Molecular and Transcriptional Regulation of Megakaryocyte Differentiation

Authors

  • Ramesh A. Shivdasani M.D., Ph.D.

    Corresponding author
    1. Departments of Adult Oncology and Cancer Biology, Dana-Farber Cancer Institute and Department of Medicine, Brigham & Women's Hospital and Harvard Medical School, Boston, Massachusetts, USA
    • Departments of Adult Oncology and Cancer Biology, Dana-Farber Cancer Institute, 44 Binney Street, Boston, Massachusetts 02115, USA. Telephone: 617-632-5746; Fax: 617-632-5739
    Search for more papers by this author

Abstract

Megakaryocytes, among the rarest of hematopoietic cells, serve the essential function of producing numerous platelets. Genetic studies have recently provided rich insights into the molecular and transcriptional regulation of megakaryocyte differentiation and thrombopoiesis. Three transcription factors, GATA-1, FOG-1, and NF-E2, are essential regulators of distinct stages in megakaryocyte differentiation, extending from the birth of early committed progenitors to the final step of platelet release; a fourth factor, Fli-1, likely also plays an important role. The putative transcriptional targets of these regulators, including the NF-E2-dependent hematopoietic-specific β-tubulin isoform β1, deepen our understanding of molecular mechanisms in platelet biogenesis. The study of rare syndromes of inherited thrombocytopenia in mice and man has also refined the emerging picture of megakaryocyte maturation. Synthesis of platelet-specific organelles is mediated by a variety of regulators of intracellular vesicle membrane fusion, and platelet release is coordinated through extensive and dynamic reorganization of the actin and microtubule cytoskeletons. As in other aspects of hematopoiesis, characterization of recurrent chromosomal translocations in human leukemias provides an added dimension to the molecular underpinnings of megakaryocyte differentiation. Long regarded as a mysterious cell, the megakaryocyte is thus yielding many of its secrets, and mechanisms of thrombopoiesis are becoming clearer. Although this review focuses on transcriptional control mechanisms, it also discusses recent advances in broader consideration of the birth of platelets.

Introduction

Mechanisms of thrombopoiesis are of considerable interest in hematology and cell biology, in part because of the variety of human thrombocytopenia syndromes and because megakaryocyte (MK) differentiation encompasses many unusual attributes. Mature MKs are polyploid cells that assemble a unique set of organelles, including alpha granules, dense bodies, and an extensive system of internal membranes. The cellular mechanisms of endomitosis in MKs have been clarified recently [1, 2], and genetic studies are rapidly bringing the molecular basis of MK organelle biogenesis into sharper focus [3-6]. Perhaps most remarkable among the processes associated with MK differentiation, and certainly specific for this lineage, is that by which a voluminous cytoplasm fragments into thousands of individual platelets. Substantial experimental evidence now supports a model, initially proposed in the 1970s and 1980s [7-10], wherein differentiated MKs extrude long cytoplasmic processes (“proplatelets”) that serve as the immediate precursors of circulating platelets [11-13]. The microtubule (MT) cytoskeleton plays a central role in thrombopoiesis [14, 15], and recent evidence suggests that platelet assembly occurs de novo within proplatelet extensions [16].

In the 1990s, significant advances were made independently in understanding the transcriptional basis of hematopoietic cell differentiation [17, 18]. Analysis of cis-regulation of cell-specific genes, isolation, and gene targeting of lineage-restricted transcription factors, and complementary studies in cell differentiation in vitro all combined to provide a glimpse of how common progenitors produce vastly different cell types. One theme to emerge from these studies is that cell-restricted expression of lineage-specific genes is achieved through use of both lineage-restricted and more widely expressed transcriptional regulators. The potential complexity afforded by combinatorial action of proteins probably allows cells considerable flexibility and sophistication in regulating gene expression. Recent advances have highlighted the particular importance of three erythro-MK transcriptional regulators, GATA-1, FOG-1, and NF-E2, in discrete stages of MK and platelet differentiation.

