Transcriptional Up-regulation of Endothelial Cell Matrix Metalloproteinase-2 in Response to Extracellular Cues Involves GATA-2*

Matrix metalloproteinase-2 (MMP-2) plays a critical role in endothelial cells during the processes of angiogenesis and vascular remodeling. Endothelial cell production of MMP-2 is greatly enhanced when cells are cultured within a three-dimensional type I collagen matrix coinciding with the increased invasive and migratory phenotype of the cells. To define the transcriptional regulation of MMP-2 in rat microvascular endothelial cells, we performed promoter-reporter assays with a series of promoter truncations. Activity of the full promoter was significantly greater in cells cultured within three-dimensional type I collagen compared with cells cultured as a monolayer (two-dimensional) on type I collagen. Truncation of the region encompassing base pairs –1562 to –1375 (relative to the start codon) of the MMP-2 promoter resulted in loss of this differential activity of the MMP-2 promoter. Analysis of this region indicated two putative GATA-2 binding domains between –1437 and –1387. Southwestern blot analysis and electrophoretic mobility shift assays confirmed the binding of GATA-2 to this region of the MMP-2 promoter. Overexpression of GATA-2 in COS-7 cells significantly increased the activity of the full-length MMP-2 promoter-luciferase construct. Endothelial cells expressed greater levels of GATA-2 protein in three-dimensional compared with two-dimensional cultures, and activity of the –1437/–1387 region of the MMP-2 promoter was significantly greater in three-dimensional cultured endothelial cells. Together, these results indicate GATA-2 regulation of the MMP-2 promoter in endothelial cells and that the GATA-2 binding domain is sufficient to drive increased activity of the MMP-2 promoter in response to an extracellular matrix stimulus.

The matrix metalloproteinases (MMPs) 1 consist of a family of zinc-and calcium-dependent endopeptidases that cleave specific subsets of extracellular matrix proteins, growth factors, and cell surface receptors (1,2). Endothelial cell production and activation of MMPs, including MMP-2, are critical for the proc-ess of angiogenesis (3,4). Angiogenesis in skeletal muscle may be initiated by growth factors, by hemodynamic forces, or by mechanical stretching forces (5). The angiogenic process requires that endothelial cells proteolyze their basement membrane and then migrate through the interstitial matrix to form a new capillary. Inhibition of MMP activity prevents basement membrane degradation and the process of capillary sprouting in activity-stimulated skeletal muscle (6). When cultured within a three-dimensional type I collagen matrix, endothelial cells gain an invasive and migratory phenotype including increased production of matrix metalloproteinases and enhanced cell motility similar to what is seen in vivo during capillary sprouting (7,8). Regulated transcription of MMP-2 is important in determining this invasive phenotype in endothelial cells.
Regulation of MMP-2 occurs at both transcriptional and post-translational levels. Although the rate of MMP-2 protein production is controlled largely through alterations in mRNA transcription, proteolytic activity is regulated through secretion, cell surface binding, and activation of the latent enzyme (2). In contrast to the large amount of information available regarding the regulation of MMP-2 activity at the protein level, transcriptional regulation of MMP-2 has not been well characterized.
MMP-2 promoter analysis has been carried out in several cell types of both rat and human origin. The MMP-2 promoter has no TATA or CAAT boxes and is considered to be under constitutive control via GC-rich promoter regions. An enhancer region, known as RE1, has been identified at base pairs Ϫ1322 to Ϫ1282 of the rat promoter sequence (9). Binding proteins within the RE1 domain are well characterized and include Y-box protein-1 (YB-1), p53, and AP-2 (10,11), as well as the transcriptional suppressor, nm23-␤ (12). These transcription factors contribute greatly to the constitutive expression of MMP-2 in rat mesangial cells. Characterization of the promoter in rat endothelial cells identified an additional enhancer region (RE2) at base pairs Ϫ1435/Ϫ1375 (13). These authors demonstrated greater binding activity to the enhancer region in endothelial cells derived from spontaneously hypertensive (SHR) versus normotensive (WKY) rats, but the binding protein(s) was never identified. The human MMP-2 promoter sequence differs from the rat sequence, but several enhancer domains appear to be common between the two species. For instance, the human MMP-2 promoter is activated by p53 (14) and AP-2 (15) in a region designated enhancer region 2 (r2) that is located at Ϫ1635/Ϫ1593 relative to the transcription initiation start.
