¹Division of Hematology/Oncology, Department of Pediatrics, Children's Hospital Boston and the Dana-Farber Cancer Institute, Boston, Massachusetts, USA

²Howard Hughes Medical Institute, Harvard Medical School, Boston, Massachusetts, USA

Correspondence to: S H Orkin, Howard Hughes Medical Institute, Harvard Medical School, Boston, Massachusetts, USA. E-mail: orkin@bloodgroup.tch.harvard.edu

Previous work has demonstrated that lineage-specific transcription factors play essential roles in red blood cell development. More recent studies have shown that these factors participate in critical protein-protein interactions in addition to binding DNA. The zinc finger transcription factor GATA-1, a central mediator of erythroid gene expression, interacts with multiple proteins including FOG-1, EKLF, SP1, CBP/p300 and PU.1. The mechanisms by which these interactions influence GATA-1 function, as well as any possible relationships between these seemingly disparate complexes, remain incompletely understood. However, several new findings have provided further insight into the functional significance of some of these interactions. Studies involving point mutants of GATA-1 have shown that a direct physical interaction between GATA-1 and FOG-1 is essential for normal human erythroid and megakaryocyte maturation in vivo. In addition, evidence has emerged that physical interaction between GATA-1 and the myeloid/lymphoid specific factor PU.1, an oncogene implicated in murine erythroleukemia, acts to functionally cross-antagonize one another. This provides a possible mechanism by which dysregulated expression of hematopoietic transcription factors leads to lineage maturation arrest in leukemias.

Oncogene (2002) 21, 3368-3376 DOI: 10.1038/sj/onc/1205326

GATA-1; FOG-1; PU.1; EKLF; transcription factor; cross-antagonism

Introduction

The hematopoietic system in vertebrates requires a continual replacement of cells since the end product, mature blood cells, have limited lifespans. This is achieved through self-renewal of pluripotential hematopoietic stem cells and their sequential commitment to progenitor and precursor cells of gradually more restricted potential. Over the past couple of decades, efforts have been made to understand the molecular mechanisms that determine these cell fate decisions. Evidence has accumulated demonstrating that lineage-specific transcription factors (likely working in conjunction with general transcription factors) play essential roles in this process (for review, see (Shivdasani and Orkin, 1996)). Studies employing targeted gene disruption in mice reveal characteristic blocks in hematopoietic maturation that occur in the absence of these different factors (Figure 1). Importantly, the majority of these factors (denoted by an asterisk in Figure 1) are also associated with chromosomal translocations or viral insertions in various human and murine leukemias. This supports a model in which the dysregulation of normal transcriptional machinery plays a causal role in these hematologic malignancies. Further complexity in our understanding of this process has recently been added by the recognition that critical protein-protein interactions modulate the activities of these transcription factors. This review will focus on recent developments that provide new insight into the significance of these interactions in erythroid development. In particular, it will address findings demonstrating the functional cross-antagonism between different lineage-specific transcription factors and its implications in leukemogenesis. Possible relationships between these different complexes will be discussed and new approaches to their study reviewed.

GATA-1

GATA-1, a zinc finger (zf) transcription factor, plays a central role in erythroid development. It was first identified by its ability to bind functionally important DNA regulatory sequences found in globin genes (Tsai et al., 1989; Evans and Felsenfeld, 1989). Since then, GATA-binding motifs ((T/A)GATA(A/G)) have been identified in the promoters and/or enhancers of virtually all erythroid and megakaryocytic-specific genes studied (Orkin, 1992; Weiss and Orkin, 1995a). GATA-1 contains two zinc fingers, both of the Cys-X₂-Cys-X₁₇-Cys-X₂-Cys configuration. The carboxyl terminal zinc finger is responsible for high-affinity DNA binding, whereas the amino terminal zinc finger stabilizes the interaction (Martin and Orkin, 1990).

Expression of GATA-1 is restricted to erythroid, megakaryocytic, eosinophilic, mast and multipotential precursor cells within the hematopoietic system (Tsai et al., 1989; Evans and Felsenfeld, 1989). The sole non-hematopoietic site of expression is the Sertoli cells of the testis (Ito et al., 1993). Gene targeting studies in mice have shown that GATA-1 is essential for normal erythropoiesis (Pevny et al., 1991). GATA-1 hemizygous male knock-out mice (GATA-1 is located on the X-chromosome) die in mid-embryonic gestation (e10.5) from severe anemia with arrest in erythroid maturation at a proerythroblast-like stage (Fujiwara et al., 1996). In vitro differentiated GATA-1^- ES cells likewise fail to mature past the proerythroblast stage and undergo rapid apoptosis, indicating a role for GATA-1 in cell survival as well as maturation (Weiss and Orkin, 1995b). Loss of GATA-1 expression in megakaryocytes also leads to defects in maturation characterized by impaired endoreduplication and granule formation, disorganized platelet demarcation membrane synthesis, and hyperproliferative growth (Shivdasani et al., 1997; Vyas et al., 1999).

