Concise Review: Erythroid Versus Myeloid Lineage Commitment: Regulating the Master Regulators§


  • Linda Wolff,

    Corresponding author
    1. Laboratory of Cellular Oncology, Center for Cancer Research, National Cancer Institute, Bethesda, Maryland, USA
    • 37 Convent Drive, Bethesda, Maryland 20892, USA
    Search for more papers by this author
    • Telephone: 301-496-6763; Fax 301-594-3996

  • Rita Humeniuk

    1. Laboratory of Cellular Oncology, Center for Cancer Research, National Cancer Institute, Bethesda, Maryland, USA
    Search for more papers by this author

  • Author contributions: R.H. and L.W.: wrote the manuscript and prepared the figures.

  • Disclosure of potential conflicts of interest is found at the end of this article.

  • §

    first published online in STEM CELLS EXPRESS April 5, 2013.


Developmental processes, like blood formation, are orchestrated by transcriptional networks. Those transcriptional networks are highly responsive to various environmental stimuli and affect common precursors resulting in increased production of cells of the erythroid lineage or myeloid lineage (granulocytes, neutrophils, and macrophages). A significant body of knowledge has accumulated describing transcription factors that drive differentiation of these two major cellular pathways, in particular the antagonistic master regulators such as GATA-1 and PU.1. However, little is known about factors that work upstream of master regulators to enhance differentiation toward one lineage. These functions become especially important under various stress conditions like sudden loss of red blood cells or pathogen infection. This review describes recent studies that begin to provide evidence for such factors. An increased understanding of factors regulating cellular commitment will advance our understanding of the etiology of diseases like anemia, cancer, and possibly other blood related disorders. STEM Cells2013;31:1237–1244


Blood cells are being continuously produced and replenished through a stepwise commitment of blood stem and progenitor cells. Proper balance in the production of blood cells is important to survival and occurs within both the white- and red-cell compartments. The anemia due to the loss of red blood cells (RBCs) is a life threatening condition, often accompanying blood diseases such as leukemias and dysplasias. The organism's response to sudden loss of red cells can be very immediate where replenishment of red cells must occur within hours for life to be sustained. Committed erythroid progenitors, present in the spleen, can respond to loss of RBCs and low oxygen by proliferating and differentiating to red cells. However, these progenitors must also be replenished from the bone marrow and they usually derive from multipotent progenitors that have the capacity to differentiate into either erythroid/megakaryocyte or myeloid cells. We are only beginning to get insight into the mechanisms that instruct progenitors to shift commitment as needed. Such investigations were only possible with the ability to separate multipotent progenitors and study their behavior. Although there is a great deal known about master regulators at bifurcations in the differentiation pathways, almost nothing is known about how the expression or function of the master regulators are promoted or enhanced when a specific lineage is required to expand rapidly. This review addresses new research on how progenitor cells with potential for both erythroid and myeloid cell fates can be influenced to take one path or the other.


Hematopoietic stem cells (HSC) give rise to the multipotent common myeloid progenitors (CMPs), which have the capacity to differentiate toward the erythroid-megakaryocyte and myeloid lineages. This cell can subsequently give rise to either a granulocyte-macrophoge progenitor (GMP) or a megakaryocyte-erythroid progenitor (MEP) (reviewed in [1]). There is also evidence for a progenitor with lymphoid-myeloid potential that has lost the ability to differentiate into the erythroid or megakaryocyte lineages [2]. Therefore, lineage bifurcations may occur at different levels in the hematopoietic hierarchy. Importantly, a mixed lineage gene expression pattern is present at low levels in stem cells and progenitors [3, 4]. As commitment progresses, the expression patterns become limited and the levels of expression of lineage-specific genes increases. Analysis of lineage commitment at the single cell level revealed that uncommitted cells, independently of each other, activate expression of various regulatory genes resulting in a low probability of lineage commitment [5]. Moreover, this study suggests a constant fluctuation of uncommitted cells between self-renewal and committed states and existence of multiple entry routes into lineage commitment. Therefore, as one might suspect, cell fate determination occurs through the function of major transcriptional regulators that act both positively to regulate lineage-specific genes and negatively to regulate genes of other lineages [5].