Following is a brief overview of the transcriptional regulation of thrombopoiesis and highlights of recent studies that illustrate the major components of our present understanding. These studies help assemble the rough outline of a putative transcriptional hierarchy, which is first constructed on the basis of the experimental findings. The second half of this review discusses the known and possible functions of other regulators of MK differentiation. For additional details about selected transcription factors, readers are directed to Kaluzhny et al. [19].

Roles of GATA-1, FOG-1, and Fli-1 in Early- and Mid-Stages of Thrombopoiesis

Background

Much of the attention in transcriptional regulation of MK genes has focused on the GATA family of zinc-finger proteins, which activate transcription by engaging the DNA sequence WGATAR in the cis-regulatory elements of many lineage-restricted genes. Important early advances included identifying GATA-1 and GATA-2 as erythro-MK transcription factors [20, 21] and finding that the rat platelet factor 4 (PF4) and human glycoprotein (GP) IIB genes were regulated in part through isolated GATA sites or the combination of juxtaposed GATA and Ets-binding cis-elements [22, 23]. Virtually every examined MK-expressed gene reveals functional GATA and Ets cis-elements in transfection analysis of artificial promoter constructs in vitro [24, 25].

Insights on GATA-1 Function from Knockout Mice and Human Patients

Mice lacking GATA-1 selectively in MKs [26] and humans with critical GATA-1 point mutations [27, 28] have defects that reveal the minimum requirement for this factor in thrombopoiesis. First, platelet counts are about 15% of the normal, platelet size is increased at least twofold, platelet shape is spherical rather than discoid, and the bleeding time is prolonged [26-29]. Second, GATA-1-null MKs proliferate exuberantly in vitro, and their substantial accumulation in vivo appears to reflect primary growth dysregulation rather than a secondary response to thrombocytopenia [26, 29]. This finding suggests that GATA-1 is a negative regulator of cell proliferation in early MK progenitors. The vastly increased numbers of tissue MKs reveal an abnormally small and immature cytoplasm that harbors few platelet granules amid highly disorganized internal membranes. Thus, absent or impaired GATA-1 function is associated with thrombocytopenia and defective platelets as a result of a unique MK differentiation arrest. At a minimum, GATA-1 must control some portion of the program of gene expression that regulates cell replication, drives MK cytoplasmic maturation, and coordinates development of platelet organelles. Other features, including proplatelet formation and regulation of platelet size and numbers, may either be directly under GATA-1 control or simply reflect failure of the cells to progress beyond an early differentiation block.

For a variety of reasons, the critical transcriptional targets of GATA-1 are not readily apparent. For one, the closely related factors, GATA-1 and GATA-2, are coexpressed and likely partially redundant in the MK lineage. Second, many MK-restricted genes for which there is substantial in vitro evidence of GATA-mediated regulation are expressed at nearly normal levels in GATA-1-deficient mouse MKs [29] (there are inconsistencies between GPIBα and β mRNA and protein levels in human platelet samples [28]). In considering the cellular phenotype, however, the inescapable conclusion is that some transcriptional targets of GATA-1 cannot be regulated by other factors, and their absence must account for the abnormalities observed in GATA-1-null MKs and platelets. The p45 subunit of the transcription factor NF-E2, which is discussed in greater detail below, might be one such gene. Expression of p45 NF-E2 is reduced in GATA-1-null MKs [29] and also depends on the GATA coactivator FOG-1 [30]; GATA sites are also essential for activity of one of the two p45 NF-E2 promoters in developing red blood cells [31]. Moreover, abnormalities seen in the absence of NF-E2 appear later in MK ontogeny than those seen with GATA-1 deficiency [32, 33] so that a cautious epistatic argument may also be made along these lines.

cDNA subtraction methods have suggested one additional candidate MK target of GATA-1 gene regulation, the inositol polyphosphate 4-phosphatase type I [34]. GATA-1-deficient MKs express considerably reduced levels of this mRNA, and its forced expression restores nearly normal growth kinetics in MK progenitors. This phosphatase catalyzes hydrolysis of membrane-bound and soluble forms of the phosphatidylinositol 3-kinase second messengers inositol bisphosphates and triphosphates, so that a direct role in regulating MK cellular proliferation is certainly plausible. It remains to be determined whether the type-1 inositol phosphatase, which is expressed widely, is a direct or indirect transcriptional target of GATA-1 in MKs.