A sea of proteins

In addition to binding DNA, transcription factors often make critical protein-protein interactions that modulate their activities. Several proteins have been reported to interact physically with GATA-1 including FOG-1 (Tsang et al., 1997), LMO-2 (Osada et al., 1995), EKLF/Sp1 (Merika and Orkin, 1995), p300/CBP (Blobel et al., 1998), and PU.1 (Zhang et al., 1999; Rekhtman et al., 1999; Nerlov et al., 2000) (Figure 2). The challenge has been to sort out which of these interactions are functionally significant, determine the mechanisms by which they modulate GATA-1 activity, and understand how they relate to one another, if at all.

FOG-1 (friend of GATA)

FOG-1 is a 998 amino acid nuclear multitype zinc finger protein that was identified in a yeast two-hybrid screen for GATA-1 interacting proteins (Tsang et al., 1997). It binds specifically to the amino zinc finger of GATA-1. FOG-1 contains nine predicted zinc fingers, four of which (zfs 1,5,6 and 9) individually mediate interactions with GATA-1 (Fox et al., 1999). FOG-1 is expressed abundantly in erythroid and megakaryocytic cells, and is co-expressed with GATA-1 during development (Tsang et al., 1997). As its interacting partner GATA-1, FOG-1 plays an essential role in erythropoiesis and megakaryopoiesis. FOG-1^-/- mice die in mid-embryonic gestation (e10.5-11.5) due to severe anemia with arrest in erythroid maturation at a stage similar to that observed in the GATA-1^- mice, providing genetic evidence that these two factors act in a common pathway (Tsang et al., 1998). However, in contrast to GATA-1^- mice, FOG-1^-/- mice exhibit complete failure of megakaryopoiesis, indicating that FOG-1 also has a GATA-1 independent role in early megakaryopoiesis.

Studies utilizing a point mutant of GATA-1 with markedly reduced affinity for FOG-1 (but normal DNA binding activity) have shown that a direct physical interaction between GATA-1 and FOG-1 is required for normal erythropoiesis in vitro (Crispino et al., 1999). A similar GATA-1 gene missense mutation was recently identified in affected members of a family with severe congenital dyserythropoietic anemia and thrombocytopenia (Nichols et al., 2000). The megakaryocytes from these patients resemble murine GATA-1 deficient megakaryocytes suggesting that a GATA-1:FOG-1 interaction is also critical for late stages of megakaryopoiesis.

Both GATA-1 and FOG-1 are members of protein families. FOG-2, a second mammalian FOG gene, is expressed in cardiac, neural and gonadal tissues (Tevosian et al., 1999; Svensson et al., 1999; Lu et al., 1999). FOG-2^-/- mice die during embryogenesis from cardiac defects characterized by thin ventricular myocardium, Tetralogy of Fallot malformation, common atrioventricular valve, and defective coronary vasculature (Tevosian et al., 2000; Svensson et al., 2000). Six members of the GATA family are known. GATA-1, -2, and -3 have roles predominantly within the hematopoietic system, whereas GATA-4, -5, and -6 are involved in non-hematopoietic tissues (Orkin, 1992). All GATA family members are capable of interacting with either FOG-1 or FOG-2 through conserved residues in their amino zinc fingers. Knock-in of a FOG non-interacting point mutation into the GATA-4 gene, one of the cardiac expressed GATA factors, results in embryonic death and a constellation of heart defects similar to that observed in the FOG-2^-/- mice (Crispino et al., 2001). This indicates that interactions between GATA and FOG proteins are critical for multiple developmental processes.

Efforts have now been focused on understanding the mechanism by which FOG proteins influence GATA-mediated processes. Cis-acting regulatory sequences of erythroid and megakaryocyte-specific genes often contain multiple GATA DNA binding sites, and these are often situated at considerable distances from one another in promoter and enhancer elements. Since a single FOG molecule can potentially interact simultaneously with multiple GATA molecules, one possibility is that FOG acts as a molecular bridge between these sites. This could bring distal enhancer elements into proximity of the promoter through a DNA looping mechanism. However, mutants of FOG-1 that are capable of binding only a single molecule of GATA-1 rescue erythroid maturation from a FOG-1 deficient cell line as well as the wildtype molecule, providing evidence against such a model in its simplest form (Cantor AB, Katz SG, and Orkin SH, submitted for publication). Another possibility is that FOG provides a transcriptional activation domain either by itself or via interaction with another transcriptional coactivator. This predicts that protein domains outside of the GATA-binding zinc fingers would be required for its activity. However, FOG-1 molecules containing extensive and overlapping deletions spanning the entire molecule (but retaining at least one GATA-binding zinc finger) also are able to rescue erythroid maturation of the FOG-1^-/- cell line (Cantor AB, Katz SG, and Orkin SH, submitted for publication). This suggests that a simple interaction between FOG-1 and GATA-1 is sufficient to activate GATA-1. This could occur through an allosteric change in GATA-1 or perhaps by the displacement of a repressor protein bound to GATA-1. In support of this latter model, Evans and Felsenfeld have demonstrated that GATA-1 has reduced transcriptional activity in hematopoietic cells compared to non-hematopoietic cells, suggesting that the hematopoietic environment somehow dampens GATA response (Evans and Felsenfeld, 1991). Testing of this directly in the FOG-1^-/- cells has been precluded by the lack of stability of GATA-binding FOG-1 zinc fingers expressed by themselves. Further studies will be required to elucidate the mechanism of action of FOG proteins.

