Context-dependent GATA Factor Function

GATA factors are fundamental components of developmentally important transcriptional networks. By contrast to common mechanisms in which transacting factors function directly at promoters, the hematopoietic GATA factors GATA-1 and GATA-2 often assemble dispersed complexes over broad chromosomal regions. For example, GATA-1 and GATA-2 occupy five conserved regions over ∼100 kb of the Gata2 locus in the transcriptionally repressed and active states, respectively, in erythroid cells. Since it is unknown whether the individual complexes exert qualitatively distinct or identical functions to regulate Gata2 transcription in vivo, we compared the activity of the -3.9 and +9.5 kb sites of the Gata2 locus in transgenic mice. The +9.5 site functioned as an autonomous enhancer in the endothelium and fetal liver of embryonic day 11 embryos, whereas the -3.9 site lacked such activity. Mechanistic studies demonstrated critical requirements for a GATA motif and a neighboring E-box within the +9.5 site for enhancer activity in endothelial and hematopoietic cells. Surprisingly, whereas this GATA-E-box composite motif was sufficient for enhancer activity in an erythroid precursor cell line, its enhancer function in primary human endothelial cells required additional regulatory modules. These results identify the first molecular determinant of Gata2 transcription in vascular endothelium, composed of a core enhancer module active in both endothelial and hematopoietic cells and regulatory modules preferentially required in endothelial cells.

Tissue-specific gene regulation is a tightly controlled process directed and altered by a plethora of cellular cues. Although numerous cell type-specific transcription factors have been identified, and common biochemical mechanisms have emerged (1,2), many questions remain unanswered regarding how ensembles of such factors function at complex mammalian loci. Considerable progress has been made in analyzing how members of the GATA transcription factor family (GATA-1 to -6) establish transcriptional networks that orchestrate developmental processes, including hematopoietic stem cell differentiation into diverse blood cell types (3,4).
In addition to conferring developmental control, GATA factors can function in certain differentiated cell types, including endothelial cells. GATA-2 was initially identified as an activator of endothelin-1 expression in endothelial cells (5,6) and also based on homology to GATA-1 (7). Subsequently, GATA-2 was detected in certain primitive and definitive hematopoietic cells, neurons, and cells of the developing liver, heart, placenta, and pituitary (8 -15). GATA-2 is implicated in activating or repressing a limited number of target genes (5, 13, 16 -25), including Gata2 itself (26 -29).
GATA factor complexes assemble at five restricted regions or "GATA switch sites" dispersed over ϳ100 kb of the Gata2 locus (26,27,29). GATA-2 occupies these sites at the transcriptionally active locus, and GATA-1 displaces GATA-2 concomitant with repression (26,27,29). Although it is unknown whether the GATA switch sites function similarly or distinctly to regulate Gata2 transcription in vivo, they exhibit certain structural and functional differences. Transient transfections in mouse erythroleukemia (MEL) 4 (30) and G1E cells (31), which exclusively express GATA-1 and GATA-2, respectively, revealed that the GATA switch sites segregate into distinct groups based on GATA motif-dependent enhancer activity (27,29). The Ϫ1.8 and Ϫ2.8 sites are active predominantly in G1E cells, the Ϫ77 and Ϫ3.9 sites are similarly active in MEL and G1E cells, and whereas the ϩ9.5 site has activity in both cells, the activity in G1E cells is considerably greater. Thus, despite the common feature of binding GATA factors, the enhancer activities of these elements in cultured hematopoietic cells differ. Furthermore, whereas GATA-1-mediated repression stimulates histone deacetylation at the Gata2 open reading frame and extending ϳ4 kb upstream, encompassing four of the five GATA switch sites, histone acetylation at the Ϫ77 kb site is unaltered (29).