GATA-Dependent and Independent Roles for FOG-1 in MKs

The MK and platelet abnormalities in humans and mice with genetic defects in GATA-1 are remarkably similar, even though the underlying molecular bases are distinct. In knockout mice, the genetic lesion in a GATA-1 cis-element abolishes MK expression [26], whereas patients harbor single point mutations, either V205M or D218G, which interfere with the protein interaction between GATA-1 and its cofactor FOG-1 [27, 28, 35]. These findings strongly suggest that most, if not all, defects observed with the absence of GATA-1 in murine MKs result from the loss of FOG-mediated GATA activity and implicate the GATA-1/FOG-1 partnership in regulating MK proliferation and platelet formation. This pair of transcription factors may also operate synergistically to regulate the p45 NF-E2 and GPIIBα promoters in MKs [30, 36]. Yet, mice with germline absence of FOG-1 lack MK progenitors completely and, hence, show an unexpected role in even earlier stages of megakaryopoiesis [37]. Low levels of MK-specific transcripts, including PF4 and GPIIBα, are detectable, which implies that early MK-lineage cells may be produced but that these cells can neither replicate nor differentiate effectively. Curiously, none of the GATA-factor knockout mice display a similar phenotype [26, 38]. Consequently, the most conservative interpretation of the data is that FOG-1 functions within a pathway unrelated to GATA proteins early in MK ontogeny and then again later in a GATA-dependent pathway of cellular maturation and thrombopoiesis. However, alternative possibilities cannot presently be ruled out. For example, in GATA-1-null mice, GATA-2 may be able to compensate for loss of GATA-1 in early MK differentiation (but not later) in a FOG-1-dependent pathway. While these issues are sorted out in mice with deficiencies of multiple GATA proteins, it is fair to conclude that both GATA-1 and FOG-1 exert a profound influence on several aspects of MK differentiation.

Fli-1 in Megakaryopoiesis

Winged helix-turn-helix proteins of the Ets family activate transcription by binding to purine-rich sequences in gene promoters. If MK-specific gene expression is commonly controlled through GATA and Ets cis-elements, it stands to reason that Ets-family transcription factors are responsible for significant aspects of MK differentiation, a prediction that is fulfilled in part by analysis of mice lacking Fli-1 [39]. The Fli-1 protooncogene, associated with Ewing's sarcoma in humans and experimental Friend virus-induced erythroleukemias in mice, has a DNA-binding specificity that distinguishes it from other Ets proteins [40]. Whereas Fli-1+/– ES cells contribute to the megakaryocytic lineage in chimeric mice, Fli-1–/– ES cells do not, which suggests either that Fli-1 is required to generate MKs or that MKs lacking this protein are at a competitive disadvantage [39]. Fli-1–/– mouse fetuses die at mid-gestation, principally as a result of vascular developmental aberrations, but careful study of MKs cultured from their fetal livers prior to death reveals several abnormalities [39]. MK progenitors are modestly increased in number compared to control littermates, and the cultured MKs have a poorly developed cytoplasm with few alpha-granules and disorganized internal membranes. Of the many examined transcripts considered to be regulated by Ets proteins, including c-Mpl, GPIIB, GPIX, and the von Willebrand factor, only GPIX mRNA levels are significantly reduced. The role of GPIX in thrombopoiesis, per se, is unclear, but there are likely to be other transciptional targets of Fli-1 as well. One caveat to these experiments is that cytology could only be examined for MKs cultured in a rich mix of recombinant cytokines, and so the biology of native MKs is uncertain.

