Chapter 18: Hematopoietic Stem Cells, Progenitors, and Cytokines

Blood cell production is an enormous and complex process. Based on the adult blood volume (5 L), the number of each of the blood cell types per microliter of blood, and their circulatory half-life, it can be calculated that each day an adult human produces 2 × 10¹¹ erythrocytes, 1 × 10¹¹ leukocytes, and 1 × 10¹¹ platelets. These numbers can all increase approximately 10-fold in states of blood cell destruction or enhanced need. Over the past four decades experimental hematologists have developed a model of blood cell production in which a hierarchical developmental progression of primitive, multipotential hematopoietic stem cells (HSCs) gradually lose one or more developmental potentials and ultimately become committed to a single cell lineage, which matures into the corresponding blood cell type.¹ Perhaps one of the most compelling arguments supporting this model of hematopoiesis is derived from extensive purification schemes using cell surface markers that yield cells at each predicted developmental stage (Fig. 18–1).² Although hematopoietic development is considered by most investigators as an irreversible stepwise and progressive loss of developmental potentials, studies now suggest that cells undergoing apparent differentiation steps might oscillate between different stages depending on their position in the cell cycle.³ But regardless of the precise relationships between different stages of hematopoietic development, the availability of this model and the data leading to its construction have provided important insights into the biology and clinical uses of hematopoietic stem and progenitor cells. This chapter focuses on our understanding of the molecular basis for blood cell development, beginning with the HSC and its offspring, the lineage-committed progenitor cells.

Blood cell production begins in the yolk sac,⁴ where extraembryonic mesoderm develops into angioblasts and primitive erythroid precursors at day 7 postcoitum of the mouse; cells of the outer layer of the undifferentiated mesoderm at this time flatten and become endothelial cells, and the inner cells round up to become clusters of erythroid precursors,⁵ termed blood islands(Chap. 7). Like in the embryo proper, there is much evidence to suggest that these two cells are derived from a common precursor (the hemangioblast).⁶ Once adjacent blood islands begin to coalesce on day 8, the endothelial cells form vascular channels, which by day 8.5 connect with the embryonic vasculature, allowing yolk sac blood cells to exit the blood islands, complete their maturation, and enucleate in the embryonic bloodstream.⁷ In both mouse and man there is a stage of embryonic development where both primitive erythrocytes (as characterized by ζ globin phenotype) and definitive red cells are produced in the yolk sac, although the former appears only very transiently. Although not as well characterized, yolk sac myelopoiesis and thrombopoiesis also occur, perhaps as part of the development of multipotent progenitors that appear by day 8.5 postcoitum. Cells capable of differentiating into multiple cell lineages become recognizable early during yolk sac hematopoiesis.⁸ However, such cells reproducibly engraft only in the marrow of myeloablated embryonic animals and not in adults,⁹ making it unlikely that such cells are true HSCs, although this topic remains controversial. By day 11 postcoitum repopulating HSCs are clearly present in the yolk sac, but the relationship of these cells and the HSCs that are clearly demonstrable a day earlier in a region of the embryonic paraaortic splanchnopleure known as the aorta-gonad-mesonephros (AGM) is not certain. By day 12.5, postcoitum hematopoiesis in the murine yolk sac is eliminated.

Although it was long believed that the developmental origin of the adult mammalian hematopoietic system was the yolk sac, subsequent research has shown that the first adult-type HSCs are derived from mesodermal cells within the ventral wall of the dorsal aorta of the AGM.^10,11,12 The AGM remains a source of hematopoiesis between days 9.5 and 11.5 postcoitum in the mouse, and days 30 and 37 in the human.^13,14 Of interest, the development of hematopoietic cells in this region (as well as in the yolk sac) occurs in a “reverse” direction, that is, single lineage-committed progenitors appear prior to multilineage progenitors, which appear prior to stem cells. This region also has cells that express a number of molecules in common with endothelial cells, including CD34, the transcription factors SCL and GATA-2, and the receptors c-KIT and FLK-1.¹⁵ Moreover, cell culture experiments have established that such cells display combined endothelial and hematopoietic potential, establishing them as “hemangioblasts,” the postulated combined endothelial cell–hematopoietic precursor.¹⁶

Approximately 2 days following the appearance of HSCs in the AGM region, hematopoiesis begins in the murine fetal liver. Careful dissection experiments of the 1970s indicate that fetal liver hematopoiesis is dependent on an exogenous source of hematopoietic cells,¹⁷ which populate the fetal liver in two waves, consisting of erythroid and multilineage progenitors around day 9 of murine gestation and committed progenitors and true HSCs at day 11.¹⁸ Although there is no direct proof, the temporal appearance of these cell types in the AGM approximately 1 to 2 days prior to their appearance in the fetal liver strongly suggests that the former is the source for populating the latter. In humans the fetal liver becomes the major source of blood cells around 5 weeks of gestation, and the marrow begins to populate with hematopoietic cells at 8 weeks of gestation. Unlike the random pattern of cells seen in the yolk sac, hematopoiesis in the fetal liver is well organized; erythroid cells are usually found in clusters surrounding a central macrophage and CD15+ myelopoietic cells localize mainly around portal triad vessels, although lymphoid precursors fail to demonstrate a specific localization pattern and are randomly found amongst hepatocytes.¹⁹ Up to 50 percent of the fetal liver is composed of hematopoietic cells at days 12 to 14 of murine embryonic life, a proportion that begins to decrease as hepatocytes replace hematopoietic cells and the latter shift to the marrow, prior to birth.

The final shift in the site of hematopoiesis occurs before birth; although the marrow begins to populate with liver derived hematopoietic cells at day 16 in the mouse and at 8 weeks gestation in the human, it is mostly myeloid in nature and contributes little to the circulating blood until just before birth.²⁰ Hematopoietic stem and progenitor cells circulate in large numbers during fetal life, as clinically witnessed by the use of umbilical cord blood as a rich source of HSCs for transplantation. However, shortly after birth neonatal blood has very few primitive hematopoietic cells, as they begin to home to and lodge in the marrow. Genetic studies reveal that marrow localization of HSCs is dependent on the chemokine CXCL12 (previously known as stromal cell-derived factor [SDF]-1)²¹ as elimination of the chemokine or its receptor (CXCR4) leads to marrow hypoplasia.²² The shifts in localization of hematopoiesis during mammalian development are likely the result of changes both in the cell surface adhesion molecules on hematopoietic stem and progenitors that occur during ontogeny, and the characteristics of stromal cells of the yolk sac, AGM, fetal liver, and adult marrow that provide the microenvironmental support of HSC survival, homing and lodgment, self-renewal, proliferative expansion, and differentiation (Chap. 7).

FUNCTIONAL DEFINITION

Although the concept of a common “mother cell” of all blood elements in the adult dates to Maximov in 1909, and its potential for participation in disease as proposed by Danchakoff in 1916,²³ the basic concepts of a hierarchical organization of stem and progenitor cells leading to mature blood cell production were coalesced by Till and McCulloch using a spleen colony-forming assay, experimentally establishing the existence of multipotential hematopoietic cells.²⁴ The capacity to transplant marrow cells and reconstitute all aspects of hematopoiesis in myeloablated recipients provided an in vivo assay for the HSC, but it was not until the development of clonal in vitro assays of lineage-committed progenitors that a coherent model of blood cell production began to emerge. The pioneering work of Pluznik and Sachs²⁵ and of Bradley and Metcalf²⁶ provided methods to enumerate and characterize marrow cells committed to the hematopoietic lineage. These investigators independently developed culture conditions that allowed colonies of leukocytes to develop from single progenitors. However, as a result of the more fastidious conditions required for erythropoiesis and megakaryopoiesis in vitro, the description of methods to culture these progenitors did not occur for another decade or more.^{27,28,29,30,31} Work using density fractionation, cell sorting, and fluorescent dye exclusion methods has yielded purified populations of stem cells,^{32,33,34,35,36} common myeloid³⁷ and lymphoid³⁸ progenitors, and lineage-restricted hematopoietic progenitors,^39,40 methods that have greatly advanced our understanding of the cell and molecular biology of blood cell development. Figure 18–1 depicts a working model of this process.

STEM CELL KINETICS

Based on transplantation data indicating that there are a remarkably similar total-body number of HSCs in mice and cats, it has been estimated that all mammals, including humans, possess 2 × 10⁴ stem cells,⁴¹ and because only a small fraction of these are cycling (and therefore contributing to blood cell production) at any given time, it is also clear that daily blood cell development from the few cycling stem cells to produce the approximately 4 × 10¹¹ mature blood cells represents a massive amplification process. However, the capacity of HSCs to contribute to hematopoiesis changes with age (Chap. 9). The number of HSCs increases with age in some but not all strains of mice.^42,43 Also, HSC differentiation in aged animals is skewed toward the myeloid rather than lymphoid lineage.⁴⁴ The molecular basis for these changes are undergoing intense study.^45,46,47,48

Another measure of stem cell kinetics is the time it takes for transplanted marrow cells to repopulate a lethally irradiated animal. Studies using retroviral markers suggest that HSCs can be divided into short-term and long-term repopulating cells, based on the timing of their appearance in the blood following intravenous transplantation (fewer than or more than 3 months following transplantation in mice).⁴⁹ However, a rapidly repopulating stem cell has been identified using a direct marrow injection strategy, a cell capable of generating large numbers of erythroid and myeloid cells within 2 weeks of injection.⁵⁰ Moreover, by transplanting luciferase-labeled single stem cells, a strategy that allows the serial tracking of the cells during life, initially detected foci were found to expand locally, seed other sites in the marrow or spleen, and then recede with different kinetics.⁵¹ From these experimental approaches it is clear that HSCs are heterogeneous.

STEM CELL ASSAYS

Transplantation Assays

Assays of Murine Stem Cells

Experimental transplantation in animals affords the clearest estimation of HSC properties as the capacity to durably regenerate all of hematopoiesis in an otherwise lethally irradiated animal remains the gold standard for the field; moreover, the technique can be made quantitative. Typically, either 2 × 10⁵ genetically marked, whole murine marrow cells, or reduced numbers of variably purified cells are infused intravenously into recipient animals who had previously received 90 to 110 cGy of whole-body irradiation. Blood cells and marrow are monitored for hematopoietic recovery in the following weeks and months, and the success of the transplant is measured by survival, and long-range contribution to hematopoiesis in the recipient. The contribution of donor cells to recovery is established by analysis of the posttransplant blood or marrow cells; the most common method of distinguishing donor from residual recipient blood and marrow cells is the use of flow cytometry against isoforms of the cell membrane-bound phosphatase CD45, present on virtually all hematopoietic cells. In a more quantitative embodiment of the strategy, limiting numbers of the genetically distinct cells (e.g., CD45.1+) are mixed with a “just adequate” (for full recovery) number of alternately marked cells (e.g., CD45.2+) and the proportion of CD45.1 to total CD45.1+ plus CD45.2+ cells is assessed following transplantation, yielding a calculation of the number of stem cells in the initial inoculum, an approach termed competitive repopulation.⁵² Because there exist both “short-term” and “long-term” repopulating cells, the degree of donor cell chimerism is tested 3 or more months following transplantation, to be certain that only the latter are evaluated. For example, transplantation of megakaryocyte-erythroid progenitor (MEP) cells allows for survival in a lethally irradiated mouse, as these cells allow sufficient time for endogenous recovery of the small number of relatively radio-resistant HSCs in the recipient mouse.⁵³ Consequently, survival alone following cell transplantation is not a sufficient measure of the presence of stem cells in a given population. Thus, with the appropriate caveats, this approach allows an assessment of the numbers or “quality” of HSCs in the test population (i.e., some genetically altered stem cell populations repopulate less robustly than wild-type cells as a consequence of defects in cytokine receptors or other genes that affect the self-renewal, survival, or proliferation of stem cells). Based on the use of these experimental tools, we know most about murine HSCs. Obviously, this approach is not available to assess human HSCs. Instead, a number of alternate experimental approaches have been developed.

