GATA-1- and FOG-dependent Activation of Megakaryocytic αIIB Gene Expression*

FOG is a multitype zinc finger protein that is essential for megakaryopoiesis, binds to the amino-terminal finger of GATA-1, and modulates the transcription of GATA-1 target genes. Presently investigated are effects of FOG and GATA-1 on the transcription of the megakaryocytic integrin gene,αIIb. In GATA-1-deficient FDCER cells (in the presence of endogenous FOG), ectopically expressed GATA-1 activated transcription 3–10-fold both from αIIbtemplates and the endogenous αIIb gene. The increased expression of FOG increased reporter construct transcription 30-fold overall. Unexpectedly, αIIb gene transcription also was stimulated efficiently upon the ectopic expression in of FOG per se. This occurred in the absence of any detectable expression of GATA-1 and was observed in multiple independent sublines for both the endogenousαIIb gene and transfected constructs yet proved to depend largely upon conserved GATA elements 457 and 55 base pairs upstream from the transcriptional start site. In 293 cells, FOG plus GATA-1 but not FOG alone only moderately stimulatedαIIb transcription, and no direct interactions of FOG with the αIIb promoter were detectable. Thus, FOG acts in concert with GATA-1 to stimulateαIIb expression but also can act via a GATA-1-independent route, which is proposed to involve additional hematopoietic-restricted cofactors (possibly GATA-2).

The course of development of hematopoietic progenitor cells is dictated, in part, by the differential expression of lineagespecifying transcription factors. Lymphopoiesis, myelopoiesis, granulopoiesis, erythropoiesis, and megakaryopoiesis, for example, are known from gene disruption experiments to depend on the expression of Ikaros (1), PU.1 (2), CCAAT/enhancerbinding protein-␣ (3), and GATA-1 (4), respectively. In addition, the abilities of such factors to control rates of target gene expression can involve interactions with additional lineagerestricted co-regulators. This is illustrated by roles for distinct Ikaros-Helios-Aiolos complexes in specifying developmental fates of T cells (1,5,6), by the regulation of Oct factor activity by the B cell-specific coactivator OBF-1/Bob1/OCA-B, (7) and during erythropoiesis by the complexing of GATA-1 with FOG (friend of GATA-1) (8), C-terminal binding protein (9), and Tal1 plus Lmo2 (10,11). GATA-1 is a Cys2/Cys2 zinc finger DNAbinding protein that binds preferentially to (A/T)GATA(A/G) elements via its carboxyl-terminal finger domain (12) and is expressed in erythrocytes, megakaryocytes, eosinophils, and mast cells (13)(14)(15)(16). GATA-1 gene disruption in mice results in embryonic lethality due to anemia (4) and to an arrest in erythroid development at a late proerythroblast stage (17). During megakaryopoiesis, important roles for GATA-1 have been illustrated by experiments wherein the targeted disruption of an upstream activating element in the GATA-1 gene results in an accumulation of early megakaryocytic progenitor cells and a deficiency in platelet production (18). FOG is a 110,000-kDa multitype zinc finger protein that was discovered in a yeast two-hybrid screen based on its ability to interact specifically with the amino-terminal zinc finger of GATA-1 (8). In FOG Ϫ/Ϫ mice (and in FOG Ϫ/Ϫ embryonic stem (ES) 1 (19), erythropoiesis is blocked at a penultimate stage, while effects on megakaryopoiesis are more dramatic, and FOG Ϫ/Ϫ yolk sac and fetal liver cells give rise to few, if any, megakaryocytes (19). This broad defect indicates that FOG expression is either essential for early commitment to this lineage and/or that FOG acts subsequently to promote the transcription of late megakaryocytic genes.

