Chapter 74: Lymphopoiesis

Blood is formed from a succession of sites during embryonic and fetal development, beginning outside the embryo in the yolk sac. Soon afterward, hematopoiesis begins in the embryo proper, initially in the para-aortic splanchnopleura (PAS) and aorto-gonad-mesonephros (AGM) regions, then the fetal liver, spleen, and finally the fetal marrow (Chap. 7). With each change of anatomical site, the range of hematopoiesis becomes progressively more complex and similar to that of the adult (Fig. 74–1).

When assigning hematopoietic function to each developmental stage, it is important to distinguish the lineage “potential” of stem and progenitor cells that arise from certain areas (i.e., the ability to generate specific lineages in vitro from immature cells removed from a region) from the spontaneous physiologic production of lineages in each region. With this distinction in mind, the onset of lymphopoiesis during embryogenesis lags behind development of the myeloid and erythroid lineages. Although myeloid, erythroid, and natural killer (NK) cells can be produced from all extraembryonic and embryonic sites, B and T lymphocytes are predominantly generated from so-called definitive hematopoietic stem cells (HSCs) in the embryo proper.¹

MURINE HEMATOPOIETIC DEVELOPMENT

Most of the studies exploring embryonic and fetal hematopoiesis have been performed using mouse models. Although the timing of each developmental stage has been carefully mapped, it has long been a source of controversy as to whether hematopoiesis in the embryo is initiated from colonizing precursors from the extraembryonic yolk sac, or whether the embryonic sites of hematopoiesis arise independently from the yolk sac.^2,3,4,5,6 This debate has implications for understanding the lineages generated at different sites of hematopoiesis and thus for tracing the ancestry of the lymphoid cells that are produced in the mammalian embryo. One reason for the difficulty in assigning the exact organ in which lineages are generated, is that each site of hematopoiesis is active during overlapping periods (see Fig. 74–1). In addition, once circulation has been established, it is difficult to rule out the possibility that stem cells and progenitors found in one location did not migrate from another. However studies using Ncx1^–/– mice, which lack both heartbeat and circulation,^7,8 are beginning to dissect the autonomous lineage potentials of these distinct embryonic hematopoietic tissues.

The first wave of hematopoiesis in the mouse begins in the extraembryonic tissue of the yolk sac by 7.5 days of gestation (E7.5), before circulation is established.^9,10 This initial stage of so-called primitive hematopoiesis produces mostly erythrocytes and macrophages. Although lymphocytes are not detectable at this time,¹⁰ the contribution of first-wave progenitors to downstream fetal lymphopoiesis has been suggested by the identification of a lymphomyeloid progenitor in the E9.5 yolk sac, which expresses Rag-1, one of the earliest lymphoid-specific events,¹¹ as well as of a distinct progenitor with B-1/marginal zone B cell potential.¹² Further studies have identified both thymic-repopulating and multipotent potential in the yolk sac,^13,14 indicating emerging changes to our understanding of primitive hematopoiesis. The murine placenta has also been identified as an autonomous source of multipotent hematopoietic cells as early as E8.5⁸; however, the direct contribution of either yolk sac or placental progenitors to definitive lymphoid development remains to be determined. Definitive HSCs that are capable of generating all lymphohematopoietic lineages first appear in the PAS/AGM region at E8.5 to E9.^3,10 High-level, multilineage reconstituting activity typical of definitive HSC can be found in the murine AGM region by E10.5. However, although AGM cells can produce all lineages, including T and B lymphocytes in vitro, lymphocytes do not spontaneously develop in the fetus until hematopoiesis has begun in the fetal liver. Rag-1 expression, one of the earliest lymphoid-specific events, can be found in the E11 murine fetal liver.¹⁰ T-cell potential has been identified in the yolk sac and PAS as early as E8.25 to E9.5 of murine gestation¹³; however, T-cell differentiation in vivo begins with the colonization of the thymus around E11 by stem or progenitor cells that migrate to the thymus from the AGM, fetal liver, and, later still, the fetal marrow.^15,16

HUMAN HEMATOPOIETIC DEVELOPMENT

Hematopoietic cells have been identified in the human yolk sac as early as day 18 of embryonic life, at which time, like the mouse, they are almost exclusively comprised of erythrocytes and, to a lesser extent, monocytes and macrophages (see Fig. 74–1).¹⁷ Although no lymphocytes are seen in the yolk sac, yolk sac progenitors do have NK cell potential under certain in vitro conditions.^17,18 The same yolk sac progenitors, however, do not possess the capacity for B- or T-cell development, even when placed in culture conditions that permit lymphoid differentiation.¹⁸

