CHAPTER 2: Hematopoiesis

In Chapter 1, we introduced the morphologically recognizable marrow progenitors and their progeny, the red cells, white cells, and platelets that make up the formed elements of the blood. Here we turn to the topic of hematopoiesis (from the Greek "to make blood"), the process by which these elements are created. As we will discuss, hematopoiesis depends on rare, morphologically inconspicuous progenitor cells that have remarkable functional properties. Unraveling how hematopoiesis is maintained and regulated has provided a paradigm for understanding the biology of tissue stem cells and has resulted in the development of effective therapies for certain cancers and genetic disorders as well as a variety of conditions in which bone marrow output is inadequate to maintain normal blood cell counts.

The developmental origins of hematopoietic cells are complex and incompletely understood. Cells with the properties of hematopoietic stem cells (HSCs; described in the following section) arise several times in different tissues during prenatal development, producing successive waves of hematopoiesis (Fig. 2-1). Hematopoiesis first appears around day 16 of gestation in the embryonic yolk sac; at this site, it is limited to the production of red cells, which are needed for oxygen transport in the newly developed circulatory system. Hematopoietic cells arise anew around 3 to 4 weeks of gestation in a portion of the ventral mesoderm referred to as the aorta-gonad-mesonephros region. HSCs derived from this region (and possibly the yolk sac as well) are believed to migrate through the blood and take up residence in the liver, which becomes a hematopoietic organ at around 6 weeks of gestation and serves as the major source of hematopoietic cells throughout much of fetal development. HSCs and some hematopoietic progenitors also appear by around 6 weeks of gestation in the placenta and umbilical cord blood and persist in these sites until birth. By 7 to 8 weeks of gestation, lymphoid progenitors derived from the liver begin to seed the newly developed thymus, which is the major site of T-lymphocyte development. Around 5 months of age, HSCs derived from the liver (and possibly other sites) home to the bone marrow, which becomes the dominant source of hematopoietic elements by birth, around the time that hepatic hematopoiesis ceases.

Under normal circumstances, hematopoiesis is confined to the marrow throughout postnatal life, but under stress (e.g., severe inherited anemias) it can persist or reappear in the liver and spread to other extramedullary sites, such as the spleen and the lymph nodes. Hematopoietic neoplasms such as chronic myelogenous leukemia also tend to involve the same organs, sometimes producing massive hepatosplenomegaly and more modest lymphadenopathy.

Hematopoiesis is a remarkably dynamic process. The marrow normally produces approximately 200 billion red cells, 100 billion platelets, and 60 billion neutrophils each day, numbers that are just sufficient to match the rate of peripheral destruction of these formed elements. Hematopoiesis must persist throughout life and be finely tuned and highly responsive to changes in peripheral blood counts, because both cytopenias (too few formed elements) and cytoses (too many formed elements) can have serious or fatal consequences.

The ability of the hematopoietic system to meet these homeostatic challenges involves a hierarchy of progenitor cells with distinct roles (Fig. 2-2). At the top of this hierarchy lies the HSC. HSCs are essential for the maintenance of hematopoiesis and are by definition multipotent, meaning they can give rise to all other hematopoietic cells. When HSCs divide, under homeostatic conditions, it is thought that at least one of the two daughter cells remains an HSC, a critical property referred to as self-renewal that maintains HSC numbers. Some HSC divisions may be symmetric, such that both daughter cells either become HSCs or begin to differentiate into progenitors (Fig. 2-3). Symmetric divisions that give rise to two HSCs may occur in the fetal liver, a stage of development during which HSC numbers increase. Symmetric divisions in which both daughter cells begin to differentiate into an early hematopoietic progenitor, a process termed commitment, may occur following periods of hematopoietic stress. Other HSC divisions are asymmetric, such that one cell remains an HSC, and the second commits to differentiation. Asymmetric cell divisions are dominant in the bone marrow, in which the number of HSCs remains fairly constant. Details still remain to be worked out, but it appears that the potential of very early committed progenitors is first restricted to myeloid (red cell, granulocyte, and megakaryocyte) or lymphoid (B cell, T cell, and natural killer cell) differentiation. With subsequent divisions, the differentiation potential of progenitors is further refined, so that it is ultimately restricted to a single cell type.