SCL (TAL1) and LMO2 (RBTN2)

SCL (TAL 1), a member of the basic helix-loop-helix (bHLH) family of transcription factors, was first identified by its involvement in recurrent chromosomal translocations in patients with T-cell acute lymphoblastic leukemia (ALL), (Begley et al., 1989; Finger et al., 1989; Chen et al., 1990). Misexpression of SCL by chromosomal translocation or other mechanisms is thought to underlie the basis for cell transformation (Bash et al., 1993). SCL, like other member of the bHLH family, heterodimerizes with E proteins, such as E2A, and binds to canonical DNA sequences, CANNTG, termed E-boxes. These sequences are found in cis-acting regulatory elements of erythroid-specific genes in conjunction with GATA-binding motifs. SCL^-/- ES cells fail to contribute to all hematopoietic lineages in adult chimeric mice, and germline SCL^-/- mice die during embryogenesis with a complete absence of yolk sac blood, indicating that SCL plays an essential role in early hematopoiesis (Porcher et al., 1996; Shivdasani et al., 1995; Robb et al., 1995).

LMO2 (RBTN2), a LIM domain-containing protein, has also been implicated in the pathogenesis of certain T-Cell ALLs. The loss of function phenotype is identical to that of SCL, consistent with the finding that SCL and LMO2 physically interact with high affinity (Warren et al., 1994; Osada et al., 1995). Using CASTing and gel electrophoresis mobility shift assays (EMSA), Rabbitts and his colleagues have provided evidence for a pentameric complex formed between SCL/E2A, LM02, GATA-1, and another LIM-containing protein, Ldb1 (Wadman et al., 1997). From these data, they have proposed a model in which LMO2/Ldb1 acts to bridge a SCL/E2A heterodimer bound to an E-box sequence and GATA-1 bound to a GATA element located about nine base pairs away (Figure 3). The precise role of this presumptive complex in hematopoietic development requires additional work in the future.

DNA-binding independent roles of SCL have also been documented. We and our colleagues have shown that a mutant of SCL that lacks a DNA-binding domain is capable of rescuing primitive erythropoiesis from in vitro differentiated SCL^-/- ES cells (Porcher et al., 1999). Moreover, DNA-binding is dispensable for ectopically-expressed SCL to induce T-cell leukemia/lymphoma in mice (O'Neil et al., 2001). This latter phenomenon may be due to interference with normal E protein function in T-cells. These findings further underscore the critical nature of protein-protein interactions of hematopoietic transcription factors in both normal and pathophysiologic processes.

EKLF (Erythroid Kruppel-like Factor)

Globin gene expression is regulated in a developmentally specific pattern. In humans, beta-like embryonic globin (hemoglobin ) is expressed first in embryonic red blood cells within the yolk sac blood islands. Thereafter, fetal globins (hemoglobin A and G) are expressed in fetal liver-derived definitive red blood cells. Finally, adult - and -globins are expressed at increasing levels around the time of birth within the bone marrow derived red cells. The genes that encodes these proteins are arranged on chromosome 11 in the order in which they are expressed. In mice, the -globin cluster is similarly arranged (5'-, h₀, h₁, _maj, and _min-3'), but unlike humans, lacks fetal-specific globin genes. Each of these genes (in both mice and humans) is regulated by proximal cis-acting sequences. However, high level expression requires the presence of a locus control region (LCR) located approximately 6 kb 5' to the gene cluster.

Erythroid Kruppel-like factor (EKLF) is a zinc finger transcription factor that was first identified by subtractive hybridization of DS-19 MEL and J774 monocyte-macrophage cell lines (Miller and Bieker, 1993). The EKLF-DNA binding site consensus sequence 5'-NCNCNCCCN-3' corresponds to a functionally important motif within the adult -globin gene promoter. Mutation of this sequence in humans is found in some patients with -thalassemia (Orkin et al., 1982). Targeted disruption of the EKLF gene in mice also results in lethal -thalassemia (Perkins et al., 1995; Nuez et al., 1995). EKLF knockout mice die from severe anemia at the fetal liver stage due to failure of adult -globin gene activation. EKLF^-/- mice containing a complete human -globin locus transgene have reduced levels of -globin, but elevated levels of -globin expression, compared to wildtype mice containing the same transgene (Perkins et al., 1996; Wijgerde et al., 1996). Taken together, these experiments suggest that EKLF participates in the switch from embryonic or fetal globin to adult -globin expression in humans. New insight into EKLF function has come from recent experiments showing that EKLF requires the presence of a SWI/SNF-related chromatin remodeling complex for its tissue-specific regulation, indicating that it may affect transcription by altering chromatin configuration (Armstrong et al., 1998). A role for EKLF in coordinating erythroid cell proliferation and hemoglobinization has also recently been revealed in studies utilizing a genetically-engineered EKLF^-/- hematopoietic cell line (Coghill et al., 2001).

Binding sites for both EKLF (and the related ubiquitously expressed protein Sp1) and GATA-1 are found in close proximity in cis-regulatory elements of erythroid-specific genes. In addition, both EKLF and Sp1 physically associate with the zinc finger region of GATA-1 and synergistically activate GATA-1 target genes in transiently expressed reporter constructs (Merika and Orkin, 1995). Thus, protein-protein interactions between EKLF and GATA-1 may be involved in facilitating the switch from fetal to adult globin expression, although there is no direct evidence to support this at this time.