As in the mouse, definitive hematopoiesis develops first in the AGM region derived from the splanchnopleura, as evidenced by the finding that the AGM is the site where CD34+ cells with the capacity for full lymphoid and myeloid differentiation are first found in the human embryo.^18,19 The AGM develops at day 27 of gestation in the human, when human HSCs are generated as clusters of two or three cells arising from the endothelium specifically on the ventral surface of the preumbilical region of the aorta. These cells are clonogenic and highly proliferative, rapidly increasing to several thousand in number and spreading further along the aorta. However, hematopoiesis exists only transiently in the AGM, disappearing entirely by day 40.¹⁷ Although lymphoid cells can be produced in culture from cells extracted from the AGM^17,18 the HSC of the AGM do not produce mature cells in situ; instead their role is to migrate and colonize the fetal liver, producing the next wave of hematopoiesis. As in the mouse, recent studies have also detected multipotent progenitors with lymphoid potential in the human placenta. These reports suggest that hematopoietic stem/progenitor cells may also be generated de novo from the large vessels of the chorionic plate as early as week 5 of gestation, with multipotent HSC detectable at 15 weeks.^20,21 However, the contribution of placental HSC to fetal liver or marrow colonization remains unclear.

Although blood cells are first detectable in the human fetal liver as early as day 23, they exist at this time only as erythroid and myeloid cells associated with hepatic sinusoids. These erythroid cells consist of megaloblasts expressing embryonic hemoglobins (globin chains ζ and ε), and no CD34+ cells are seen in the fetal liver during this early phase. It is likely that this first stage of fetal liver hematopoiesis is secondary to colonization of more mature cells from the yolk sac. By day 30, AGM-derived CD34+ cells appear in the fetal liver,²² and by day 32, these cells are able to maintain long-term hematopoiesis in vitro.²² Erythroid cells in the fetal liver at this later stage of definitive hematopoiesis consist of enucleated macrocytes producing fetal hemoglobin (globin chains α and γ). In normal development, as with the yolk sac and AGM, hematopoiesis in the fetal liver is transient, disappearing by 20 weeks of gestation.¹

The final wave of hematopoietic development takes place in the fetal marrow, starting around 11 weeks of gestation. The initial cells seen in the marrow are CD15+ myeloid cells and glycophorin A+ erythroid cells, and hematopoiesis is again associated with the endothelium, taking place in the medullary sinusoids before osteoblast formation.²³ Eventually, CD34+ cells are found in the fetal marrow and behave functionally as true HSC, generating B, T, NK, and myeloid and erythroid lineages.¹ HSC have found their final niche, and lifelong, self-renewing lymphohematopoiesis resides permanently in the marrow thereafter.

THYMIC DEVELOPMENT

The human thymic microenvironment begins to develop at approximately 4 weeks’ gestation and then undergoes three developmental phases.²⁴ The first phase occurs between 4 and 8 weeks’ gestation, with the appearance of thymic epithelium arising as a product of the third and fourth pharyngeal pouches²⁵ and the expansion of thymic epithelial cells. The second phase occurs between 9 and 15 weeks’ gestation and is characterized by the development of subcapsular, cortical, and medullary regions.²⁴ Thymic colonization by fetal liver-derived progenitors and lymphocyte production begins at approximately week 9.²⁵ The ability of thymocytes to respond to the mitogen phytohemagglutinin is detectable as early as 10 weeks’ gestation,²⁶ and alloreactive, phenotypically mature T cells can be found by 13 to 16 weeks’ gestation.²⁷

The third phase occurs from 16 weeks’ gestation until age 1 to 2 years and is characterized by robust intrathymic T-cell maturation (Chaps. 6 and 76).

An exhaustive study of 136 human postnatal thymuses ranging from neonatal life to more than 90 years old, found that essentially all postnatal thymic growth (based on weight and volume) occurs during the first postnatal year, mostly in the first few months of life.²⁸ From the age of 12 months, the human thymus undergoes steady involution, with a reduction of thymocytes and thymic epithelium, particularly the medulla, and a corresponding increase in fatty infiltration of the perivascular space.²⁸

Whereas mice lose approximately 90 percent of the wet weight of the thymus during life, the increasing fatty infiltration in the perivascular space in the aging human thymus maintains the total size of the thymus in healthy people into late life.²⁸ Radiologic studies using computed tomography (CT) scans have confirmed that total thymic size remains stable in humans throughout life although parenchymal tissue atrophies dramatically (~95 percent; Chaps. 6 and 9).²⁹

B-CELL DEVELOPMENT

The hallmark characteristic of a mature B cell circulating in blood or residing in secondary lymphoid tissue is the expression of cell-surface immunoglobulin (Ig). The cell-surface Ig consists of μ, δ, γ, α, or ε heavy chains disulfide-linked to κ or λ light chains (Chap. 75). The cell-surface Ig and the associated signaling molecules Igα (CD79a) and Igβ (CD79b) are referred to as the B-cell receptor (BCR). Progenitor (pro-) B cells are defined by the absence of both cytoplasmic μ heavy chains and cell-surface BCR. Precursor (pre-) B cells are defined by the presence of cytoplasmic μ heavy chains in the absence of cell-surface BCR. This minimal definition of pro-B, pre-B, and B cells forms the basis of the current detailed model of human B-cell development.³⁰ B-cell development can be divided into two stages: an antigen-independent stage that occurs primarily in fetal liver and fetal and adult marrow, and an antigen-dependent stage that occurs primarily in secondary lymphoid tissue, such as spleen and lymph node.