Interestingly, the Fli-1 gene locus on the long arm of human chromosome 11 is usually deleted in patients with the rare Jacobsen syndrome, a contiguous-gene trait that results in thrombocytopenia, mild mental retardation, and characteristic cardiac and facial anomalies; the recently described Paris-Trousseau syndrome, in which platelet alpha-granules are abnormally large [41] and a microdeletion of chromosome 11 is present, may be a variant of the Jacobsen syndrome [42]. Both the Ets-1 and Fli-1 genes map near these deletions, and the MK abnormalities seen in mice in the absence of Fli-1 may reflect the same pathophysiology [39].

NF-E2, an Essential Regulator of Platelet Release

Background and Gene-Targeting Studies

Like GATA-1, the heterodimeric basic-leucine zipper transcription factor NF-E2 was also identified on the basis of its presumptive role in regulating gene expression in maturing erythrocytes. However, knockout mice lacking the 45kD hematopoietic-restricted subunit of NF-E2 (p45) display only a mild erythroid cell phenotype [43-45], whereas defects in thrombopoiesis are dramatic [32]. In the absence of NF-E2 function, early MK differentiation, including endomitosis and subsequent expansion of cytoplasmic volume, is overtly normal [32], although proliferation of MK progenitors is mildly reduced [44]. In contrast, platelet release is hampered severely, perhaps absolutely, and NF-E2-deficient animals lack normal circulating platelets. Their MKs reveal the stigmata of impaired platelet release, including enlarged cytoplasm, disorganized internal membranes, and failure to form proplatelets in culture; this phenotype is reproduced when lethally irradiated wild-type mice are rescued by p45 NF-E2-deficient fetal livers [32, 33].

NF-E2 is a heterodimeric factor, and the p45 subunit appears to be free to associate with any of three 18-20kD proteins that constitute the small-Maf subfamily and are expressed widely, in partially overlapping patterns [46, 47]. Of the three small-Maf proteins, only MafG and MafK are expressed in MKs, with MafG predominating in cells with advanced differentiation [46, 48]. MafK–/– mice have no measurable defects [47, 49] and isolated absence of MafG leads to mild thrombocytopenia [46] whereas MafG–/–MafK–/– compound mutant mice virtually phenocopy the absence of p45 NF-E2 with profound thrombocytopenia in vivo and loss of proplatelets in cultured MKs [50]. Considered together with the detailed characterization of p45 NF-E2-null MKs, these observations point to late-expressed MK genes that mediate platelet release as the key transcriptional targets of the NF-E2 complex.

Regulation of MK Transcription by NF-E2

Only two presumptive transcriptional targets of NF-E2 have been identified to date, thromboxane synthase (TXS) and β1 tubulin [51, 52]. The murine TXS gene locus encompasses a functional NF-E2 cis-element [53], and its expression is reduced in the absence of either p45 NF-E2 [51] or a combination of small-Maf proteins [54]. However, treatment with aspirin, a potent cyclooxygenase inhibitor, does not interfere with proplatelet formation in vitro [55], and our limited understanding of basic mechanisms does not suggest an obvious function for TXS within present models of thrombopoiesis. In contrast, β1 tubulin has an immediately plausible role in platelet release. Of the known vertebrate β-tubulin isoforms, mammalian β1 and its ortholog chicken β6 are the most divergent, are restricted in expression to blood cells, and are found within peripheral MT rings in platelets and nucleated erythrocytes, respectively [56-58]. Mammalian β1 tubulin is exquisitely restricted to mature MKs, where it localizes in proplatelet shafts and the prospective platelet marginal band, and, to a lesser extent, to embryonic erythrocytes; expression in both sites is dependent on NF-E2 [52]. Moreover, knockout mice lacking β1 tubulin are thrombocytopenic as a result of defects in MK proplatelet formation, and their platelets carry structurally defective marginal MT bands [59]. Although it remains unclear whether the β1 tubulin gene is a direct transcriptional target of NF-E2, its characterization as an essential gene that is specifically and highly downregulated in the absence of NF-E2 in vivo provides an immensely satisfying entree into the molecular analysis of platelet release.

