Interaction between hex and GATA transcription factors in vascular endothelial cells inhibits flk-1/KDR-mediated vascular endothelial growth factor signaling.

Recent evidence supports a role for GATA transcription factors as important signal intermediates in differentiated endothelial cells. The goal of this study was to identify proteins that interact with endothelial-derived GATA transcription factors. Using yeast two-hybrid screening, we identified hematopoietically expressed homeobox (Hex) as a GATA-binding partner in endothelial cells. The physical association between Hex and GATA was confirmed with immunoprecipitation in cultured cells. Hex overexpression resulted in decreased flk-1/KDR expression, both at the level of the promoter and the endogenous gene, and attenuated vascular endothelial growth factor-mediated tube formation in primary endothelial cell cultures. In electrophoretic mobility shift assays, Hex inhibited the binding of GATA-2 to the flk-1/KDR 5'-untranslated region GATA motif. Finally, in RNase protection assays, transforming growth factor beta1, which has been previously shown to decrease flk-1 expression by interfering with GATA binding activity, was shown to increase Hex expression in endothelial cells. Taken together, the present study provides evidence for a novel association between Hex and GATA and suggests that transforming growth factor beta-mediated repression of flk-1/KDR and vascular endothelial growth factor signaling involves the inducible formation of inhibitory Hex-GATA complexes.

gift from Dr. Stuart H. Orkin (Harvard Medical School, Boston, MA). Human GATA-3 and -6 and Hex cDNA fragments were amplified using PCR from reverse-transcribed HUVEC total RNA. To generate the Hex expression vector, Hex cDNA was subcloned into the pcDNA3 vector (Invitrogen). To generate the plasmids expressing Gal4 DNA-binding domain (GBD) fused with GATA-2, -3, and -6, each of the three GATA cDNA fragments was inserted into pGBKT7 (Clontech, Palo Alto, CA). To construct the plasmid expressing Gal4 activation domain (GAD) fused to Hex, the Hex cDNA fragment was subcloned into pGAD424 (Clontech). To generate a FLAG-tagged Hex (pFLAG-Hex) and FLAGtagged GATA-2 (pFLAG-GATA2), Hex and GATA-2 cDNA fragments were subcloned into pFLAG (Sigma), respectively. For construction of pGEM-hflk, a 266-bp human flk-1/KDR cDNA fragment was amplified from reverse-transcribed HUVEC total RNA and subcloned into pGEM-T-easy (Promega). Similarly, pGEM-hHex and pGEM-hGAPDH were derived by ligating a 296-bp human Hex cDNA fragment and a 283-bp human GAPDH fragment into pGEM-T-easy. Orientation was confirmed by automated DNA sequencing.
Yeast Two-hybrid Screening and ␤-Galactosidase Assays-A HUVEC cDNA library was constructed using a two-hybrid cDNA library construction kit (Clontech). AH109 yeast were transfected sequentially with the GATA-6 bait vector and cDNA library and then spread on synthesized dropout medium plate in the absence of tryptophan, leucine, and histidine. After 5 days of incubation at 30°C, colonies complemented by histidine autotrophy were isolated and confirmed to be positive by ␤-galactosidase assay according to the manufacturer's instruction (Clontech; yeast protocols handbook). The plasmids from the positive colonies were purified, and the inserts were subsequently sequenced. To analyze the protein-protein interaction, Y190 yeast were co-transfected with pGAD-Hex and a plasmid in which GBD was fused either with GATA-2, -3, and -6. Similarly, Y190 yeast were co-transfected with pGBD-GATA2 and a plasmid containing GAD fused with Hex. The cells were spread and incubated on the synthesized dropout medium plate without tryptophan and leucine.