Assays of Human Hematopoietic Stem Cells

Severely immunocompromised mice can be engrafted by human HSCs, provided their survival can be supported in a strictly controlled animal care environment and that the experiments take place prior to the development of other untoward effects in such animals (e.g., tumor formation). The first assay employing this strategy relies on the combined immunodeficiency created by the severe combined immunodeficiency (SCID) and nonobese diabetic (NOD) genetic mutations.⁵⁴ Subsequently, these mice were found to bear some ability to reject or alter the developmental characteristics of human cell repopulation, leading other investigators to add genetic defects to the NOD-SCID background that improve the engraftment of normal and pathologic human marrow cells, such as NOD-SCID/β₂-microglobulin null,⁵⁵ or the most commonly used NOD-SCID/γc null,⁵⁶ or by crossing with mice that also express human hematopoietic cytokines.⁵⁷ Such animal models have allowed the assessment of (1) stem cell numbers in human CD34+ cells from mobilized blood or umbilical cord blood,⁵⁸ (2) assessment of the effects of gene therapy vectors,^59,60 cell-cycle inhibitors,⁶¹ or cytokine cocktails designed to expand stem cell numbers^{58,59,60,61,62,63,64} on the retention of repopulating capacity, or (3) the study of fundamental biologic properties of human HSCs in vivo, such as the cell-cycle restriction of repopulating cells.⁶⁵

Surrogate In Vitro Stem Cell Assays

Although in vivo assays remain the gold standard, NOD-SCID and more severely immunocompromised mice are difficult to maintain and remain expensive and quite cumbersome methods to assess human HSC quality and quantity. As a result, a number of culture-based methods were developed to more quickly and quantitatively evaluate human HSC function. Generally, each relies on long-term cell growth in culture and other special features to establish their validity as a model of the human HSC.

The ability to grow marrow cells in culture for extended periods of time provided an important tool to explore HSC biology.⁶⁶ In long-term cultures human or murine marrow is incubated in serum-containing medium under defined conditions, and after several weeks the stromal layer that has developed is recharged with fresh marrow cells, which then produce mature blood cells and their progenitors for many months. Cell fractionation studies show that the HSC resides adherent to the stromal cell layer in such cultures,⁶⁷ and that enzymatic disruption of the stromal layer will allow one to reseed a secondary stromal cell layer with the capacity to produce hematopoietic cells for a period of weeks to months, thereby defining an in vitro assayable cell termed the long-term culture-initiating cell (LTC-IC).⁶⁸ A second assay that was developed based on similar principles is the cobblestone area-forming cell (CAFC), which, when evaluated by phase-contrast microscopy, gives rise to complex colonies of multiple hematopoietic cell types under the stromal cell layer of long-term cultures.⁶⁹ Unfortunately, when careful comparisons are made between these assays and transplantation studies, the true HSCs comprise only a fraction of the repopulating cells found in marrow. Thus, conclusions about stem cell behavior from such in vitro assays cannot be considered rigorous.

CELL SURFACE PHENOTYPE

Numerous investigators have used monoclonal antibodies to an increasing number of hematopoietic cell surface proteins to negatively and/or positively enrich for stem and primitive hematopoietic progenitor cells. Although the function of only a few of these stem cell markers is known, it has not impeded their use for research and/or therapeutic benefit. Others have taken advantage of the capacity of primitive hematopoietic cells to extrude fluorescent organic chemicals or on their buoyant density to obtain purified populations of these scarce marrow cells; most successful stem cell purification strategies employ several such techniques.

The antigenic proteins and glycoproteins that exclusively or predominantly present on HSCs include (1) CD34, a 90- to 110-kDa type I glycoprotein that is postulated to mediate cell adhesion and/or cell-cycle arrest^70,71,72; (2) CD90 (Thy1),⁷³ a heavily glycosylated glycophosphoinositol-linked protein that participates in T-cell adhesion to stromal cells⁷⁴; (3) CD117 (the c-Kit receptor),⁷⁵ which supports primitive hematopoietic cell survival and proliferation^76,77; (4) AA4,³⁴ a murine molecule homologous to the human phagocyte C1q complement receptor⁷⁸; (5) Sca1,⁷⁹ a murine surface molecule shown by knockout studies to be necessary for normal stem cell development⁸⁰; (6) CD133,⁸¹ a 115-kDa pentaspan cell-surface glycoprotein expressed on the apical surface of neuroepithelial and HSCs that has been proposed to function in establishing or maintaining plasma membrane protrusions⁸²; (7) CD164,⁸³ a cell-surface sialomucin that is present in several alternately spliced isoforms and that enhances blood cell homing and inhibits CD34+/CD38− cell proliferation,⁸⁴ and CD150, a member of the signaling lymphocyte activation molecule (SLAM) family of lymphocyte proliferation receptors⁸⁵; and (8) CD110 (the thrombopoietin [TPO] receptor c-Mpl)⁸⁶ present on virtually all repopulating HSCs,⁸⁷ and established to be vital for human HSC physiology as genetic elimination of the receptor leads to congenital amegakaryocytic thrombocytopenia at birth and aplastic anemia shortly thereafter (Chap. 117).

Many or most of the surface membrane proteins found on HSCs are also present on cells that have begun to differentiate toward specific lineages, precluding the exclusive use of positive selection alone for stem cell purification. Thus, a number of stem cell purification strategies include negative selection, based on cell surface markers absent on HSCs but present on mature blood cells and their corresponding unilineage-committed progenitors. Typically, cocktails of negatively selecting antibodies include CD38, HLA-DR, CD3, CD4, CD5, or CD8 for T lymphocytes; CD11b, CD14, or Gr-1 to exclude macrophages and granulocytes; CD10, CD19, CD20, or B220 to eliminate B lymphocytes; and glycophorin A or Ter119 to remove erythroid cells. The products that result from the use of such combinations of negative-selecting antibodies are termed Lin− cells.

A particularly difficult problem is presented by separating true HSCs from their progeny committed to the lymphoid or myeloid lineage, but not differentiated beyond that stage. Several studies clarify the cell surface profile of the common lymphoid progenitor (CLP) as Lin−/interleukin (IL)-7R(receptor)α+/Thy1−/Sca-110^w/c-kit^low37 and the common myeloid progenitor (CMP) as Lin−/IL-7Rα−/c-Kit+/Sca-1−.³⁶ The cell surface phenotype of human HSCs includes CD34+/CD38−/KDR(VEGFR2)+/Thy1+/CD133+/Lin−, although most of these markers require careful clinical assessment before their widespread use in patients can be considered. At present, at least for murine cells, that problem has been solved. The most effective cell sorting method for purifying murine HSC is the E-SLAM approach, negatively selecting for CD48 and positively selecting for CD45, CD201, and CD150. It is estimated that such a cell population is at least 50 percent pure HSC based on single-cell transplantation experiments in mice.

STEM CELL INTEGRINS

Integrins are a family of heterodimeric single-pass transmembrane proteins (18 α and 8 β subunits form at least 24 different cell surface adhesion receptors in humans) characterized by multiple immunoglobin (Ig)-like extracellular domains that allow two-way communication between a cell and its environment.⁸⁹ A large number of cell types require contact for survival; in vitro, this is usually manifest as integrin-dependent cell adhesion, either to extracellular matrix protein(s) or to other cells. In such cultures, disruption of adherence causes programmed cell death; for example, endothelial cells undergo apoptosis upon forced detachment in vitro, as a result of disruption of multiple integrins.⁹⁰ Integrins also influence the proliferation of cells by affecting the G₁ to S phase transition of the cell cycle.⁹¹ These effects also operate in vivo; α₁ integrin (a component of the α₁β₁ collagen receptor) null mice have a hypoplastic dermis, and the growth of α₁ −/− fibroblasts on collagen is substantially reduced.⁹²

Hematopoietic stem and progenitor cells express multiple integrins, including α₄ β₁ (also termed very-late antigen [VLA] 4), which binds to either vascular cell adhesion molecule (VCAM) 1 or fibronectin, and α₅  β₁ (VLA5), which binds to a region of fibronectin distinct from the β₁β₁ binding domain. Moreover, primitive hematopoietic cells are thought to express integrin αIIbβ₃, the platelet fibrinogen receptor, based on the death of multiple hematopoietic lineages in mice expressing a suicide transgene under control of the integrin αIIb promoter.⁹³ However, the physiologic significance of this finding is uncertain at present.

The avidity of progenitor cell–integrin interactions can be altered by external effectors; numerous cytokines and chemokines, including cytokines critical for stem cell function (stem cell factor [SCF], TPO, and CXCL12), enhance integrin-mediated binding.⁹⁴–⁹⁶ Counterreceptors for both integrins, such as VCAM1 and fibronectin (FN), are highly expressed in the marrow matrix and on marrow stromal cells (see “Matrix Proteins” in “The Hematopoietic Microenvironment” later). Integrin-based interactions with the stroma are responsible for homing and retention of stem and primitive progenitor cells in the marrow, as antibodies that interfere with the interaction can mobilize stem and progenitor cells into the blood.⁹⁷ However, it is uncertain whether integrins can influence the survival or growth of HSCs, or affect their ultimate developmental fate.

METABOLIC CHARACTERISTICS

One of the hallmarks of HSCs is their resistance to chemotherapy-induced cytotoxicity. A primary reason for this property is high-level expression of drug efflux pumps of the multidrug resistance class of proteins.⁹⁸ The presence of these verapamil-sensitive efflux pumps has enabled the separation of HSCs based on their low-level retention of various fluorescent markers such as rhodamine 123 and Hoechst 33342, the “Rh^lo/Ho^lo” population of murine cells and the side population (SP) of cells in human marrow.⁹⁹ However, before such maneuvers can be used for clinical stem cell enrichment procedures, the lack of toxicity of the fluorescent dyes must be confirmed. Nevertheless, such experimental strategies continue to shed important insights into HSC biology.

As a consequence of residing physically removed from the marrow vasculature, the stem cell microenvironment is hypoxic (see “Anatomy” in “Hematopoietic Microenvironment”). The metabolic consequence is that HSCs display a low metabolic state, with most of its ATP generation derived from glycolysis. As the HSC rarely undergoes cell division, it can “afford” the low metabolic state of cells dependent on glycolysis. One of the transcription factors critical for HSC physiology, MEIS1,¹⁰⁰ appears responsible for this metabolic profile, by driving expression of hypoxia-inducible factor (HIF)-1α, that upregulates expression of the glycolytic enzyme machinery.¹⁰¹ In addition, HIF-1α also induces pyruvate dehydrogenase kinases (PDK1-4) that prevent mitochondrial pyruvate oxidation by suppressing pyruvate dehydrogenase complex, the first step that fuels the Krebs cycle. Stem cell down-modulation of mitochondrial oxidative phosphorylation acts to reduce the generation of reactive oxygen species (ROS), which is highly toxic to HSCs. It has been suggested that this property, the avoidance of ROS generation, contributes to the longevity of the stem cell pool in any individual.

As HSCs mature into committed progenitors, they migrate toward the marrow vasculature, and into a higher oxygen tension atmosphere. This would appear to be required to gain the additional metabolic rate afforded by oxidative phosphorylation. As one component of this “metabolic switch,” marrow hematopoietic progenitor cells express lower levels of PDKs, reducing PDK-mediated suppression of oxidative phosphorylation seen in HSCs.¹⁰²

CELL-CYCLE CHARACTERISTICS

Adult hematopoietic cells display altered engraftment capacity dependent on their phase in the cell cycle. Using primitive hematopoietic cell populations several investigators have demonstrated that only quiescent G₀/G₁ phase cells engraft into lethally irradiated recipient animals; cells in the S and early G₂ phase display minimal engraftment capacity,^102,103 a situation that can be experimentally manipulated; elimination of p21, a key cell-cycle progression gene, enhances stem cell expansion.¹⁰⁴ This finding correlates well with findings that the profile of expressed genes in a highly selected population of primitive hematopoietic cells shifts when they are induced from G_0/1 phase into the cell cycle.¹⁰⁵ However, even though this cell-cycle dependence of engraftment of stem cells is true for adult cells, the corresponding cell populations derived from umbilical cord blood or fetal liver is not cell-cycle dependent.¹⁰⁶ A better understanding of these findings is very likely to shed important new insights into the genes that regulate engraftment.