cells differentiated in vitro)
Since FOG is a co-factor for GATA-1 (8) and since functional GATA elements occur within the promoters of most megakaryocytic genes studied to date (20 -25), FOG might act as an obligatory GATA-1 co-factor. However, GATA-1 mutants that fail to bind FOG have been shown to activate the expression of the EKLF, heme-regulated eIF-␣-kinase, and FOG (26) genes in GATA-1-deficient ES cells. Thus, GATA-1 and/or FOG also may act in combination with alternate co-factors to regulate erythromegakaryocytic gene expression. With regards to megakaryocytic genes, investigations of roles for FOG are limited to two studies to date. In 416B cells, ectopically expressed FOG and GATA-1 increased the frequency of cells expressing acetylcholinesterase (8), and in 3T3 fibroblasts expression of FOG plus GATA-1 significantly activated transcription from a 7-kb upstream region of the erythromegakaryocytic gene p45 NF-E2 (8). To further determine how FOG might affect megakaryocytic gene expression, we presently have investigated whether FOG might regulate the expression of the megakaryocytic integrin subunit, ␣IIb. ␣IIb expression is restricted to megakaryocytes, platelets, and their progenitors (27) and, together with a more broadly expressed subunit ␤ 3 , forms an integrin receptor that functions in platelet aggregation (28). In the promoter domains of the rat and human ␣IIb genes, upstream as well as TATA box-positioned GATA-1 ele-ments previously have been shown to be important for transcription (22,29,30). Flanking each of these two GATA elements are elements for Ets factor binding that likewise contribute to efficient transcription from the proximal promoters of the rat and human ␣IIb genes (28,30,31). Together with a Ϫ14 bp element for Sp1 (32), these elements (which lie within a 600-bp promoter domain) have been proposed to direct the lineage-specific expression of ␣IIb, and similarly distributed elements also occur within the promoters of several additional megakaryocytic-specific genes including the Tpo receptor (23), chemokine PF4 (20), GPIb␣ (24), and GPIX (25) genes. The present investigation focuses on ␣IIb gene expression and provides evidence that FOG acts as an important positive regulator via both GATA-1-dependent and independent routes.
Transcriptional Reporter Assays-In transfections of FDCER and derived cell lines, exponentially growing cells were adjusted to 3 ϫ 10 5 cells/ml and transferred to six-well plates (3 ml/well). For each single transfection, 12 l of FuGENE-6 liposomes (Roche Molecular Biochemicals) were added to 88 l of Opti-MEM I medium, and this mixture then was combined with 1.8 g of reporter plasmid DNA plus 0.2 g of pCMV-SEAP (Tropix, Bedford, MA). Complexes were incubated at 23°C for 15 min and added to cells. At 24 h of culture, transfected cells were collected (200 ϫ g for 10 min), washed in PBS, and lysed in reporter lysis buffer (1% Triton X-100, 2 mM 1, 2-diaminocyclohexane-N,N,NЈ,NЈ-tetraacetic acid, 2 mM dithiothreitol, 10% glycerol, 25 mM Tris phosphate, pH 7.8) (Promega). Cleared supernatants were assayed for protein concentration (BCA protein assay; Pierce) and for luciferase activity. To control for limited variability in transfection efficiencies, secreted alkaline phosphatase (SEAP) activities in culture medium were assayed (Phospha-light kit; Tropix, Bedford MA). The activities of reporter plasmids in 293 cell lines were assayed as follows. Cells (30% confluent, 100-mm dishes) were transfected using calcium phosphate (Life Technologies), 4.5 g of reporter plasmid DNA, 0.2 g of pCMV-␤gal, and 15 g of sheared and purified salmon sperm DNA. At 48 h of culture, transfected cells were collected (200 ϫ g for 10 min), washed in phosphate-buffered saline (140 mM NaCl, 2.7 mM KCl, 1.8 mM KH 2 PO 4 , and 8.1 mM Na 2 HPO 4 , pH 7.2), and lysed in reporter lysis buffer. Cleared supernatants (10 min at 8000 ϫ g) were assayed for luciferase activities (luciferase assay reagent; Promega, Madison, WI) and for ␤-galactosidase activities (39).

GATA-1-dependent Activation of ␣IIb Gene Transcription in FDCER Cell
Lines-In primary studies, roles for GATA-1 and FOG on endogenous ␣IIb gene transcription were tested via their stable expression in FDCW2-derived cell lines. Recently, our laboratory has shown that these cells do not express GATA-1 at detectable levels, yet support the ability of exogenous GATA-1 to (auto)activate the de novo expression of the endogenous GATA-1 gene (38). As shown in Fig. 1A, Northern blot analyses of FDCER-GATA-1 cells revealed that exogenous GATA-1 expression also activated the expression of the endogenous ␣IIb gene. To confirm that this result was not a fortuitous clonal effect, ␣IIb transcript expression in FDCER-G1 clones c.10, c.9, and c.11 (i.e. three independent clones) was analyzed by 32 P-labeled reverse transcriptase-PCR (Fig. 1B). In each clone, ␣IIb transcript expression was elevated severalfold due to the expression of exogenous GATA-1 (as compared directly with parental FDCER cells).