CREB-binding protein (CBP)

CREB-binding protein (CBP) is a ubiquitously expressed histone acetlytransferase (HAT) that interacts with a large variety of proteins (Goodman and Smolik, 2000). Co-immunoprecipitation of endogenous proteins in mouse erythroleukemia cells has shown that CBP physically associates with GATA-1 (Blobel et al., 1998). This interaction occurs through the zinc fingers of GATA-1 and the E1A binding domain of CBP. Moreover, CBP markedly stimulates GATA-1's transcriptional activity in transient transfection experiments in non-hematopoietic cells. Since acetylation of histone proteins correlates with transcriptionally active chromatin, a simple model posits that GATA-1 recruits HAT activity (through CBP) to critical regions of chromatin associated with erythroid-specific genes and thereby facilitates their expression. This may be an oversimplification, as GATA-1 protein itself is acetylated on conserved lysine residues by CBP, a modification which appears to enhance its transcriptional activity (Hung et al., 1999; Boyes et al., 1998). Mutation of the acetylated lysine residues to arginine (which can not be acetylated) markedly impairs GATA-1's ability to restore erythroid maturation in a GATA-1 deficient cell line, implying that this modification plays an important functional role. The exact mechanism of this enhanced activity is controversial in that data are conflicting as to possible effects on DNA-binding (Hung et al., 1999; Boyes et al., 1998). Additional circumstantial evidence for a role of CBP in hematopoiesis has been provided by the observation that haploinsufficiency of CBP in mice leads to a bone marrow failure syndrome and a strikingly high incidence of hematologic malignancies (Kung et al., 2000).

PU.1

PU.1, a member of the ets family of transcription factors, plays an essential role in granulocytic, monocytic and lymphoid development (Scott et al., 1994; McKercher et al, 1996; Hromas et al, 1993). It has previously been shown that Friend virus causes murine erythroleukemia by the proviral insertion of the spleen focus forming component into the PU.1 gene locus. This leads to dysregulated expression of PU.1 in erythroid lineage cells, which has been thought to underlie the mechanism of cellular transformation (Moreau-Gachelin et al., 1989). Additional evidence has emerged from observations that transgenic mice that overexpress PU.1 in erythroid cells develop erythroleukemias at higher rates than non-transgenic animals, and retrovirally expressed PU.1 immortalizes bone marrow-derived erythroblasts at high efficiency (Moreau-Gachelin et al., 1996; Schuetze et al., 1993). Conversely, ectopic expression of GATA-1 in myelomonocytic cells transforms them into erythroid, megakaryocytic and eosinophilic cells (Kulessa et al., 1995; Visvader et al., 1992). New reports from several groups have suggested a mechanism to explain these phenomena (Rekhtman et al., 1999; Zhang et al., 1999, Zhang et al., 2000; Nerlov et al., 2000). These investigators demonstrate that PU.1 and GATA-1 functionally cross-antagonize one another through direct physical interaction.

Thus, a block in GATA-1 function by PU.1 may explain, at least in part, the block in erythroid maturation in these erythroleukemias. Skoultchi and his colleagues have provided strong evidence for this model (Rekhtman et al., 1999). They demonstrated that ectopic expression of PU.1 in erythroid cells blocks their maturation in vivo. However, overexpression of GATA-1 in these cells restores normal erythroid differentiation.

Putting the puzzle together

As illustrated by the preceding discussion, numerous tissue-restricted transcription factors have been identified and their role in erythropoiesis validated by in vivo experimental systems. Strong evidence has been provided demonstrating that these factors participate in critical protein-protein interactions and, as such, function as multiprotein complexes. The next important step to understanding the transcriptional regulation of erythropoiesis will be to elucidate how these different complexes relate to one another. It is reasonable to speculate that the composition of these complexes changes throughout erythroid development depending on both spatial and temporal contexts. It also seems likely that additional factors, perhaps as yet unknown, may also participate in these complexes. Perhaps the discovery of such factors will help clarify the seemingly disjointed array of complexes and facilitate a better understanding of the molecular basis of erythroid differentiation.

New approaches to study transcriptional regulation of hematopoiesis

Classical gene-targeting techniques in mice have provided invaluable insights into the role of different transcription factors in erythropoiesis. However, these whole animal approaches by themselves are limited by the time required to generate such animals, limited biologic material in cases of embryonic lethality, inability to assess functional roles at stages past a primary block, and in vivo compensatory mechanisms that may mask effects. New approaches are needed to complement these techniques.

Cre-lox P recombinase system

Cre recombinase is a bacterial enzyme that specifically catalyzes recombination at characteristic short DNA sequences referred to as loxP sites (for review, see Rajewsky et al., 1996). These sequences can be inserted at regions flanking a gene or gene segment of interest using classical gene targeting approaches. Mice containing these 'floxed' alleles are then crossed to mice engineered to express Cre recombinase under a tissue specific and/or inducible promoter. This approach allows for the study of gene function in a tissue-specific and/or temporal-specific manner, circumventing the problems of multi-system effects or early lethality due to loss of gene expression. Such a system, utilizing transgenic mice expressing Cre recombinase under the control of GATA-1 regulatory sequences, has recently been employed to develop a mouse model for paroxysmal nocturnal hemoglobinuria type II (Jasinski et al., 2001).