We have shown that MMP-2 production in rat microvascular endothelial cells is sensitive to the extracellular matrix environment and is largely unresponsive to growth factor stimula-tion (8). The mechanism by which extracellular matrix stimulation enhances endothelial cell transcription of MMP-2 is not yet known. In this study, we examined the rat MMP-2 promoter activity in rat microvascular endothelial cells under control and three-dimensional collagen-stimulated conditions. Our results provide strong evidence that the transcription factor GATA-2 is important in enhancing transcription of the MMP-2 promoter in response to three-dimensional collagen. These data suggest that GATA-2 may be a critical transcription factor regulating MMP-2 under pro-angiogenic conditions.

MATERIALS AND METHODS
Cell Culture-Microvascular endothelial cells were isolated from rat extensor digitorum longus muscles using a protocol adapted from Madri and Williams (16). Briefly, the muscles were isolated, minced, and digested with 0.5% collagenase type I (Invitrogen) in sterile HEPESsaline buffer (140 mM NaCl, 10 mM HEPES, pH 7.4, 10 mM KCl, 0.1 mM CaCl 2 , 0.2 mM MgCl 2 ) in a shaking water bath (37°C for 30 min). The tissue was then centrifuged, resuspended twice in HEPES-saline buffer containing 5% bovine serum albumin, then resuspended in 1 ml of HEPE-saline with 1% bovine serum albumin and layered over a Percoll gradient and centrifuged at 13,000 ϫ g for 20 min. The layer corresponding to the microvascular fragments was removed, washed with HEPES-saline-1% bovine serum albumin, and resuspended in complete Dulbecco's modified Eagle's medium (10% fetal calf serum, sodium pyruvate, L-glutamine, penicillin, streptomycin) containing 5 ng/ml recombinant vascular endothelial cell growth factor (Calbiochem) and plated onto gelatin-coated flasks. Cells were subcultured every 3-4 days and used for experiments between passages 4 and 10. Endothelial identity was confirmed by positive endothelial cell nitric-oxide synthase and von Willebrand's factor staining, and uptake of acetylated low density lipoprotein. Monolayer (two-dimensional) and three-dimensional cultures were set up as described previously. Briefly, 1 ϫ 10 6 cells were plated on 35-mm dishes pre-absorbed with a thin layer of type I collagen (two-dimensional) or embedded within a three-dimensional collagen matrix (2.5 mg/ml acid-soluble type I collagen (ICN) buffered with Earle's salt and neutralized with sterile NaOH) at a density of 1 ϫ 10 6 cells/ml collagen and cultured for 24 or 48 h in complete Dulbecco's modified Eagle's medium without vascular endothelial cell growth factor. COS-7 cells were purchased from ATCC (Manassas, VA) and cultured in complete Dulbecco's modified Eagle's medium.
Gelatin Zymography-Cells were lysed in 120 mM Tris-HCl (pH 8.7), 0.1% Triton X-100, and 5% glycerol. Lysates were pelleted to clear cellular debris, and protein was quantified using bicinchonic acid assay (Pierce). 10 g of protein were size-fractionated through 8% SDS-polyacrylamide gel impregnated with 0.02% gelatin under nonreducing conditions. Gels were washed in 2.5% Triton X-100 and then incubated for 20 h at 37°C in 50 mM Tris-HCl, pH 8.0, containing 5 mM calcium chloride. Gels were fixed and stained with Coomassie Blue R-250 as described previously (8). Gels were imaged using a FluorChem gel documentation system, and densitometry was performed using Alpha-Ease software. Cleared bands were quantified by densitometry to calculate total MMP-2 protein (sum of latent and active bands), and % active MMP-2, calculated as (active band/total MMP-2) ϫ 100.