Bone marrow HSCs normally spend most of the time in a resting state referred to as quiescence, only "awakening" to divide, at most, every few months. Quiescence may help to maintain the multipotent state and protect HSCs against the acquisition of mutations that could lead to transformation and cancer development. Under conditions of increased hematopoietic demand, however, HSCs in the marrow divide more frequently and are more likely to divide symmetrically, expanding their numbers. In extreme circumstances, substantial numbers of HSCs and early progenitors may leave the marrow and migrate to the liver, spleen, and lymph nodes, where they can produce extramedullary hematopoiesis.

Given that HSCs have the ability to circulate in the blood, why is postnatal hematopoiesis normally confined to the marrow? One important contributing factor is active homing of HSCs. Circulating HSCs adhere to and migrate across bone marrow capillaries at sites where there are high levels of the chemokine CXCL12. Upon entering the interstitium, the HSCs nestle between blood vessels and osteoblasts (cells that line trabecular bone), which create a specialized microenvironment, or niche. These "niche" cells release factors (such as CXCL12) that regulate HSC behavior in ways that are not yet completely understood. HSCs are resistant to stimulation by hematopoietic growth factors (described in the following section), possibly because niche factors actively promote quiescence. It may be that HSCs expand only when growth factors increase and quiescence factors decrease concomitantly. Another idea posits that HSC numbers are regulated by niche "vacancy," such that HSCs expand their numbers only when open niches are available. Under conditions of severe hematopoietic stress, it is possible that secondary niches appear in sites of extramedullary hematopoiesis, such as the liver.

Because HSCs are rare cells that are morphologically indistinguishable from lymphocytes, special means must be used to identify them. HSCs express particular surface markers such as CD34 and actively pump out certain dyes, properties that can be used to identify populations of cells that are enriched for HSCs. However, proof that viable HSCs are present in a sample requires functional testing. If a cell preparation can completely reconstitute long-term hematopoiesis when transfused into a host that has had its own marrow cells destroyed (e.g., by high doses of radiation), then it must contain HSCs, which can be quantified by serial dilution of the sample. Although this procedure, termed stem cell transplantation, was originally developed (and is still widely used) for experimental purposes, it was rapidly adopted as a means to treat a variety of diseases (described later in this chapter). It remains the only form of stem cell therapy that is widely used in clinical practice.

Myelopoiesis (the production of myeloid elements—red cells, granulocytes, and platelets) is regulated at the level of myeloid progenitors by hematopoietic growth factors. As differentiation proceeds, myeloid progenitors lose multipotency and the capacity for self-renewal, but in turn acquire two other key properties: 1) an increased capacity for cell division, and 2) surface expression of specific receptors for hematopoietic growth factors. By controlling the growth and survival of committed myeloid progenitors, growth factors regulate the production of red cells, granulocytes, and platelets from the marrow. Some growth factors, such as stem cell factor (also known as c-KIT ligand) and interleukin (IL)-3, have growth- and survival-promoting effects on multiple types of progenitors, whereas other factors, such as erythropoietin and thrombopoietin, have effects that are restricted to progenitors committed to a single line of differentiation (Fig. 2-4).

The marrow can increase its production of myeloid elements as much as 10-fold in response to growth factor stimulation. Three lineage-restricted growth factors are particularly important regulators of myeloid element production.

• Erythropoietin (Epo), a growth factor secreted by cells primarily located in the interstitium of the kidney, is a critical growth factor for early erythroid progenitors. Epo expression in the kidney is regulated by oxygen tension through the transcription factor hypoxia-inducible factor (HIF), described further in Chapter 3, such that when delivery of oxygen to tissues falls, HIF activity and Epo production rise (Fig. 2-5).