Specification of macrophage/neutrophil cell fate require two major transcription factors, the Ets transcription factor, PU.1 and C/EBPα. Although PU.1 is found to be expressed in both myeloid and lymphoid cells, C/EBPα is found predominantly in myeloid cells [6–8]. Mice lacking PU.1 have a multilineage defect in generation of monocytes, granulocytes, B and T cells [9, 10]. Analysis of CMP and GMP from mice with conditional null alleles revealed that PU.1 is required for normal development of macrophages and granulocytes at all stages tested [11, 12]. PU.1 is also required to maintain the HSC pool in the bone marrow and to generate the earliest myeloid and lymphoid progenitors [12].

C/EBPα is expressed in granulocytes, monocytes, and eosinphils. Low levels of the transcription factor can be detected in HSCs and its expression increases with development of CMPs and GMPs [7]. As the cells develop into mature neutrophils or monocytes, expression is diminished [7, 13]. Conditional C/EBPα (−/−) mice were used to demonstrate that disruption of the gene blocks the transition from CMP to GMP [13]. In support of the roles of C/EBPα and PU.1 in specifying the myeloid lineage, when either is expressed from a retroviral vector in B-lineage cells, they induce monocytic development [14]. Interestingly, it has been shown that C/EBPα and PU.1 itself regulate PU.1 transcription indicating that there is a feedback mechanism to enhance and lock in commitment [15, 16].

PU.1 and C/EBPα are required for both macrophage and neutrophil specification, however, the relative levels of these two transcription factors regulate macrophage versus neutrophil cell fate choice [3, 17, 18]. Interestingly, although high levels of PU.1 induce the formation of macrophages in the presence of Interleukin-3 (IL-3), they induce neutrophils in the presence of the granulocyte-colony stimulating factor [18]. This indicates that cytokines can, at least in some instances, alter the course of cell fate as discussed in more detail later. As PU.1 expression is increased, Egr-1,2 and Nab-2 levels also increase and act as secondary cell fate determinants, while Gfi-1, an activator of neutrophil specification, is repressed [3, 19].


GATA-1, a gene that is conserved among vertebrates [20, 21], is considered to be the master regulator of erythropoiesis. GATA-1 expression levels are very low in HSC and CMPs, however, the presence of GATA-1 in these multipotent cells is probably important for lineage priming [4]. When GATA-1-GFP transgenic mice were used to analyze changes in GATA-1 expression during erythroid differentiation [22], it was found that the level of expression of GATA-1 increases as cells progress to MEPs and early erythroid blasts. Expression is highest in the late-stage erythroid progenitors and decreases gradually with further differentiation. Embryonic stem cells or erythroid cells of mice deficient in GATA-1 fail to develop into mature erythroid cells [23, 24]. Embryonic lethality results from the GATA-1 null mutation between days 10.5 and 11.5 due to severe anemia [25]. GATA-1 not only regulates many target genes such as the α- and β-globins, and Bcl-XL, a survival factor, but also regulates itself, Epo and the Epo receptor (EpoR) [26]. These latter functions guarantee perpetuation of erythropoiesis.

GATA-1 requires the cooperation of other transcription factors to specify the erythroid lineage through the upregulation of erythroid-specific genes and the downregulation of nonerythroid genes. Recently, Chromatin Immunoprecipitation (ChIP)-seq was used to generate chromatin occupancy maps of GATA-1 and some proteins that cooperate with GATA-1(reviewed in [27–29]). The results of these studies have been very informative regarding the real targets of GATA-1 and its regulatory partners. Several studies determined that the majority of GATA-1 binding sites are at distal regulatory elements with only 10%–15% of the binding sites found proximal to promoters [30–33]. These binding sites are equally found at intragenic and intergenic regions. When data related to gene expression and genome occupancy were combined, following GATA-1 activation, it was determined that about 300–700 are direct targets of the transcription factor. However, around 5,000 genes are actually differentially expressed with activation of GATA-1, because most genes are indirectly regulated. About half of the genes regulated by GATA-1 are activated, while the others are repressed [30, 31, 33]. When an analysis of motifs at the GATA-1 activated genes was performed, it was discovered that binding sites for SCL/TAL1 were highly enriched [34, 35]. There was a reciprocal finding that enrichment of GATA-1 binding sites was found near SCL/TAL1 sites at genes activated by SCL/TAL1 [35]. SCL/TAL1 forms a complex in erythroid cells with the ubiquitous Basic helix-loop-helix (bHLH) protein E2A and with cofactors LMO2 and LDB1. In conjunction with GATA-1, these form a pentameric complex [27]. KLF1, another transcription factor that has a role in erythropoiesis, is involved primarily in activation of genes [36, 37]. It was found to bind within 1 kb of GATA-1 48% of the time, but these regions are not co-occupied by SCL [37]. This suggests that regulation of genes by GATA-1/SCL or GATA-1/KLF1 is mutually exclusive. The importance of the transcription factors SCL/TAL1, LMO2, LDB1, and KLF1 to erythropoisis had previously been well documented by cell-based ex vivo assays and in studies of strains of mice with knockouts of each of these genes [29].