Northern Blotting-Total cellular RNA was extracted from two-dimensional or three-dimensional cultured endothelial cells using TRIzol (Invitrogen) according to the manufacturer's directions. The RNA concentration was determined by spectrophotometer. 10 g of total RNA was electrophoresed through a 1% agarose-formaldehyde gel as described previously (8). Gels were stained with SYBR green to visualize ribosomal subunits. The cDNA probe encoding mouse MMP-2 was labeled with ␣-[ 32 P]dCTP using random primer labeling (New England Biolabs) and separated from unincorporated nucleotides using G-50 spin columns (Amersham Biosciences). Blots were hybridized overnight, washed, and then exposed to Hyperfilm (Amersham Biosciences) at Ϫ80°C. Films were imaged with the FluorChem system, and intensities of MMP-2 mRNA bands were normalized to 28 S densities to control for loading.
Nuclear Extracts-Nuclear extracts of endothelial cells cultured for 24 h on two-dimensional collagen (12.5 g/cm 2 ) or three-dimensional collagen were prepared according to a previous study (17) and quantified using bicinchonic acid assay (Pierce).
Western and Southwestern Blots-For Western blots, nuclear extracts (10 g) were separated using 12% polyacrylamide gels under reducing conditions and then transferred to polyvinylidene difluoride membranes. After blocking, blots were incubated overnight with anti-bodies for AP-2 or p53 (Santa Cruz Biotechnology) or YB-1 (kindly provided by Dr. David Lovett, San Francisco Veterans Affairs Medical Center, University of California). After secondary antibody incubation, bands were detected using the SuperSignal West Pico (Pierce) and FluorChem gel documentation systems (AlphaInnotech). Densitometry was performed using AlphaEase software. For Southwestern blots, nuclear extracts (15 g) were separated using 12% polyacrylamide gels under reducing conditions, which were then transferred to nitrocellulose membranes. After renaturing and blocking overnight at 4°C in TNED buffer (50 mM Tris-HCl, pH 7.5, 50 mM NaCl, 0.1 mM EDTA, and 1 mM dithiothreitol) containing 5% skim milk powder, blots were washed with TNED buffer and then hybridized for 5 h at room temperature using TNED buffer containing 10 g/ml poly(dI⅐dC) and 8 pmol of ␥-[ 32 P]ATP-labeled double-stranded probe corresponding to base pairs Ϫ1435 to Ϫ1387 of the MMP-2 promoter (synthesized by Sigma). Blots were washed twice for 15 min with TNED and then exposed to x-ray film at Ϫ80°C using an enhancing screen.
Electrophoretic Mobility Shift Assays-Consensus GATA-2 binding double-stranded oligonucleotides (5Ј-CCTGGCTTATCTCCGGCTGC-3Ј; Panomics) and complimentary oligonucleotides corresponding to base pairs Ϫ1435 to Ϫ1387 or Ϫ1329 to Ϫ1293 of the MMP-2 promoter were annealed and end-labeled using ␥-[ 32 P]ATP. An enriched source of human GATA-2 in vitro transcribed and translated from pMT2-GATA2 (kindly provided by Dr. Stuart Orkin) with the Promega TNT system was used to define the binding pattern of this transcription factor to consensus GATA-2 binding domains and to the MMP-2 promoter region. Gel shift reactions consisted of a 5 ϫ 10 4 cpm oligonucleotide probe, binding buffer (10 mM Tris-HCl, pH 7.5, 50 mM NaCl, 1 mM EDTA, 0.5 mM dithiothreitol, 10% glycerol), 1 g of poly(dI⅐dC), 500 ng of salmon sperm DNA, 1 l of TNT reaction, or 5-10 g of nuclear extract. For supershift assays, 1-4 l of antibody (anti-GATA-2 from Santa Cruz Biotechnology or anti-GATA-2 kindly provided by Dr. Stuart Orkin) was added to the reaction and incubated for 15 min prior to gel loading. Reactions were electrophoresed through a 4% nondenaturing polyacrylamide gel followed by autoradiography.