Next, to test whether this effect was mediated by cis elements within the ␣IIb promoter, an extended upstream region of the murine ␣IIb gene was cloned, sequenced, and used to prepare promoter-luciferase reporter constructs. Within this approximately 1000-bp promoter region, elements at Ϫ457 bp and Ϫ55 bp exist together with flanking consensus Ets factor binding elements (at Ϫ508 to Ϫ501 and Ϫ44 to Ϫ37 bp) (Fig. 2). Within the human and rat ␣IIb promoters (45), each of these elements are positionally conserved. Within the previously undescribed upstream region, no additional consensus elements for these or other possible transfactors were apparent. Extended and truncated ␣IIb promoter-reporter constructs were prepared (i.e. p␣IIb910-Luc and p␣IIb545-Luc), and their activities first were assayed in FDCER-G1 cells versus control parental FDCER cells (Fig. 3). Exogenous GATA-1 (in the presence of low to moderate levels of endogenous FOG; see below) stimulated transcription from p␣IIb545-Luc and p␣IIb910-Luc approximately 5.4-and 2.9-fold, respectively. Maximal rates of transcription from each construct in FDCER-G1 cells were comparable, but transcription from p␣IIb910-Luc in parental FDCER cells was more pronounced. No effects of GATA-1 expression on low level transcription of the promoterless control template pGL2Basic were observed. For p␣IIb545-Luc, essentially equivalent results were obtained in repeated transfections of independent clonal lines of FDCER-G1 cells (Fig. 3B).
FOG Amplifies GATA-1-dependent ␣IIb Gene Transcription in FDCER-G1 Cells-In FDCER-G1 cells, possible effects of FOG on ␣IIb gene transcription next were tested by increasing FOG expression in these lines via stable transfection. In FD-CER-G1-FOG, FDCER-FOG, FDCER-G1, and parental FD-CER cells, Northern blotting first was used to assay levels of FOG and GATA-1 (as well as GATA-2) transcript expression (Fig. 4). As a point of comparison, levels of these transcripts in erythroid B6SUt.EP cells (and lymphoid CTLL2-ER cells) were co-analyzed. FOG transcript levels in FDCER cells were appreciable yet below those observed in B6SUt.EP cells. In FD-CER-G1 cells, the ectopic expression of GATA-1 interestingly led to an estimated 3-fold increase in FOG transcript levels. In contrast, levels of GATA-2 transcript expression in FDCER-G1 and FDCER-G1-FOG cells were diminished. With regard to ␣IIb expression, ectopic expression of FOG in FDCER-G1 cells proved to stimulate rates of ␣IIb transcription to levels at least 5-fold above levels in FDCER-G1 cells and 38-fold above levels in parental FDCER cells (Fig. 5A). This result also was observed in repeated independent experiments in FDCER-G1-FOG cell lines. Based on these results (and the knowledge that FOG does not affect GATA-1's DNA binding activity) (26), it was predicted that levels of FOG in FDCER-G1 cells might limit ␣IIb expression. If so, further increases in ectopic GATA-1 expression in FDCER-G1 cells might squelch rather than enhance the activity of FOG-GATA-1 complexes. To test this prediction, FDCER-G1 cells were transfected stably with a second GATA-1 expression vector (pCINeoG1), and effects on transcription from m-␣IIb reporter constructs were assayed. Increased expression of exogenous GATA-1 in FDCER-G1-pCG1 cells proved to inhibit transcription from p␣IIb545-Luc (and p␣IIb910-Luc) approximately 3-fold as compared with FDCER-G1 cells (Fig. 5B). Results are representative of three independent experiments (and increased levels of GATA-1 expression in FDCER-G1-pCG1 cells have been documented previously) (33). This apparent squelching effect demonstrates that levels of GATA-1 in FDCER-G1 cells do not limit ␣IIb transcription and is at least consistent with the notion that, when overexpressed, GATA-1 instead may sequester an apparently limiting co-factor such as FOG.