• Thrombopoietin (Tpo) is an important growth factor for megakaryocytes, the large polyploid marrow cells that shed platelets into the circulation. Tpo is secreted constitutively by hepatic parenchymal and endothelial cells as well as by bone marrow stromal cells. Unlike that of Epo, the activity of Tpo is not regulated by changes in its expression but rather through a competition between platelets and megakaryocyte progenitors for Tpo (Fig. 2-6). Platelets and marrow progenitors both express Tpo receptors, such that as platelet levels fall, the level of free Tpo available to stimulate megakaryocytic precursors in the marrow rises. Thus, platelet production is regulated by the total body platelet mass. This relationship explains why the sequestration of platelets in enlarged spleens fails to stimulate increased platelet production; these platelets still bind Tpo, keeping free Tpo levels relatively low even in the face of thrombocytopenia in the peripheral blood.

• Granulocyte colony stimulating factor (G-CSF) is an important growth factor for neutrophil progenitors. Like Tpo, G-CSF is secreted constitutively by several different cell types, including macrophages, endothelial cells, and fibroblasts. The production of G-CSF increases over basal levels in response to inflammatory cytokines such as IL-1 and tumor necrosis factor. By stimulating granulocytic progenitors in the marrow, G-CSF can markedly increase neutrophil production.

Epo, Tpo, and G-CSF are glycoproteins that bind to specific members of the cytokine receptor family. When activated by growth factors, the Epo, Tpo, and G-CSF receptors all trigger signaling through several downstream pathways, including the JAK-STAT, Ras, and AKT pathways (Fig. 2-7). These pathways have effects on metabolism and gene expression that enhance the growth and survival of committed progenitor cells, the proliferating precursors that ultimately give rise to the formed myeloid elements. Following growth factor stimulation it generally takes about 10 days for the expanded pool of progenitors to mature, resulting in a lag between the elevation of growth factor levels and the appearance of increased numbers of newly formed elements in the peripheral blood.

Regulation of lymphocyte production (lymphopoiesis) is more complicated than myelopoiesis, because it occurs at both the level of lymphocyte progenitors and mature lymphocytes. In the marrow, early progenitors with lymphoid potential expand under the influence of growth factors such as IL-7. These cells may differentiate into B cells, T cells, or natural killer cells, as follows:

• Some lymphoid progenitors remain in the marrow and differentiate into B cells (Fig. 2-8). This process involves sequential productive rearrangements of the immunoglobulin (Ig)H heavy chain gene followed by the kappa or lambda Ig light chain genes, which permits naïve B cells to express IgM and IgD on their cell surfaces. Upon emerging from the marrow, naïve mature B cells home to peripheral lymphoid tissues, where they can persist in a resting state for long periods. When stimulated by antigen, some of these cells proliferate and differentiate directly into IgM-secreting plasma cells or IgM-expressing memory B cells. However, most antigen-stimulated B cells migrate into germinal centers, a specialized niche found in peripheral lymphoid tissues such as lymph nodes. Many of these germinal center B cells die, but those that make high affinity antibodies survive and undergo class-switching, a process that allows B cells to express other types of immunoglobulin. These cells then leave the germinal center and become either long-lived memory cells or terminally differentiate into plasma cells.

• Other marrow-derived lymphoid progenitors migrate to the thymus. Here, these cells differentiate into T progenitors and rearrange their T-cell receptor genes (Fig. 2-9). The only cells expressing T-cell receptors that survive are those that bind antigen with low affinity in the context of major histocompatibility complex class I (also known as human leukocyte antigen [HLA] class I) or class II (HLA class II) antigens; this process is referred to as positive selection. Those that recognize self-antigens with high affinity die by apoptosis (negative selection), which helps to prevent the development of autoimmunity, whereas cells that fail to rearrange their T-cell receptors productively or make receptors that fail to bind antigen die by neglect. Most cells emerging from the thymus express αβ T-cell receptors on their surfaces as well as the co-receptors CD4 or CD8. These cells circulate in the blood and home to lymphoid tissues throughout the body. When exposed to antigen in the proper cellular context, T cells become activated, enlarge, and begin to divide. Like B cells, activated T cells can either undergo terminal differentiation into antigen-specific effector cells (either CD4-positive Th1 or Th2 T-helper cells or CD8-positive cytotoxic T cells) or become long-lived memory T cells. A smaller number of progenitors in the thymus express a different type of T-cell receptor composed of γ and δ chains. These γδ T cells fail to express CD4 or CD8 and tend to home to the skin and the gut.