Genetically engineered cell lines

Murine ES cells can be differentiated in vitro to multipotential hematopoietic precursors as well as mature blood cells (Keller et al., 1993). Retroviral expression of Hox-11 or combinations of v-raf and v-myc have been shown to immortalize hematopoietic precursor cells derived in this manner (Keller et al., 1998; Coghill et al., 2001). These cell lines retain both growth factor dependency and differentiation capacity. Adaptation of this system using genetically modified ES cells provides cellular based assay systems in well-defined microenvironments to study hematopoietic gene function (Cantor AB, Katz SG, and Orkin SH, submitted for publication; Coghill et al., 2001).

Chromatin immunoprecipitaion assays

Chromatin immunoprecipitation assays represent a valuable tool to investigate the role of transcription factor and/or transcription factor associated proteins at specific gene regulatory sequences. This approach offers the advantage of examining interactions in vivo and in the context of chromatinized DNA. Chromatin is cross-linked by chemical agent, sheared, and then immunoprecipitated with specific antibodies directed against the molecule of study. Cross-linking is then reversed and polymerase chain reaction used to amplify gene sequences of interest.

Functional genomics

Remarkably, many of the same transcriptional molecules used in mammalian hematopoiesis have orthologues in organisms with more rudimentary blood systems. This provides an opportunity to exploit the powerful genetic systems already established for some of these organisms to identify novel genes and test function of known genes involved in mammalian hematopoiesis. Large scale genetic screens in Zebrafish (Danio rerio) using chemically-induced mutants have already uncovered several genes with homology to known mammalian erythroid-specfic genes including ALAS-E and erythroid -spectrin, as well as novel genes such as ferroportin1 (Brownlie et al., 1998; Liao et al., 2000; Donovan et al., 2000).

Drosophila melanogaster contain two hematopoietic-like cell lineages, plasmatocytes/macrophages and crystal cells, which participate in innate immunity. Development of both lineages requires the drosophila GATA factor serpent. Recent work has shown that the orthologue of another mammalian transcription factor, AML-1, is required for crystal cell formation (Lebestky et al., 2000). Conversely, U-shaped, a FOG orthologue, represses the crystal cell lineage (Fossett et al., 2001). Thus, the basic machinery of blood cell formation appears to have been well conserved throughout evolution and warrants the study of mammalian hematopoiesis in these model organisms.

Cross-antagonism of lineage-specific transcription factors: implications for leukemogenesis

As discussed earlier, misexpression of the myeloid/lymphoid transcription factor PU.1 in erythroid lineage cells leads to erythroleukemia in mice. This occurs through the direct interaction and cross-antagonism of PU.1 and GATA-1. Additional examples of cross-antagonism between lineage-specific transcription factors have begun to emerge. Nerlov and his colleagues have shown that the myeloid transcription factor C/EBP- cross-antagonizes FOG-1 during eosinophil development (Querfurth et al., 2000). C/EBP- transcriptionally downregulates FOG-1 expression in multipotential precursors, a prerequisite for development of the eosinophil lineages. Conversely, enforced expression of FOG-1 in eosinophilic lineage cells transforms them into more primitive precursors, a process that requires a protein-protein interaction with GATA-1 to transcriptionally repress C/EBP- target genes.

Pax 5, a paired domain-containing transcription factor, plays an essential role in B lymphoid development (Adams et al., 1992). Busslinger and his colleagues have demonstrated that pro B cells derived from Pax 5-/- mice develop into multiple other tissues including macrophages, osteoclasts, dendritic cells, granulocytes, and natural killer cells after culture in appropriate growth factors (Nutt et al., 1999). However, restoration of Pax 5 expression in these cells restricts them to the B-cell lineage implying that Pax 5 represses these alternate lineage choices under normal conditions.

Likewise, dissociated muscle cells from mice containing homozygous disruption of the muscle-specific Pax 7 gene produced about a 10-fold increase in the number of hematopoietic colonies compared to cells from wildtype mice suggesting that Pax 7 may normally repress the hematopoietic lineage (Seale et al., 2000).

Thus, a new view of the transcriptional regulation of hematopoietic development has begun to emerge. In contrast to a model in which lineage-specific transcription factors play principally positive roles in activating gene programs, these observations suggest that they simultaneously exert inhibitory effects on alternate lineage gene programs. This arises through direct cross-antagonism of alternate lineage-specific transcription factors (Figure 4). This paradigm may provide insight into mechanisms of maturation arrest seen in acute leukemias. Namely, misexpression of an alternate lineage-specific transcription factor (through chromosomal translocation, epigenetic phenomenon, viral insertion, etc.) may block maturation by directly antagonizing the activity of normal transcription factors (Figure 5). Exploitation of these mechanisms may provide therapeutic approaches to malignancy aimed at differentiating cancer cells by manipulation of transcription factor interactions. Support for this notion comes from the work of Tenen and colleagues who have shown that the fusion protein AML1-ETO, the product of a chromosomal translocation found in blast cells of patients with a subtype of acute myelogenous leukemia, downregulates the essential myeloid transcription factor C/EBP (Pabst et al., 2001). However, conditional expression of C/EBP in a cell-line model of this disorder restores normal neutrophilic differentiation. Thus, a better understanding of the multiple partnerships of hematopoietic transcription factors should open up new possibilities for rational drug design in the treatment of leukemias and lymphomas.