Since GATA factors have both distinct and overlapping biological activities (28,(32)(33)(34)(35)(36), it is instructive to consider the molecular circuitry underlying such diversity. In principle, context-dependent functions can arise at the level of chromatin occupancy or postchromatin occupancy. Although GATA motifs are abundant throughout the genome, only a small subset are occupied in cells (29,(37)(38)(39). Parameters that govern GATA motif occupancy include intrinsic features of the motifs, their relationship to nearest neighbor cis-elements, and the surrounding chromatin environment (39). Since four of the five GATA switch sites can be occupied by either GATA-1 or GATA-2 at distinct stages of erythropoiesis (27,29), it seems likely that context-dependent GATA switch site enhancer activities reflect different functions post-GATA factor chromatin occupancy. Context-dependent post-chromatin occupancy activities can involve the combinatorial arrangement of GATA motifs with neighboring cis-elements; a particularly instructive example involves an Ets motif adjacent to a GATA motif. This configuration allows a GATA-1-FOG-1 complex to activate the megakaryocytic ␣IIB promoter in a transient transfection assay; without the Ets motif, the GATA-1-FOG-1 complex represses the reporter (40). In addition, a GATA-E-box composite element, which mediates assembly of a multiprotein complex containing GATA-1, LMO2, LDB1, SCL/TAL1 and E2A in erythroid cells (41)(42)(43)(44), is implicated in activation of certain GATA-1 target genes. SCL/TAL1 is required for development of all hematopoietic lineages (45), reflecting important roles in HSC generation (46) and hematopoietic commitment of hemangioblasts (47). Considerably less is known about factors that function with GATA-2, but classical transcription factors, such as Sp1 (48) and AP-1 (49), appear to facilitate GATA-2 function at promoters, and Ets factors function in concert with GATA-2 in endothelial cells (16,50,51).
Enhancer activities in transfection assays often do not recapitulate transcriptional mechanisms in vivo, and therefore we tested whether GATA switch sites at the Gata2 locus have distinct enhancer activities in transgenic mouse embryos. We demonstrate that the ϩ9.5 GATA switch site functions autonomously as a strong enhancer in endothelial cells and the fetal liver, a major site of erythropoiesis during embryogenesis. By contrast, the Ϫ3.9 GATA switch site, which binds GATA-1 and GATA-2 in erythroid precursor cells (27,29), lacks autonomous activity in vivo and has little to no activity in cultured endothelial cells. Mechanistic analyses revealed that the ϩ9.5 enhancer critically requires a core module, consisting of a GATA motif and a neighboring E-box, in both endothelial cells and erythroid precursor cells. Surprisingly, although the core module was sufficient for enhancer activity in the erythroid precursor cell line, its enhancer function in primary human endothelial cells required additional regulatory modules. These studies provide evidence that the combinatorial usage of enhancer modules can establish context-dependent GATA factor functions.
Plasmid Constructs-GATA-2 sequences were cloned from a murine 129SV bacterial artificial chromosome DNA isolated by Research Genetics/Invitrogen. Primers used to amplify genomic regions of Gata2 for the creation of the plasmid constructs used herein are available upon request. The integrity of cloned sequences was confirmed by DNA sequence analysis. The pGL3basic luciferase reporter plasmid was obtained from Promega. For LacZ reporter constructs, sequences identical to the respective transient construct were cloned into the pSV␤ vector (Clontech).
Transient Transfection Assay-HUVECs and HAECs were plated 1 day prior to transfection and were ϳ60 -70% confluent at the time of transfection. An equal amount of each plasmid (2 g) was added to 100 l of Opti-MEM (Invitrogen) reduced serum medium, incubated with Lipofectin reagent (6 l/1 g of DNA; Invitrogen) for 15 min at room temperature, and then added to the cells. The cells were incubated with the transfection mixture for 3 h before the readdition of Medium 200. Cell lysates were harvested 48 h post-transfection and were assayed for luciferase activity using the Luciferase Assay System (Promega). G1E cell transfections were conducted as described previously (29). The luciferase activity for each sample was normalized to the protein concentration of the lysate, as determined by a Bradford assay (Bio-Rad) using ␥-globulin as a standard. At least two independent preparations of each plasmid were analyzed.