The first B cells detectable in the human fetus are found in the fetal liver²⁵ at approximately 8 weeks’ gestation, with the appearance of cytoplasmic IgM+ pre-B cells; by 10 to 12 weeks, surface IgM+ B cells are seen in the fetal liver³¹ and fetal omentum.³² B-cell and IgM production move to the fetal marrow and spleen by 17 weeks of gestation (Chaps. 6, 7, and 75).^33,34 From the end of the second trimester throughout adult life, marrow is the exclusive origin of B-cell development.³⁵ The frequency of early B-lineage cells as a percentage of the total nucleated lymphohematopoietic cell pool is higher in fetal than in adult marrow. However, the ratio between pro-B, pre-B, and immature B cells and the mitotic activity within these fractions is relatively constant.³⁶

In murine B-cell development, two functionally and immunophenotypically distinct types of B cells, B-1 and B-2, have been described.³¹ Most B cells in adult mice are B-2 cells, which form part of the adaptive immune system by their ability to interact with T cells and undergo immunoglobulin heavy chain class switching. The B-1 cells make up approximately 5 percent of adult murine lymphocytes, but demonstrate a far less diverse immunoglobulin repertoire than the B-2 cells, responding to carbohydrate antigens and other T-cell–independent immunogens and forming part of the innate immune system. Murine B-1 cells are marked by their expression of CD11b, and are found in multiple sites, including the spleen, intestine, and the pleural and peritoneal cavities.^37,38 The B-1 cells can be further divided into B-1a cells (which secrete immunoglobulins spontaneously) and B-1b cells (in which immunoglobulin production is induced) based on the expression of the marker CD5. The demonstration of a B-1a/marginal zone B-cell–restricted progenitor in the E9 extraembryonic yolk sac suggests the B-1 lineage may emerge prior to the development of the adaptive immune system.¹² At present, there is no clear evidence that humans have similar subpopulations of B-1 and B-2 cells during development.³¹

NATURAL KILLER CELL DEVELOPMENT

Functional NK cells can be detected in the human fetal liver as early as 9 to 10 weeks of gestation,²⁶ but NK cell differentiation can be induced in vitro from progenitors derived at all stages of hematopoietic development, even those from the yolk sac.^1,17,18 Thus, the onset of NK potential is not equivalent to the full lymphoid potential of definitive hematopoiesis, as NK potential can be assigned to a range of progenitor types that exist at different stages, including primitive hematopoiesis. NK cell production can be considered as providing an essential defense mechanism for the developing mammalian embryo prior to development of more complex pathways of adaptive immunity.

DENDRITIC CELL DEVELOPMENT

Dendritic-like cells which express class II major histocompatibility (MHC) antigens are produced at all stages of embryonic and fetal hematopoiesis, being first detected in the human yolk sac and mesenchyme as early as 4 to 8 weeks, before development of the fetal marrow or thymus.³¹ Dendritic cells (DCs) are detectable at each site of hematopoiesis as soon as they become active, in the human fetal thymus at 11 to 14 weeks, marrow at 14 to 17 weeks, spleen at 16 weeks, and tonsils at 23 weeks.^39,40 DC and macrophages are closely related and phenotypically similar (Chap. 67), both expressing MHC class II. Thus, clear discrimination of these two cell types can be difficult, particularly as many studies that provided information on human DC development were conducted before all the molecular and antibody tools for analysis of DC were available.

The conceptual framework for how the lymphocyte lineages are generated from HSC was developed largely from studies using genetically engineered mice and murine transplant models. Although necessary and useful as a starting point, caution should be exercised in translating the results of the murine studies to human lymphopoiesis, or in assuming for any species that only one pathway to lymphopoiesis exists at all stages of ontogeny.^41,42 Additionally, the conclusions about lineage relationships of isolated populations are influenced by limitations inherent in any of the in vitro or in vivo assays employed to examine differentiation potential.⁴³