Acknowledgements

AB Cantor was supported by an NCI K08 Mentored Clinical Scientist Award (CA 82175-02)

Adams B, Dorfler P, Aguzzi A, Kozmik Z, Urbanek P, Maurer-Fogy I, Busslinger M. (1992). Genes Dev., 6: 1589-1607. MEDLINE

Armstrong JA, Bieker JJ, Emerson BM. (1998). Cell, 95: 93-104. MEDLINE

Bash RO, Hall S, Timmons CF, Crist WM, Amylon M, Pullen J, Carrol AJ, Buchanan GR, Smith RG, Baer R. (1993). Blood, 86: 666-676.

Begley CG, Aplan PD, Denning SM, Haynes BF, Waldmann TA, Kirsch IR. (1989). Proc. Natl. Acad. Sci. USA, 86: 10128-10132. MEDLINE

Blobel GA, Nakajima T, Eckner R, Montminy M, Orkin SH. (1998). Proc. Natl. Acad. Sci. USA, 95: 2061-2066. Article MEDLINE

Boyes J, Byfield P, Nakatani Y, Ogryzko V. (1998). Nature, 396: 594-598. Article MEDLINE

Brownlie A, Donovan A, Pratt SJ, Paw BH, Oates AC, Brugnara C, Witkowska HE, Sassa S, Zon LI. (1998). Nat. Genet., 20: 244-250. Article MEDLINE

Chen Q, Cheng J-T, Tsai L-H, Schneider N, Buchanan G, Carroll A, Crist W, Ozanne B, Siciliano MJ, Baer R. (1990). EMBO J., 9: 415-424. MEDLINE

Coghill E, Eccleston S, Fox V, Cerruti L, Brown C, Cunningham J, Jane S, Perkins A. (2001). Blood, 97: 1861-1868. Article MEDLINE

Crispino JD, Lodish M, Mackay JP, Orkin SH. (1999). Mol. Cell, 3: 219-228. MEDLINE

Crispino JD, Lodish Maya , Thurberg BL, Litovsky SH, Collins T, Molkentin JD, Orkin SH. (2001). Genes Dev., 15: 839-844. MEDLINE

Donovan A, Brownlie A, Zhou Y, Shepard J, Pratt SJ, Moynihan J, Paw BH, Drejer A, Barut B, Zapata A, Law TC, Brugnara C, Lux SE, Pinkus GS, Pinkus JL, Kingsley PD, Palis J, Fleming MD, Andrews NC, Zon LI. (2000). Nature, 403: 776-781. Article MEDLINE

Evans T, Felsenfeld G. (1989). Cell, 58: 877-885. MEDLINE

Evans T, Felsenfeld G. (1991). Mol. Cell Biol., 11: 843-853. MEDLINE

Finger LR, Kagan J, Christopher G, Kurtzberg J, Hershfield MS, Nowell PC, Croce CM. (1989). Proc. Natl. Acad. Sci. USA, 86: 5039-5043. MEDLINE

Fossett N, Tevosian SG, Gajewski K, Zhang Q, Orkin SH, Schulz RA. (2001). Proc. Natl. Acad. Sci. USA, 98: 7342-7347. MEDLINE

Fox AH, Liew C, Holmes M, Kowalski K, Mackay J, Crossley M. (1999). EMBO J., 18: 2812-2822. Article MEDLINE

Fujiwara Y, Browne CP, Cunniff K, Goff SC, Orkin SH. (1996). Proc. Natl. Acad. Sci. USA, 93: 12355-12358. Article MEDLINE

Goodman RH, Smolik S. (2000). Genes Dev., 14: 1553-1577. MEDLINE

Hromas R, Orazi A, Neiman RS, Maki R, Van Beveran C, Moore J, Klemsz M. (1993). Blood, 82: 2998-3004. MEDLINE

Hung H-L, Lau J, Kim AY, Weiss MJ, Blobel GA. (1999). Mol. Cell Biol., 19: 3496-3505. MEDLINE

Ito E, Toki T, Ishihara H, Ohtani H, Gu L, Yokoyama M, Engel JD, Yamamoto M. (1993). Nature, 362: 466-468. MEDLINE

Jasinski M, Keller P, Fujiwara Y, Orkin SH, Bessler M. (2001). Blood, 98: 2248-2255. Article MEDLINE

Keller G, Kennedy M, Papayannopoulou T, Wiles MV. (1993). Mol. Cell. Biol., 13: 473-486. MEDLINE

Keller G, Wall C, Fong AZC, Hawley TS, Hawley RG. (1998). Blood, 92: 877-887. MEDLINE

Kulessa H, Frampton J, Graf T. (1995). Genes Dev., 9: 1250-1262. MEDLINE

Kung AL, Rebel VI, Bronson RT, Ch'ng L-E, Sieff CA, Livingston DM, Yao T-P. (2000). Genes Dev., 14: 272-277. MEDLINE