Transfection of COS-7 Cells and Immunoprecipitation Assays-COS-7 cells were co-transfected with either pFLAG-Hex and pMT 2 -GATA2, or pFLAG-GATA2 and pcDNA3-Hex expression plasmids using the FuGENE 6 reagent (Roche Applied Science) as instructed by the manufacturer. Two days later, the transfected cells were freeze-thawed three times and incubated for 30 min on ice with cell lysis buffer (0.1% IGEPAL CA-630, 50 mM Tris-HCl, 150 mM NaCl, 5 mM EDTA, 0.5 g/ml pepstatin A, 10 g/ml leupeptin, 2 g/ml aprotinin, 200 M phenylmethylsulfonyl fluoride, pH 7.5), followed by centrifugation at 20,000 ϫ g for 5 min. The supernatant was incubated with anti-FLAG polyclonal antibody (Sigma) overnight at 4°C. The resulting mixture was then mixed and incubated with protein G-Sepharose (Amersham Biosciences) for 1 h at 4°C. The immobilized beads were washed five times with 1 ml of cell lysis buffer containing 1.5% IGEPAL CA-630. Each sample was separated on 12% SDS-polyacrylamide gel and transferred to a nitrocellulose membrane (Amersham Biosciences). The membrane was incubated either with anti-FLAG M5 monoclonal antibody (Sigma), anti-Myc monoclonal antibody (Invitrogen) or anti-GATA-2 antibody (Santa Cruz, CA).
Alternatively, subconfluent HUVEC (2 ϫ 10 7 cells) were harvested for nuclear extracts according to the mild nitrogen cavitation method (31) to keep intact the protein-protein associations. 1 g of nuclear extracts was precleaned by centrifugation with 4 g of control IgG (Santa Cruz). The resulting supernatant was incubated with 30 g of agarose-conjugated anti-GATA-2 monoclonal antibody (Santa Cruz, sc-267 AC) or an identical amount of isotype-matched mouse control IgG (Santa Cruz, sc-2343) overnight at 4°C. The immobilized samples were separated and transferred to a polyvinylidene difluoride membrane. The membrane was then probed for Hex and GATA-2 by Western blot analysis, using anti-Hex antibody (generated by Dr. Tamio Noguchi, Japan) and anti-GATA-2 antibody (Santa Cruz, sc-16044, and generated by Dr. Stuart H. Orkin, Harvard Medical School, Boston), respectively. The complexes were visualized with an ECL advance Western blotting detection kit (Amersham Biosciences).
Transfection of HEK-293 Cells or HUVEC and Analysis of Luciferase Activity-HEK-293 cells and HUVEC were transfected as described previously (30). Briefly, either 2 ϫ 10 5 cells/well of HEK-293 cells or 1 ϫ 10 5 cells/well of HUVEC were seeded onto 12-well plates 18 -24 h before transfection. 0.05 pmol of the luciferase reporter plasmid (either KDR-, KDR (GATA mut)-, or KDR (SP1 mut)-luc), 50 ng of pRL-CMV (Promega), 0.075 pmol of the GATA expression vector and either 0.0375 or 0.075 pmol of Hex expression vector were incubated with 2 l of Fu-GENE 6. As a negative control, empty vector (pMT 2 and pcDNA3) was transfected instead of GATA-2 and Hex expression vector, respectively. 24 h later for HEK-293 cells and 48 h later for HUVEC, the cells were washed with phosphate-buffered saline, lysed, and assayed for luciferase activity using the dual luciferase reporter assay system (Promega) and a Lumat LB 9507 luminometer (Berthold, Gaithersburg, MD).