For several decades, our understanding of hematopoiesis has been built on a hierarchical schema in which all the pathways of differentiation lead away from a multipotent HSC, and progress through discrete progenitor stages that mark each branch-point of lineage commitment (Chap. 18). In the classical paradigm, the earliest differentiation “decision” made by an HSC is to enter one of two pathways, marked by either a common lymphoid progenitor (CLP) or a common myeloid progenitor (CMP), which has full myeloid and erythromegakaryocytic differentiation potential (Fig. 74–2).⁴⁴ With each successive stage of differentiation, lineage-specific cell-surface markers and transcription factors are upregulated, and alternative lineage potentials are lost. Consequently, the CLP is defined as a single cell that can give rise to all lymphoid lineages (B, T, and NK), but cannot generate myeloid, erythroid, or megakaryocytic lineages. The concept of progenitor populations marking two mutually exclusive differentiation pathways, one limited to myeloid and erythromegakaryocytic potential and the second defining lymphoid commitment, was held long before cells that satisfied the criteria of CLP were identified.^45,46,47 In contrast, the existence of single clonogenic cells with myeloid, erythroid, and megakaryocyte lineages was shown more than three decades ago through the use of in vitro clonal assays to demonstrate so-called colony-forming unit-granulocyte erythromyeloid megakaryocyte,⁴⁸ and later confirmed using markers to prospectively isolate such cells in mice⁴⁹ and humans.⁵⁰ The lymphocyte lineages were assumed to be closely related because of a number of associations, for example, common anatomical sites of T and B lymphopoiesis (spleen, lymph nodes; Chap. 6), similar molecular mechanisms that regulate T-cell receptor and B-cell immunoglobulin rearrangements (Chaps. 75 and 76), and severe B- and T-lymphoid defects that result from single genetic mutations in mice (Chap. 80).⁵¹

Development of flow cytometry (synonym: fluorescence-activated cell sorting [FACS]) made possible the isolation of rare hematopoietic cell populations and the subsequent interrogation of lineage potential using in vitro cultures and in vivo reconstitution studies (Chap. 18). Primitive multilymphoid progenitors with little or no clonogenic myeloid or erythroid potential have now been isolated from human tissue using flow cytometry with combinations of various cell-surface markers.^52,53,54 However, it seems likely that lineage relationships are less-strictly organized than once believed. Studies in mice show that the erythroid and megakaryocytic lineages can branch off at an earlier point in hematopoiesis, and that lymphoid (i.e., T, B, and NK) and myeloid (or at least monocytic) lineages can arise from the same pathway through a so-called lymphoid-primed multipotent progenitor (LMPP).⁵⁵ It remains unclear which of these lineage differentiation pathways are most physiologically significant during steady-state hematopoiesis, but it is likely that more than one pathway can exist simultaneously and alternative pathways may predominate during different stages of ontogeny and from different sites of hematopoiesis.

MURINE LYMPHOID PROGENITORS

In 1997, investigators working with murine marrow cells, isolated progenitors that possessed no myeloid or erythromegakaryocytic potential, but when transplanted into irradiated recipients could rapidly restore T-, B-, and NK cell lineages.⁵⁶ Clonal in vitro and in vivo studies showed that all lymphoid lineages were derived from a single common progenitor, thus proving the existence of the long-assumed CLP and supporting the classical model of lymphopoiesis. This study isolated cells based in part on expression of interleukin-7 receptor alpha (IL-7Rα).⁵⁶ The IL-7Rα+ CLP do not express hematopoietic markers associated with fully differentiated lineages (they are called “lineage negative” or “lin^neg” cells). As an indication that they are more differentiated than multilineage HSCs, expression of certain HSC-related cell-surface markers (Sca-1, Thy-1, c-kit) is downregulated.⁵⁶ Thus the full immunophenotype assigned to the murine CLP is Lin^neg IL-7Rα+ Thy-1^neg Sca-1^lo c-kit^lo. This contrasts the murine CLP immunophenotype with that of the murine HSC, which is found within the Lin^neg IL-7Rα^neg Thy-1^lo Sca-1^hi c-kit^hi population.⁵⁶

Work with murine marrow has prompted a reexamination of when the lymphoid lineage pathways diverge from those of the myeloid and erythroid lineages (see Fig. 74–2). A population from murine marrow cells, defined largely by expression of the receptor FLT3 (Fms-like tyrosine kinase 3), has been shown to possess full lymphoid and some myeloid potential, but not erythroid or megakaryocytic potential.^55,57 These lin^negsca-1+c-kit+CD34+FLT3^hi (synonym: LSK CD34+FLT3^hi) cells are primed for lymphoid commitment in that they have downregulated genes involved in erythromegakaryocytic differentiation and upregulated lymphoid-associated genes.⁵⁸ They have thus been termed LMPPs.^55,58 Although they are able to generate monocytes and granulocytes in vitro, their differentiation potential is nonetheless skewed heavily to lymphoid cells; after transplantation into irradiated recipients, LSK CD34+FLT3^hi cells rapidly reconstitute B and T lymphopoiesis. However, unlike the multipotent LSK CD34+FLT3^neg cells, reconstitution of myeloid lineages in vivo from LSK CD34+FLT3^hi is very limited, and importantly lacks granulocytic potential.⁵⁵ Indeed, subsequent in vivo fate-mapping studies have questioned the physiologic importance of a lymphomyeloid differentiation pathway to steady-state myeloid development, both in the marrow and thymus.^59,60

One confusing factor in defining lineage potential is the ability of both myeloid and lymphoid progenitor populations to differentiate into DCs, at least in vitro (Chap. 21).^61,62,63 As in vitro-derived DC express many cell-surface markers common to myeloid antigen-presenting cells, irrespective of the lineage of origin, the extent to which identification of in vitro myeloid potential is confounded by a DC program is unclear.