Distinct Functions of Gata2 Locus GATA Switch Sites in Vivo-
GATA-1 and GATA-2 occupy five highly conserved regions ( Fig. 1) of the Gata2 locus in erythroid precursor cells, and these GATA switch sites confer qualitatively and quantitatively distinct enhancer activities in GATA-1-and GATA-2-expressing hematopoietic cells in vitro (26,27,29). To determine if the differential activities in vitro reflect unique activities in vivo, we analyzed the ϩ9.5 and Ϫ3.9 sites in F 0 transgenic mouse embryos. These elements, located at ϩ9.5 and Ϫ3.9 kb relative to the Gata2 1S promoter, were cloned upstream of the 1S promoter fused to LacZ, injected into mouse oocytes, and implanted into recipient females. Embryos were harvested at E11 and stained with X-gal to reveal LacZ expression. The (ϩ9.5)1SLacZ transgene displayed a reproducible expression pattern in 7 of the 31 embryos containing the LacZ transgene ( Fig. 2A). The remaining embryos showed no detectable transgene expression. Transverse sections of all seven expressing embryos revealed expression throughout vascular endothelium, in endocardial cells lining the interior of the heart, and in a subset of cells in the fetal liver ( Fig. 2A and supplemental Fig. 1), which is heavily colonized with GATA-1-and GATA-2-expressing hematopoietic precursor cells at E11 (53,54). Although quantitative differences in expression were apparent (compare whole mount images of two representative (ϩ9.5)1SLacZ embryos in Fig. 2A), no ectopic staining was detected. Thus, the ϩ9.5 element confers enhancer activity in multiple sites of endogenous Gata2 expression.
By contrast to the ϩ9.5 site, analysis of (Ϫ3.9)1SLacZ in 15 E11 embryos containing the transgene revealed no endothelial, endocardial, or fetal liver staining (Fig. 2B). Based on the frequency of obtaining transgene-positive embryos expressing the (ϩ9.5)1SLacZ construct and the failure to detect expression of the (Ϫ3.9)1SLacZ construct in 15 transgene-positive embryos, one would predict that at least 98% of any additional transgenepositive embryos would also not express the transgene. Since the Ϫ3.9 site enhancer functions in GATA-1-expressing MEL cells, but not in GATA-2-expressing G1E cells, differing from the ϩ9.5 site that has highest activity in G1E cells, the transgenic results (Fig. 2) strongly support the notion that these elements have fundamental functional differences.
Due to the activity of the ϩ9.5, but not the Ϫ3.9, site in vascular endothelium in vivo, we tested whether the ϩ9.5 site is preferentially active in two primary human endothelial cell subtypes, HUVECs and HAECs, relative to the Ϫ3.9 site. The ϩ9.5 site conferred strong enhancer activity, recapitulating its in vivo activity, whereas the Ϫ3.9 site had little to no activity in HUVECs and HAECs (Fig. 2C). Previously, we demonstrated that the Ϫ1.8 GATA switch site conferred modest enhancer  activity in G1E, but not MEL, cells (27). To determine if any GATA binding element with higher activity in G1E versus MEL cells has endothelial enhancer activity, we also tested the Ϫ1.8 construct in HUVECs and HAECs. However, the Ϫ1.8 site lacked activity in these cells (Fig. 2C), indicating that the ϩ9.5 site has unique determinants permissive for function in endothelial cells.
Gata2 ϩ9.5 Site Endothelial Enhancer Activity Requires a Composite GATA-E Box Motif-Mutation of the ϩ9.5 site GATA motifs abrogates enhancer activity in erythroid precursor cells, but the importance of these motifs in other cultured cells and in vivo has not been analyzed. This is important, since the ϩ9.5 site activity in vascular endothelium might not be GATA factor-dependent. Therefore, we tested the activity of a construct in which the three GATA motifs of the ϩ9.5 region were mutated ((ϩ9.5 mtG-1,2,3)1SLacZ) in F 0 transgenic mouse embryos. Of the 10 embryos that genotyped positive for the construct, transverse sections revealed no LacZ expression in vascular endothelium or fetal liver (Fig. 3A), providing strong evidence that ϩ9.5 site enhancer activity in endothelial cells and fetal liver is GATA motif-dependent.
Of the three GATA motifs within the ϩ9.5 site, only the central motif has a canonical WGA-TAR sequence; the additional motifs are imperfect (nGATAR and WGATAn, respectively) ( Fig.  1). Moreover, an E-box resides between the first and second GATA motifs, 8 bp upstream of the WGATAR. This arrangement meets the criteria predicted to be important for assembly of a multimeric complex containing GATA-1 and E-proteins (41). To assess the functional importance of the individual motifs, the WGATAR motif and E-box were independently mutated and compared with the activity of the ϩ9.5 triple GATA mutant. Mutation of the WGATAR abrogated the strong enhancer activity in HUVECs, identical to the effect of mutating all three GATA sites (Fig. 3B). Similarly, mutation of the E-box alone abolished enhancer activity in HUVECs (Fig. 3B). These data strongly implicate the WGATAR and E-box as being critical for ϩ9.5 endothelial enhancer activity.