Sandwich Tube Formation Assays-400-l aliquots of type-I collagen gel (Koken) containing EGM-2 MV medium (Clonetics) without basic fibroblast growth factor were added to 24-well plates and allowed to gel FIG. 1. Hex interacts the GATA family proteins. A, schematic representation of Hex. The black bar indicates the isolated fragment by two-hybrid screening. B and C, the yeast strain Y190 was co-transformed with HEX-pGAD and with human GATA-2, GATA-3, or GATA-6 in pGBD. The cells were plated in synthesized dropout medium lacking tryptophan, leucine, and histidine. ␤-Galactosidase assays and calculation of the activity units were performed as described under "Experimental Procedures." The values are the means Ϯ S.E. of three separate experiments. a.a., amino acids for 1 h at 37°C. HUVEC infected with Adeno-Blank or Adeno-Hex were seeded at 1 ϫ 10 5 cells/well and incubated for 24 h in 5% CO 2 . The medium was removed, and the HUVEC were covered with 400 l of the gel. The plate was incubated for 30 min at 37°C. The cells were incubated with 1 ml of EGM-2 MV media in the absence of basic fibroblast growth factor, in the presence or absence of SU1498 (Calbiochem, San Diego, CA). Two days later, a branched capillary network was visualized under a microscope. Images from at least three different areas in each well were captured by a digital camera under a microscope.
FIG. 5. Hex reduces binding of GATA motif on the flk-1/KDR and GATA-binding proteins. A, electrophoretic mobility shift assays were performed with 32 P-labeled 5Ј-UTR GATA probe and 10 g of nuclear extract from HUVEC infected with adenovirus expressing IRES-mediated EGFP (Adeno-Blank) or adenovirus expressing IRES-mediated Hex and EGFP (Adeno-Hex). The open and closed arrows indicate specific DNA-protein complexes. In competition assays, 10-, 100-, and 500-fold molar excesses of unlabeled self-probe and mutant 5Ј-UTR GATA probe BAS-1800 (Fuji Film, Japan). The signals were quantified with NIH Image.

Identification of Hematopoietically Expressed Hex as a GATA-interacting Protein by Yeast Two-hybrid Screening-The
GATA family of transcription factors has been implicated not only in the early differentiation of endothelial cells but also in the transduction of extracellular signals in the adult endothelium. Several members of the GATA family have been identified in endothelial cells, including GATA-2, -3, and -6. Previous studies have shown that GATA-1 and -2 interact with partner proteins in erythroid and megakaryocyte cells (34). Our goal was to identify the proteins that interact with GATA transcription factors in endothelial cells. To that end, we employed a yeast two-hybrid system in which full-length human GATA-6 served as bait. From a screen of 8.6 ϫ 10 6 clones, a total of 58 clones demonstrated histidine auxotroph. After elimination of false positives, 16 clones were selected, one of which encoded a fragment (amino acids 75-270) of Hex (Fig. 1A). To confirm the specificity of interaction between Hex and GATA-6, constructs containing the Gal4-activating domain fused with full-length Hex (pGAD-Hex) and either the Gal4-binding domain fused with GATA-6 (pGBD-GATA6) or the Gal4-binding domain alone (pGBD) were co-expressed in yeast AH109. As shown in Fig. 1B, co-expression of pGAD-Hex and pGBD-GATA6 resulted in a 4.2-fold increase in ␤-galactosidase activity compared with co-expression of pGBD and pGAD-Hex.
We next wished to determine whether Hex interacts with human GATA-2 and -3. To that end, pGAD-Hex was co-expressed in yeast AH109 with constructs containing the Gal4binding domain fused either with GATA-2 or -3 (pGBD-GATA2 or GATA3, respectively) As shown in Fig. 1C, co-expression with human GATA-2 and -3 resulted in 5.8-and 4.7-fold induction of the ␤-galactosidase activity, respectively, compared with co-expression with Gal4-binding domain alone (pGBD). Taken together, these results suggest that Hex interacts with GATA-2, -3, and -6.