HUMAN LYMPHOID PROGENITORS

The CD34 cell-surface marker is expressed on human HSCs and on a variety of different types of hematopoietic progenitors, including those restricted to lymphoid development (Chap. 18).^64,65 CD34 has been combined with additional cell-surface markers to identify human multilymphoid progenitors, for example, CD10,^52,66 CD7,^53,54,67,68 and CD45RA.^52,67 Similar to the mouse CLP, in addition to the upregulation of expression of the aforementioned markers, lymphoid commitment is accompanied by downregulation of certain HSC cell-surface markers, such as c-kit and Thy-1.⁴¹

As in murine studies, no one marker used in isolation is able to define a human lymphoid progenitor.⁶⁵ For example, although the expression of CD7 can be used to define a subset of CD34+lin^negCD38^neg cells in cord blood that are multilymphoid progenitors without myeloid or erythroid potential,^53,54 CD34+lin^negCD38+CD7+ cells from cord blood have full lineage (lymphoid, myeloid, and erythroid) potential. Furthermore, when comparing the progenitor populations identified in human studies with those described from murine experiments it is important to recognize that species differences exist between cell-surface markers.⁴¹ For example, IL-7Rα expression is used to define murine CLP,⁵⁶ but CD34+lin^negCD38^negCD7+ multilymphoid progenitors in human cord blood do not express IL-7Rα,⁵³ and CD34+lin^negCD38+IL-7Rα+ cells in human cord blood have both myeloid, lymphoid and even some erythroid potential. A different ontogeny and source of hematopoietic cells will also introduce unexpected variations of progenitor immunophenotype and function.⁴¹ Whereas most murine studies have been conducted with adult murine marrow, most human studies have been performed with umbilical cord blood, a more logistically available source containing progenitors that are significantly more proliferative than marrow.⁶⁹ Again using the example of the CD34+lin^negCD38^negCD7+ multilymphoid progenitor, although this immunophenotype can be used to identify multilymphoid progenitors in cord blood,⁵³ the same markers cannot be used in human marrow because CD34+lin^negCD38^neg marrow cells do not express CD7. Two candidates for the human equivalent of the murine LMPP have been described, both of which coexpress FLT3: one in the marrow identified as CD34+lin^negCD38+CD45RA+ and high expression of CD62L (L-selectin)⁷⁰; and another in the marrow and cord blood characterized as a CD34+lin^negCD38^negThy-1^lo/negCD45RA+ multilymphoid progenitor (MLP).⁷¹ The hierarchical relationship of these two populations to each other, or to the marrow CD34+lin^negCD38+CD45RA+CD10+ CLP is unclear; and as in the murine system, the physiologic contributions of a lymphomyeloid developmental pathway to steady-state hematopoiesis has yet to be determined. Adding to the concept of possible independence of dendritic cell development, DC potential has been found in all the primitive human lymphoid progenitors reported.^{52,53,67,70,71,72}

THYMIC PROGENITORS

It was long assumed that lymphoid commitment in the marrow precedes thymic seeding and T-cell development. However, despite the clear existence of lymphoid-committed progenitors within the marrow, the dominant cell type that migrates from the marrow and seeds the thymus to initiate thymopoiesis is still a matter of controversy. As described above, a variety of marrow-derived lymphoid-restricted progenitors and LMPPs are each able to generate T cells in vitro and in vivo. However, careful examination of the thymus has revealed primitive progenitors that have not only lymphoid, but also myeloid and erythroid potential. Such rare cells have been identified in murine thymus, where they are referred to as early thymic progenitors (ETP),⁷³ and also in human thymus, where they have the phenotype CD34+lin^negCD1a^negCD7^neg.^74,75 The lineage potential of such cells as well as the sharing of many cell-surface markers and similar gene expression profile to HSCs, suggest strongly that HSCs or at least multipotent progenitors are able to seed the thymus directly without a preceding stage of lymphoid commitment in the marrow. Which of these alternative progenitor types are dominant in terms of their contribution to steady-state thymopoiesis is yet to be determined⁷⁶; however, it is likely that early thymic progenitor lineage potential is itself dynamic, based on colonization of the murine thymus with temporally distinct waves of both lymphoid-restricted and multipotent thymic-seeding progenitors during embryonic development.⁷⁷