As mentioned above, GATA-1 has been shown to repress as well as activate genes. The GFI-1b protein in combination with the LSD1-CoREST corepressor complex has been implicated in gene silencing [38, 39]. In addition, the well-known GATA-1 interacting, FOG-1, is most likely involved in repression by recruiting the NuRD complex [38, 40]. Surprisingly, it was reported by Blobel and colleagues that the NuRD complex recruited by FOG-1 to GATA-1 binding sites is capable of activating, as well as, repressing genes [41, 42]. It is not clear how this complex, previously known for its repressing function, through deacetylase activity, could activate genes. The authors speculate on possible mechanisms. For example, H3K4 methylation might interfere with NuRD's ability to deacetylate histones or FOG-1 might be post-translationally modified in a way that interferes with its repressive activity.

To date, most studies on transcription regulation of genes that control erythroid fate have involved transcription initiation via cell specific transcription complexes. However, it has become increasingly evident that another level of control involving cell specific factors is at the transcriptional elongation step. Based upon experiments in TIF1γ mutant zebrafish and biochemical studies in human cells, Bai et al. [43] hypothesize a model whereby TIF1γ can regulate erythroid cell fate through effects on elongation. Pol II elongation is inhibited by PAF and DSIF and their genetic studies uncovered a functional antagonism between TIF1γ and PAF/DSIF. TIF1γ releases paused Pol II in the presence of the inhibitor proteins by recruiting pTEFb and FACT through interaction with the SCL complex. Elongation is facilitated by the kinase function of P-TEFb, which phosphorylates DSIF and Pol II CTD.


Transcriptional determinants of hematopoietic cell fate can be considered master regulators if they have several characteristics (a) they are required for development [44] into a specific lineage as shown by knockout experiments; (b) they antagonize the opposite lineage programs; and most important (c) they are capable, when introduced into cells, of directing or even changing cell fate. For this reason, they have also been referred to as “motors for reprogramming.” GATA-1 and PU.1 are considered master regulators of erythroid and myeloid cell fate, respectively. However, based upon these criteria, FOG-1 and C/EBPs have also been demonstrated to act as master regulators. It was found that HSCs and premegakaryocyte/erythroid cells, that are deficient in FOG-1, can undergo reprogramming to specify myeloid cells [45].

It is well-documented through numerous studies that GATA-1 and PU.1 direct cell fate as master regulators, by both regulating erythroid and myeloid lineage genes, respectively, as well as functioning as antagonists to each other's activity. In early avian studies, it was found that GATA-1 was capable of reprogramming myelomonocytic cell lines into eosinophils, thromboblasts, and erythroblasts [46]. In addition, GATA-1, when enforced in the myeloid cell line M1, induced megakaryocyte/erythroid differentiation [47]. In other studies, it was discovered that PU.1 expression through retroviral insertional mutagenesis or by expression in transgenic mice resulted in erythroleukemia (MEL) [48, 49]. In support of antagonism between GATA-1 and PU.1, it was found that GATA-1 could induce differentiation of MEL cells toward the erythroid lineage and that PU.1, when expressed in MEL cells via an expression vector, could block this differentiation [50, 51].

