Lineage Commitment and Maturation in Hematopoietic Cells: The Case for Extrinsic Regulation

Commitment can be defined as the decision a cell makes to enter, or generate progeny that enter, a particular maturation lineage at some future time. This decision need not necessarily be accompanied by any immediate change in morphology or expression of novel membrane proteins or regulator receptors.

The existence of a lineage-committed state in many hematopoietic progenitor cells is substantiated by the ability of such cells, when in semisolid cultures, to generate colonies of a single lineage even when stimulated by a mixture of growth factors that should have permitted cells of other lineages to survive and proliferate had they been generated in the clones.1 

In embryonic development, the principle is firmly established that commitment is extrinsically regulated by position effects or inductive gradients and does not occur by random chance. There is clear evidence that commitment events in the initiation and continued production of hematopoietic populations follow the same principle.2 

Hematopoietic commitment depends on, and is presumably initiated by, activation of a succession of nuclear transcription factors, beginning with SCL/TAL-1 and LMO2 and then involving more lineage-restricted transcription factors such as GATA-1 or PU-1.3 The extrinsic agents responsible for activation of these nuclear transcription factors are at present unknown.

The earliest of these events occur in cells not expressing hematopoietic receptors and are thus not subject to control by such regulators. However, there are hematopoietic precursors such as stem cells and multipotential and bipotential progenitor cells that are still able to make commitment decisions and that do express receptors for many, but not all, regulators.4 Can these commitment decisions be modulated by the action of particular regulators? The window of opportunity for such regulators is quite narrow. The commitment of single-lineage progenitor cells appears to be irreversible and not subject to change by the action of hematopoietic growth factors. Thus, insertion of the erythropoietin receptor into macrophage precursors from the marrow or fetal liver allows erythropoietin to stimulate macrophage colony formation, without the development of erythroid cells in such colonies.5Conversely, insertion of the macrophage colony-stimulating factor (M-CSF) receptor into erythroid precursors allows M-CSF then to stimulate the formation of pure erythroid colonies.6 

For cells still able to make commitment choices, it is unfortunate that documentation of commitment usually requires a readout involving the production of maturing cells, because these two processes may be unrelated in their molecular nature. An essential requirement in such studies is that all subsequent progeny of the cell under study be able to be accounted for, because apparent commitment may be spurious if selective survival has occurred.

Recloning studies on dividing blast or multipotential colony-forming cells indicated that commitment is an asymmetrical event, that virtually random combinations of committed progeny can be generated,7 8 and that lineage commitment does not follow a structured sequential pattern. However, the stochastic (random) nature of these events does not permit the conclusion that the pattern of commitment is incapable of being skewed by extrinsic signaling. Indeed, the growth factors used to stimulate the initial cell divisions might themselves have imposed a particular pattern of commitment.

In this context, the behavior of biopotential granulocyte-macrophage progenitor cells suggested that hematopoietic regulators could influence commitment decisions. When one daughter was cultured with one growth factor and the other with a differing growth factor, such as granulocyte-macrophage colony-stimulating factor (GM-CSF) and M-CSF, the lineage of the progeny generated appeared to be significantly influenced by the growth factor initially used.9 Because both GM-CSF and M-CSF can support the proliferation of granulocytes and macrophages,10 there seems little risk that selective loss of progeny cells biased the results observed.

A somewhat similar outcome was observed when blast colonies were stimulated to develop by the combination of stem cell factor (SCF) and interleukin-3 (IL-3). Compared with the content of eosinophil progenitors in SCF-stimulated blast colonies, the SCF/IL-3 combination resulted in a significant increase in the proportion of eosinophil-committed progenitors.11 However, in this example, the colonies were large and it is difficult to exclude selective loss of eosinophil progenitors in SCF-stimulated colonies because SCF has little action on eosinophil precursors.

A comparable study failed to obtain convincing evidence that the pattern of progenitor cell formation in developing human multipotential colonies could be altered when using various combinations of growth factors.12 

In the case of the multipotential cell line FDCP-Mix A4, selective lineage commitment appeared to be achieved by particular growth factors,13 but in this study the key experiments were performed using suspension cultures and the progeny could not be traced back to individual cells.

There are two experimental systems in which the induced change is undoubtedly due to the action of extracellular regulatory factors. The difficulty lies in establishing whether either system involves the occurrence of genuine differentiation commitment.

Cells of murine leukemic lines WEHI-3B or M1 grow autonomously as undifferentiated blast cells. The cells are highly clonogenic (>80% to 90%) and self-renewal at cell division is essentially 100%. When cultured in the presence of granulocyte colony-stimulating factor (G-CSF) for WEHI-3B cells or of IL-6, leukemia inhibitory factor (LIF), or oncostatin M for M1 cells, self-renewal is markedly and irreversibly suppressed and the few progeny produced in clonal cultures exhibit maturation in the macrophage, or less often the granulocytic, lineage.14,15 This response is described as differentiation induction. However, it is not clear that the leukemic and normal stem cell systems are necessarily biologically equivalent or that the molecular basis for the changes is necessarily the same. Suppression of leukemic self-renewal often is accompanied by cell death, unlike commitment in normal stem cells. Furthermore, the leukemic cell lines appear not to be multipotential; thus, suppression of self-renewal is not associated with much evidence of lineage commitment choices, although with agents such as IL-6 acting on M1 or WEHI-3B cells, exclusive macrophage maturation can be striking.15 

In the second system, blood monocytes or culture-generated macrophages can be converted to phenotypically and functionally distinct dendritic cells by culture with GM-CSF plus tumor necrosis factor α (TNFα) or IL-4.16 The system is highly reproducible and does not overtly involve selective cell death. The question is whether this phenotypic switch warrants the label of differentiation commitment?