Physical Interaction between Hex and GATA-2 in Mammalian Cells-Having identified the interaction between GATA and Hex in the yeast two-hybrid system, we wished to determine whether this interaction occurs in mammalian cells. Of the various members of the GATA family of transcription factors, GATA-2 is expressed most abundantly in cultured endothelial cells (data not shown) and is believed to play a predominant role in endothelial cell biology. Therefore, we chose to focus on the interaction between Hex and GATA-2. To that end, COS-7 cells were transiently transfected with expression plasmids for human GATA-2 (pMT 2 -GATA2), FLAG-tagged human GATA-2 (pFLAG-GATA2), FLAG-or Myc-tagged human Hex (pFLAG-Hex or pMyc-Hex), or vector alone (pMT 2 , pMyc or pFLAG) and then processed for immunoprecipitation. The transfected cells expressed high levels of GATA-2 and Hex (Fig.  2, A and B, lanes 1-3). In co-transfections (pMT 2 -GATA2 and pFLAG-Hex), GATA-2 co-precipitated with the anti-FLAG antibody ( Fig. 2A, lane 6, closed arrowhead) but not with the nonimmune control ( Fig. 2A, lane 7). In contrast, in experiments in which either (pMT 2 -GATA2 and pFLAG) or (pMT 2 and pFLAG-tagged Hex) were transfected, GATA-2 did not co-precipitate ( Fig. 2A, lanes 4 and 5). Furthermore, in co-transfections (pFLAG-GATA2 and pMyc-Hex), Myc-tagged Hex co-precipitated with the anti-FLAG antibody (Fig. 2B, lane 6, closed arrowhead). Although in experiments in which either (pFLAG and pMyc-Hex) or (pFLAG-GATA2 and pMyc) were transfected, Hex did not co-precipitate (Fig. 2B, lanes 4 and 5).
Next, we wished to identify the physical interaction between endogenous GATA-2 and Hex in endothelial cells. To that end, nuclear extracts were prepared and processed for immunoblotting and co-immunoprecipitations. As shown in Fig. 2C, endogenous GATA-2 and Hex were detected in nuclear extracts from HUVEC but not COS-7. Endogenous Hex was co-precipitated with anti-GATA2 antibody but not with isotype-matched control IgG (Fig. 2D, lanes 2 and 3). In semi-quantitative calculations from two independent experiments, anti-GATA-2 antibody immunoprecipitated 75.2% of total GATA-2 in HUVEC. Moreover, 8.29% of total cellular Hex was physically associated with GATA-2. Collectively, these findings suggest that GATA-2 and Hex specifically interact with one another in cultured endothelial cells.
Hex Inhibits GATA-2-mediated flk-1/KDR Promoter Activity-We next wished to study the functional relevance of the interaction between GATA-2 and Hex. We have previously shown that GATA-2 binds to a GATA motif in the 5Ј-UTR region of the flk-1/KDR promoter and that this effect is necessary for full expression (35). To determine the effect of Hex on GATA-2-mediated activation of flk-1/KDR, transactivation assays were carried out in HEK-293 cells (nonendothelial cells) and HUVEC (endothelial cells) co-transfected with KDR-luc (Ϫ115 and ϩ296 flk-1/KDR coupled to luciferase) and an expression plasmid containing either human GATA-2 (pMT 2 -GATA2) or Hex (pcDNA3-HEX). As a negative control, the cells were co-transfected with vector alone (pMT 2 or pcDNA3). Consistent with our previous findings, the basal level of flk-1/KDR promoter activity in HEK-293 cells was significantly transactivated by 4.2-fold with overexpression of GATA-2 (Fig. 3A). GATA-2-mediated stimulation of promoter activity was completely abrogated by co-expression of Hex in a dose-dependent manner. Moreover, overexpression of Hex did not change the basal level of the promoter activity (Fig. 3A). In HUVEC, a high level of flk-1/KDR promoter activity (8.7-fold higher than SV40 promoter plus enhancer construct (pGL2-control), not shown) occurred, and overexpression of GATA-2 resulted in 2.8-fold transactivation of the promoter (Fig. 3B). More importantly, overexpression of Hex resulted in huge reduction of the flk-1/ KDR promoter activity (15.9-fold compared without expression of GATA-2 and Hex). Co-expression of GATA-2 failed to recover the Hex-mediated attenuation of the flk-1/KDR promoter activity (2.3-fold induction compared Hex expression alone) (Fig.  3B). To confirm that the Hex-mediated down-regulation of the promoter activity was mediated by the 5Ј-UTR GATA motif on the flk-1/KDR gene, either the GATA element or the SP1 element (as a control) point-mutated