from allogeneic HSCT was suggested by minimal efficacy of allogeneic HSCT grafts that were either T-cell depleted or from identical twin donors, as observed in the first identical twin bone marrow transplants for leukemias in the 1950s described above and systematically studied by Horowitz et al. (1990). Furthermore, it was observed that donor lymphocyte infusion alone could promote durable remissions in CML and ALL (Kolb et al. 1990; Poon et al. 2013). These studies supported the existence of GVL in allogeneic HSCT and thereby allowed for the development of reduced intensity conditioning chemotherapy regimens that could extend allogeneic HSCT eligibility for older or less medically fit patients (Slavin et al. 1998; Dreger et al. 2000; Nagler et al. 2000). They also suggested that donor lymphocytes are the cellular effectors of GVL.

Investigating the cellular mechanisms of GVL has continued to yield clinical and preclinical cancer cell therapy advances. Cytotoxic CD8⁺ T lymphocytes were the first cells identified to be capable of causing GVL, but at the expense of GVHD, which occurs through graft recognition of minor and unmatched major histocompatibility antigens on host leukemic cells and normal tissues, respectively (Tsoi et al. 1980; Falkenburg et al. 1991; Niederwieser et al. 1993; Warren et al. 1998; Bonnet et al. 1999). These results spurred investigations to synthetically manipulate graft T cells to maximize GVL without triggering greater GVHD. Principal among these is genetic manipulation to express CARs. CARs are single-chain cell-surface receptors that combine fragments of multiple different immune cell surface receptors to create a single construct that can be transfected into patient cells that can then be transplanted back to mediate GVL in the patient (Fig. 5). CARs typically contain an extracellular cancer cell–recognition domain (e.g., a single-chain variable fragment antibody) and an intracellular signaling domain usually containing both costimulatory (CD28, 4-1BB) and stimulatory (CD3ζ) fragments (Fig. 5, inset). Patient-derived CAR T cells have so far been used for refractory hematologic malignancies including ALL (Maude et al. 2018; Park et al. 2018), DLBCL (Neelapu et al. 2017; Schuster et al. 2019), and multiple myeloma (Raje et al. 2019). Whereas CAR T cells have induced impressive response rates in trials, their use can cause potentially fatal cytokine-release syndrome, and they are further limited by the expense of the need to collect and transfect a recipient's cells each time to create an autologous product to avoid GVHD. Another attempt to improve GVL without increasing GVHD relies on donor natural killer cells (NKs), which are large granular lymphocytes that express MHC class I–specific killer-cell immunoglobulin-like receptors (KIRs) such that NK cells will kill targets lacking their allotypic MHC-I. Ruggeri et al. (1999, 2007) showed that NK alloreactive grafts (grafts with KIR mismatch between donor and recipient) were impressively correlated with lower relapse rates and survival in allogeneic HSCT for AML—without increased GVHD reactions. Because they do not mediate GVHD in an MHC class II–dependent fashion, the possibility exists of using standardized, expanded NK cells as an “off-the-shelf” cell therapy for cancer. Therefore, clinical trials (Hodgins et al. 2019) and improvements to NK cell cytotoxicity using iPS technology and modification with NK-specific CARs (Li et al. 2018a) are currently underway.

Figure 5.

Creation of patient-specific chimeric antigen receptor (CAR) T-cell therapy, in which patient T cells after delivery of the CAR cassette—a synthetic transmembrane (TM) receptor recombining multiple functional domains (inset, right)—can target cells bearing tumor-specific antigens. (scFv) Single-chain variable fragment.

CONCLUDING REMARKS

HSC regeneration is the complex process by which rare cells of the bone marrow that normally maintain a long-term dormant state through conservative metabolism and external signaling from cells of the bone marrow niche are awakened by proliferative niche factors released in the context of bone marrow injury by radiation, drugs, infection, or trauma. These same factors can be harnessed to mobilize and collect HSCs for transplant. The story of HSC research and bone marrow transplant has progressed from its origin in the exigencies of radiation injury and preliminary transplants of unpurified bone marrow cell populations in animals and humans to advanced single-cell science on mechanisms of HSC regeneration and the use of bone marrow transplant for tens of thousands of patients today. HSC research and its intertwined history and practice in bone marrow transplant shows the synergy of basic and applied science. In the future, single-cell methods will continue to deepen our understanding of the mechanisms of HSC regeneration. Techniques such as single-cell RNA sequencing are providing enhanced understanding of transcriptional heterogeneity within the HSC cellular compartment as well as cells of the niche. These tools can provide useful molecular information to further subclassify HSCs through their occupancy of different transcriptional states, to understand the heterogeneity and different functions of cells of the bone marrow niche, and to uncover dynamic signaling factors and mechanisms expressed by subsets of cells invisible to methods conducted on bulk populations. Finally, this improved understanding of the heterogeneity and signaling of HSCs and their niche will enable further developments in engineering HSCs and their derivatives for treatment of a wide and growing range of therapeutic indications.