• The third type of lymphocyte is the natural killer cell. These cells also arise from lymphoid progenitors in the marrow, where their production is promoted by the cytokine IL-15, and subsequently migrate to peripheral lymphoid tissues. They do not express either antibodies or T-cell receptors on their surfaces but can be activated by certain cytokines or exposure to cells that have been infected by viruses or have abnormal patterns of surface antigen expression, such as cancer cells. Like CD8-positive cytotoxic T cells, natural killer cells have azurophilic granules containing enzymes that are capable of killing target cells.

Once an early progenitor commits to differentiate, what determines which kind of cell it becomes? Two models have been proposed. One supposes that factors produced in the microenvironment instruct progenitors to differentiate along certain lines. In some instances, this appears to be true. For example, lymphoid progenitors exposed to factors that activate the Notch pathway become T cells, whereas, in the absence of Notch signals, these progenitors mainly become B cells instead. The other model supposes that progenitors randomly (stochastically) become competent to adopt particular fates and that hematopoietic growth factors act on this pool of cells to control their growth and survival. This model appears to hold for myeloid progenitors, which (as described earlier) are regulated by hematopoietic growth factors.

Both models require that progenitors turn on the expression of a set of genes that allows hematopoietic differentiation to proceed. Experimental work with "knockout" mice has shown that certain transcription factors, proteins that associate with DNA and control gene expression, are critically important in directing differentiation along particular lines (Fig. 2-10). For example, the loss of the transcription factor PAX5 specifically blocks B-cell development, while leaving other lineages intact. Similarly, loss of Notch1, a unique type of receptor that also acts directly as a transcription factor, selectively blocks T-cell development, while mutations in C/EBPα block granulocytopoiesis. Other factors are required in early hematopoietic progenitors, and, as a result, their loss causes a complete failure of hematopoiesis. MLL is an example of one such factor. On the other end of the developmental spectrum, some factors have no role in early progenitors or in lineage determination but are instead required for the further differentiation of mature cells. For example, loss of the transcription factor BCL6 prevents antigen-stimulated B lymphocytes from maturing into germinal center B cells. Thus, hematopoiesis is orchestrated by a complex interplay between extrinsic factors (hematopoietic growth factors and local factors produced in the microenvironment) and factors intrinsic to hematopoietic progenitors (growth factor receptors and hematopoietic transcription factors).

One important group of disorders in which these regulatory mechanisms go awry is the various kinds of hematologic malignancies, cancers of hematopoietic cells. These cancers are commonly associated with acquired mutations that alter the function of the same transcription factors that control differentiation. In fact, mutations in particular transcription factors tend to be found in tumors composed of cells that correspond to the stage in development at which the affected transcription factor normally acts. For example, PAX5 mutations are found in tumors composed of early B-cell progenitors, whereas BCL6 mutations are confined to tumors derived from germinal center B cells. In general, cancer-specific mutations in transcription factors interfere with differentiation, holding cells in an immature state. In addition to transcription factor mutations, mutations in one or another component of the signaling pathways that are normally activated by growth factors are often found in hematopoietic cancers. These mutations typically stimulate signaling even in the absence of growth factors, permitting tumors to proliferate in a growth factor–independent fashion. These themes are expanded upon in later chapters describing the hematopoietic neoplasms.

Stem cell transplantation (SCT) is described in detail in Chapter 26; we present only a brief overview here. The ability of HSCs to home to the bone marrow niche and completely reconstitute hematopoiesis makes it feasible to deliver these cells by simple transfusion into the peripheral blood. Tissue sources of HSCs used in SCT include bone marrow, the umbilical cord blood of newborns (a rich source of HSCs), and the peripheral blood. SCT may be autologous (in which case the HSCs are from the same patient) or allogeneic (in which case the HSCs are obtained from another individual). The source of the HSCs and the type of SCT that is performed vary according to the clinical indication.