GENE EXPRESSION PROFILE

It can be argued that the most critical feature of the HSC is its ability to quantitatively balance its three fates—apoptosis, self-renewal, and differentiation—into the mature elements of the blood. Moreover, the undifferentiated cell must express (at the least) the initiating genes responsible for all possible developmental lineages. A useful conceptual framework for this process can be constructed by considering the gene expression profiles of stem and committed hematopoietic progenitors that develop into the multiple hematopoietic differentiation pathways. At each developmental step genes associated with the adopted pathway should remain expressed or be upregulated, while the genes that specify the alternate lineage(s) are likely silenced. A thorough understanding of these gene expression profiles should help to explain the circuitry of specific aspects of hematopoiesis, and of developmental biology in general.

Initial studies using immortalized multipotent hematopoietic cell lines reinforced this conceptual framework; pluripotency is characterized by the expression of multiple genes associated with multiple cell fates.¹⁰⁷ Studies of purified HSCs and lineage-committed progenitors have also strengthened this hypothesis, revealing coexpression of several different lineage-affiliated gene sets in single primitive hematopoietic cells.¹⁰⁸ In contrast, the downstream progenitors of HSCs were found to express only lineage-appropriate transcripts, such as for the granulocyte colony-stimulating factor receptor (G-CSF-R) in granulocyte-macrophage progenitors (GMPs), or β-globin and the erythropoietin receptor (EPOR) in committed erythroid progenitors.³⁶ Similar findings were reported for lymphoid committed cells, although some promiscuity was detected in B-cell progenitors.¹⁰⁹

With these principles established, more ambitious efforts to catalogue all the genes expressed by each stage of hematopoietic development have been made possible by advances in microarray and whole transcriptome sequencing approaches to gene expression.¹¹⁰ On an even broader scale, and as might be expected, comparisons of different types of stem cells (e.g., liver, skin, neural) reveals an overlap in the expressed genes, supporting the hypothesis that the mechanisms responsible for critical stem cell properties, such as self-renewal, are shared among the cells derived from multiple organs.¹¹¹ This observation also provides a powerful tool to identify such proteins. Such studies have also begun to identify novel genes expressed in HSCs, potentially allowing our better understanding of their role in hematopoiesis.

TRANSCRIPTION FACTOR PROFILE

An important goal of modern cell biology is to provide a molecular explanation for the gene or sets of genes required to orchestrate specific developmental events. Fundamental to this process is an understanding of the proteins present in cells that regulate gene transcription in a lineage-, ontogenic stage-, and developmental level-specific manner. Unlike what is claimed for many organ-specific programs, no single lineage-unique family of master regulators exerts executive control over hematopoiesis. Rather, an assemblage of specific and nonunique factors and signals converge to determine lineage and differentiation patterns. Several transcription factors have been identified in stem cell populations or have been shown to affect stem cell differentiation into the lymphoid and myeloid lineages. In addition to transcription factors that regulate HSC expansion, a number of epigenetic and microRNA changes have been identified that affect gene expression in these cells. The BMI1 gene encodes a protein that forms part of a polychrome group repressor complex, which represses a number of important target genes including the cell-cycle regulator p16/INK4a, a pathway that regulates HSC function in normal and malignant hematopoiesis.¹¹² In addition, methylation can affect HSC gene expression, as the DNA methyltransferases DNMT3A and DNMT3B affect HSC self-renewal.¹¹³ And microRNA (miRNA) species are regularly being identified that regulate transcription and translation of critical HSC genes. For example, 9 miRNA were overexpressed and 22 downregulated in CD34+/CD38− HSCs compared with CD34+/CD38+ cells. Among the most upregulated miRNAs in the more primitive cells was miR-520h, predicted to target ATP-binding cassette, subfamily G (ABCG2) gene, known to be involved in stem cell maintenance. Transduction of miR-520h into CD34+ cells increased the numbers of several progenitor cell types (colony-forming unit–erythroid [CFU-E], burst-forming unit–erythroid [BFU-E], and colony-forming unit–granulocyte-macrophage [CFU-GM]) as well as the total number of CD34+ cells (reviewed in Ref. 114).

Hematopoietic Stem Cell Self-Renewal and Expansion

Members of the Hox family of transcription factors are important regulators of hematopoietic cell decisions, at least at the level of self-renewal/expansion, based on (1) a similar role in multiple organ systems¹¹⁵; (2) their lineage- and differentiation-stage-specific expression pattern in hematopoietic cells¹¹⁶; (3) disruption of their usual level or pattern of expression that leads to hematologic expansion or malignancies¹¹⁷; and (4) their elimination,¹¹⁸ or elimination of the gene(s) that regulate them,¹¹⁹ which leads to significant defects in hematopoiesis. In addition, members of the extradenticle family of homeodomain-containing proteins serve as cofactors for Hox proteins, altering their cellular localization, DNA-binding affinities, and specificities. Like Hox genes, genetic elimination of some of these cofactor proteins can lead to HSC defects. For example, Pbx1 null mice display greatly reduced numbers of CMPs,¹²⁰ and overexpression or altered expression of MEIS1 is associated with hematologic malignancy.¹²¹

Hematopoietic Stem Cell to Common Lymphoid Progenitor Commitment

The Ikaros gene encodes a family of lymphoid-restricted zinc-finger transcription factors related to the Drosophila hunchback gene.¹²² All isoforms of Ikaros contain a highly conserved carboxyl-terminal activation domain and two zinc-finger domains that mediate their dimerization. However, only isoforms 1 to 3 of the six known alternately spliced forms contain more than three of the four N-terminal zinc fingers required for DNA binding to the consensus DNA core motif GGGA.¹²³ The PU.1 gene is 1 of approximately 30 members of the Ets family of transcription factors that bind to the purine-rich sequence 5′-GGAA-3′.¹²² Genetic elimination of the Ikaros and PU.1 genes have established their critical role in commitment of HSCs to the lymphoid lineage; fetal stem cells in Ikaros −/− mice fail to generate any definitive T- or B-lymphocyte precursors,¹²⁴ and although thymocyte precursors can be identified postnatally, they undergo aberrant differentiation or fail to develop into the CD4, dendritic, and some γδT-cell subsets in adult mice. Consequently, Ikaros is essential for all of lymphopoiesis early during ontogeny, and for several subsets of lymphocytes later in life. In a similar fashion, PU.1-deficient mice also lack any definitive T- and B-cell precursors in their lymphoid organs at birth (and myeloid cells; see “HSC to CMP Commitment” below),¹²⁵ and if knockout mice are maintained on antibiotics and survive the first 48 hours of life, they begin to develop normal-appearing T cells 3 to 5 days later. In contrast, mature B cells and macrophages remain undetectable in the older mice, indicating absolute tissue dependence for this lineage.

Hematopoietic Stem Cell to Common Myeloid Progenitor Commitment

The SCL (stem cell leukemia) gene encodes one of the transcription factors responsible for the initial stages of myeloid development, a gene first identified at the site of chromosomal rearrangement in a patient with SCL.¹²⁶ SCL belongs to the helix-loop-helix family of transcription factors, which form dimers and bind DNA at consensus E-box motifs (CANNTG).¹²⁷ Although initially identified as a gene rearranged in T-cell acute lymphocytic leukemia, an essential role for SCL in hematopoietic development was established by gene ablation studies, which revealed a complete absence of primitive blood cells and lethality in scl−/− embryos at day 9.5 postcoitum.¹²⁸ Consistent with this panhematopoietic phenotype, previous studies showed that SCL is downregulated in differentiating granulocytic and monocytic progenitor cells and that forced expression of the gene in hematopoietic cell lines inhibits cytokine-induced granulocytic and monocytic differentiation.^129,130 Consistent with their respective roles in promoting stem cell and mature cell survival and proliferation, SCF sustains SCL expression in primary CD34+ cells, maintaining them in an undifferentiated state, whereas granulocyte-monocyte colony-stimulating factor (GM-CSF) downregulates SCL levels and favors granulocyte and monocyte differentiation.^130,131 Together, these results suggest that SCL expression is required for HSC and CMP maintenance, and that down-modulation of the transcription factor is essential for myeloid differentiation.

The GATA transcription factor family contains six members possessing a highly related DNA-binding domain composed of two conserved zinc-finger motifs.¹³² GATA1 and GATA2 are present in hematopoietic cells, GATA2 is found in the same cells as SCL, with GATA1 expression restricted to latter stages of erythroid/megakaryocytic (MEP) differentiation. Because genetic elimination of GATA2 is lethal as a result of numerous nonhematopoietic defects, and because individual hematopoietic lineage-specific knockouts have not yet been engineered, the role of GATA2 in early hematopoiesis is uncertain. However, like SCL, elimination of GATA2 expression is required for hematopoietic cell maturation.¹³³

As noted above, numerous lines of evidence indicate that HSCs express the TPO receptor, c-Mpl, as best exemplified by its expression on all AA4+/Sca+ cells that are capable of long-term hematopoietic repopulation.⁸⁷ Several investigators have shown that the 5′ flanking region of the c-mpl gene contains a functionally important GATA site and that GATA1 transactivates the gene in hematopoietic cell lines.¹³⁴ Because GATA1 does not appear in hematopoietic cells until they have lost their repopulating capacity, it is possible that GATA2 fulfills this role in HSCs, although there is no evidence yet available establishing that this protein can transactivate the c-mpl GATA site.¹³⁵

STEM CELL AGING

While blood cell counts do not change substantially in elderly mammals, a number of alterations are demonstrable in HSCs derived from older individuals. In aged mice, the HSC compartment expands, although each HSC has reduced capacity to expand.¹³⁶ Upon transplantation, aged marrow HSCs display a myeloid skewing, generating reduced numbers of T- and B-lymphoid precursors, that along with thymic involution helps to explain the immune depletion seen in older adults. Similar findings were reported when aged human stem cells were transplanted into immunocompromised mice.¹³⁷ This topic has been reviewed.¹³⁸

It has been estimated that the concentration of cells within the marrow is 10⁹/mL; as a result, multiple cell–cell and cell–matrix interactions occur. A major advance in experimental hematology has been the capacity to grow hematopoietic cells in long-term culture. When high concentrations of marrow cells are placed in serum-containing cultures, a stromal cell layer and extracellular proteinaceous matrix form, and when subsequently recharged with fresh marrow cells, these long-term cultures (LTCs) are capable of supporting hematopoiesis for months with simple demidepletion and replacement of culture medium. It is assumed that the cell–cell and cell–matrix interactions that develop in such cultures more closely resemble those found in vivo, helping to explain the longevity of such cultures and their capacity to maintain hematopoietic stem and primitive progenitor cells far longer ex vivo than do nonstromal cell-containing cultures. The molecular basis for the improved hematopoietic environment of LTCs is thought to rely on stromal cell surface molecules that promote cell–cell contact, prevent programmed cell death, and regulate growth.

The microenvironmental effects on HSCs have far reaching clinical implications as well; our ability to mobilize marrow stem cells for transplantation has greatly changed the way we treat hematologic and other malignancies, and ultimate success in the efforts of experimental hematologists to expand HSCs ex vivo with cocktails of cytokines and stromal cells for applications in gene therapy and regenerative medicine will undoubtedly derive only from a thorough understanding of the molecular bases for the interaction of HSCs with their microenvironment (Fig. 18–2).