CHALLENGES IN FUNCTIONAL CHARACTERIZATION OF LYMPHOID PROGENITORS

The accurate assignment of lineage potential to immunophenotypically defined progenitors requires clonal analysis. Although clonal assays for myelo-erythromegakaryocytic progenitors have existed for more than 30 years,⁴⁸ the ability to differentiate HSCs along lymphoid pathways has been relatively recent, particularly for human studies.⁴³ In vitro assays for human lymphoid potential became available when it was observed that selected murine stromal cell lines were capable of supporting B-cell, NK cell, and DC differentiation from primitive human HSCs.^78,79,80 T-cell differentiation systems are more complex, requiring an in vitro model that recapitulates the unique environment of the thymus. Originally this was only possible using the fetal thymic organ culture method, a system in which large numbers of murine or human progenitors are seeded into whole thymic lobes in so-called hanging drop cultures.⁸¹ A more efficient in vitro system for studying murine and human T-cell differentiation has been developed using a murine stromal monolayer that expresses the Notch ligand Delta-like 1 (“OP9-DL1 stroma”).⁸² However, none of the in vitro T-cell culture systems simultaneously support B-cell development, making proof of full T- and B-lymphoid potential at a clonal level technically problematic. In vivo transplantation of a single murine HSC can prove multilineage potential at a clonal level, but this also is technically difficult, especially when studying progenitor populations that are not self-renewing. In vivo studies with human cells are particularly challenging as they rely on xenogeneic transplant models with low engraftment efficiency.^41,43

CYTOKINES IN LYMPHOPOIESIS

The many cytokine pathways that regulate lymphoid development, differentiation, and function are too numerous and complex for a full description here. However, the cytokine receptors of the common gamma (γ_c) chain family should be mentioned particularly because of their biologic importance in lymphopoiesis and their clinical relevance in primary immune deficiency disease. The γ_c subunit is a signaling component of six different cytokine receptors, interleukin (IL)-2,⁸³ IL-4,^84,85 IL-7,^86,87 IL-9,⁸⁸ IL-15,⁸⁹ and IL-21,⁹⁰ all of which act on different stages and pathways involved in lymphopoiesis.^51,91,92 All six γ_c-dependent receptors are unique in their activation of the Janus kinase 3 (JAK3) tyrosine kinase, a molecule that directly interacts with γ_c to mediate signaling.⁹³ In addition to the γ_c subunit, each of these receptors are comprised of an α subunit through which specific ligands bind; IL-2R and IL-15R also share a common β subunit.⁵¹

Null mutations of γ_c result in severe combined immunodeficiency (SCID) syndromes in mice and humans. However differences in the specific lineages affected reveal important species differences in cytokine dependency.⁵¹ The most important of these differences is in the requirement for IL-7 signaling in human and murine B-cell development. Adult murine B-cell development has an absolute requirement for IL-7 to IL-7R interaction and subsequent downstream signaling involving the γ_c subunit of the IL-7R and JAK3.⁹⁴ In contrast, IL-7 is not essential for human B-cell development. X-linked SCID patients with mutations in the γ_c cytokine-receptor subunit exhibit profound thymic hypoplasia and an absence of NK cells but normal or elevated numbers of B cells.⁵¹ SCID patients with mutations in JAK3^95,96 or the IL-7R⁹⁷ also have normal numbers of blood B cells. Although B-cell numbers are normal, B-cell function in patients with γ_c-deficient SCID is not normal and patients are hypogammaglobulinemic, presumably partly as a result of the role of IL-4 in B-cell function and the absence of T-cell interactions in antibody production. These collective results indicate IL-7 is not essential for at least the numerically normal development of human B cells.

NK cells are absent in patients with γ_c-deficient and JAK3-deficient SCID, but are normal in IL-7Rα deficiency.^65,97,98 NK cells are also absent in mice deficient in IL-15,⁹⁹ IL-15Rα,¹⁰⁰ or IL-2Rβ (a subunit shared by IL-2R and IL-15R),¹⁰¹ demonstrating the essential role of IL-15, but not IL-7, in NK cell development. Although no null mutations for IL-15 or its receptor have been described in humans, a familial NK cell deficiency has been described in humans in which the response to IL-15 and IL-2 appears to be subnormal.¹⁰²

The production of both B and NK cells in patients with IL-7Rα deficiency, shows that in humans IL-7 is not required for the earliest stages of lymphoid commitment or growth of CLPs. This point is further supported with the finding that multilymphoid CD34+CD38^negCD7+ progenitors in human cord blood do not express IL-7Rα,⁵³ and that early lymphoid progenitor subsets are preserved in the marrow of γ_c and JAK3-deficient patients.¹⁰³ In contrast to B cells and NK cells, however, T-cell development is absolutely dependent on IL-7 in both mice and humans.⁹¹ In both species, mutations of any portion of the IL-7 signaling pathway, that is, γ_c, IL-7Rα, or JAK3, completely prevents T-cell development.⁵¹ IL-2, in contrast, although an important cytokine in proliferation and function of mature T cells, is not essential for thymopoiesis; mutations in IL-2,¹⁰⁴ IL-2Rα, or IL2Rβ¹⁰⁵ result in functional T-cell defects, but T cells are not absent.