A series of recent reports allows further comment on MK gene regulation by NF-E2. Most relevant to the mechanism of thrombopoiesis is the observation that isolated expression of β1 tubulin is insufficient to rescue the proplatelet defect in p45 NF-E2–/– MKs [52]. Indeed, a subtraction library of transcripts that are downregulated in NF-E2-null MKs contains a number of other genes, many of which encode structural or regulatory components of the MT cytoskeleton (Lecine and Shivdasani, unpublished data). Hence, it is reasonable to speculate that NF-E2 regulates a broad program of gene expression in maturing MKs that is required to reorganize the MT cytoskeleton and initiate proplatelet formation. Other cellular defects in NF-E2-deficient MKs may also help illuminate underlying mechanisms. For example, p45 NF-E2–/– MKs fail to develop normal inside-out signaling through the αIIbβ3 integrin, as inferred from their inability to bind fibrinogen in response to platelet agonists [60]. Because humans or mice with loss of αIIbβ3 are not thrombocytopenic [61], however, it is unclear whether this finding reflects the general arrest in differentiation of NF-E2-null MKs or points, more interestingly, to a subtle role for αIIbβ3 in some aspect of proplatelet formation.

Second, some components of a transcriptionally active protein complex that includes NF-E2 have been characterized, and post-translational modifications of NF-E2 that may regulate its function are being elucidated. p45 NF-E2 interacts directly with the TATA-binding protein-associated factor TAFII130 [62] and with the cAMP-response element-binding protein-binding protein (CBP) [63]. Both phosphorylation of the p45 subunit by cAMP-dependent protein kinase [64] and acetylation of the small-Maf subunit by CBP [65] may modulate NF-E2 function. Although the latter post-translational modifications were uncovered in erythroid cell lines, similar regulatory factors are likely to operate within developing MKs as well. Incidentally, acetylation also modulates the transcriptional activity of GATA-1 [66, 67].

Finally, there is at least one additional layer of regulation of NF-E2-dependent genes in vivo. Homodimers or heterodimers of small-Maf proteins can bind to NF-E2 cis-elements and function as transcriptional repressors [68]. Indeed, both proplatelet formation and thromboxane synthase mRNA levels are exquisitely sensitive to cellular doses of small-Maf proteins [54], and this finding has important implications for the timing of activation of NF-E2-regulated genes. Although p45 NF-E2 is present early in MK ontogeny, its most obvious and important functions are evident only in terminally differentiated cells. One intriguing possibility is that prior to the need for coordinated expression of NF-E2 target genes, they are silenced by virtue of engaging small-Maf dimers and de-repressed later, when protein stability or post-translational modifications favor the activity of p45-p18 heterodimers over those of other small-Maf protein complexes.

Molecular Regulation of MK and Platelet Differentiation

General Considerations

Creating and releasing large numbers of blood platelets, the raison d'etre of mammalian MKs, relies on orchestrated biosynthesis of lineage-specific granular organelles and their precise assembly into anucleate cells with a sophisticated cytoskeletal architecture. GATA-1 and NF-E2 provide one level of control over this process, possibly within a linear hierarchy of transcriptional regulation. These transcription factors might coordinately activate numerous genes that function in concert to mediate MK cytoplasmic reorganization, membrane recruitment, and organelle assembly and transport (Fig. 1). Additional transcription factors undoubtedly regulate the many steps that lay between the commitment of a cell to the MK lineage and fragmentation of the mature MK cytoplasm into individual blood platelets. To understand the molecular basis of thrombopoiesis, it is, therefore, useful to consider the functions of MK-restricted transcription factors as well as the transcriptional regulation of essential genes that drive MK differentiation. In this section, recent advances in the genetic analysis of selected thrombocytopenic disorders that provide important clues about the molecular underpinnings of platelet biogenesis are highlighted.

Figure Figure 1..