A variety of disorders are treated with SCT, including:

• Inherited blood disorders that qualitatively or quantitatively impair the function of HSCs or their progeny.

• Acquired bone marrow failure (aplastic anemia).

• Certain forms of cancer, particularly those of hematopoietic origin.

In patients with inherited or acquired conditions that lead to defective or deficient hematopoiesis, SCT serves to supply normal HSCs that can produce adequate numbers of fully functional formed elements. In the setting of cancer, SCT is used to "rescue" patients from bone marrow aplasia induced by very high doses of radiation or chemotherapy, which are needed to kill certain types of cancer cells.

Except for identical twins, the recipients of allogeneic SCT are reconstituted with cells that are genetically distinct from the host. In this situation, the transplanted HSCs will be recognized as foreign and rejected by the patient's immune system unless a "conditioning" regimen consisting of radiation and/or chemotherapy is given. Conditioning serves two purposes: 1) it suppresses the recipient's immune system, helping to prevent rejection of the transplanted cells, and 2) it destroys or displaces the recipient's HSCs, creating vacancies in the marrow niche for the transplanted HSCs. The successful reconstitution of allogeneic SCT recipients with HSCs from another individual has several important immunologic consequences. On the one hand, because lymphocytes derived from the transplanted HSCs recognize the recipient's cells as foreign, recipients will develop potentially fatal graft-versus-host disease unless immunosuppressive drugs are given. On the other hand, when allogeneic SCT is performed in patients with cancer, donor lymphocytes derived from the transplanted HSCs also recognize host cancer cells as foreign, producing a beneficial graft-versus-tumor effect. The means by which clinicians attempt to walk the fine line between the beneficial and deleterious immunologic effects of allogeneic SCT are described in Chapter 26.

An attractive but as yet unfulfilled use of SCT is as a means to deliver "gene therapy" to individuals with inherited hematopoietic disorders. In principle, it should be possible to repair genetic defects (for example, a defective β-globin gene encoding sickle hemoglobin) in stem cells in the laboratory and then reconstitute the patient with "corrected" HSCs through autologous SCT. Recent advances in reprogramming of somatic cells into stem cells with the capacity for hematopoiesis provide a reason for optimism about the long-term prospects of this approach.

Growth factors are used clinically in a variety of settings to enhance the marrow output of formed myeloid elements.

• Epo is the most widely used growth factor. It is most effective when given to treat anemias associated with inappropriately low levels of Epo. These include the anemia of renal failure, in which the production of Epo is diminished by damage to the kidney parenchyma, and the anemia of chronic inflammation, in which inflammatory cytokines suppress the production of Epo. Anemia of inflammation is common in patients with certain inflammatory disorders and various forms of cancer (Chapter 7). Epo is also used with varied success in patients with hematopoietic neoplasms, such as myelodysplastic syndromes (Chapter 20), that are associated with ineffective hematopoiesis.

• G-CSF is primarily used in the treatment of drug induced neutropenia, particularly following high-dose chemotherapy, which causes a transient marrow aplasia that lasts roughly 21 days. G-CSF lessens the period of neutropenia, thereby decreasing the frequency of fevers related to infection. Granulocyte-macrophage colony stimulating factor, another growth factor that enhances the production of both granulocytes and macrophages, is also used in this setting and appears to have similar efficacy to that of G-CSF. G-CSF is also used in patients with congenital disorders associated with low neutrophil counts (neutropenia), as described in Chapter 18.

• Tpo and other factors that stimulate the production of megakaryocytes and shedding of platelets have proven to be beneficial in patients with immune thrombocytopenia (Chapter 14) but are not effective in patients with low platelet counts (thrombocytopenia) following chemotherapy or in those with primary disorders of the marrow that impede platelet production.