FOG Activation of ␣IIb Gene Transcription via a GATA-1independent Route-In control experiments, FOG per se also was expressed in FDCER cells, and levels of ␣IIb gene transcription were assayed. Initially, this was tested using p␣IIb545-Luc. Somewhat unexpectedly, the expression of FOG at levels 2-3-fold above endogenous levels (see above; Fig. 4) increased rates of p␣IIb545 transcription in FDCER-FOG cells to levels essentially equivalent to those supported by GATA-1 in FDCER-G1 cells (Fig. 6A). This was observed in clonal as well as in polyclonal FDCER-FOG cell lines and suggested that FOG might promote ␣IIb gene transcription in the absence of GATA-1. To critically test this possibility, 32 P-labeled reverse transcriptase-PCR was used to assay endogenous GATA-1 and ␣IIb transcript levels (Fig. 6B). In FDCER-FOG cells, no GATA-1 transcripts were detected. However, levels of endogenous ␣IIb gene expression in all clones tested were increased to levels approximating those induced by GATA-1 in FDCER-G1 cells.
The extent to which FOG-stimulated transcription of the m-␣IIb gene depended upon intact Ϫ457 bp and/or Ϫ55 bp TATA box position GATA elements next was tested. First, roles for these elements in supporting p␣IIb545-Luc transcription in FDCER-G1 cells versus parental FDCER cells were examined. One, the other, or both GATA elements were mutated to the nonfunctional sequence CATA (46), and activities of the derived constructs p␣IIb545-⌬5ЈG-Luc, p␣IIb545-⌬3ЈG-Luc and p␣IIb545-⌬5Ј⌬3ЈG-Luc were assayed. Mutation of the Ϫ457 bp GATA element inhibited transcription from the m-␣IIb promoter in FDCER-G1 cells 5.5-fold (20,900-to 3800-unit decrease); mutation of the TATA box-positioned Ϫ55 bp GATA element inhibited transcription 2.7-fold (20,900-to 7800-unit decrease); and the mutation of both GATA elements inhibited transcription 9.1-fold (20,900-to 2300-unit decrease) (Fig. 7A,  upper panel). Similar results were observed in all independent clonal lines of FDCER-G1 cells tested. 2 These effects on m-␣IIb gene transcription of disrupting Ϫ457 and Ϫ55 bp GATA elements are similar to effects observed previously for analogous mutations of the rat ␣IIb promoter as assayed in transfected rat marrow cells (30) and suggest that these two GATA elements and their associated trans-factors act in a somewhat more than additive fashion to promote m-␣IIb gene transcription. Effects of mutating GATA elements on FOG-stimulated m-␣IIb transcription next were tested in FDCER-G1-FOG cells (Fig. 7A, lower panel). Consistent with data in Fig. 5, the increased expression of FOG in these cells stimulated transcription from the wild-type p␣IIb545-Luc construct 4-fold above levels in FDCER-G1 cells (and 42-fold above levels in To further test this possibility, transcription from the mutant construct p␣IIb545-⌬5Ј⌬3ЈG-Luc was assayed in FDCER-FOG cells (Fig. 7B). Residual transcription from p␣IIb545-⌬5Ј⌬3ЈG-Luc in these GATA-1-deficient cells was somewhat higher than in FDCER cells, again suggesting that FOG might stimulate m-␣IIb gene expression at least to a limited extent via a GATA element-independent route.
Finally, to test whether transcription from ␣IIb proximal promoter constructs would be stimulated efficiently by ectopically expressed FOG, GATA-1, and/or Ets-1 in nonhematopoietic cells, these factors first were expressed stably in 293 fibroblasts to yield 293-G1, 293-FOG, 293-G1-FOG, and 293-G1-FOG Ets-1 cells. Transfections with p␣IIb545-Luc then were performed, and activities were assayed in triplicate (with pCMV-␤Gal as a co-reporter). As shown in Fig. 8A (upper  panel), ectopically expressed GATA-1 per se only slightly increased rates of p␣IIb545-Luc transcription in 293-G1 cells (approximately 2-fold above parental 293 cells), while ectopically expressed FOG (in 293-FOG cells) per se had no detectable positive effect. In combination, however, these factors in 293-G1-FOG cells reproducibly stimulated transcription from the m-␣IIb gene proximal promoter approximately 6-fold above levels in parental 293 cells. Ectopically expressed Ets-1, in contrast, did not significantly stimulate m-␣IIb transcription in this reconstituted system in the absence or presence of GATA-1 (or GATA-1 plus FOG) (Fig. 8A, lower panel). Data shown are representative of three independent experiments in which essentially equivalent effects of these trans-factors on transcription from the m-␣IIb proximal promoter were observed and similar activities were observed for p␣IIb910-Luc. 2 In advance, Western and Northern blotting were used to identify matched sublines in which expression levels were highly comparable (Fig. 8B). These results demonstrate the positive co-action of 2 P. Gaines, unpublished observation. FIG. 2. Features of the murine ␣IIb promoter. The 5Ј domain of m-␣IIb was cloned, sequenced, and aligned with previously sequenced regions of the rat (r) and human (h) ␣IIb genes (45). Shown are consensus elements for GATA-1 (boxed) and Ets (bracketed), together with an Ebox-like element (broken box). Differences from a previously reported sequence (58) are nucleotides C at position Ϫ534 and T at position Ϫ527 (previously assigned as T and C, respectively).