GATA-2, GATA-3, and GATA-6 mRNA can be detected in HUVECs by Northern blotting (55). However, since HUVECs of different passages and sources can have distinct phenotypes, real time PCR was used to confirm whether our HUVECs exhibited a similar pattern of GATA factor expres- For embryos expressing (ϩ9.5 mtG-1,2,3)1SLacZ, histological sections show complete loss of endothelial staining in the dorsal aorta (DA) and endocardium (EC) and also in hematopoietic staining in the fetal liver (FL). B, influence of cis-element mutations on endothelial enhancer activity. HUVEC cells were transiently transfected with reporter plasmids derived from the pGL3 luciferase vector containing the Gata2 1S promoter cloned upstream of luciferase (1SLuc) with or without the wild-type or mutant ϩ9.5 site. The sites mutated (GATA motifs 1-3 and the E-box) are indicated in red text. The plot depicts luciferase activities of the cell lysates normalized by the protein concentrations of the lysates. The activity of the 1SLuc construct was designated 1.0 (means Ϯ S.E.). The 1SLuc, (ϩ9.5)1SLuc, (ϩ9.5 mtG-1,2,3)1SLuc, (ϩ9.5 mtG-2)1SLuc, and (ϩ9.5 mtE)1SLuc constructs were analyzed in 10, 10, 9, 3, and 4 independent experiments, respectively. In each experiment, transfections were performed in triplicate. *, p Ͻ 0.05 with respect to (ϩ9.5)1SLuc.
sion. This analysis in HUVECs and HAECs revealed GATA-2 and GATA-6 expression, lower GATA-3 and GATA-4 expression, and undetectable levels of the other GATA factors, thus implicating these GATA factors as potential mediators of ϩ9.5 endothelial enhancer activity (data not shown).
Differential Importance of Enhancer Modules for Endothelial Versus Hematopoietic Enhancer Activity-Given the dual requirement of GATA motifs and the E-box for ϩ9.5 site endothelial enhancer activity, we reasoned that the enhancer activity requires consolidation of the GATA motifs with the E-box to yield a composite element. Based on this prediction, replacing the GATA motifs of the Ϫ3.9 site with the ϩ9.5 site GATA motif-E-box module would be expected to endow the Ϫ3.9 site with enhancer activity in endothelial cells. The GATA motifs and E-box in their native arrangement were excised from the ϩ9.5 site and substituted for the two GATA motifs in the endothelial-inactive Ϫ3.9 site. This chimeric GATA switch site was cloned upstream of the 1S promoter fused to a luciferase reporter and analyzed in HUVECs. The (ϩ9.53Ϫ3.9)1SLuc chimera had little to no enhancer activity in HUVECs (Fig. 4). Furthermore, substituting the Ϫ3.9 site GATA motifs for the ϩ9.5 site GATA-E-box composite element ((Ϫ3.93ϩ9.5)1SLuc) abolished ϩ9.5 site endothelial enhancer activity (Fig. 4), further demonstrating the functional importance of the ϩ9.5 sequences.
The failure of the ϩ9.5 GATA-E-box composite element to function in the context of the Ϫ3.9 sequences suggests that either additional Ϫ3.9 sequences dominantly suppress activity of the ϩ9.5 site GATA-E-box composite element in endothelial cells, or the composite element requires accessory ϩ9.5 sequences to function as an endothelial enhancer. To distinguish between these possibilities, we tested whether the 5Ј and 3Ј arms of the ϩ9.5 site are required for the GATA motifs and E-box to function. Deletion of the 3Ј arm ((ϩ9.5 ⌬3Ј)1SLuc) attenuated enhancer activity in HUVECs (58% decrease), whereas deletion of the 5Ј arm ((ϩ9.5 ⌬5Ј)1SLuc) more severely reduced enhancer activity (78% decrease) (Fig. 4). Transient transfection experiments in HAECs also revealed ϩ9.5 enhancer activity, which was critically dependent on the 5Ј arm (data not shown). Deletion of the poorly conserved region of the 5Ј arm ((ϩ9.5 5Ј⌬1)1SLuc) did not significantly affect enhancer activity. However, additional truncations of the 5Ј arm resulted in incremental, but significant, inhibition of enhancer activity, with removal of the final 36 bp upstream of the first GATA motif ((ϩ9.5 ⌬5Ј)1SLuc) most strongly compromising the enhancer.