Transient Transfections-The rat MMP-2 promoter (72-PGL2) was provided kindly by Dr. David Lovett. The full-length promoter (Ϫ1686 base pairs to the ATG codon) was subcloned into pGL3Basic (Promega). Truncations were constructed using restriction enzyme double digests (KpnI-AflII, KpnI-AlwN1, KpnI-PstI, and KpnI-SmaI) followed by gel purification and ligation of the plasmid, transformation into bacteria, and purification of selected colonies using Qiafilter Midi Preps (Qiagen). LipofectAMINE-based transient transfections of MMP-2 promoter sequences and Renilla luciferase were performed on COS-7 cells or on primary cultures of rat microvascular endothelial cells (passages 6 -10) according to the manufacturer's directions (Invitrogen). For COS-7 transfections, cells were lysed 48 h after transfection. For endothelial cell transfections, 24 h after transfection, cells were trypsinized and divided into two-dimensional and three-dimensional conditions. 48 h after two-dimensional and three-dimensional culturing, cells were lysed and assayed for luciferase activity according to the Promega Dual Luciferase Reporter System. Luminescence was detected with a Berthold 9501 tube luminometer. Data were normalized for activity of Renilla luciferase to account for transfection efficiency and then expressed as -fold increase in comparison with the light output of the pGL3 Basic or pGL3 promoter constructs. Results from four independent experiments, each with duplicate wells, were averaged and presented as mean Ϯ S.E.
Statistics-Statistical analyses of two-dimensional and three-dimensional conditions were performed using Student's paired t tests with significance established as p Ͻ 0.05.

RESULTS
Rat skeletal muscle microvascular endothelial cells cultured for 24 h within a three-dimensional collagen type I matrix produced significantly greater MMP-2 mRNA and protein when compared with cells cultured on a thin monolayer coating of collagen type I (Fig. 1, A and B). Likewise, the MMP-2 promoter analysis demonstrated greater transcriptional activity of the fulllength promoter in endothelial cells cultured in three-dimensional compared with two-dimensional conditions (Fig. 1C). Truncation of the promoter region (from Ϫ1562 to Ϫ1375) removed the responsiveness to the three-dimensional collagen culture condition. Further deletion of Ϫ1365 to Ϫ510 again reduced promoter activity but did not have a differential effect on cells in two-dimensional versus three-dimensional culture (Fig. 1C).
The region between Ϫ1322 and Ϫ1282 has previously been designated as RE1 by Lovett and colleagues (9). This region contains binding sites for transcription factors YB-1, AP-2, and p53 (11). We tested whether these transcription factors were induced in three-dimensional culture by Western blotting of nuclear extracts. There was no change in p53 (0.91 Ϯ 0.32), a slight decrease in AP-2 (0.80 Ϯ 0.07), and a strong decrease in YB-1 (0.11 Ϯ 0.1, p Ͻ 0.05), in three-dimensional culture compared with their two-dimensional levels (Fig. 2, A-C). Gel shift assays using double-stranded oligonucleotides corresponding to base pairs Ϫ1327 to Ϫ1293 of the MMP-2 promoter showed two shifted complexes; the faster mobility complex was of similar intensity in nuclear extracts of cells in two-dimensional compared with three-dimensional cultures, whereas the slower mobility complex was greatly reduced in three-dimensional compared with two-dimensional nuclear extracts (Fig. 2D).