TRANSCRIPTIONAL REGULATION IN LYMPHOPOIESIS

The hierarchical differentiation pathways that lead irreversibly to the diverse array of functionally specialized mature lymphocytes are regulated by groups of genes expressed and repressed in a complex, precisely orchestrated sequence. As with cytokine regulation, our understanding of which transcriptional factors control each stage of differentiation has been developed using a combination of gene expression analyses in isolated progenitors and precursors, and an examination of the functional consequences of genetic mutations in mice and humans. The review in this chapter focuses on genes that regulate the earliest commitment decisions in the production of lymphoid progenitors; regulation of later differentiation stages in each lineage is discussed in Chaps. 75 to 77, respectively.

The complex interplay between groups of genes involved in hematopoietic differentiation has been likened to a multidimensional network whose “regulatory space” is formed by a dynamic balance between certain transcriptional regulators.¹⁰⁶ Expression analysis of multiple genes in defined progenitor populations demonstrates levels of promiscuity at early stages of hematopoiesis, that preclude assignment of any unique gene expression pattern to each stage.^106,107,108 As differentiation proceeds, a more specific “genetic fingerprint” for each lineage develops.

Regulation of Early Lymphoid Commitment

Ikaros Although no single gene has been identified as a lymphoid-specific master regulator, several transcription factors have been shown to be essential for the early stages of lymphopoiesis. The gene Ikaros, which encodes a family of DNA-binding zinc finger proteins, was identified in murine knockout studies as essential for all fetal lymphopoiesis.^109,110 However, in the postnatal setting, the role of Ikaros is more complex and less specific. Adult Ikaros^null mice completely lack B cells, and although T cells are produced, their differentiation is abnormal.¹¹¹ A murine study has suggested that Ikaros is not required for the initial lymphomyeloid versus myeloerythroid commitment decision, and that not only lymphoid differentiation, but also certain fate choices in the myeloerythroid pathway are affected by Ikaros.¹¹² As the expression of two key lymphoid cytokine receptors, FLT3 and IL-7Rα, is dependent on Ikaros, and as these markers are used to isolate murine LMPP and CLP, respectively, it is still not completely clear at which exact lymphoid progenitor stage Ikaros exerts its effects.¹¹² In addition to lymphoid progenitors, Ikaros isoforms are also expressed in HSCs, and myeloid lineages in mice^{112,113,114,115,116} and humans.^116,117 Although Ikaros may act as a typical transcription factor in some settings, Ikaros also affects gene expression through its role in chromatin formation.¹¹⁸

Pu.1 The transcription factor PU.1 is essential for normal B- and T-lymphocyte development, but its effects are highly dose dependent. At high levels of PU.1, key myeloid regulatory genes are upregulated and macrophage differentiation is induced preferentially over lymphoid differentiation.¹¹⁹ Low-level expression of PU.1, however, is essential for lymphopoiesis.^120,121 Mice in which PU.1 is completely absent lack B cells and have abnormal fetal thymopoiesis. However, studies with mice in which PU.1 is deleted specifically in B-lineage cells show that PU.1 is not essential for B-cell differentiation beyond the pre-B stage.¹²² It is likely that the critical role for PU.1 in murine lymphopoiesis lies in its upregulation of expression of the receptor for IL-7, which as mentioned above, is a key cytokine in both B and T lymphopoiesis in mice.¹²⁰

E2A E2A (encoded by TCF3) generates two basic helix-loop-helix proteins, E12 and E47, through differential splicing.¹²³ Murine studies suggest that E2A is necessary for lymphoid priming of multipotent progenitors and that the E2A proteins prime expression of a number of lymphoid-associated genes.¹²⁴ There is a dose-dependent requirement for E2A expression in the development of LMPP and CLP.¹²⁴ Both B- and T-lineage commitment are severely reduced in the absence of E2A, but Ikaros and PU.1 expression are normal.^124,125,126 E2A affects B lymphopoiesis in part through upregulation of early B-cell factor (EBF)¹²⁷ and T lymphopoiesis through upregulation of expression and function of the key T-cell specification factor Notch 1.¹²⁸

Regulation of B-Cell Commitment

The transcription factors Ikaros, PU.1, E2A, EBF, and Pax5 are essential for normal B-cell differentiation. Mice that have functional deletions in any one of these genes have severely abnormal B-cell development; however, of these genes, only EBF and Pax5 are B-cell specific within the hematopoietic system.