FOG and GATA-1 in fibroblastic cells. These effects, however, were blunted as compared with those in FDCER-G1-FOG cells, and this is at least consistent with possible roles for alternate hematopoietic factors in activating the ␣IIb gene. DISCUSSION As introduced above, the disrupted expression of FOG in mice blocks the formation of megakaryocytes and erythrocytes. As shown initially in studies by Crispino et al. (26), however, FOG appears to be dispensable for the activation of at least certain GATA-1 target genes and has been proposed to act differentially with GATA-1 at distinct subsets of erythroid and megakaryocytic genes (19,26). In addition, a FOG homologue in Xenopus recently has been discovered and demonstrated through ectopic expression and explant assays to repress the transcription of at least certain erythroid genes (possibly via interactions with C-terminal binding protein) (47). In separate studies, FOG also has been shown to inhibit GATA-1-dependent transcription from the eosinophil-and basophil-specific gene, eosinophil major basic protein (48). These reports suggest that FOG does not act simply as a GATA-1 co-activator and that its activity depends upon not only lineage but also promoter contexts. Despite FOG's essential role in megakaryopoiesis (19), it also is noted that studies of its effects on megakaryocytic genes are limited to date to the demonstrated ability of FOG to stimulate transcription from a 7000-bp promoter domain of the erythromegakaryocytic p45 NFE2 gene in transiently transfected 3T3 fibroblasts (8). Such considerations prompted the present investigations of roles for FOG (and GATA-1) in ␣IIb gene expression.
As a hematopoietic cell line that is GATA-1-deficient and expresses endogenous FOG at moderate levels, FDCER cells proved to be an advantageous model in which to test dosage effects of these (co)factors on ␣IIb gene expression. With regard to GATA-1 effects per se, with the exception of the observation in chicken HD50M myeloblastic cells that exogenous GATA-1 can promote the outgrowth of thromboblastic-like cells (including a subline that stained with an antibody thought to be specific for avian ␣II/␤ 3 integrins) (49), the present study is the first to demonstrate GATA-1-dependent activation of endogenous ␣IIb gene expression. Consistent with the results of previous experiments, promoter-reporter transfection experiments in FDCER-G1 cells showed this to depend upon GATA elements positioned at Ϫ457 and Ϫ55 bp within the proximal ␣IIb promoter. In several additional megakaryocytic genes including mpl (23), GPIb␣ (24), GP-1X (25), and PF4 (20), GATA elements likewise occur within 90 bp of transcription start sites and have been proposed to substitute for canonical TATA boxes by binding to a multisubunit TFIID complex, which may contain GATA-1, an Ets factor, Sp1, and (based on the present findings) possibly FOG. In the present studies, however, this Ϫ55 GATA element contributed meaningfully to GATA-1-stimulated (and FOG-stimulated) ␣IIb transcription yet proved to be somewhat less important than a Ϫ457 bp element. Flanking each of these GATA elements are sites for the binding of one or more Ets family transcription factors, and these sites also have been demonstrated to support transcription at the human and rat ␣IIb proximal promoters (22, 30 -32, 50). In FDCER and FDCER-G1 cells, Northern blot analyses of Fli-1, Spi-1, and Ets-1 transcripts revealed each to be expressed at appreciable levels, and in FDCER-G1 lines Ets-1 levels were increased approximately 2-fold. 2 Interestingly, Ets-1 (and Ets-2) recently has been shown to bind to C-terminal binding protein/p300 (51,52), and based on the ability of GATA-1 to bind a nonequivalent region of C-terminal binding protein/p300 (9,53), it is possible that FOG might also tether at least indirectly to one or another Ets factor.