We thus focused on the region of the MMP-2 promoter that lies between Ϫ1562 and Ϫ1375. A transcription factor search on this region using the TRANSFAC data base (18) identified two sites located at Ϫ1431/Ϫ1422 and Ϫ1399/Ϫ1390 that shared considerable homology to the nominal binding sequence for the transcription factor GATA-2 (Fig. 3A). This region corresponds to a portion of the enhancer region described by Bottles et al. (13). We then examined protein binding to this sequence first through Southwestern blotting. Using endothelial cell nuclear extract as a protein source, this technique Results represent means Ϯ S.E. of five independent experiments. In B, Northern blotting showed similar increases in MMP-2 mRNA (normalized for loading to 28 S ribosomal RNA). Results are from three independent experiments. Asterisks represent significance compared with twodimensional control (p Ͻ 0.05). C summarizes the transcriptional activities of full-length and truncated MMP-2 promoter-luciferase constructs when transiently transfected into microvascular endothelial cells. The transfected cells were then cultured in two-dimensional (hatched bars) and three-dimensional (solid bars) collagen matrices for 48 h, and luciferase activities were measured. The position of truncation (relative to translation start site) is illustrated on the left. On the right, ratios of three-dimensional to two-dimensional mean luciferase activities are provided for each promoter construct. Results (mean Ϯ S.E. of four independent experiments) are ratios of firefly luciferase compared with Renilla activities expressed as a -fold increase above to Basic (promoterless) activity. identified a weak band (of ϳ53 kDa) that interacted with the labeled MMP-2 promoter sequence (Fig. 3B). This size is consistent with GATA-2, as has been demonstrated by other researchers using Western blotting (19). We tested specifically for GATA-2 protein binding ability to the MMP-2 promoter sequence using gel shift assay with in vitro transcribed and translated GATA-2 protein. Two shifted bands were detected when binding to the MMP-2 promoter sequence, whereas one band was detected when binding to the consensus GATA-2 oligonucleotide (Fig. 3C). Incubation of the protein-DNA complex with anti-GATA-2 antibody resulted in a supershifted complex, a decreased intensity of the slower mobility protein-DNA complex (solid arrow), and increased intensity of the faster mobility protein-DNA complex (open arrowhead) (Fig.  3D). It is likely that both bands correspond to GATA-2 binding and arise from variable levels of protein interaction to the two GATA-2 binding sites present in the MMP-2 promoter sequence (compared with the single site on the consensus oligonucleotide).
We tested for direct evidence that GATA-2 trans-activates the MMP-2 promoter by cotransfecting COS-7 cells with GATA-2 or empty expression vector and the full-length MMP-2 promoter-luciferase construct. Western blotting confirmed the effectiveness of the overexpression procedure in elevating levels of GATA-2 protein in the COS cells (Fig. 4A). MMP-2 pro-moter activity was enhanced 3.3-fold (p Ͻ 0.05) in the presence of GATA-2 overexpression compared with the empty vector (Fig. 4B).
To assess the role of GATA-2 in the transcriptional control of MMP-2 in endothelial cells, we first examined endogenous levels of GATA-2 in cells cultured in two-dimensional or threedimensional type I collagen. By Western blotting, GATA-2 protein was increased in the nuclear extracts of endothelial cells cultured in three-dimensional collagen compared with twodimensional collagen (Fig. 5A). Similarly, gel shift assays showed formation of GATA-2 complexes in the presence of both two-dimensional and three-dimensional nuclear extracts, but these complexes were more intense in the presence of the three-dimensional cultured endothelial cell nuclear extracts (Fig. 5B). These complexes were identical to those seen using the in vitro transcribed and translated GATA-2 (Fig. 3, C and D) and were competed with 10 -50-fold excess cold oligonucleotide and supershifted with anti-GATA-2 antibody (data not shown). We confirmed the differential activity of the GATA-2 binding region in response to extracellular matrix cues through transient transfection of microvascular endothelial cells with the GATA-2 binding sequence of the MMP-2 promoter (Ϫ1435/ Ϫ1387) coupled to the pGL3 promoter followed by culturing the transfected cells in two-dimensional or three-dimensional culture conditions for 48 h. Extracts from cells in the three- FIG. 2. Known transcriptional activators are not increased in response to three-dimensional collagen. Western blotting of endothelial cell nuclear extracts following 24 h of culture in twodimensional (2D) and three-dimensional (3D) conditions showed no change in p53 protein levels (A), a slight decrease in AP-2 (B), and a significant decrease in YB-1 (C). Results are presented as mean Ϯ S.E. for three independent experiments. Asterisk indicates significance compared with two-dimensional control (p Ͻ 0.05). Electrophoretic mobility shift assay of endothelial cell nuclear extracts incubated with labeled oligonucleotide corresponding to base pairs Ϫ1327 to Ϫ1293 of the MMP-2 promoter indicates two shifted bands, the slower of which decreases in three-dimensional conditions compared with two-dimensional conditions, whereas the faster shifted band remains unchanged between two-dimensional and three-dimensional conditions (D). dimensional culture condition had significantly greater transactivation of the GATA-2 binding sequence compared with those from two-dimensional culture conditions (1.8-fold above two-dimensional conditions, p Ͻ 0.05; Fig. 5C). DISCUSSION This study shows that the transcription factor, GATA-2, enhances the transcription of MMP-2 in microvascular endothelial cells cultured within a three-dimensional type I collagen matrix. These results provide the first mechanistic evidence of extracellular matrix-driven transcriptional regulation of the MMP-2 gene.