Pax5 Pax5 is expressed specifically in B-lineage–committed progenitors and is required for normal expression of the B-lineage genes CD19 and CD79a.¹²¹ Pax5–/– mice are blocked at the pro-B cell stage, but express most early B-cell–related genes.¹²⁹ Although Pax5 can activate a small subset of B-lineage genes, its main function in B-cell differentiation appears to be the suppression of T-cell and myeloid transcriptional programs at the murine pro-B–cell stage, thus enforcing commitment to the B lineage.^121,129,130 Consistent with this role, PU.1, E2A, and EBF function earlier than Pax5 in B lymphopoiesis, and forced expression of Pax5 does not rescue the B-cell defect seen in EBF–/– mice or PU.1–/– mice.¹²¹

Ebf EBF (encoded by EBF1) is a helix-loop-helix zinc finger protein that activates a B-lineage transcriptional program, and induces B lymphoid in preference to myeloid development, in part by antagonizing the expression of genes encoding alternative lineages such as C/EBPα (CCAAT/enhancer binding protein), Id2, and PU.1,¹³¹ and, in part, by inducing Pax5 expression.¹²¹ Ebf1–/– lymphoid progenitor populations from mice lack the ability to generate B cells but retain the ability to generate T, NK, and myeloid cells.¹³¹ Overexpression of EBF in multipotent progenitors promotes B-cell production at the expense of myeloid differentiation.¹³¹ EBF and E2A function cooperatively in early B lymphopoiesis¹²⁴; however, overexpression of EBF can rescue B-cell differentiation in E2A-deficient mice, including activation of Pax5.¹³² Pax5 overexpression however cannot rescue the B-cell defect in EBF–/– mice,¹²¹ demonstrating a critical, Pax5-independent role of EBF in early B-cell fate decisions.

Regulation of T-Cell Commitment

Notch Upon arrival into the thymus, multipotent progenitors from the marrow become rapidly committed to the T- and NK-cell pathways. The most important environmental cue for T-cell commitment is delivered by the thymic epithelium in the form of the Notch ligands, Delta-like 1 (DLL1) and Delta-like 4 (DLL4).¹³³ Binding of one of these ligands to the Notch 1 receptor expressed on the surface of thymocyte precursors causes activation of intracellular Notch and a series of transcriptional programs turn on to switch lineage fate toward the T lineage at the expense of B-cell development.¹³³ In mice, Notch is absolutely required for T-cell differentiation and proliferation, including β selection.¹³⁴ Analogous to control of early B cell differentiation by E2A, Notch signaling activates a transcriptional network which includes factors critical for lineage specification (GATA-3, TCF-1), and commitment (BCL11b).¹³⁵ However, although Notch signaling is necessary for murine thymopoiesis it is not sufficient for activation of the full complement of T-cell genes.¹³⁶ The ability of hematopoietic progenitors to respond to Notch signaling and commit to T-lineage fate depends on a balance between positive and negative regulators. Combinations of at least four other transcription factors are required to initiate T-cell development: PU.1, Ikaros, Runx family factors, and E2A.^124,133 In addition, leukemia-lymphoma–related factor (LRF/Pokemon, encoded by Zbtb7a) must be downregulated to allow Notch signaling to induce T-cell fate decisions.¹³⁷ Notch signaling also plays important roles at later stages of thymocyte differentiation.¹³³

The effects of Notch signaling have been extensively studied in mice, but the exact stages and processes regulated by Notch appear to differ between mice and humans. For example, using in vitro studies of human T-cell development, it appears that while Notch is essential for early thymocyte proliferation, it is not required for β selection or T-cell receptor αβ differentiation.^138,139 As with so much of the information described in this chapter, the most important challenge that lies ahead is to translate the detailed mechanistic framework developed from murine studies into careful investigations of human lymphopoiesis.

GATA-3 GATA-3 is a key transcriptional factor for T-cell development, and is essential at various stages of differentiation. However, in addition to T cells, GATA-3 is also expressed in uncommitted HSCs, CLPs, and even in nonhematopoietic cells, and its effects are complex and highly dose-dependent.^33,135,140

Tcf-1 TCF-1 (encoded by the TCF7 gene) is a transcription factor essential for T-cell development, and is directly activated by Notch signaling.^141,142 In ETPs, TCF-1 promotes cell survival as well as activation of T-lineage specific genes, including Gata3 and Bcl11b.^141,142 Induction of a T-cell specific transcriptional program by TCF-1 can occur even in the absence of Notch signaling; however, it cannot activate the essential T-lineage gene Ptcra,¹⁴² indicating that, as in B-cell development, T-cell specification occurs through both hierarchical and combinatorial transcription factor interactions.

Bcl11b BCL11B was identified as a transcription factor required for the normal generation of αβ T cells during β selection; however, upregulation of Bcl11b first occurs at the earlier CD4^negCD8^neg (double negative)-2 (DN2) stage, likely through transcriptional activation by TCF-1.¹³⁵ In DN2 cells, Bcl11b appears to contribute minimally to the T-lineage specification program governed by Notch/E2A/GATA-3/TCF-1 activity, but rather is required for the suppression of stem/multipotent progenitor-associated genes, which marks the loss of myeloid potential and final commitment to the T-cell lineage.¹⁴³