Studies using truncated or mutated receptors have identified a C-terminal domain in the G-CSF,17,18 thrombopoietin (TPO),19 and GM-CSF¹⁵ receptors whose loss or reduced function results in failure to suppress self-renewal in leukemic cells or failure to induce maturation in leukemic or immortalized cells. However, analysis using different leukemic cell lines has suggested that the domain signaling suppression of self-renewal need not be identical to the domain inducing maturation,15 suggesting that these processes may differ in their signaling requirements in different cell types. If so, the compelling data on regulator-inducible suppression of self-renewal, particularly in leukemic cell lines, may not be transferrable to the commitment process occurring in normal stem cells.

With appropriate immortalized cell lines, regulators such as G-CSF clearly can suppress self-generation and induce granulocytic generation.13 17 However, is this genuine commitment, as defined by a decision to enter a particular lineage, or is it merely that suppression of self-renewal allows the activation by default of a pre-existing maturation program that was previously suppressed by self-generation? The same question can be raised for the various leukemic cell line models.

Possible differences between lineage commitment, differentiation induction, and phenotypic switching will only be resolved when the molecular events and genetic basis of the various changes are established.

When a hematopoietic regulator is used to stimulate colony formation, maturation accompanies the cell proliferation and is commonly assumed to be due to the action of the regulator. However, there is no fixed link between the number of cell divisions and the progress of maturation. Indeed, high growth factor concentrations increase colony size with an accompanying delay in maturation to postmitotic cells.1 

Despite this, there are multiple lines of evidence indicating that hematopoietic regulators play an active role in maturation. A C-terminal region of the G-CSF receptor has been shown to be required for granulocyte maturation in cell lines17,18 and the maturation arrest of congenital neutropenia is often associated with mutations in the C-terminus of the G-CSF receptor.20 TPO provides a clear example of the mandatory action of a regulator in achieving maturation. Although IL-3 is a better proliferative stimulus for megakaryocytes than is TPO, only TPO can achieve full cytoplasmic maturation and platelet formation.21 Another obvious example of hematopoietic regulator action is the distinct phenotypic difference imposed on the morphology and behavior of macrophage colony cells when part of the colony is subsequently stimulated by M-CSF and part stimulated by GM-CSF.22 

However, it has been questioned whether hematopoietic regulators are necessary to initiate maturation. The receptor insertion studies referred to earlier indicate that the actual maturation programs executed are not dependent on the particular hematopoietic regulator used as the proliferative stimulus. Furthermore, when growth factor was withdrawn from cultures of an immortalized cell line, but cell survival was ensured by overexpression of Bcl-2, many cells were able to undergo substantial maturation.23 A similar phenomenon can be observed in clonal cultures of normal marrow cells when developing clones are transferred to factor-free cultures. Cell division ceases, but some maturation to granulocytic or macrophage cells can be observed.24 However, in these experiments it is unclear what earlier role may have been played by the growth factor before it was withdrawn. Can growth factors initiate maturation programs that remain arrested while active cell division is being stimulated but are able to become activated after factor withdrawal and proliferation ceases?

In vivo, there are clear examples of what appears to be a lack of any rational control of differentiation commitment, resulting in an apparently bizarre inefficiency in the process of hematopoiesis. Inactivation of the genes for G-CSF or TPO leads to selective deficiencies in mature granulocytes or platelets25,26 that can be corrected by the injection of G-CSF or TPO. However, in both cases, depletion of the lineage-specific growth factor results in a severe depletion of progenitor cells in all lineages25,27and, conversely, the injection of either factor results in an equally inappropriate expansion of progenitor cells in all lineages.28,29 The pattern is strongly reminiscent of the cloning results from blast colonies that suggested that commitment is random and may not be subject to extrinsic control. In vitro studies have indeed shown that an agent such as G-CSF, when acting with SCF, amplifies the production of many inappropriate cells whose further proliferation cannot be stimulated by this combination.30Because other growth factors are present in the body, these may have synergized with the agent being studied to produce the results observed. These observations do not therefore actually exclude the possibility that an agent can induce a more selective extrinsic modulation of differentiation commitment.

Hematopoietic commitment is likely to be extrinsically regulated, but there is only limited evidence, and probably only a limited opportunity, for hematopoietic regulators being involved in these events. The major extrinsic agents initiating commitment may in fact be novel molecules yet to be detected and possibly with no proliferative actions on hematopoietic cells. A variety of evidence implicates hematopoietic growth factors as playing an important role in some aspects of maturation. However, once established, maturation programs seem not to be qualitatively altered by the particular growth factor used to activate mature cell production.