Lebestky T, Chang T, Hartenstein V, Banerjee U. (2000). Science, 288: 146-149. Article MEDLINE

Liao EC, Paw BH, Peters LL, Zapata A, Pratt SJ, Do CP, Lieschke G, Zon LI. (2000). Development, 127: 5123-5132. MEDLINE

Lu J-R, Mckinsey TA, Xu H, Wang D-Z, Richardson JA, Olson EN. (1999). Mol. Cell. Biol., 19: 4495-4502. MEDLINE

Martin DIK, Orkin SH. (1990). Genes Dev., 4: 1886-1898. MEDLINE

McKercher SR, Torbett BE, Anderson KL, Henkel GW, Vestal DJ, Baribault H, Klemsz M, Feeney AJ, Wu GE, Paige CJ, Maki RA. (1996). EMBO J., 15: 5647-5658. MEDLINE

Merika M, Orkin SH. (1995). Mol. Cell Biol., 15: 2437-2447. MEDLINE

Miller IJ, Bieker JJ. (1993). Mol Cell Biol., 13: 2776-2786. MEDLINE

Moreau-Gachelin F, Ray D, Mattei MG, Tambourin P, Tavitian A. (1989). Oncogene, 4: 1449-1456. MEDLINE

Moreau-Gachelin F, Wendling F, Molina T, Denis N, Titeux M, Grimber G, Briand P, Vainchenker W, Tavitian A. (1996). Mol. Cell. Biol., 16: 2453-2463. MEDLINE

Nerlov C, Querfurth E, Kulessa H, Graf T. (2000). Blood, 95: 2543-2551. MEDLINE

Nichols KE, Crispino JD, Poncz M, White JG, Orkin SH, Maris JM, Weiss MJ. (2000). Nat. Genet., 24: 266-270. Article MEDLINE

Nuez B, Michalovich D, Bygrave A, Ploemacher R, Grosveld F. (1995). Nature, 375: 316-318. MEDLINE

Nutt SL, Heavey B, Rolink AG, Busslinger M. (1999). Nature, 401: 556-562. Article MEDLINE

O'Neil J, Billa M, Oikemus S, Kelliher M. (2001). Oncogene, 20: 3897-3905. MEDLINE

Orkin SH, Kazazian HHJ, Antonarakis SE, Goff SC, Boehm CD, Sexton JP, Waber PG, Giardina PJ. (1982). Nature, 296: 627-631. MEDLINE

Orkin SH. (1992). Blood, 80: 575-581. MEDLINE

Osada H, Grutz G, Axelson H, Forster A, Rabbitts TH. (1995). Proc. Natl. Acad. Sci. USA, 92: 9585-9589. MEDLINE

Pabst T, Mueller BU, Harakawa N, Schoch C, Haferlach T, Behre G, Hiddemann W, Zhang D-E, Tenen DG. (2001). Nat. Med., 7: 444-451. Article MEDLINE

Perkins AC, Gaensler KM, Orkin SH. (1996). Proc. Natl. Acad. Sci. USA, 93: 12267-12271. MEDLINE

Perkins AC, Sharpe AH, Orkin SH. (1995). Nature, 375: 318-322. MEDLINE

Pevny L, Simon CM, Robertson E, Klein WH, TSai SF, D'Agati V, Orkin SH, Costantini F. (1991). Nature, 349: 257-260. MEDLINE

Porcher C, Liao EC, Fujiwara Y, Zon LI, Orkin SH. (1999). Development, 126: 4603-4615. MEDLINE

Porcher C, Swat W, Rockwell K, Fujiwara Y, Alt FW, Orkin SH. (1996). Cell, 86: 47-57. MEDLINE

Querfurth E, Schuster M, Kulessa H, Crispino JD, Doderlein G, Orkin SH, Graf T, Nerlov C. (2000). Genes Dev., 14: 2515-2525. MEDLINE

Rajewsky K, Gu H, Kuhn R, Betz UAK, Muller W, Roes J, Schwenk F. (1996). J. Clin. Invest., 98: 600-603. MEDLINE

Rekhtman N, Radparvar F, Evans T, Skoultchi AI. (1999). Genes Dev., 13: 1398-1411. MEDLINE

Robb L, Lyons I, Li R, Hartley L, Kontgen F, Harvey RP, Metcalf D, Begley CG. (1995). Proc. Natl. Acad. Sci. USA, 92: 7075-7079. MEDLINE

Schuetze S, Stenberg PE, Kabat D. (1993). Mol. Cell. Biol., 13: 5670-5678. MEDLINE

Scott EW, Simon CM, Anastasi J, Singh H. (1994). Science, 265: 1573-1577. MEDLINE

Seale P, Sabourin LA, Girgis-Gabardo A, Mansouri A, Gruss P, Rudnicki MA. (2000). Cell, 102: 777-786. MEDLINE

Shivdasani RA, Fujiwara Y, McDevitt MA, Orkin SH. (1997). EMBO J., 16: 3965-3973. Article MEDLINE

Shivdasani RA, Mayer E, Orkin SH. (1995). Nature, 373: 432-434. MEDLINE

Shivdasani RA, Orkin SH. (1996). Blood, 87: 4025-4039. MEDLINE

Svensson EC, Tufts RL, Polk CE, Leiden JM. (1999). Proc. Natl. Acad. Sci. USA, 96: 956-961. Article MEDLINE

Svensson ED, Huggins GS, Lin H, Clendenin C, Jiang F, Tufts R, Dardik FB, Leiden JM. (2000). Nat. Genet., 25: 353-356. Article MEDLINE

Tevosian SG, Deconinck AE, Cantor AB, Reiff HI, Fujiwara Y, Corfas G, Orkin SH. (1999). Proc. Natl. Aca. Sci. USA, 96: 950-955.