Graphical representation of the suspected or established roles of individual transcription factors (grey boxes on the right) and some candidate (grey boxes) or presumed (white boxes) transcriptional targets involved in megakaryocyte (MK) differentiation.Transcriptional control of the genesis of bipotential erythroid-MK (Ery-MK) progenitors remains mysterious, whereas FOG-1-null mice suggest an essential role for this transcriptional regulator in propagation or differentiation of committed MK progenitors. Direct studies on mouse or human MK colonies point to positive regulation of proliferation of MK progenitors by CBFA2 and NF-E2, and negative regulation by GATA-1. Inositol polyphosphate 4-phosphatase Type I likely acts downstream of GATA-1 in this function. Further cytoplasmic maturation is impaired in the absence of Fli-1 or GATA-1 in mice and when human GATA-1 is mutated such that it interacts poorly with its cofactor FOG-1. The p45 subunit of NF-E2 may be a direct target of GATA-1 gene regulation, although other genes are undoubtedly also involved. Finally, platelet release is regulated by NF-E2, probably through coordinate activation of many genes that reorganize the cytoskeleton and transport organelles into proplatelets. One essential candidate target gene of NF-E2 is an MK- and platelet-specific isoform of β-tubulin (β1).

Insights from Analysis of Platelet Disorders

In the gunmetal strain of mutant mice, which shows thrombocytopenia and coat color dilution, the MK ultrastructure superficially resembles that in mice lacking NF-E2—granules are sparse and demarcation membranes are frequently disorganized in isolated clusters within an expanded cytoplasm [69]. The mutation responsible for the gunmetal phenotype greatly reduces the activity of Rab geranylgeranyl transferase, the enzyme that attaches lipid groups to the Rab family of small Ras-like GTPases and, hence, targets them to specific membrane compartments [6]. Selected Rab proteins, which as a group mediate a variety of membrane and vesicular transport processes, are thus directly implicated in important aspects of thrombopoiesis. In a parallel development, three groups recently reported that mutations causing a rare group of inherited human macrothrombocytopenic disorders, including the May-Hegglin anomaly and the Sebastian and Fechtner syndromes, reside in a nonmuscle myosin heavy chain gene [70-72]. Together, these observations underscore the importance of organellar transport and actin-based motors in thrombopoiesis and identify some of the key mediators. However, the transcriptional regulation of these clinically important genes is unknown, and future studies should focus on clarifying the extent to which they depend on NF-E2, GATA-1, or other MK transcription factors.

The molecular basis for the majority of human platelet storage pool diseases, or the role of transcription factors in these disorders, is not known [73]. In laboratory mice, by contrast, platelet storage pool deficiencies have yielded rich insights into the molecular regulation of organelle biosynthesis. The mocha and pearl mutations result from defects in the δ and β3a subunits, respectively, of the adaptor protein-3 (AP-3), which participates in clathrin-mediated budding of vesicles from the trans-Golgi network [4, 74]. Beige mice and patients with the Chediak-Higashi syndrome have mutations in the lysosomal traffic regulator Lyst [3], whereas pallid mice carry mutations in a novel gene that interacts physically with syntaxin-13, a t-SNARE protein that localizes to endosomal membranes and mediates vesicle membrane fusion [5]. Organelle biogenesis is a cardinal feature of thrombopoiesis and understanding its transcriptional basis is of obvious interest. It is worth noting, however, that most of the genes implicated in these steps are expressed widely and that mutations also result in melanocyte, leukocyte, and neuronal abnormalities. Hence, mechanisms for their transcriptional regulation may either be specific in MKs or shared with other cell types.

Equally relevant from both scientific and clinical perspectives are the molecular defects associated with distinct human thrombocytopenic syndromes. Patients with congenital amegakaryocytic thrombocytopenia phenocopy mice with targeted loss of c-Mpl, the surface receptor for thrombopoietin (TPO), and uniformly harbor point mutations in the c-Mpl gene [75-77]. Although the genetic defect in the equally rare syndrome of thrombocytopenia with absent radii (TAR) remains to be elucidated, MKs and platelets in this disorder also reveal abnormalities in the TPO/c-Mpl signaling axis [78, 79]. Both groups of patients highlight the predictably central role of c-Mpl in regulating thrombopoiesis and raise interest in understanding the transcriptional control of c-Mpl and other signaling components in this pathway. At least some portion of MK-restricted c-Mpl gene expression appears to be regulated by GATA- and Ets-family transcription factors [80, 81], but little is known about other signaling intermediates, many of which are widely expressed.