Marrow stromal cells influence hematopoiesis in a number of ways, by producing several cytokines that positively or negatively affect hematopoietic cell growth,^{139,140,141,142} including some, like SCF, that are expressed on their cell surfaces, resulting in enhanced biologic activity.¹⁴³ Stromal cells are the origin of a number of extracellular matrix proteins that either directly affect hematopoietic cells, or do so indirectly by binding growth factors and presenting them in a functional context.¹⁴⁴ They also bear the Jagged/Delta family ligands that stimulate Notch proteins to undergo cleavage and translocation into the nucleus, events that are critical mediators of cell fate decision making,^145,146 including for hematopoietic cells.¹⁴⁷ Cell–cell interactions mediated by integrins present on hematopoietic cells and counterreceptors on stromal cells are also very important for hematopoiesis.^65,72 In addition to bringing hematopoietic cells into close proximity to cells producing soluble or cell-bound cytokines, and hence raising the local concentration of these growth promoting proteins, integrin engagement leads to intracellular signaling, usually promoting entry into the cell cycle and preventing programmed cell death.¹⁴⁸ Reflecting the vital and sometimes lineage specific roles of the hematopoietic microenvironment, the extracellular matrix and stromal cells reside in a highly organized structure (Chap. 5).

ANATOMY

Hematopoiesis is highly compartmentalized within areas of red marrow, with erythropoiesis occurring in clusters surrounding a central macrophage,¹⁴⁹ granulocyte development associated with stromal cells,¹⁵⁰ and megakaryopoiesis occurring adjacent to the endothelial sinusoidal cells.¹⁵¹ In the adult marrow, the specialized niche in which HSCs develop into differentiated progeny has been termed the hematon by Peault, a structure that includes Str01+ mesenchymal cells, desmin-positive perivascular lipocytes, Flk1+ endothelial cells, macrophages, and hematopoietic progenitors.¹⁵² From these structures can be derived all lineages of committed colony-forming cells (e.g., CFU-GM and BFU-E) and primitive cells that score positive in CAFC assays, LTC-IC, and high proliferative potential colony-forming cell assays (Chap. 5).

One consequence (or perhaps cause) of this anatomical arrangement is that the stem cell microenvironment is quite hypoxia. It is estimated that the O₂ level of the stem cell niche is approximately 5 percent. The HSC response to hypoxia is discussed in “Metabolic Characteristics” above.

STROMAL CELLS

The marrow microenvironment is composed of multiple cell types. Fibroblasts are perhaps the best-studied of the marrow stromal cells, and can bind to primitive hematopoietic cells by engaging cell-surface integrins.¹⁵³ Marrow endothelial cells also support primitive hematopoietic cells, including LTC-IC.¹⁵⁴ The CXCL12–abundant reticular (CAR) cells, which surround the sinusoidal endothelial cells in vivo, are also likely to play the critical niche function of the vascular wall.^155,156 However, based on their ability to increase the number of HSCs when experimentally increased, osteoblasts, which line trabecular bone and reside adjacent to primitive hematopoietic cells,¹⁵⁷ are thought to provide a critical role in serving as the HSC supportive niche.¹⁵⁸ The origin of all of these cell types is thought to reside in the mesenchymal stromal cell (MSC), a functionally defined entity that under specific conditions can be induced to form fibroblasts, endothelial cells, CAR cells, and osteoblasts, amongst others,¹⁵⁹ and hold promise to therapeutically manipulate hematopoiesis.¹⁶⁰ MSCs are discussed more extensively in Chap. 30.

Marrow stromal cells affect HSCs in multiple ways. Each of these cells is known to produce a number of cytokines critical for primitive and mature hematopoietic cell development. For example, although a number of organs produce TPO constitutively,¹⁶¹ marrow stromal cells are induced to produce the hormone in states of thrombocytopenia.^162,163 Stromal cells produce SCF constitutively in both soluble and membrane bound forms,⁷⁶ and FLT-3 ligand (FL) is produced both constitutively by stromal cells and lymphocytes and can be induced to high levels in the presence of pancytopenia.¹⁶⁴

Besides growth factor production, stromal cells are also known to display counterreceptors for the integrins present on hematopoietic cells, including VCAM-1,¹⁶⁵ interactions that promote cell survival and proliferation in several ways.¹⁶⁶ Osteoblast-derived annexin II serves as an adhesion molecule for HSCs.¹⁶⁷ Stromal cells also elaborate extracellular matrix components, including collagen, laminin, FN, heparins, hyaluronan, and tenascin, which display important effects on HSCs (see “Matrix Proteins” later). These substances, in turn, engage a number of HSC integrins and other cell-surface molecules, and form a solid matrix on which hematopoietic cells firmly attach. Of considerable clinical interest, it appears that interference with cell–matrix interactions,¹⁶⁸ or digestion of the extracellular matrix itself,¹⁶⁹ is involved in mobilizing HSCs by some agents such as granulocyte colony-stimulating factor (G-CSF) and IL-8.

It has long been known that the marrow is innervated by the autonomic nervous system,¹⁷⁰ which influences HSCs in several ways, such as directing HSC trafficking by acting on nestin-positive microenvironmental cells.¹⁷¹ One or more of these functions appear to be critical for HSC homeostasis, as marrow nerve injury impairs hematologic recovery following chemotherapy-induced injury.¹⁷²

Cytokines

The regulation of stem cell survival, proliferation, and differentiation has been difficult to address because of the rarity of stem cells and the requirement that they be assessed using cumbersome transplantation assays. Several cytokines are able to exert effects on HSCs. The pursuit of the cytokines that affect HSCs is of more than pure physiologic interest, as the availability of the right combination of such proteins could allow expansion of the cells for therapeutic use without sacrificing their pluripotent and self-renewal capacities. Three proteins—SCF, FL, and TPO—and their corresponding receptors (c-Kit, Flt3, and c-Mpl, respectively) exert important effects on the number and/or growth of HSCs both in vitro and in vivo (Table 18–1).

Stem Cell Factor

The molecule termed SCF, steel factor, mast cell growth factor, or c-Kit ligand was cloned by several groups based on its binding to a cell surface receptor encoded by the protooncogene c-Kit,⁷⁶ previously identified as responsible for the severe defects in hematopoiesis, pigmentation, and gametogenesis in W mice. As the phenotype of mice bearing alleles of W was quite similar to those of steel (Sl), but in transplantation studies one strain displayed a stem cell autonomous defect (W) while the other was not (Sl), it had been hypothesized that the two genes represented the receptor for a growth factor and the cytokine itself, respectively, a tenet proven true with the cloning of SCF.

The extracellular domain of c-Kit is composed of five immunoglobulin-like domains, which leads through a typical transmembrane domain to the intracellular domain that bears a split domain-type tyrosine kinase. A single molecule of SCF binds to the first three immunoglobulin domains (D₁D₂D₃) of two c-Kit receptors. The two D4 domains of a dimeric c-Kit receptor display substantial electrostatic repulsion toward each other, precluding the juxtaposing of the two transmembrane domains, and by extension, the two intracellular kinase domains. Once SCF binds to c-Kit the affinity for dimer formation overcomes the D4 electrostatic repulsion and the two kinase domains are brought together to initiate signaling.¹⁷³ The intracellular mediators activated by SCF binding to c-Kit include phosphoinositol 3′-kinase (PI3K), mitogen-activated protein kinases (MAPKs), phospholipase C gamma (PLCγ), and c-Src (Chap. 17; reviewed in Ref. 174).

SCF is synthesized by marrow fibroblasts and other cell types. Soluble SCF is a highly glycosylated 36-kDa protein released from its initial site on the cell membrane by proteolytic processing. An alternatively spliced form of SCF messenger RNA (mRNA), that does not encode the cleavage site, remains on the cell membrane, and is a more potent stimulus of c-Kit-receptor-bearing cells.¹⁴⁰ The ratio of soluble to membrane encoding SCF mRNA varies widely in different tissues, ranging from 10:1 in the brain, to 4:1 in the marrow, to 0.4:1 in the testis.

The importance of SCF to hematopoiesis is easily demonstrated; although nullizygous mice (Sl/Sl) are embryonic lethal because of a number of developmental defects, the presence of a partially functional allele (Sl^d) allows compound heterozygotes (Sl/Sl^d) to survive into adulthood, albeit with severe anemia because of diminished numbers/quality of HSCs. In addition to its critical role in the development of embryonic and fetal hematopoiesis, treatment of adult mice with an antibody that neutralizes the SCF receptor, c-Kit, also results in severe pancytopenia, indicating an important hematopoietic role for the receptor/ligand pair throughout life.

When present in culture SCF alone can maintain the long-term repopulating ability of murine Sca-1+/Rh^lo/Lin− hematopoietic cells, suggesting that the cytokine can promote the survival of HSCs in vitro.¹⁷⁵ However, alone, SCF is only a weak stimulator of cell proliferation, primarily inducing the development of mast cells both in vitro and in vivo. Nevertheless, in the additional presence of IL-3, IL-6, IL-11, G-CSF, or TPO, SCF exerts profound effects on the generation of hematopoietic progenitor cells of all lineages,^176,177,178 pointing to primitive hematopoietic cells as critical targets. The molecular mechanisms of such synergy are beginning to emerge.¹⁷⁹ A physical association of c-Kit and EPOR has been detected following SCF stimulation of cells bearing both receptors, an event that is essential for their functional synergy.¹⁸⁰

Flt3 Ligand

FL was cloned as the binding partner for the then newly identified novel orphan receptor Flt3,¹⁸¹ a protein most closely related to the receptors for macrophage colony-stimulating factor (M-CSF) (hence the term flt = fms-like tyrosine kinase), and c-Kit. FL is expressed by T lymphocytes and marrow stromal cells.^164,181 The Flt3 receptor is a 160-kDa cell-surface molecule expressed primarily on primitive hematopoietic cells.¹⁸² Like c-Kit, activation of Flt-3 results in activation of several signaling mediators, including the p85 subunit of phosphatidylinositol 3-kinase, SHP, PLCγ, and a guanosine 5′-triphosphatase (GTPase)-activating protein, activating Ras.¹⁸³ Normal Flt3 signaling also activates the MAPKs extracellular regulated kinase (ERK)-1 and ERK2 but leads to only weak phosphorylation of signal transducer and activator of transcription (STAT)-5, in contrast to an oncogenic form of the receptor, identified in 25 percent of patients with myelodysplastic syndromes or acute myelogenous leukemia.^184,185 In the leukemic cells of such patients Flt3 bears an internal tandem duplication of the kinase domain, resulting in the constitutive activation of the receptor. Clinically, this is associated with reduced likelihood of patient survival; hence, this observation has led to an attempt to control the growth of such mutant-receptor-bearing cells with specific Flt3 kinase inhibitors,¹⁸⁶ with some success (Chap. 88).