By contrast to other loci in which the GATA-E-box composite motif suffices to confer tissue-specific transcription and despite the apparent requisite 8 -10-bp spacing between the GATA motif and the E-box, the ϩ9.5 site uniquely requires additional sequences to function in endothelial cells. Importantly, this result portends the existence of autonomous and nonautonomous GATA-E-box composite motifs in gene regulatory regions.
Given the surprising finding that the GATA-E-box composite element was insufficient to function in endothelial cells, we considered whether the requirement for additional cis-elements is a hallmark of the ϩ9.5 enhancer in all GATA-2-expressing cells. Alternatively, the additional sequences might be differentially required in distinct GATA-2-expressing cells (e.g. endothelial versus hematopoietic cells). Since the ϩ9.5 site confers enhancer activity in the fetal liver (an important hematopoietic site) ( Fig. 2A), we tested whether the 5Ј and 3Ј arms are required for enhancer activity in G1E erythroid precursor cells expressing endogenous GATA-2. By contrast to the major inhibition upon removing the 5Ј and 3Ј arms on endothelial cell enhancer activity, the 5Ј arm deletion reduced enhancer activity in G1E cells by only 36%. Similarly, removal of the 3Ј arm only reduced activity by 22% (Fig. 5). Whereas the reduced activity upon loss of the 5Ј arm was statistically significant (p ϭ 0.017), the reduction associated with the 3Ј arm deletion was insignificant (p ϭ 0.120). Nonetheless, the magnitude of the reductions after 5Ј and 3Ј arm deletions in G1E cells were significantly less (p Ͻ 0.0001 for both) than the reductions in HUVECs (Fig. 4). Therefore, although the GATA motifs and E-box are equally crucial in both cell types (Figs. 4 -6), the 5Ј and 3Ј arms are considerably more important for enhancer activity in endothelial cells (Figs. 4 -6).
Since the 5Ј arm is preferentially required for ϩ9.5 enhancer activity in HUVEC versus G1E cells, we tested whether this regulatory module is competent to function as an autonomous enhancer in HUVECs. The 5Ј arm was cloned upstream of the SV40 promoter and tested by transient transfection analysis in HUVECs and G1E cells. The 5Ј arm slightly increased activity (Ͻ2-fold) of the SV40 promoter in both cell types (supplemental Fig. 2), strongly suggesting that this sequence lacks endothelial cell-specific enhancer activity. Thus, in the transient transfection assay, the 5Ј arm functions collectively with other components of the ϩ9.5 site to generate maximal enhancer activity.
endothelial, but not hematopoietic, cells critically requires additional regulatory modules. In G1E cells, the 5Ј and 3Ј regulatory modules exert only modest modulatory functions, since their deletion reduces reporter activity 36 and 22%, respec-tively. However, deletion of these regulatory modules severely reduces enhancer activity in HUVECs and HAECs.
The identification of the ϩ9.5 site as an endothelial cell enhancer in vitro and in vivo represents the first delineation of a regulatory region within the Gata2 locus that confers activity in vascular endothelium. Since Gata2 is expressed in at least subregions of the vasculature (5,6,24,25,49,55), based on our results, it is attractive to propose that the ϩ9.5 site is an important determinant of endogenous Gata2 expression in the vasculature. Moreover, the ability of the ϩ9.5 site to concurrently confer expression in regions enriched in hematopoietic precursors during early development, namely the fetal liver and dorsal aorta, might reflect an involvement of the ϩ9.5 site in regulating Gata2 expression in multipotent hematopoietic precursors and hemangioblasts (19,56). However, definitive proof in this regard will require its deletion from the endogenous locus. Intriguingly, although the Ϫ3.9 site resembles the ϩ9.5 site in erythroid precursor cells, in that both GATA-1 and GATA-2 occupy these sites, the extent of GATA factor occupancy at the Ϫ3.9 site in erythroid cells does not correlate with competence to regulate transcription in endothelial cells.