Many of these early lineage reprogramming experiments were carried out in vitro in cell lines, but they had the disadvantage that the cells were transformed. Transformation inherently affects cell-renewal and differentiation properties of cells, therefore, it was never entirely clear if these reprogramming activities of transcription factors reflected their normal abilities. Therefore, elegant studies using primary cells were also performed. Nerlov and Graf [52] were able to induce myeloid lineage commitment by activation of PU.1 in avian MEPs. Furthermore, in an examination of the plasticity of murine myeloid progenitor cells, Heyworth et al. [53] discovered that bipotential cells committed to neutrophils and monocytes could be reprogrammed by GATA-1 to take on characteristics associated with erythroid, eosinophil, and basophil-like cells. Following enforced expression of GATA-1, progenitor cells formed colonies resembling mixed lineage colonies that normally develop in semisolid media from mulitpotential progenitors.

Up to this point, most reports showing the effects of GATA-1 on lineage determination had used overexpression. There were none showing that loss of GATA-1 affected cell fate, although GATA-1 loss had been demonstrated to be required for terminal erythroid differentiation. A study in zebrafish by Galloway et al. [54] was significant, because it was the first to demonstrate that loss of Gata1 altered myeloid differentiation during ontogeny. Gata1 morpholinos, which caused anemia, resulted in persistent pu.1 expression in the intermediate cell mass, a finding that correlated with conversion of blood precursors to myeloid cells. Another study in zebrafish identified myeloerythroid progenitor cells that are probably equivalent to the mammalian CMPs. They showed reciprocal negative regulation of pu.1 and gata1 through the use of knockout and transplant experiments [55].

Following the observations that the transcription factors GATA-1 and PU.1 antagonize each other's activity to restrict cells to a specific fate, investigations into the actual mechanism of this antagonism at the DNA level began to appear (reviewed in [44]). It was discovered that PU.1 represses GATA-1 target genes by binding to GATA-1 and recruiting pRb, Suv39h, and HP1α [56, 57]. GATA-1 represses PU.1 by displacing the critical coregulator c-Jun at the Ets domain of the protein [58, 59].


As discussed above, there is a large body of work supporting the idea that dominance of either GATA-1 or PU.1 is central to the decision to specify erythroid and myeloid lineage cells. These transcription factors positively regulate lineage-specific genes while antagonizing each other by repression. But the question that remains is how do bipotential progenitor cell populations shift the commitment to erythroid or myeloid cells at the expense of the other lineage when a need arises. Although regulation by GATA-1:PU.1 dominance is crucial to our understanding of lineage restriction in hematopoietic cell differentiation, it does not explain how the balance is re-established under stress. It could be presumed that there are many pathways or molecular events that could feed into this GATA-1 versus PU.1 control, especially for rapid transitions. Research on this has already uncovered some intriguing mechanisms by which the GATA-1:PU.1 balance could be altered. Since these two master regulators positively autoregulate themselves, a small change in a single cell that alters transcription factor dominance could be amplified subsequently to lock in a programmatic developmental change. Some ways that the balance could be tipped are signal transduction induced changes in transcription factors, transcription elongation changes, and epigenetic alterations. As discussed below there are examples for each of these in erythroid/myeloid homeostasis. In support of the idea that some events must precede the true lineage commitment programs were findings of Pina et al. [5] who carried out single cell expression analysis in undifferentiated EML cells. Their data support the idea that erythroid commitment occurs with expression of only a limited number of lineage components and the expression pattern varies from cell to cell. This may indicate the existence of different entry points for lineage commitment. Even more interesting was the finding that early commitment may involve minimal or no expression of GATA-1. The authors propose that the heterogeneous state resolves to a coordinated terminal differentiation program.