More remarkable are, first, the overall 30 -40-fold increase in levels of ␣IIb promoter transcription stimulated by ectopic co-expression of GATA-1 plus FOG in FDCER-G1-FOG cell lines and, second, the ability of FOG to activate ␣IIb gene transcription in GATA-1-deficient FDCER cells. Increases in ␣IIb transcription due to exogenous FOG in FDCER-G1-FOG cells are suggested to reflect FOG's role as a limiting factor in GATA-1-dependent ␣IIb gene activation, and direct interactions between these co-factors are the most straightforward to propose as a mechanism underlying observed effects on ␣IIb transcription. However, opportunities also exist for GATA-1 and possibly FOG to modulate by secondary routes the expression of other potential regulators of ␣IIb gene expression. The case for direct action mechanisms is underlined by the apparent ability of exogenous GATA-1 to squelch ␣IIb transcription when expressed at elevated levels in FDCER-G1-pCG1 cells (see Fig. 5B) and by the major dependence of FOG activity in FDCER-G1-FOG cells on the intactness of Ϫ457 and Ϫ55 bp GATA elements. Nonetheless, several observations also are consistent with alternate mechanisms of FOG action in addition to those mediated by interactions with GATA-1. These include FOG's ability to activate ␣IIb transcription in the ap- , and/or mEts1 (pAPuroEts-1), and the following cell lines stably expressing these factors (separately or in combination) were isolated: 293, 293-G1, 293-FOG, 293-G1-FOG, 293-Ets1, 293 Ets1-G1, and 293-Ets1-G1-FOG cells. The ability of each subline to support transcription from p␣IIb545-Luc then was assayed. pCMV␤gal was co-transfected, and samples were normalized for ␤-galactosidase activity to correct for limited variability in transfection efficiencies. Plotted are the activities (mean relative light units Ϯ S.D.) of triplicate transfections. Shown in parentheses are -fold increases in luciferase activity supported by the specified transcription factors. B, levels of GATA-1, FOG, and/or Ets-1 expression in the above 293 cells and derived cell lines were assayed by Western blotting (for GATA-1; upper panel) or by Northern blotting (for FOG and Ets-1; lower panels). Equivalence in RNA loading was confirmed by hybridization to a 32 P-labeled glyceraldehyde-3-phosphate dehydrogenase probe (GAPDH). hematopoietic factors (other than Ets-1). FDCER cells normally express FOG at readily detectable levels (see Fig. 4). Thus, the moderate ectopic increase in FOG expression, while not predicted (in the absence of GATA-1) to significantly affect m-␣IIb gene expression, proved to stimulate the transcription of p␣IIb545-Luc and the endogenous m-␣IIb gene in FDCER-FOG cells at levels comparable with those supported in FD-CER-G1 cells by exogenous GATA-1. Recently, GATA-2 has been shown to be capable of binding via its amino-terminal zinc finger to FOG (8,54) and is known to possess DNA binding properties highly similar to those of GATA-1 (55,56). Also, GATA-2 is expressed at appreciable levels in FDCER cell lines (see Fig. 4), and it therefore presently is speculated that FOG activation of m-␣IIb expression in FDCER-FOG cells might be facilitated by its partnering with GATA-2. This raises questions as to whether FOG might also interact with or otherwise regulate GATA-2 in other cells, including immature hematopoietic cells, which require high level GATA-2 expression for their early development (57). Consistent with this notion, Deconinck et al. recently have hypothesized that proliferation of hematopoietic progenitor cells in Xenopus might involve effects of FOG on GATA-2 expression (47). In this context, the downregulation of GATA-2 and up-regulation of FOG due to GATA-1 expression in FDCER-G1 cells (see Fig. 4) is again noted. Finally, it also is possible that a presently identified E-box-like element immediately 3Ј to the Ϫ457 bp GATA element in the murine and human ␣IIb promoters (see Fig. 2) might also recruit FOG-GATA-1 (and/or FOG-GATA-2) complexes via its potential to bind Tal1-Lmo2 complexes (11). Each of the above possible architectures is consistent with recently mapped interactions among these transcription factors (9,11,53), and in future experiments, it should be of interest to discover which of these architectures might provide for the selective high level expression of ␣IIb in megakaryocytic but not erythroid cells (each of which are believed to express all of the above-mentioned factors).