Comparative genomic, chromatin immunoprecipitation, and in vivo functional analyses indicate that tissue-specific expression of developmentally important genes is controlled by cis-elements dispersed throughout noncoding regions of a locus (29,(57)(58)(59)(60). These conserved, distal elements can direct transcription in both shared and unique cell types. For enhancers functional in multiple cell types (e.g. the ϩ9.5 site), enhancer activity can be derived from the coordinated actions of a "core" module, which mediates activity in the full spectrum of cell types, and one or more "regulatory" modules, which establish cell type-specific permutations of the core activity (Fig. 6). The collaboration of diverse regulatory proteins at enhancers is FIGURE 5. Differential molecular determinants for Gata2 ؉9.5 site enhancer activity in endothelial versus hematopoietic cells. The graph depicts results from transient transfection analysis of constructs containing the wild-type or mutated ϩ9.5 kb site in G1E erythropoietic precursor cells. G1E cells were transiently transfected with reporter plasmids derived from the pGL3 luciferase vector containing the Gata2 1S promoter cloned upstream of luciferase (1SLuc). The plot depicts luciferase activities of the cell lysates normalized by the protein concentrations of the lysates. The activity of the 1SLuc construct was designated 1.0 (means Ϯ S.E.). The 1SLuc, (ϩ9.5)1SLuc, (ϩ9.5 mtG-1,2,3)1SLuc, (ϩ9.5 mtG-2)1SLuc, (ϩ9.5 mtE)1SLuc, (ϩ9.5 ⌬5Ј)1SLuc, (ϩ9.5 ⌬3Ј)1SLuc, and (ϩ9.5 ⌬5Ј⌬3Ј)1SLuc constructs were analyzed in 8,8,4,4,4,8,4, and 4 independent experiments, respectively. In each experiment, transfections were performed in triplicate. *, p Ͻ 0.05 with respect to (ϩ9.5)1SLuc. FIGURE 6. Combinatorial transcriptional control via a shared core enhancer module and cell type-specific regulatory modules. Whereas the core module is similarly required in HUVECs and G1E cells, the 5Ј and 3Ј regulatory modules are preferentially required in HUVECs. Although the model depicts several factors bound to the regulatory module, deletion analysis of the 5Ј arm suggests that the respective cis-elements constituting this activity are distributed over ϳ167 bp of sequence directly upstream of the core GATA sites and E-box. In the chart below, the relative importance of each regulatory module/component is expressed with respect to the loss of enhancer activity after its mutation/deletion: ϩϩϩ, 76 -100% decrease; ϩϩ, 51-75% decrease; ϩ, 26 -50% decrease; Ϫ, 0 -25% decrease (not statistically significant). an established concept (39,61) and forms the basis for a model in which factors with qualitatively distinct activities converge at an enhancer that controls Drosophila development (52). In this model, "selector" proteins, which regulate development of specific cell and tissue types, function through enhancers in concert with signal-dependent factors, which lack the capacity to define developmental fate. Although the ϩ9.5 core module contains cis-elements that bind GATA factors and E-proteins, which can be important determinants of both hematopoiesis and vasculogenesis, our mutational analysis of the 5Ј regulatory module did not reveal a restricted cis-element mediating its activity. The absence of a single element, the high conservation of the 5Ј regulatory module sequences (Fig. 4), and abundance of conserved motifs capable of binding factors in the context of naked DNA, suggest that factors function through sequences distributed throughout the regulatory module. Thus, the regulatory activity might reflect the sum of multiple protein-DNA interactions.
What mechanisms underlie the autonomous and nonautonomous functions of the ϩ9.5 core enhancer module in hematopoietic versus endothelial cells? The GATA factor-E-protein complex at the composite element in hematopoietic cells might be considerably more stable than the endothelial cell complex. Accordingly, factors occupying the regulatory module might ensure stable complex assembly in endothelial cells rather than directly contributing to the activation function of the enhancer. Alternatively, the constraints involved in recruiting Pol II to the Gata2 promoter might differ in endothelial versus hematopoietic cells, and therefore factors binding the regulatory module might contribute a unique activation function required to overcome the endothelial cell-specific constraint, rather than serving an architectural function to regulate GATA-E-protein complex assembly. Our description of a new endothelial cell enhancer that functions in vivo, the discovery of a core enhancer module required for function in two distinct cell types, and surrounding regulatory modules mediating cell typespecific functions provides a strong foundation for dissecting mechanisms underlying context-dependent GATA factor function in the hematopoietic and vascular systems.