Other Transcription Factors Associated with Human Diseases of MKs and Platelets

Several homeobox genes are expressed in hematopoietic cells, often in patterns that change dramatically with lineage differentiation [82]. These observations raise the possibility of important functions that remain, for the most part, mysterious. The HoxA10 gene, encoded in the HoxA cluster, is coexpressed in MKs and in the developing forelimb, and its overexpression in mouse bone marrow cells selectively drives expansion of MK progenitors [83]. Heterozygous mutations in the neighboring human gene HoxA11 were recently identified in two independent kindreds with an extremely rare syndrome of amegakaryocytic thrombocytopenia that is associated with radioulnar synostosis but is distinct from the TAR syndrome [84]. Curiously, HoxA11 mRNA is undetectable in normal human platelets or TPO-stimulated stem cells, and these patients retain one wild-type HoxA11 allele; the relationship between the molecular and cellular defects is therefore unclear. Nevertheless, the putative association between a Hox gene and MK and skeletal abnormalities is provocative because it hints at a role for other Hox genes in the more common but enigmatic TAR syndrome. Mice with various Hox gene disruptions manifest discrete skeletal anomalies, but very few Hox genes have an established function in hematopoiesis. In particular, the HoxA9 gene is fused to a nucleoporin gene in rare cases of acute myeloid leukemia (AML) [85], and HoxA9–/– mice show myelolymphoid cell abnormalities, but these mice have normal platelet counts and normal numbers of bone marrow MKs [86].

Gene loci disrupted by recurring chromosomal translocations in acute leukemias have been a rich source of insight into normal mechanisms of development and hematopoiesis [17, 87]. The t(1;22) translocation, associated exclusively with infant megakaryoblastic leukemia, combines two widely expressed and previously uncharacterized genes to create a fusion product [88]. One of the fusion partners, christened MAL-22 (for megakaryoblastic acute leukemia-chromosome 22), is related by sequence to two other predicted human proteins (mal-16 and mal-17) and to a Drosophila protein (D-mal) implicated in low-specificity binding to AT-rich regions of DNA. The second fusion partner, named OTT (for one twenty-two) and also related to a Drosophila protein (split-end), possesses features suggestive of a role either as a classical transcription factor or as an RNA-binding protein. Thus, although the functions of the oncogenic fusion protein and of the parent genes remain obscure, it is reasonable to speculate that one or both of them plays a special role in driving early hematopoietic progenitors along the MK lineage.

Along the same lines, it is worth noting that patients with Down syndrome have a particular predilection to develop acute megakaryoblastic leukemia, ordinarily a rare disease [89]. This clinical observation may implicate genes encoded on human chromosome 21 in fundamental aspects of early MK ontogeny, although it is unknown whether this reflects the functions of transcription factors per se. Nevertheless, recent reports have pointed to the role of a well-characterized transcription factor subunit encoded on chromosome 21 in familial platelet disorder with predisposition to AML (FPD-AML), a distinct, rare disorder of autosomal dominant thrombocytopenia associated with ∼30% likelihood of developing adult acute myeloid (non-MK) leukemia. Patients with this disorder carry germline mutations in a single copy of the CBFA2/AML-1 gene [90] and the case that it results from haploinsufficiency is compelling. Although this gene is rightly regarded as a central regulator of early hematopoiesis because of its frequent targeting in leukemogenic chromosomal translocations [91] as well as the phenotypes of knockout mice [92, 93], an apparently separate role in thrombopoiesis is surprising and intriguing. The moderate thrombocytopenia can be explained by the observation that MK colonies cultured from patients with FPD-AML are fewer and smaller than normal [90]. However, these patients manifest additional defects in platelet aggregation, which suggests that, like GATA-1 and NF-E2, the core-binding factor (CBF) may also regulate a subset of genes required for proper platelet assembly or function. CBFA2–/– mice have early embryonic lethality [92, 93] and CBFA2+/– mice do not show platelet defects (D.G. Gilliland, personal communication), so it may only be possible to understand the mechanisms by which CBF influences platelet function through further investigation of human samples.