plasmid was transfected into HUVEC. KDR promoter activity from SP1 point-mutated plasmid was down to 63.1% and markedly reduced to 3.4% by the co-expression with Hex. In contrast, the promoter activity from GATA point-mutated plasmid was down to 33.1%, whereas there was no significant reduction in the presence of Hex (Fig.  3B). Previous studies have shown that Hex directly binds to a (lanes 3 and 9, lanes 4 and 10, lanes 5 and 11, and lanes 6 and 12, respectively) were added to the reaction mixture. B, electrophoretic mobility shift assays were performed with 10 g of nuclear extract from HUVEC infected with adenovirus expressing IRES-mediated EGFP (Adeno-Blank) (lane 1) or adenovirus-expressing IRES-mediated Hex and EGFP (Adeno-Hex) (lanes 2-4). To test the effect of antibodies on the DNA-protein complexes, antibody against GATA-2 (lane 3) or against p65 (lane 4) was added to the reaction mixture. C, electrophoretic mobility shift assays were performed with 32 P-labeled flk-1/KDR SP1 probe and 10 g of nuclear extract from HUVEC infected as above. The arrows indicate specific DNA-protein complexes. In competition assays, a 100-fold molar excess of unlabeled self-probe or mutant SP1 probe (lanes 2 and 5 and lanes 3 and 6, respectively) was added to the reaction mixture. The results are representative of two independent experiments. consensus element (5Ј-CAATTAAA-3Ј) in the promoter region of its target genes, resulting in transcriptional activation (36). A search of the 4-kb 5Ј-flanking region and 3-kb first intron of the flk-1/KDR failed to reveal such a motif. Taken together, these results suggest that Hex represses the GATA-mediated flk-1/KDR promoter activation.
Hex Suppresses flk-1/KDR mRNA Expression in Primary Human Endothelial Cells-To determine whether Hex modulates the expression of the endogenous flk-1/KDR gene, HUVEC were infected with adenovirus expressing either IRES-mediated-EGFP (Blank) or IRES-mediated-rat Hex and EGFP (Hex) at a multiplicity of infection of 20. Using this approach, over 80% of the cells were infected as determined by EGFP expression. Western blot assays of infected cells demonstrated high levels of Hex protein (Fig. 4A). More importantly, overexpression of Hex in HUVEC resulted in a 85% reduction (mean of three independent experiments) in flk-1/KDR mRNA by RNase protection assay (Fig. 4B, compare lanes 4 and 5). These findings indicate that Hex suppresses flk-1/KDR activity not only at the level of the promoter but also at the level of the endogenous gene.
Hex Inhibits Binding of GATA-2 to the flk-1/KDR 5Ј-UTR GATA Motif-Based on the above findings, we hypothesized that Hex inhibits flk-1/KDR expression by interfering with GATA binding to the 5Ј-UTR. To test this hypothesis, we performed electrophoretic mobility shift assays in which nuclear extracts derived from HUVEC either expressing IRES-mediated EGFP (Adeno-Blank) or IRES-mediated Hex and EGFP (Adeno-Hex) were incubated with a radiolabeled probe spanning the 5Ј-UTR GATA motif (ϩ98 to ϩ122). As shown in Fig.  5 (A and B), incubation of nuclear extract from Adeno-Blank infected HUVEC with the 32 P-labeled probe resulted in the appearance of two specific DNA-protein complexes (closed and open arrows). These DNA-protein complexes were inhibited by the addition of 10-, 100-, and 500-fold molar excess unlabeled self-competitor (Fig. 5A, lanes 3-5) but not by 500-fold molar excess GATA mutant competitor (Fig. 5A, lane 6). The mobility shift pattern was identical in Hex-overexpressing cells (the more slowly migrated complexes designated with the open arrows appeared with longer exposure time). However, Hex overexpression resulted in a significant reduction (54.2% reduction by densitometry) in the intensity of the GATA-binding complexes (Fig. 5, A, compare lanes 3-5 and 9 -11, and B, compare  lanes 1 and 2). As previously reported, the DNA-protein complexes were inhibited by preincubation with the anti-GATA-2 antibody (34). In contrast, the complex was not inhibited by preincubation with anti-p65 antibody (Fig. 5B, lanes 3 and 4). Compared with control HUVEC, Hex overexpression did not significantly alter SP1 DNA binding activity (Fig. 5C). Together, these results suggest that Hex-mediated down-regulation of the flk-1/KDR is associated with an inhibition of GATA-2 binding.