FL was initially cloned using a soluble form of the receptor to identify ligand-bearing cells.¹⁸⁷ As their receptors bear a number of common structural features, it was not surprising to find that FL shares significant structural homology, as well as biologic properties with both M-CSF and SCF. Like the other two cytokines, FL displays a 4α-helix bundle tertiary structure and exists in both membrane-bound and soluble states, the result of alternate splicing of the primary transcript that does or does not include a cleavage site for its release from the cell membrane.¹⁸⁸

Unlike SCF levels that remain relatively static regardless of blood cell counts,⁷⁶ blood concentrations of FL can rise more than 25-fold in response to pancytopenia.¹⁸⁹ Interestingly, only pancytopenia, and not individual lineage deficiencies cause an increase in blood FL concentrations, suggesting that the cytokine is a bona fide regulator of stem or primitive hematopoietic cells. Consistent with this conclusion, transplantation data indicate that HSCs from Flt3-deficient mice do not effectively reconstitute the hematopoietic system,¹⁹⁰ being three- to eightfold less efficient in repopulation as wild-type cells, a conclusion reinforced by its genetic combination with c-Kit mutant mice.¹⁹⁰

Like SCF, FL appears to act on HSCs only in synergy with other hematopoietic cytokines,^191,192 a finding particularly true for its combination with TPO.^193,194 In addition, FL is a potent stimulus of B lymphopoiesis and granulocyte-macrophage proliferation and development, particularly of the latter toward the dendritic cell lineage.^195,196

Thrombopoietin

TPO is a 45- to 70-kDa hormone that was cloned by both traditional biochemical purification and expression cloning strategies based on the use of a then orphan class I cytokine receptor, first identified as the cellular homologue of the murine-transforming oncogene v-mp1.¹⁹⁷ TPO bears extensive sequence homology to erythropoietin (EPO), sharing 20 percent identity and an additional 25 percent similarity. The hormone is produced in several organs, including the liver, kidney, skeletal muscle, and the marrow stroma. Based on murine liver transplantation studies about half of steady-state TPO production occurs in that organ,¹⁹⁸ but in states of thrombocytopenia the marrow stroma increases production substantially.^160,163 The hormone acts on megakaryocyte (MK) progenitors to enhance their survival and proliferation and on immature MKs to promote their differentiation, but surprisingly not on mature cells during platelet formation.¹⁹⁹ Multiple lines of evidence also indicate that TPO can exert profound effects on the HSC. The hormone also supports the survival of candidate HSC populations, and acts in synergy with IL-3 and SCF to induce these cells into the cell cycle and increase their output of both primitive and committed hematopoietic progenitor cells of all lineages.^200,201 These properties are also seen in vivo. For example, administration of the hormone to myelosuppressed animals leads to more rapid recovery of all hematopoietic lineages, including primitive cells,^{202,203,204,205} and genetic elimination of TPO or its receptor severely reduces the number of marrow stem and progenitor cells of all lineages to 15 to 25 percent of normal values.^87,206,207 In addition, as noted in “Flt3 Ligand” above, TPO acts in synergy with FL to expand primitive hematopoietic cells in suspension culture, and when used to supplement LTC, the hormone maintained HSC numbers for up to 2 months,²⁰⁷ compared to standard LTCs in which repopulating HSCs are no longer detectable at this time.

The TPO receptor, the product of the cellular protooncogene c-Mpl, is a member of the cytokine receptor family that includes EPO, G-CSF, growth hormone, leptin, and many others. Upon binding TPO, the homodimeric c-Mpl activates its tethered Jak2 kinases, leading to phosphorylation of three of the cytoplasmic domain tyrosine residues. These phosphotyrosine residues then act as docking sites for several secondary signaling molecules, including STATs, MAPKs, and PI3K, ultimately leading to the expression of a number of transcription factors (e.g., homeobox-containing proteins, HIFs) and cell-survival molecules (e.g., BclXL). A more complete discussion of the molecular mechanisms by which TPO affects the HSC is found in Chap. 17.

CXCL12 (Previously Termed Stromal Cell-Derived Factor 1)

CXCL12 is produced by a number of the cells that occupy the hematopoietic microenvironment, and has profound effects on HSC localization to the stem cell niche.²² However, this chemokine is also thought to display direct effects on the survival and proliferation of hematopoietic stem and progenitor cells, both alone and in synergy with other hematopoietic cytokines.^208,209

Notch Ligands

The human homologue of Drosophila Notch was identified as an altered gene product in T-cell leukemia.²¹⁰ The discovery that the hematopoietic microenvironment displays Notch ligands, and that Notch isoforms appear on primitive hematopoietic cells,^211,212 opened the possibility that Notch affects HSCs. This assertion has been directly proven: The Notch ligands Delta1 and Delta4 expand primitive hematopoietic cells.^213,214 It is possible that the favorable effect of marrow osteoblasts on HSCs is a result of their expression of Notch ligands, as inhibition of Notch processing blocks the expansion in HSCs seen in mice in which osteoblasts have been experimentally expanded.¹⁵⁸

Wnt Proteins

A role for Wnt proteins in hematopoiesis was suggested by their localization at sites of fetal blood cell production and their ability to expand hematopoietic progenitor cells.²¹⁵ Wnt 3a has been shown to expand long-term repopulating HSCs.^216,217 As Wnt proteins are expressed on primitive hematopoietic cells,²¹⁸ it is also possible that in addition to classical paracrine signaling, Wnts could act in an autocrine fashion in HSC biology.

Transforming Growth Factor β

The transforming growth factor (TGF) family of ligands (TGF-β, activins, bone morphogenetic proteins [BMP]) bind to members of the TGF-β receptor family and trigger activation of the SMAD (Sma- and Mad-related protein) group of intracellular mediators.²¹⁹ Unlike the cytokines discussed above, TGF-β members inhibit HSC cycling,^220,221 and so blunt cell expansion, at least in vitro. Nevertheless, the situation in vivo is complex; genetic elimination of TGF-β does not alter HSC self-renewal or regeneration in vivo,²²² likely because of redundancy in the TGF-β system of ligands.²²³ In contrast, genetic elimination of several of the SMAD proteins disrupts normal HSC homeostasis.^224,225 Recent data suggests that BMP4 might be the critical member of the TGF family that affects HSC biology.²²⁶

The mechanisms by which these cytokines exert their effects on HSCs are only now beginning to be understood at the molecular level, but it is already clear that effects on the transcription factors that govern HSC survival, self-renewal, and expansion likely play critical roles. It has long been understood that Wnt proteins act to stimulate an increase in intracellular levels of β-catenin, a nascent transcription factor. Upon being liberated from proteasomal degradation in the presence of Wnt, β-catenin translocates to the nucleus and alters transcription of genes displaying the T-cell factor (TCF)/lymphoid-enhancer binding factor (LEF) consensus sequence.²²⁷ Moreover, TGF-β–induced alterations in SMAD protein phosphorylation affects their ability to activate transcription directly.²²⁸ However, most of the cytokine receptors that affect HSCs do not directly affect transcription factors; rather, several cytokines affect signaling pathways that alter the expression, activity, or subcellular localization of HSC transcription factors.

As discussed in “Hematopoietic Stem Cell to Common Myeloid Progenitor Commitment” above, SCL is a helix-loop-helix transcription factor critical for hematopoiesis. SCF enhances the survival of primitive hematopoietic cells in culture by maintaining their expression of SCL,¹³¹ which enhances expression of the SCF receptor c-Kit.²²⁹ Two additional transcription factors that play vital roles in HSC expansion, HOXB4 and HOXA9, are both affected by cytokines. Exogenous expression of HOXB4 to levels only twice normal are associated with a marked and rapid expansion of transduced HSCs on their transplantation into lethally irradiated recipients.¹¹⁶ In both model cell lines and primitive hematopoietic cells TPO doubles the expression of HOXB4, in a p38 MAPK-dependent fashion.²³⁰ Of probably greater significance is the effect of TPO on HOXA9, a gene that also induces rapid expansion of HSCs on its introduction into these cells, and whose genetic elimination leads to a profound deficit in numbers of HSC in vivo.¹¹⁸ Although the hormone fails to affect total cellular levels of HOXA9 in either model cells or primary primitive murine HSC populations, TPO greatly enhances HOXA9 nuclear translocation by inducing expression of its translocation partner, MEIS1, and leading to ERK1/2 MAPK-induced MEIS1 phosphorylation.²³¹

A third mechanism by which cytokines affect HSC expansion is through global inhibitors of signaling. In addition to its direct effects on HSC survival and self-renewal pathways, TPO has been shown to interact with the adaptor protein LNK,²³² which inhibits signaling pathways derived from a broad range of hematopoietic cytokines,^233,234 including TPO.²³⁵ From these data it appears that TPO and LNK alternately regulate HSC expansion and each other.²³⁶

MATRIX PROTEINS

Fibronectin

FN is a 450-kDa fibril-forming glycoprotein composed of two subunits that is a major component of the hematopoietic microenvironment. FN is produced by both marrow stromal (endothelial cells and fibroblasts) and blood cells,²³⁷ and is implicated in marrow homing of hematopoietic cells.²³⁸ Distinct domains of FN have been identified that interact with different integrins, for example, those for integrin α₄β₁ and for integrin α₅ β₁.¹⁴⁸ HSCs display multiple integrins and their engagement contributes to cell survival and/or expansion. For example, ex vivo culture of human CD34+ cells on FN maintains the repopulating capacity of HSCs, whereas growing the cells in suspension obliterates their ability to repopulate hematopoiesis.²³⁹ FN binding to α₄β₁ integrins also enhances the generation of large numbers of committed hematopoietic progenitors²⁴⁰ and LTC-IC²⁴¹ from primitive precursors. Multiple molecular mechanisms for the effects of FN on integrin-bearing cells have been identified, and serve as a paradigm for the supportive effects of this entire class of microenvironmental signals.

Integrin engagement by FN triggers a number of intracellular signaling events that affect the cellular cytoskeleton and transcriptional events. Complexes composed of kinases, adaptors, and cytoskeletal components are recruited to sites of integrin engagement, initiated by interactions with integrin cytoplasmic domains.⁸⁹ A critical molecule for integrin-based signaling is paxillin, a 68-kDa protein that contains a number of protein–protein binding domains, and which binds to the cytoplasmic domain of the integrin.²⁴² Additional binding partners also help trigger intracellular signaling, including focal adhesion kinase (FAK) and the closely related Pyk2 kinase. Upon recruitment, FAK and Pyk2 are activated and initiate Tyr phosphorylation of paxillin and other associated molecules, creating additional protein binding sites and activating tethered secondary messenger molecules. One vital signaling pathway downstream of FAK and Pyk2 is PI3K, which is mediated by the association of its regulatory p85 subunit with the adhesion kinases (Chap. 14).²⁴³ FAK also directly activates a pathway that results in upregulation of the cyclin D promoter,²⁴⁴ affecting cell proliferation. Integrin engagement also leads to Src activation, engagement of Grb2, and activation of Ras,²⁴⁵ pathways also activated by SCF and TPO, and potentially providing a mechanism by which diverse extrinsic stimuli of HSCs may converge.

Hyaluronan

Another stromal cell matrix glycoprotein is hyaluronan, which binds to two hematopoietic cell-surface receptors, RHAMM and CD44. Although most CD34+ marrow cells express CD44, only a fraction of them adhere to hyaluronan,²⁴⁶ a process that can be mediated by cytokines, as a result of either increased surface expression of CD44 or an alteration in its conformation. Consistent with the latter notion, certain epitopes on CD44 have been shown to be inducible,²⁴⁷ and antibodies to CD44 can alter the adherence of CD34+ cells to marrow stroma.²⁴⁸ Nevertheless, other data suggests that RHAMM is the primary receptor for hyaluronan.²⁴⁹ It is also of considerable interest that primitive hematopoietic cells also express hyaluronan, and that it plays an important role in their lodgment in the marrow and subsequent proliferation.²⁵⁰

Heparan Sulfate

LTCs that support hematopoiesis develop a heparan sulfate proteoglycan layer. Immunochemical analysis has shown that marrow stromal cell lines synthesize and secrete numerous members of the syndecan family of heparan sulfate, including glypican, betaglycan, and perlecan.¹⁸ Evidence is accumulating that heparan sulfate-containing proteoglycans may be vital components of the stem cell niche. For example, the structure of the heparan sulfate secreted from stromal cell lines that support long-term hematopoiesis are significantly larger and more highly sulfated than heparan sulfate from nonsupportive stromal cell lines, and when used alone in LTCs, the former can support LTC-IC, whereas desulfated heparan sulfate cannot.²⁵¹

Tenascin

Tenascins are large, extracellular matrix (ECM) glycoproteins found in several tissues, synthesis of which is upregulated in response to tissue regeneration. Tenascins are multimeric proteins composed of numerous modules. For example, tenascin-C is composed of six subunits linked like spokes in a wheel by their C-terminal fibrinogen-like domains, each subunit being composed of multiple epidermal growth factor (EGF)-like and FN type III modules. Two forms of tenascin of molecular mass (Mr) 280 and 220 kDa are also expressed at high levels by marrow stromal cells.²⁵² Marrow cells can adhere to tenascin-C within the fibrinogen-like domain and to two sets of the FN type III-like repeats, and when so engaged, they undergo a proliferative response.²⁵³ Genetic elimination of tenascin leads to modest deficiencies in marrow hematopoietic progenitor cells,²⁵⁴ although as the levels of FN in such mice are also reduced, it is unclear if direct tenascin engagement of hematopoietic cells is responsible, or the defect is a result of the secondary reduction of FN engagement of β₁ integrins.