p15Ink4b may promote erythroid differentiation at the expense of myeloid differentiation by events that precede upregulation of GATA-1 [60–62]. p15Ink4b activity is needed to maintain the proper lineage commitment of progenitors at erythroid/myeloid bifurcations and rapid RBC replenishment following stress. It was found that p15Ink4b is expressed more highly in committed MEPs than GMP [62]. Furthermore, because mice lacking p15Ink4b have lower numbers of primitive red-cell progenitors, they develop a life threatening anemia in response to hematopoietic stress induced by both 5-fluorouracil and phenylhydrazine. Intriguingly, introduction of p15Ink4b into multipotential progenitors rendered cells them more permissive to erythroid commitment and less permissive to myeloid commitment as shown by alterations in the ability to form erythroid and myeloid colonies in semisolid media. Although p15Ink4b was originally discovered to be a cyclin-dependent kinase inhibitor, its function in cell fate appears to depend upon its ability to activate a signaling pathway involving Erk phosphorylation [62, 63]. In addition, the introduction of the gene in EML cells, a stem-cell like line, resulted in a reduction in the levels of GATA-2 protein. This is an intriguing finding, since during early erythropoisis, GATA-1 replaces GATA-2 at shared GATA motifs in the chromatin. This is called the “GATA switch” [64–68]. GATA-2 is an instable protein compared to GATA-1 and, therefore, it would be reasonable to hypothesize that signal transduction could account for GATA-2 degradation by p15Ink4b, followed by its replacement by GATA-1 at the shared target sites (Fig. 1). This would result in upregulation of GATA-1 transcription and transcription of erythroid genes including EpoR. Interestingly, GATA-2 has been shown previously to be phosphorylated by MAPK [69]. Transcription of GATA-2 itself is affected by this switch because GATA-2 sites upstream of GATA-2 positively regulate this gene and this positive regulation is abrogated in conjunction with the switch [70, 71]. The switch mechanism is consistent with the finding of reciprocal expression of GATA-1 and GATA2 seen during development of RBCs [72]. Consistently, GATA-2+/− mouse marrow has fewer GMPs, demonstrating the involvement of this gene in myeloid progenitor fate decision [73]. The fact that p15Ink4b reduces the level of GATA-2 raises the possibility that other mechanisms could also impinge on the GATA-2's susceptibility to degradation to jump start the “GATA switch”.

Figure 1.

Proposed mechanism by which p15Ink4b facilities a “GATA switch” thereby shifting the differentiation potential of uncommitted hematopoietic progenitors toward the erythroid fate. (A): Gata-2 is highly expressed in hematopoietic stem cell and uncommitted progenitors and occupies DNA sites involved in erythroid lineage specification. Since GATA-2 shares binding sites with GATA-1 it can prevent GATA-1 binding and, therefore, GATA-1 transcriptional regulation [74]. (B): Elevation of p15Ink4b expression in uncommitted hematopoietic progenitors induced rapid phosphorylation of MEK/ERK MAPK [62]. This could lead to phosphorylation of Gata-2 that renders Gata-2 potentially unstable resulting in its rapid degradation. This is consistent with the observation that inhibition of MEK/ERK signaling as well as 26S proteasome inhibitor MG132 prevents observed decrease in GATA-2 [62]. In addition, the GATA switch has been shown to be dependent on proteasome activity [67]. (C): GATA-1 is free to bind the vacated sites and transactivate the erythroid lineage specific genes.

TIF1γ, which has been implicated in regulation of transcription elongation at erythroid genes, may play a role in cell fate decisions through effects on GATA-1 expression [43]. In this regard, work carried out in zebrafish is intriguing. This gene was found to be required for erythroid differentiation from HSCs and, in its absence, myeloid cells significantly increased in number [75]. Tif1γ control of gata1 and pu.1 levels in the caudal hematopoietic tissue was shown to be responsible for these observations. Whereas Tif1γ positively regulates gata1 expression, it negatively regulates pu.1, and, therefore, modulates the erythroid versus myeloid fate outcomes from HSCs [75]. More studies will need to be carried out in mammalian cells to determine how TIF1γ is regulated under different states of hematopoietic stress.

Both p15Ink4b and TIF1γ shift the erythroid/myeloid balance toward the erythroid lineage (Fig. 2). In contrast, as reported by Lau et al. [76] [74], signaling by Desert Hedgehog (Dhh), a Hedghog family member, is a negative regulator of erythropoisis. Differentiation from CMPs to GMPs was decreased in Dhh-deficient bone marrow, whereas differentiation from CMPs to MEPs was increased.

Figure 2.