Hex Suppresses VEGF-mediated Tube Formation in Primary Human Endothelial Cell Cultures-VEGF interaction with Flk-1/KDR is a critical determinant of endothelial cell proliferation and angiogenesis (37). In the next set of experiments, we wished to determine whether Hex inhibits the VEGF-Flk-1/ KDR signaling axis. To that end, we carried out sandwich tube formation assays on collagen gel. In the absence of basic fibroblast growth factor, VEGF stimulated tube formation of HU-VEC infected with adenovirus expressing IRES-mediated-EGFP (Adeno-Blank), an effect that was 50.8% inhibited by the preincubation with the Flk-1/KDR inhibitor, SU1498 (Fig. 6). Interestingly, VEGF-mediated tube formation was significantly (58.5%) inhibited by overexpression of Hex and profoundly (77.0%) abrogated when combined with the presence of SU1498 (Adeno-Hex ϩ SU1498) (Fig. 6). These findings suggest that Hex-mediated inhibition of flk-1/KDR gene expression results in a down-regulation of VEGF signaling. The protected fragment (296 bp) represents the human Hex transcript. An [␣-32 P]UTP-labeled GAPDH riboprobe was hybridized with total RNA as an internal control. B, shown is the quantification of RNase protection assays. Densitometry was used to calculate the ratio of Hex and GAPDH signals (arbitrary expression level). The data represent the averages from two independent experiments. C, a confluent HUVEC were serum-starved and then incubated in the absence or presence of 2.5 or 10 ng/ml TGF-␤ 1 for 24 h, at which time total RNA was isolated. In RNase protection assays, an [␣- 32  Hex Induces TGF-␤ Expression in Primary Human Endothelial Cells-Previous studies have shown that whereas tumor necrosis factor ␣, VEGF, and thrombin induce flk-1/KDR expression in endothelial cells (38 -40), TGF-␤1 has the opposite effect (41). In keeping with these findings, the addition of 2.5 and 10 ng/ml TGF-␤1 to HUVEC for 24 h resulted in 49 and 58% reduction of flk-1/KDR mRNA, respectively (Fig. 7, A,  lanes 3-5, and B). Interestingly, incubation of HUVEC with 2.5 and 10 ng/ml TGF-␤1 for 18 h resulted in 2.1-and 2.4-fold stimulation of Hex mRNA, respectively. Taken together, these results suggest that TGF-␤1 may exert its inhibitory effects through a Hex-GATA-2-dependent pathway. DISCUSSION The GATA family of transcription factors has been implicated not only in the early differentiation of the endothelial cells but also in the transduction of extracellular signals. For example, insulin-like growth factor 1, tumor necrosis factor ␣, and thrombin have each been shown to induce GATA-2 activity, whereas estrogens and TGF-␤ may actually inhibit GATA binding (6,30,35,42). Together, these studies suggest that GATA transcription factors may behave like immediate early genes, serving to couple short changes in the extracellular environment to long term changes in downstream gene expression.