Laminins

Laminins are heterotrimeric (αβγ) extracellular proteins that regulate cellular function by adhesion to integrin and nonintegrin receptors. At present, 5 α chains, 3 β chains, and 2 γ chains have been characterized, which combine to form at least 12 distinct laminin isoforms.²⁵⁵ Laminins containing γ₂ and either β₁ and α₅ chains are expressed in marrow, but only the latter (laminin-10/11) binds to α₆ β₁ integrin on primitive hematopoietic cell lines²⁵⁶ and to primary human CD34+/CD38− stem and progenitor cells.²⁵⁷ A second, nonintegrin laminin receptor (LR) also binds laminins, as well as other components of the ECM, such as FN, collagen, and elastin, and is composed of an acylated dimer of 32-kDa subunits.²⁵⁸ Although not an integrin, the LR associates with integrins (e.g., integrin α₆ β₄) to modulate laminin binding.²⁵⁹ Functionally, aminin-10/11 facilitates SDF-1α–stimulated transmigration of CD34+ cells,²⁶⁰ and displays mitogenic activity toward human hematopoietic progenitor cells.²⁵⁵ The nonintegrin LR associates with the GM-CSF receptor (GM-CSF-R) to modulate its signaling properties, down-modulating receptor signaling in the absence of laminin, and releasing the inhibition when bound by its ligand.²⁶¹ This arrangement could provide a novel molecular explanation for how laminins affect cell proliferation; whether this physiology extends to other cytokines that affect HSCs is under investigation.

Collagen Types I, III, V, and VI

Collagen types I, III, IV, and VI have been identified in LTC or in situ from marrow sections by a number of methods.^35,262 Most of the marrow-derived collagen types are assembled into long fibrils, which form the fine, background reticulin staining seen on marrow biopsies, although type IV collagen is assembled into a meshwork seen most commonly as part of basement membranes. Collagens also interact with laminins in the marrow. Collagen types I and VI are strong adhesive substrates for various hematopoietic cell lines and marrow mononuclear cells, including committed myeloid and erythroid progenitors.²⁶² Classic collagen receptors on blood cells are of two types, the β₁ integrins (α₁β₁ and α₂β₁) and the nonintegrin glycoprotein VI, present predominantly on platelets.

THE AGING MARROW MICROENVIRONMENT

Like the HSC itself (see “Stem Cell Aging” earlier), the HSC niche undergoes several changes with aging. Although the number of osteoblasts decreases, they generate higher levels of ROS, inducing p38 MAPK signaling, potentially accounting for the reduction in self-renewal capacity of HSCs derived from older mammals. The number of adipocytes increases as a result of the skewed differentiation of aged MSCs; increased adiposity and reduced osteogenesis lead to decreased CXCL12 levels in the aged marrow. This finding could be responsible for altered HSC mobilization in elderly individuals. In contrast, increased levels of the CC-chemokine ligand 5 (CCL5; also known as RANTES [regulated upon activation, normal T-cell expressed and secreted]), in the niche could contribute to the altered myeloid/lymphoid skewing seen in HSCs of older individuals.²⁶³ This topic has been reviewed but clearly requires additional study.¹³⁸

CONTROVERSIES IN HEMATOPOIESIS

LINEAGE FATE DETERMINATION

One of the most contentious issues in hematopoiesis is the origin of stem cell commitment to specific blood cell lineages. Two schools of thought exist: extrinsic and intrinsic control. The former, championed by Metcalf and others,²⁶⁴ argues that cytokines, ECM, or other stimuli instruct the hematopoietic stem or progenitor cell to differentiate into specific cell types. In contrast, Dexter and others²⁶⁵ argue that a hierarchy of transcription factors direct a cell toward a specific lineage, mechanistically explained by a stochastic rise in one or more of a mutually antagonistic set of transcription factors, that drive developmental pathways by enhancing expression of the genes that characterize that pathway, and by interfering with the levels or function of the transcription factors that drive the alternate lineage fate choice.

The Case for Transcription Factors

A strong case has been made for intrinsic control of stem cell lineage determination.²⁶⁵ As Enver and colleagues state: “Simply put, the question is this: Is unilineage commitment the result of a cell-autonomous, internally driven program, or rather is it the consequence of a cell responding to an external, environmentally imposed agenda?” These and several other investigators argue that the stochastic rise in one or another lineage determining transcription factor in the multilineage progenitor leads to its ultimate lineage commitment.

It is abundantly clear that transcription factors can direct lineage commitment in hematopoietic cells. A partial list of transcription factors restricted to specific hematopoietic lineages includes Pax5 (B cells),²⁶⁶ Ikaros (B/T cells),²⁶⁷ PU.1 and C/EBPα (myeloid and B cells),^268,269 GATA1 (erythrocytes and MKs),^132,270 Fli1 (MKs),²⁷¹ and C/EBPε (granulocytes).²⁷² A number of loss-of-function studies have revealed the nonredundant role of these proteins in development of the corresponding cell lineage. For example, genetic elimination of Pax5 eliminates B cells²⁷³; elimination of Ikaros leaves a mouse devoid of fetal T cells, fetal and adult B cells, and their progenitors¹²⁴; and loss of C/EBPα leads to absolute neutropenia.²⁷⁴ Moreover, the exogenous expression of several transcription factors in lineage committed progenitor cells can redirect cell fate. For example, C/EBPα is expressed in myeloid progenitor cells, and introduction of a regulatable C/EBPα gene into purified erythroid progenitors causes their switch to the myeloid lineage.²⁷⁵ In further support of this hypothesis, several lines of evidence have been gathered, including the finding that forced expression of the antiapoptotic gene bc1₂ in a growth factor-dependent multipotential hematopoietic cell line resulted in growth factor independence and spontaneous differentiation into all of the possible cell lineages that develop when the corresponding growth factor(s) are added to the wild-type cells.²⁷⁶

In addition to providing these and other arguments in favor of a transcription factor–based intrinsic regulatory mechanism of stem cell fate, proponents of the intrinsic hypothesis point to feed-forward switch-like molecular mechanisms in which a stochastic increase in one of a binary set of such transcription factors reduces the level or activity of those transcription factors responsible for alternate cell fates. An example of this physiology is illustrated by the mutually antagonistic effects of the erythroid transcription factor GATA1 and the myeloid transcription factor PU.1; GATA1 acts to inhibit the myeloid activation potential of PU.1,²⁷⁷ and PU.1 blocks the binding of GATA1 to its genetic target sites.²⁷⁸ Thus, when the level of GATA1 stochastically rises above that of PU.1 in a CMP, the granulocyte-macrophage potential would be extinguished and the MEP potential of the cell would march forward, unfettered. Alternately, CMPs in which PU.1 levels rise above that of GATA1 would develop along the myeloid lineages, both through the direct stimulation of myeloid gene expression by PU.1, and indirectly by the blockade of GATA1-mediated erythroid and megakaryocytic gene expression programs.

The Case for Humoral Mediators

Although much evidence has been garnered in favor of an intrinsic mechanism of stem and progenitor cell fate determination, proponents of an extrinsic instructive hypothesis have also generated a large amount of compelling evidence in favor of the importance of extrinsic signals. One illustrative example of the capacity of certain extrinsic signals to impact specific patterns of differentiation is that the exogenous expression of an IL-2Rβ transgene in CLPs induces their differentiation into myeloid cells.²⁷⁹ Subsequent studies revealed that the presence of the exogenous receptor leads to upregulation of the GM-CSF-R in the CLP, and that exogenous expression of GM-CSF-R could also lead a CLP toward monocyte/macrophage development.² In separate studies, other cytokines were shown to direct myeloid lineage fate determination; compared to the differentiation profile seen when marrow cells were cultured with SCF alone, an antiapoptotic stimulus, the addition of IL-5 greatly enhanced the number of marrow progenitor cells that gave rise to eosinophilic colonies, whereas the addition of TPO induced a predominance of megakaryocytic colonies, without significant changes in the number of apoptotic cells in any of the three culture conditions. These results were interpreted to indicate that while the SCF could keep nearly all progenitor cells alive under the cell culture conditions employed, the second cytokine directed the multilineage progenitors into specific cell fates.²⁸⁰

A number of external signaling events have been found to directly impact the transcriptional apparatus of the cell. For example, as previously noted, two transcription factors that lead to the self-renewal and expansion of HSCs, HOXB4, and HOXA9 are induced to higher levels of expression or to translocate into the nucleus of stem cells in response to TPO.^235,236 Moreover, SCL, a transcription factor that when expressed in maturing hematopoietic cells inhibits cytokine-induced granulocytic and monocytic differentiation, maintaining them in an undifferentiated state, is enhanced by SCF and down-modulated by GM-CSF.¹³¹ And the level of c-Myb, which determines whether a MEP develops an erythroid or MK fate is affected by TPO, mediated by its induction of miR150.²⁸¹ Thus, strong evidence supporting both extrinsic and intrinsic control of lineage determination has been presented, and like the case for most conflicts in biology, it is most likely that elements of both mechanisms operate in hematopoiesis.

Like nearly every other cell type, the HSC is subjected to a number of noxious stimuli, and must possess mechanism to survive such insults. Moreover, the regulation of HSC levels is carefully regulated, such as seen following transplantation, through both cell expansion and programmed cell death. The mechanisms that drive HSC survival are beginning to be determined.

A number of cytokines affect cell survival by regulating inhibitors of apoptosis, such as the Bcl proteins. In addition, in response to sublethal DNA damage, cytokines can also act to trigger DNA repair mechanisms. For example, TPO acts to increase the efficiency of DNA-protein kinase-dependent nonhomologous end-joining in response to irradiation or chemotherapy, helping to repair double-stranded DNA breaks.²⁸²

The ability to divide symmetrically to generate identical daughters is a feature of most cells, including HSCs. However, the multipotent stem cell possesses an added ability to undergo asymmetric cell divisions, yielding one committed progenitor daughter and one stem cell daughter, or two differentiating progeny; regulating the balance between symmetric and asymmetric stem cell divisions becomes critical in maintaining proper HSC numbers and in meeting the demand for differentiated cells. The beginnings of the molecular origins of asymmetric division in the HSC are under study.²⁸³

A question related to the previous discussion of whether intrinsic or extrinsic factors determine HSC lineage fate, is whether intrinsic or extrinsic factors determine the possible outcomes for a dividing HSC (two HSC progeny [stem cell expansion], one HSC and one differentiating cell [a self-renewal division], or two differentiating progeny). It is clear that feedback mechanisms exist that govern the size of the stem cell pool, as following myeloablation and transplantation of a limited number of HSCs, the pool expands toward that seen in a normal individual but not beyond, even when subjected to forced overexpression of genes that enhance HSC expansion.²⁸⁴ HSCs do not appear to have a limit on their capacity for expansion; experiments using serial transplantation of marrow cells revealed that even after four such maneuvers the transplantation of a limiting number of HSCs was associated with a 10-fold expansion in the recipient,²⁸⁴ a level of expansion remarkably consistent from one serial transplant to the next. Thus, there does not appear to be an intrinsic limit on HSC expansion that sets the size of the stem cell pool. Rather, evidence from quantitative transplants suggests that there exist both intrinsic and extrinsic controls on the size of the stem cell pool.