Transcriptional networks that orchestrate erythroid versus myeloid cell fate decision of hematopoietic progenitors are regulated such that in many circumstances increases in commitment of one lineage results in decreases in another lineage. Erythroid and myeloid specific master regulators GATA-1 and PU.1 inhibit each other's transcription. Research on the factors or conditions that drive the commitment to one lineage or another, and influence GATA-1 and PU.1, has only begun recently. Included in this figure are some factors implicated in influencing these cell fate decisions and include p15Ink4b, tif1γ, Dhh, and epigenetic modifiers. Abbreviations: CMP, common myeloid progenitor; Dhh, Desert Hedgehog; HSC, hematopoietic stem cell; MPP, multipotent progenitor.

The role of cytokines in the lineage fate determination has been debated in the literature since their discovery. It was technically difficult to prove the instructive function of cytokines and moreover to dissociate it from the permissive role, selective survival, or proliferation of the progeny [77]. The macrophage colony stimulating factor (M-CSF) is a good example of the instructive cytokine action that also serves as a fascinating paradigm for cytokine/transcription factor interactions [78]. MafB is a bZip type transcription factor that was demonstrated to restrict the ability of (M-CSF) to instruct myeloid divisions in HSCs. In its absence, there was a specific enhancement of sensitivity to M-CSF and activation of PU.1. Although HSCs showed a competitive repopulation advantage, that was dependent upon M-CSF and PU.1, T lymphocyte and erythroid compartments showed no competitive advantage [79]. Hence, MafB levels determine the sensitivity of uncommitted progenitors to M-CSF and directly regulate myeloid lineage commitment. Although erythropoietin is a cytokine known to play a major role in the survival and proliferation of early erythroid progenitors, its role in cellular commitment remains to be established. Flt-3 kinase has been shown to instruct myeloid versus erythroid lineage commitment and specifically dendritic cell differentiation [77]. Cytokines provide an excellent mode of communication between the cell and its environment. Future research will shed light on the regulation of cytokine/transcription factor interactions, like the one described for M-CSF, MafB, Pu.1, and M-CSF receptor in response to hematopoietic stress.

Epigenetic modifications have long been implicated in hematopoietic lineage specification based on the coordinated changes in gene expression, chromatin state, and DNA methylation [80, 81]. Mutations in epigenetic modifiers, including TET2, IDH1, IDH2, ASXL1, EZH2, and DNMT3A have been implicated in the pathogenesis of various myeloid malignancies, namely Myeloproliferative neoplasms (MPN), Myelodysplastic syndromes (MDS), and Acute myeloid leukemia (AML) [82]. However, role of these genes in myeloid/erythroid lineage commitment is largely enigmatic due to lack of functional studies. Tet-2 knockout mice have been shown to develop MDS-like disease with marked expansion of erythroid progenitors [83]. ASXL1 has been shown to directly interact with PRC2 complex and may be important for its recruitment and stability in myeloid cells. The HOXA cluster of genes has been identified as a specific target of ASXL1 [84]. Conditional Ezh2 knockin mice have shown, specific expansion of the GMP compartment with increased sensitivity of progenitor cells to cytokines [85]. Interestingly, EZH2 controls expression of INK4B, highlighting again the importance of this gene in hematopoietic differentiation [86, 87]. L3MBTL1, a polycomb group protein, located within the region commonly deleted in patients with polycythemia vera (PV, MDS, and AML, has been shown to regulate erythroid differentiation of human hematopoietic progenitor cells [88].


Many of the factors that play a role in the maintenance of balance between the progenitors of the erythroid and myeloid compartments have also been implicated in MDS and hematological malignancies [89]. Mutations of both of the master regulators, GATA-1 and PU.1, have been observed in leukemia. Somatic mutations in GATA-1 contribute to the development of acute megakaryocyte leukemia (AMKL) in children with Down Syndrome (DS) [90, 91]. The fetal liver is most likely the origin of the leukemia initiating cell since these mutations occur in utero [92]. A subset of children with DS will develop AMKL by the age of 1–4 years with an accompanying myeloid and erythroid dysplasia. The mutations prevent the formation of full length GATA-1. The resulting protein, called GATA-1s, is lacking the N-terminal activation domain necessary for proper megakaryocyte differentiation and growth control [93, 94]. Ten percent of infants with DS also develop a transient myeloproliferative disorder, which appears to be a clonal preleukemia characterized by an accumulation of immature megakaryoblasts in the fetal liver and peripheral blood [90]. Gain-of-function mutations of GATA-2 have been found in acute myeloid transformation [95]. This may be due to the fact that GATA-2 prevents the switch to GATA-1 at many regulators of erythropoiesis, thus maintaining more primitive blast cells.