An important clue to understanding the multifaceted role of a transcription factor is found in its repertoire of proteinprotein interactions. Indeed, previous studies have uncovered an array of partner proteins that interact with GATA transcription factors to regulate gene transcription (8, 14 -18, 20, 22, 23, 25, 27). In the present study, we have extended the list of protein partners by demonstrating a novel interaction between GATA proteins and the homeobox protein Hex in primary human endothelial cells. Hex is a member of the homeobox transcription factors and has been shown to bind to a consensus motif (5Ј-CAATTAAA-3Ј) in the promoter region of downstream target genes, resulting in repression of gene transcription (43). In addition, Hex may indirectly modulate gene expression through protein-protein interactions. For example, Hex has been shown to associate with cAMP-responsive element-binding protein to induce the SMemb/nonmuscle myosin heavy chain B gene in vascular smooth muscle cells (44). Moreover, Hex-Jun protein interactions have been reported to suppress c-Jun, JunB, and JunD-mediated gene activation (45).
In the present study, we have demonstrated GATA-2, -3, and -6 that have been identified in endothelial cells were physically interacted with Hex in cultured cells by immunoprecipitation analysis. GATA-2, -3, and -6 have highly conserved zinc finger domain and the region involved the usual protein-protein interactions (2,10,13,15,18,19,20,29,37,39). Further studies will be required to determine whether these motifs are responsible for mediating the binding of Hex to GATA-2, GATA-3, and GATA-6.
In a previous study, we demonstrated that TGF-␤ inhibits flk-1/KDR expression through a mechanism that involves reduced binding of GATA-2 to a palindromic GATA site in the 5Ј-UTR (35). In keeping with the role for GATA-2 in mediating flk-1/KDR expression, Hex overexpression resulted in reduced binding of GATA-2 to 5Ј-UTR GATA motif and a dose-dependent inhibition of flk-1/KDR promoter activity. It is interesting to speculate that the specific association between Hex and GATA-2 interferes with zinc finger domain-mediated DNA binding activity. Adenovirally mediated overexpression of Hex did not completely abrogate (54.2% inhibition) GATA-2 binding activity but did result in a more profound reduction of flk-1/ KDR mRNA and the promoter activity (Figs. 3 and 4). It has been shown that Hex and Jun interaction inhibits Jun-medi-ated transactivation without interfering with the Jun-DNA binding activity (45). In addition, Hex contains the transrepression domain at the N terminus. These observations suggest that Hex may exert its inhibitory effect through two mechanisms, namely through competitive inhibition of GATA-2-DNA to the flk-1/KDR promoter and direct repression of transcription.
The results of the present study provide new insight into how TGF-␤ mediates its inhibitory effect on the GATA-2-flk-1/KDR signaling axis. The observation that TGF-␤ treatment of endothelial cells resulted in concomitant up-regulation of Hex mRNA and down-regulation of flk-1/KDR in endothelial cells suggests that TGF-␤ signaling promotes transcriptionally inactive or inhibitory complexes between GATA-2 and Hex. The functional consequence of this interaction was borne out in in vitro studies of angiogenesis, in which the overexpression of Hex abrogated VEGF-flk-1/KDR-dependent endothelial cell tube formation.
The mechanism by which TGF-␤1 induces Hex expression remains to be elucidated. TGF-␤1 is known to induce Smad-2 and -5 activity in endothelial cells (46). Moreover, Smad-mediated signaling has been implicated in the control of Hex expression (47). Mice that are null for Smad-2 deficiency are embryonic lethal at E7.5 and lack detectable levels of Hex (48). Taken together, these observations raise the possibility that TGF-␤1 signaling is coupled to Hex-GATA-mediated inhibition of flk-1/KDR through a Smad-2-dependent pathway.
Most recently, Nakagawa et al. (49) reported that Hex acts as a negative regulator of angiogenesis. In the latter study, Hex was shown to completely abrogate the VEGF-mediated proliferation, migration, and invasion of HUVEC (49). Although our data are consistent with those of Nakagawa et al., they are novel in that they: 1) provide a link between a natural inhibitor of angiogenesis (TGF-␤) and Hex and 2) reveal a mechanistic connection between Hex, GATA-2, and flk-1/KDR expression. Indeed, based on these findings, we propose that Hex, as well as GATA-2, represent new therapeutic targets for anti-angiogenesis therapy.