Following transplantation the degree to which a limiting number of transplanted HSCs expand depends on the source of the cells; fetal liver cells expand to a far greater degree than a similar number of adult marrow-derived HSCs, indicating that an intrinsic mechanism governs stem cell expansion divisions. However, evidence for an extrinsic mechanism that regulates stem cell expansion also exists, as the transplantation of a smaller number of either fetal liver or adult marrow HSCs resulted in slower marrow recovery but ultimately greater levels of HSC expansion than did infusion of larger numbers of cells. These results were interpreted to suggest that the more rapid recovery of marrow function associated with the administration of a larger marrow inoculum, with its increased numbers of stem cells, prematurely shut down HSC expansion, calling attention to an extrinsic regulatory mechanism. Moreover, the differences in expansion capacity amongst fetal and adult stem cells might also reflect the influence of extrinsic factors. When adult human marrow cells are transplanted, they retain their stem cell capacity only if quiescent at the time of transfer. In contrast, fetal liver and cord blood stem cells contribute to long-term hematopoiesis regardless of the phase of the cell cycle in which they reside at the time of harvest.²⁸⁶ It is postulated that this latter property of fetal stem cells depends on the fetal hematopoietic microenvironment, making it likely that extrinsic factors play the key role in the decision to self-renew or differentiate. Strong experimental evidence has been generated indicating that stem cell numbers in an individual are governed by the number of hematopoietic niches.²⁸⁷ This data strongly suggests that, at the very least, the availability of stem cell niches places an upper limit upon HSC expansion.

Clues from the developmental biology of lower organisms may also shed important insights into the mechanisms that regulate the decision between symmetric and asymmetric HSC divisions.²⁸⁸ Within the niche of developing Drosophila gonadal tissue exist hierarchies of cells. When female gonadal stem cells divide, the cell directly contacting the niche supportive cells remains a stem cell, the daughter that loses contact differentiates and initiates oogenesis. A similar niche architecture also sets the stage for gonadal stem cell retention in the fly testis and in many tissues of many organisms. The developing principle is that a stem cell in contact with the stem cell determining niche stromal cell, or residing in a region of the niche possessing the highest concentration of a stem cell–determining soluble factor, will remain a stem cell, and those removed from contact or soluble factor will differentiate. In such a niche, the axis of stem cell division then determines cell fate; if the axis of cell division is parallel to the front of stem cell–determining contact or soluble mediator gradient, the proximal cell will remain a stem cell while the distal cell differentiates; if the axis of cell division is perpendicular, both cells will remain under the influence of the “stemness” factor(s), and remain stem cells. Consequently, spindle-polarizing signals could be responsible for the fate of the daughters of stem cell division, a focus of much research, but at present, few established mechanisms.

STEM CELL PLASTICITY

A remarkable observation has been repeatedly made in patients who underwent sex-mismatched (male into female) marrow transplantation, suffered subsequent organ damage, and were carefully studied until the time of their death. In such settings, Y chromosome-bearing cells were identified at the site of repair of previous myocardial infarctions, strokes, and other organ damage. These observations suggest that hematopoietic cells can contribute to the replacement of damaged cells of multiple organs. More direct experimentation has lent additional support to this idea; several investigators have found that marrow cells are capable of giving rise to cells of multiple organs, including nerve,^289,290 liver,^291,292 skeletal muscle,²⁹³ and cardiac muscle,²⁹⁴ in a process termed transdifferentiation. However, direct evidence establishing this conclusion is lacking, as most such studies have assayed only partially purified cell populations that might also contain alternate types of stem cells,²⁹⁵ and almost none have been performed using single cells, a requirement for robust proof of their multipotency. An alternate explanation for the presence of marked hematopoietic cells at nonhematopoietic sites of organ damage has been termed cell fusion. It has long been appreciated that marrow cells (especially macrophages) can fuse with other cells, and spontaneous in vitro fusion of embryonic stem cells with marrow-derived cells yields hybrids that display stem cell function^296,297; further experimentation is required to prove or disprove the concept of HSC plasticity,³⁰⁰ a proof that will have far reaching implications for regenerative medicine.

The loss of one or more developmental potentials of the HSC results in a progenitor committed to any number of specific hematopoietic cell lineages. Besides the loss of pluripotency, committed hematopoietic progenitors display a number of characteristics that differ from their parents, including the lack of capacity for self-renewal, a higher fraction of cells traversing the cell cycle, reduced ability to efflux foreign substances, and a change in their surface protein profile. On the genetic level, the transition of HSCs to committed progenitors is marked by the downregulation of a large number of HSC-associated genes and progressive upregulation of a limited number of lineage-specific genes. This section highlights some of the features of specific lineage-committed progenitors that allow for their purification, characterization, and, potentially, their manipulation for therapeutic benefit. Details of the morphologic, biochemical, and genetic aspects of the differentiation of each of these progenitors is found in the chapters corresponding to their mature blood cell types.

PROGENITOR CELL ASSAYS

Assays for most hematopoietic progenitor cells consist of marrow or (occasionally) blood cells, either unfractionated or purified to varying degrees, a semisolid support (either methylcellulose or agar, which prevents cellular migration), and a source of hematopoietic growth factors. The cultures are incubated in a humidified environment at 37°C for 2 to 7 days for murine cells, or 5 to 14 days for human cells, during which time the vast majority of the cells that began culture as mature blood cells die, allowing the few hematopoietic progenitors present to proliferate and differentiate into mature blood cells. As the cells in such culture systems are immobilized by the semisolid supporting matrix, all of the progeny in the resultant colonies are derived from a single progenitor, allowing one to retrospectively determine the developmental capacity of that cell, termed a colony-forming cell (CFC) or unit (CFU). The requirement for a source of hematopoietic growth factors was initially fulfilled by using cellular underlayers containing fibroblasts, lymphocytes, or monocytes, or tissue culture medium conditioned by a variety of normal and neoplastic cellular sources, but essentially all the requisite growth factors are now available in purified recombinant form. Despite substantial progress in our understanding of the developmental requirements of committed hematopoietic progenitors, we still do not have an adequate in vitro colony-forming assay for some well-characterized hematopoietic progenitor cells (e.g., those committed to the T lymphocytic or natural killer [NK] cell lineages) that still require more complex assays (e.g., fetal thymus explant assay).

CHARACTERISTICS OF SPECIFIC PROGENITOR CELL TYPES

Lymphoid Progenitors

Common Lymphoid Progenitors

The existence of a population of cells committed to all lymphoid lineages but devoid of myeloid capacity was theorized to exist based on a number of analyses. For example, patients with adenosine deaminase deficiency, or mice with genetic elimination of the γ_C receptor, the signaling kinase JAK3, or the transcription factor Ikaros, lack T and B lymphocytes and have few, if any, myeloid defects, arguing that the defects in these disorders might affect a CLP; however, this does not prove the existence of a cell common for all. Work using cell sorting for CD10+/CD34+/Thy−/c-Kit−/Lin− human marrow cells revealed the capacity to develop into T, B, NK, and lymphoid dendritic cell progenitors,²⁹⁸ but the report did not demonstrate a common progenitor capable of giving rise to each lineage on a clonal level.

More recent work, based upon the importance of IL-7 for all single-lineage lymphoid progenitor cells, and on the severe lymphopenia seen when the gene was eliminated in mice,²⁹⁹ has indicated that the IL-7R marks a CLP that can be used in flow cytometry to isolate a population of IL7R+/Lin−/Thy−/Sca^lo/Kit^lo cells that engrafts all of lymphopoiesis but no myelopoiesis in congenic mice.³⁷ For example, the injection of 2000 such CD45.1+ cells plus 1 × 10⁵ whole-marrow CD45.2+ cells into lethally irradiated CD45.2 recipients lead to 3 to 20 percent CD45.1+ B and T lymphocytes, which disappear after approximately 6 months. In contrast, this strategy never results in the appearance of CD45.1+ myeloid cells. When limiting dilution studies were performed, approximately 1 in 20 such cells could give rise to short-term B-lymphopoiesis when injected intravenously, and an equal number could give rise to T-lymphopoiesis when injected into the thymus. In colony-forming assays using IL-7, SCF, and FL, approximately 20 percent of such cells gave rise to pre-B and pro-B cell colonies in vitro. Thus, given the low likelihood of proper homing when injected into mice, it is almost certain that the CLP exists and is IL7R+/Lin−/Thy−/Sca^lo/Kit^lo. When a genetic expression analysis was performed comparing HSCs to CLPs, the latter demonstrated a down-modulation of many molecules associated with HSCs, such as the cell-surface receptors c-Mpl, β₁-integrin, and Tie2, and the transcription factors HOXA9 and EGR1, and upregulation of the IL-7R and the recombination activating protein recombination activating gene (RAG) 2.³⁰⁰

T Lymphocyte/NK Cell, T Lymphocyte, and NK Cell Progenitors

Simple colony-forming assays for mixed T/NK cell progenitors have not been developed, but the existence of the bipotent progenitor can be inferred from studies in which CD44+CD25−FcγRII/III− fetal thymic cells are cultured with genetically marked, deoxyguanosine treated fetal thymic lobes under 70 percent oxygen at 37°C. Without cytokine supplementation, such cultures yield primarily CD3+/Thy1+ T cells, but if IL-2 plus IL-15 are added, the NK cell potential (CD3−/NK1.1+) of these cells is realized, and if IL-7 is also added, the number of single cells that yield cells of both lineages increases significantly.³⁰¹ As this type of readout is possible from single day 12 fetal thymus cells, such studies establish that bipotent T/NK cell progenitors exist.

E box–binding proteins consisting of HEB, E2–2, and the E2A gene products E12 and E47 form a distinct subgroup within the large family of basic helix-loop-helix transcription factors.³⁰² Heterodimers or homodimers form between family members through their helix-loop-helix (HLH) region, and through their basic regions bind to canonical E-box DNA sequences and thereby affect gene expression, including T-cell targets such as CD4³⁰³ and the pre-Tα,³⁰⁴ and assist in recombination of the γδ T-cell receptors.³⁰⁵ It is now clear that E2A is required for the transition from the bipotent T/NK progenitor to committed T-cell progenitors as its genetic elimination leads to preservation of the former but elimination of the latter.³⁰¹

Another important subgroup of HLH proteins is the Id family, which contain an HLH region but lack a DNA binding domain, thereby acting as a sink for functional HLH proteins and thus negatively regulating the function of E proteins.³⁰⁶ Id proteins appear to be essential for NK cell development, as genetic elimination of Id2 leads to a profound loss of NK cell progenitors³⁰⁷ and forced overexpression of Id3 leads to a shift of T/NK cells preferentially into the NK cell lineage.³⁰⁸

It is also clear that Notch activation plays a vital role in T-cell lineage commitment from the CLP. Overexpression of active Notch1 directs marrow stem cells into immature CD4+/CD8+ T cells and inhibits B-lymphocyte development.³⁰⁹ Overexpression of Notch1 in RAG-deficient precursors also results in differentiation to the T-cell lineage, although only to the immature CD4−/CD8− stage, indicating that Notch cannot substitute for pre–T-cell-receptor signaling.³¹⁰

B-Lymphocyte Progenitors

B-cell progenitors include the pro-B cell, the earliest cell irreversibly committed to the lineage, which are CD34+/CD10+/CD38+/CD19+/CD20+, the pre-B cell, which displays the initial stages of immunoglobulin rearrangement, expresses immunoglobulin heavy chains in their cytoplasm and are CD34−/CD10+/CD19+/CD20+/CD38−, and immature B cells which begin immunoglobulin light-chain production, express cell-surface IgM, and are CD10+/CD19+/CD20+. Pro-B cells can be detected in a simple colony-forming assay.³¹¹ Normally, B-cell precursor development occurs in contact with the hematopoietic microenvironment, mediated by precursor cell integrin β₁β₁ and stromal cell VCAM or matrix FN. A number of cytokines affect B-cell progenitor proliferation,³¹² including IL-7,³¹³ insulin-like growth factor (IGF)-1,³¹⁴ SDF-1,³¹⁵ and SCF,³¹⁶ although based on genetic knockout studies, B cells are absolutely dependent only on IL-7²⁹⁹ and SDF-1.²²

A number of cytokines also inhibit B-cell precursor development, including interferon (IFN) α/β,³¹⁷ IFN-γ,³¹⁸ IL-4,³¹⁹ and TGF-β.³²⁰ The role of these and other inhibitory cytokines in B lymphopoiesis is complex, as in some situations a cytokine can inhibit one stage and stimulate another stage of development, and some might act indirectly.