PU.1 mutations were found in about 7% of patients with AML, particularly, FAB subtypes M0, M4, and M5 [96]. For the most part, the mutations, which consisted of deletions, decreased PU.1's ability to cooperate with interacting proteins such as AML1 or c-Jun. The deletions resulted in loss of sequences in the DNA binding domain, PEST domain, and transactivation domains. Although these mutations were heterozygous, loss of one allele did not lead to development of leukemia in mice. However, it has been proposed that the additive effects of PU.1 haploinsufficiency and other functional reductions on the PU.1 activity by oncogenic proteins may reduce overall PU.1 expression and/or function to levels that contribute to malignancy [89]. Interestingly, Rosenbauer et al. [97] showed that, mice with a hypomorphic allele that reduced PU.1 expression by 80%, developed a precancerous state and AML associated with clonal chromosomal changes. Importantly, PU.1 restoration rescued myeloid cell differentiation in AML blasts. In another study Durual et al. [98] overexpressed PU.1 in a human AML cell line and primary AML via a lentiviral vector and showed that it caused increases in myelomonocytic surface antigen expression, decreased proliferation rates, and increased apoptosis. Another more recent paper also supports the notion that loss of PU.1 expression or function can contribute to myeloid malignancy [99]. It was shown that a single nucleotide polymorphism (SNP) can lead to misregulation of PU.1. The SNP is localized to a distal upstream regulatory element and leads to loss of regulation of PU.1 by the chromatin-remodeling transcriptional regulator, special AT-rich sequence binding protein.

C/EBPα is another transcriptional regulator of myelopoiesis which is frequently dysregulated during AML. It is mutated in approximately 5%–15% of patients with the disease and there are two categories of mutations [100, 101]. Alterations in the N-terminus produce a truncated dominant negative C/EBPαp30 isoform and mutations in the C-terminal produce a protein with a modified leucine zipper. The latter prevents dimerization and DNA binding. Many patients have both types of mutations with one on each allele. Suppression of C/EBPα expression is also affected in patients with AML1-ETO, BCR-ABL, FLT3-ITD, or CEBPA promoter methylation (reviewed in [102]).

Factors that modify chromatin are likely to account for shifts in cell fate since mutations of polycomb or polycomb-related factors are associated with MDS/myeloproliferative disease (MPD) and transformation to AML [103–106] (Fig. 2). ASXL1 mutations are associated with chronic myelomonocytic leukemia, whereas mutations in EZH2 are associated with diseases accompanied by increased platelet counts. Since the loss of these repressor functions during normal development results in activating expression of genes involved in regulating lineage commitment, it is hard to imagine how they are involved in transformation to AML with its block in differentiation. However, it has been proposed that, because AML is a disease characterized by self-renewal of progenitors that are lineage specific, loss of the functions of these epigenetic regulators allows cells to differentiate from the MDS/MPD stem cell stage to the progenitor stage. This is most likely accompanied by other genetic alterations that block further differentiation as AML progresses [107].

The p15INK4b gene is hypermethylated in 50%–60% of patients with myelodysplastic syndrome and 70%–80% of patients with AML [108]. Aberrant p15INK4b methylation has been generally associated with poor prognosis [109, 110]. Furthermore, its loss in MDS patients is associated with increased risk of transformation to AML [111, 112]. The recent discovery of a novel function of p15Ink4b in enhancing differentiation to the erythroid lineage at the expense of the myeloid lineage, and the fact that its loss in mice causes increased numbers of myeloid progenitors in the bone marrow offers new insights into its tumor suppressor activity.


In conclusion, recent data indicate that cell fate decisions at the erythroid/ myeloid axis involve not only the master regulators such as GATA-1 and PU.1, but many other factors that, in turn, regulate these transcription factors. Much more will be learned in the future as to what additional factors play a role in this regulation, and which ones are used to maintain normal homeostasis versus stress responses. Indeed, further understanding of these mechanisms will lead to therapeutic measures to alter disease-induced imbalances in hematopoietic cell production.


The authors indicate no potential conflicts of interest.