Other Transcription Factors Expressed or Implicated in MKs

Despite the apparently central role of Ets transcription factors in regulating MK-expressed genes, to date only the mouse knockout of Fli-1 has revealed a phenotype to suggest a role in platelet differentiation, and in this case only a single MK gene, GPIX, is downregulated. Thus, either additional key MK Ets factors remain to be identified or there is some degree of functional redundancy among family members. Although independent experiments in bipotential cell lines have implicated Fli-1 and Spi-1/Pu.1 as potential effectors of an MK program of gene expression [94-96], at least Pu.1–/– mice fail to reveal defects in the MK-platelet axis [97]. The activity of many MK promoters is likely to be influenced by precise combinations of GATA and Ets factors and their cognate coactivators, and it will be important to understand the underlying mechanisms.

A number of other transcription factors are known to be expressed with relative, albeit variable, selectivity in maturing or terminally differentiated MKs, but their functions in this context are largely unknown. This group includes the proto-oncogenes SCL/Tal-1, c-Myc, c-Myb, and the sno/ski family [98]. Among these, c-Myb may be of lesser interest because MKs are uniquely spared in the c-Myb–/– phenotype of hematopoietic failure [99], while c-ski attracts attention by virtue of its selective expression in bipotential erythro-MK cells [100]. Combinations of these and known or yet unidentified transcription factors probably cooperate to generate MK-specific patterns of gene expression.

Future Directions and Open Questions

Transcriptional Control of Cell Fate

The most interesting functions of lineage-restricted transcription factors are exercised at the branch points of cellular hierarchies, where cell-specific programs of gene expression are activated or reinforced to permit emergence of a lineage identity. While there is increasing evidence that MKs derive from a bipotential erythro-MK progenitor [101], there are presently few clues about the transcriptional basis for defining the MK lineage. Perhaps most frustrating is the repeated observation that important aspects of differentiation of both erythroid cells and MKs depend, to varying degrees, on the same handful of hematopoietic transcription factors: GATA proteins, FOG-1, and NF-E2. Hence, the factors responsible for distinguishing early committed MKs from their erythroid siblings remain elusive. It is, of course, possible that the essential differences between sister lineages arise entirely from subtle qualitative and/or quantitative changes in the status of these and other shared transcriptional regulators. If that is the case, then satisfying solutions to the open questions must await the development of new experimental techniques, because present analytical methods usually preclude exploring nuances in transcriptional regulation.

Transcriptional Control of Cell Biological Processes Unique to MKs

Many cellular processes unique or largely restricted to MKs, including endomitosis, biogenesis of specific granules, and formation and fragmentation of proplatelets, are exceedingly intricate and must be coordinated through transcriptional control of comparable complexity. To date, our insights into these processes have derived principally from genetic studies, but in the future they will increasingly be informed by biochemical extension of the genetic findings. Just as some of our present understanding of platelet synthesis originated in experiments designed to study erythrocytes or melanocytes, so too will a dedicated examination of MK differentiation shed light on cell biological processes shared with other lineages. Studies directed to clarify mechanisms of thrombopoiesis may hence be harnessed not only to therapeutic ends but also to elucidate general aspects of organelle assembly and cellular morphogenesis.

Acknowledgements

Supported in part by NIH grant R01-HL63143. I am grateful to Harald Schulze and Sanjay Tiwari for critical comments on the manuscript. R.A.S. is a Scholar of the Leukemia and Lymphoma Society.

Ancillary