A number of transcription factors are required for mature B cell function, including PU.1, nuclear factor-κB (NF-κB), early B-cell factor (EBF), interferon regulatory factor 4 (IRF4), and Oct2, many of which bind to the promoters and enhancers involved in immunoglobulin gene expression. In contrast to these relatively later stage effects, E2A is required for commitment to the lineage.³²¹ The marrow of E2A-deficient mice is devoid of CD19 B cells, as well as most B-cell lineage-specific genes, including RAG1/2, Pax5, EBF, and VpreB. Moreover, no immunoglobulin rearrangement is detectable, and there are no IL-7 responsive cells. Reintroduction of E2A into the marrow cells of null mice reconstitutes pre–B-cell development.³²²

E2A sits upon a hierarchy of B-cell lineage-specific genes and transcription factors³²³; E2A directly regulates the expression of RAG1, λ5, D-J_H, V-Jκ, and the transcription factor EBF, the latter, in turn, regulating VpreB, mb-1, D-J_H, V-Jκ, and the transcription factor Pax5, which, in turn, regulates CD19 and LEF1 and shuts down genes associated with alternate lineages, such as M-CSF-R (monocytic), myeloperoxidase (neutrophilic), GATA1 (MEP), and pTα (T lymphocytic). As noted earlier in the section “Notch Ligands”, a critical condition for B-cell commitment is the absence of Notch signaling.

Myeloid Progenitors

Common Myeloid Progenitors

Flow cytometry has also been extensively used to purify myeloid progenitors; an IL-7R−/Lin−/c-Kit+/Sca-1− population of murine marrow cells, which by virtue of being Sca1− excludes HSCs, develop into all myeloid lineages.³⁶ Based on expression of CD34 and the Fcγ RII/III, three distinct subpopulations can be identified by further flow cytometry, IL-7Rα−/Lin−/c-Kit+/Sca-1−/CD34+/FcRγ^lo, IL-7Rα−/Lin−/c-Kit+/Sca-1−/CD34−/FcRγ^lo, and IL-7Rα−/Lin−/c-Kit+/Sca-1−/CD34+/FcRγ^hi. When tested in colony-forming assays in the presence of SCF, FL, IL-11, IL-3, GM-CSF, EPO, and TPO, each cell population yielded distinct mature cell types.³⁶ IL-7Rα−/Lin−/c-Kit+/Sca-1−/CD34+/FcRγ^lo cells give rise to all myeloid colony types, including CFU-Mix, BFU-E, CFU-megakaryocyte (CFU-MK), MEP, CFU-GM, CFU-granulocyte (CFU-G), and CFU-macrophage (CFU-M), consistent with that expected for the CMP. In contrast, IL-7Rα−/Lin−/c-Kit+/Sca-1−/CD34+/FcRγ^hi cells form only CFU-M−, CFU-G−, and CFU-GM–derived colonies in response to any of the growth factors, alone or in combination, and thus represent granulocyte/macrophage lineage-restricted progenitors (GMP). Finally, IL-7Rα−/Lin−/c-Kit+/Sca-1−/CD34−/FcRγ^lo cells form only BFU-E−, CFU-MK−, and mixed megakaryocyte-erythroid colonies, leading to their designation as MEP. To demonstrate their capacity to differentiate in vivo, limiting numbers of each cell population were transplanted into congenic mice; in such studies, cell fate outcomes correspond strictly with those of the in vitro colony assays. For example, 6 days after injection of 5000 CMPs, both donor-derived Gr-1+/Mac-1+ myelomonocytic cells and TER119+ erythroid cells were detectable in recipients. In contrast, when 5000 GMPs were transplanted, only Gr-1+/Mac-1+ cells were recovered, and only for a transient period of time. Likewise, MEPs reconstituted only TER119+ cells in similar experiments, and the genetically marked progeny from each of these progenitor populations disappear within 4 weeks of transplantation, indicating their limited self-renewal capacity.

Erythroid/Megakaryopoietic, Erythroid, and Megakaryopoietic Progenitors

Culture conditions necessary for in vitro erythropoiesis have been known for nearly 35 years,^324,325 with colony morphologies ranging from small compact clusters of 20 to 50 erythrocytes developing with 2 to 5 days in murine and human marrow plasma clot cultures (CFU-E), to large highly complex colonies containing up to thousands of cells taking from 7 to 14 days to develop in methylcellulose or agar (BFU-E). The cytokine requirement for the former is simple—EPO, whereas a cytokine that stimulates earlier cells, such as IL-3 or SCF, is required for the latter progenitor cell type.

Culture conditions that support the proliferation of MK progenitors have been established for both mouse and humans.^29,31 Using either methylcellulose, agar, or a plasma clot, two colony morphologies that contain exclusively MKs have been described. The CFU-MK is a cell that develops into a simple colony containing from 3 to 50 mature MKs, larger, more complex colonies that include satellite collections of MKs and contain up to several hundred cells are derived from the burst-forming unit–megakaryocyte (BFU-MK). Because of the difference in their proliferative potential and by analogy to erythroid progenitors, BFU-MK and CFU-MK are thought to represent primitive and mature progenitors restricted to the MK lineage. And like their erythroid counterparts, the cytokine requirements for CFU-MK are simple: TPO stimulates the growth of 75 percent of all CFU-MK, with IL-3 being required along with TPO for the remainder,⁷⁷ whereas IL-3 or SCF is required alone with TPO for more complex, larger MK colony formation from their more primitive progenitors.

Progenitors for erythrocytes and MKs display many common features: they share a number of transcription factors (SCL, GATA1, GATA2, NF-E2), cell-surface molecules (TER119), and cytokine receptors (for IL-3, SCF, EPO, and TPO), and most erythroid and MK leukemia cell lines display, or can be induced to display, features of the alternate lineage.³²⁶ Moreover, the cytokines most responsible for development of these two lineages—EPO and TPO—are the two most closely related proteins in the hematopoietic cytokine family,¹⁴⁸ and display synergy in stimulating the growth of progenitors of both lineages.⁷⁷ For these and other reasons it has been postulated that erythropoiesis and megakaryopoiesis share a common progenitor cell,³²⁷ a hypothesis now established³⁶ with the identification of IL-7Rα−/Lin−/c-Kit+/Sca-1−/CD34−/FcRγ^lo cells.

Like other primitive hematopoietic cells, bipotent MEP progenitors resemble small lymphocytes but can be distinguished by a specific pattern of cell-surface protein display. As noted above, MEPs are IL-7Rα−/Lin−/c-Kit+/Sca-1−/CD34−/FcRγ^lo. Cells committed to the MK lineage then begin to express CD41 and CD61 (integrin αIIbβ₃), CD42 (glycoprotein Ib), and glycoprotein V. Those that are committed to the erythroid lineage begin to express CD41 and the transferrin receptor (CD71), and as they mature lose CD41 expression but express the thrombospondin receptor (CD36), glycophorin, and, ultimately, globin.³²⁸ These and other cell-surface markers provide experimental hematologists several strategies to purify committed MK^39,329 and erythroid³³⁰ progenitors. Another useful method to identify megakaryoblasts is histochemical staining for von Willebrand factor, and in rodents, acetylcholinesterase.³³¹

The transcription factors expressed by erythroid and MK progenitors that allow for their commitment to the lineage are becoming increasingly well understood. GATA1 is an X-linked gene encoding a 50-kDa polypeptide that contains two zinc fingers required for DNA binding.²⁷⁰ Genetic elimination of the transcription factor established the critical role of this transcription factor in hematopoiesis; GATA1 −/− mice are embryonic lethal as a consequence of failure of erythropoiesis,³³² and MK-specific elimination of GATA1 leads to severe thrombocytopenia as a consequence of dysmegakaryopoiesis.³³³ GATA1 acts in concert with another protein that affects transcription without binding to DNA, friend of GATA (FOG).³³⁴ The importance of this interaction to megakaryopoiesis is clear: several different mutations of the site on GATA1 responsible for FOG binding lead to congenital thrombocytopenia.³³⁵

The Ets family of transcription factors includes about 30 members that bind to a purine box sequence, proteins that interact in both positive and antagonistic ways. For example, PU.1, initially termed Spi-1 based on its association with spleen focus-forming virus-induced erythroleukemias, blocks erythroid differentiation, although it appears important for MK development.³³⁶ Moreover, the Ets factor Fli-1 is essential for megakaryopoiesis³³⁷ and mutations in the transcription factor are also associated with congenital thrombocytopenia in man.²⁷¹

Granulocyte/Monocytic, Granulocyte, and Monocytic Progenitors

As noted above, GMPs (CFU-GM) are IL-7Rα−/Lin−/c-Kit+/Sca-1−/CD34+/FcRγ^hi, reflecting their beginning differentiation toward phagocytic cells (i.e., FcRγ-positive). In the human, GMP are CD34+/CD33+/CD13+ markers, which are of clinical significance. For example, CD33 is also termed Siglec-2, a member of a family of sialic-acid-binding surface membrane proteins of the immunoglobulin superfamily that are involved in cell–cell interactions and signaling. Although the role of CD33 is not yet known with certainty, it has become a therapeutic target because of its high-level expression on the blasts of several forms of acute myelogenous leukemia³³⁸; the use of gemtuzumab ozogamicin, in which a humanized anti-CD33 monoclonal antibody has been fused to N-acetyl-gamma calicheamicin 1,2-dimethyl hydrazine dichloride, a potent antitumor antibiotic, has resulted in a complete remission rate of 15 to 20 percent as a single agent in patients with relapsed disease.³³⁹ These initial successes have prompted its testing in earlier stage disease along with other active agents.³⁴⁰ When marrow grafts are purged of CD33-bearing cells durable engraftment occurs, but is often quite delayed, indicating that CD33 is not present on the HSC, but that the presence of GMPs in a transplantation product is vital for rapid engraftment.

CD13 is also termed aminopeptidase N, an ectopeptidase present in many organs other than the marrow, a member of a family of proteases that play an important role in cell growth by virtue of their cleavage of biologically important peptides, in some cases inactivating, and in some cases activating them, and by serving a cell adhesive function as well. CD13 is present on early hematopoietic cells, including myeloid and lymphoid lineage progenitors, but disappears from the latter class of cells and its expression rises as monocytes mature. Although it functions to scavenge peptides in the intestinal brush border and degrade endorphins and enkephalins in the synaptic cleft, its role in hematopoiesis is less clear, although IL-8 is a substrate of its proteolytic activity.

Once bipotent GMPs differentiate, they further restrict their developmental potential. Monocytic progenitors are characterized by a predominance of PU.1, whereas granulocytic cells by members of the C/EBP family—C/EBPα and C/EBPε—are vital for the expression of neutrophil and eosinophil granule proteins.^272,341 A recent study suggests that the developmental decision of a bipotent GMP into each of the two lineages might be mediated by alterations in the relative levels of PU.1 and C/EBP expression³⁴²; haploinsufficiency of PU.1 (PU.1+/−) results in a reduction in CFU-M frequency in the marrow and an increase in CFU-G levels, even ameliorating the neutropenia seen in G-CSF null mice. Moreover, by increasing expression of C/EBPα, a transcription factor that drives granulocytic differentiation, G-CSF further influences the choice between the granulocytic and monocytic lineages. However, it is also clear that PU.1 plays an important role in both lineages, and it is likely that additional investigation will yield new insights into the molecular mechanisms that establish the ordered process we term myelopoiesis.