Tevosian SG, Deconinck AE, Tanaka M, Schinke M, Silvio LH, Izumo S, Fujiwara Y, Orkin SH. (2000). Cell, 101: 729-739. MEDLINE

Tsai SF, Marting DIK, Zon LI, D'andrea AD, Wong GG, Orkin SH. (1989). Nature, 339: 446-451. MEDLINE

Tsang AP, Fujiwara Y, Hom DB, Orkin SH. (1998). Genes Dev., 12: 1176-1188. MEDLINE

Tsang AP, Visvader JE, Turner AC, Fujiwara Y, Yu C, Weiss M, Crossley M, Orkin SH. (1997). Cell, 90: 109-119. MEDLINE

Visvader JE, Elefanty AG, Strasser A, Adams JM. (1992). EMBO J., 11: 4557-4564. MEDLINE

Vyas P, Ault K, Jackson CJ, Orkin SH, Shivdasani R. (1999). Blood, 93: 2867-2875. MEDLINE

Wadman IA, Osada H, Grutz GG, Agulnick AD, Westphal H, Forster A, Rabbitts TH. (1997). EMBO J., 16: 3145-3157. Article MEDLINE

Warren AJ, Colledge WH, Carlton MBL, Evans MJ, Smith AJH, Rabbitts TH. (1994). Cell, 78: 45-58. MEDLINE

Weiss MJ, Orkin SH. (1995a). Exp. Hematol., 23: 99-107. MEDLINE

Weiss MJ, Orkin SH. (1995b). Proc. Natl. Acad. Sci. USA, 92: 9623-9627. MEDLINE

Wijgerde M, Gribnau J, Trimborn T, Nuez B, Philipsen S, Grosveld F, Fraser P. (1996). Genes Dev., 10: 2894-2902. MEDLINE

Zhang P, Behre G, Pan J, Iwama A, Wara-Aswapati N, Radomska HS, Auron PE, Tenen DG, Sun Z. (1999). Proc. Natl. Acad. Sci., USA, 96: 8705-8710. Article MEDLINE

Zhang P, Zhang X, Iwama A, Yu C, Smith KA, Mueller BU, Narravula S, Torbett BE, Orkin SH, Tenen DG. (2000). Blood, 96: 2641-2648. MEDLINE

Figure 1 Transcription factor requirements in hematopoiesis. Schematic representation of hematopoietic lineage pathways from pluripotential stem cells to mature blood elements. Red bars represent the location of maturation arrest observed in the absence of the corresponding transcription factor. Transcription factors associated with chromosomal translocations or viral insertions in human and murine leukemias are denoted by an asterisk. e, eosinophil; n, neutrophil; m, monocyte/macrophage; b, basophil/mast cell; M, megakaryocyte; E, erythrocyte; T, T-lymphocyte; B, B-lymphocyte

Figure 2 Multiple protein-protein interactions of GATA-1. Schematic representation of the GATA-1 molecule. Black boxes represent the two zinc fingers. Proteins documented to physically interact with GATA-1 are labeled and their binding sites on GATA-1 indicated by horizontal black bars

Figure 3 Pentameric complex formed between SCL, E2A, Lmo2, Ldb1 and GATA-1. Schematic representation of cis-acting regulatory DNA from a hypothetical erythroid target gene containing an E-box motif (CAGGTG) and GATA-binding motif (GATA) separated by nine nucleotides. SCL and its heterodimeric partner, E2A, are shown interacting with an E-box sequence and GATA-1 interacting with a GATA motif. These complexes are connected via interactions with LMO2 and Ldb1 (adapted from Wadman et al. (1997))

Figure 4 Cross-antagonistic model of lineage factor activities in hematopoietic development. Schematic representation of cell fate determination of cell types A and B from a common bipotential precursor cell. Lineage-specific transcription factors are shown here (as 'A' and 'B') acting in a positive manner to produce cell types A and B, respectively. In addition, they simultaneously inhibit alternate lineage gene programs through direct cross-antagonism of opposing transcription factors (denoted as opposed 'T's)

Figure 5 Model of transcription factor cross-antagonism and leukemia. (A) Normal cell. A lineage-specific transcription factor ('A') is depicted positively regulating a target gene. (B) Leukemia cell. Misexpression of a lineage-specific transcription factor from an opposing lineage ('B') due to chromosomal translocation, viral insertion, or epigenetic phenomenon results in functional cross-antagonism of transcription factor 'A' (indicated by opposed 'T's). This blocks expression of transcription factor A target genes and produces an arrest in cell maturation