Despite considerable evidence for regulation of the MMP-2 promoter by YB-1, AP-2, and p53 interactions in rat mesangial cells (10,11), we failed to detect a role for this region in conferring extracellular matrix responsiveness to the MMP-2 promoter. These transcription factors may contribute to overall promoter activity, as a truncation that removed a region of the promoter that included the RE1 region (Ϫ1322 to Ϫ1282) greatly reduced promoter activity in both two-dimensional and three-dimensional cultured endothelial cells (Fig. 1). We observed by Western blotting that the protein level of YB-1 was significantly lower in nuclear extracts of three-dimensional compared with two-dimensional cultured endothelial cells. Ybox binding proteins, including YB-1, can interact with both DNA and RNA resulting in transcriptional or translational activation or repression depending on the genes in question (20). Whereas the Y-box protein dpbB is reported to contribute to the transcriptional activation of thrombin-induced genes in endothelium (21), the regulation of YB-1 expression and function in endothelial cells has not yet been reported. Interest-ingly, the collagen type ␣1(I) gene in fibroblasts is repressed strongly by YB-1 interaction with the proximal promoter (22). Our results may be consistent with the role of YB-1 as a transcriptional repressor of MMP-2 in endothelial cells, because lower levels of YB-1 protein are present in three-dimensional cultured cells. It is possible that YB-1 may act to repress or interfere with transcriptional activation of MMP-2 by GATA-2, whereas the relief of YB-1 in three-dimensional conditions contributes to the overall positive transcriptional response. However, this remains to be clarified through additional experimentation as we have not specifically examined the effects of YB-1 overexpression on the MMP-2 promoter.
Rather, we defined a role for GATA-2 direct interaction with the MMP-2 promoter region Ϫ1435/Ϫ1387 utilizing analyses of in vitro DNA-protein complex formation (Figs. 3 and 5) and confirmed the trans-activation capability of GATA-2 through measurements of MMP-2 transcriptional activity in COS-7 and microvascular endothelial cells (Figs. 4 and 5). Together, these experiments provide clear evidence that GATA-2 interacts with the MMP-2 promoter and that this interaction enhances promoter activity. Increased levels of GATA-2 protein and enhanced binding of GATA-2 to the MMP-2 promoter occur in response to culture within a three-dimensional collagen matrix, implying that this domain acts as a response element for extracellular matrix-driven signaling events.
Although we have not tested for GATA-2 binding to the human MMP-2 promoter sequence, it is notable that the dual GATA-2 binding sites with 21 nucleotides separating them are highly conserved between human and rat promoters. The equivalent binding domains are located at base pairs Ϫ1434/ Ϫ1425 and Ϫ1403/1394 of the human MMP-2 sequence, as published by Bian and Sun (14).
GATA-2 is a member of the GATA family of DNA binding proteins that interacts with a DNA sequence motif (WGATAR) through highly conserved C 2 C 2 zinc finger domains (23,24). This family of transcription factors initially was recognized to play an important role in regulating expression of multiple genes in cells of erythroid origin, although it is now known that these transcription factors are also expressed in cells of other origin such as fibroblasts, embryonic brain, cardiac muscle, liver, and kidney (25). Human GATA-2 was cloned from endothelial cells following discovery of its critical role in regulating transcription of the preproendothelin-1 gene (25,26). Since that time, studies have demonstrated that GATA-2 regulates multiple endothelial cell genes including endothelial cell nitricoxide synthase (27), vascular cell adhesion molecule-1 (28), platelet/endothelial cell adhesion molecule-1 (29), and the vascular endothelial growth factor receptor flk1/KDR (30).
Interestingly, many of the genes regulated by GATA-2 are endothelial cell-specific, and, in fact, cellular dedifferentiation of endothelial cells is linked to reduced expression of several transcription factors including GATA-2 (31). However, GATA-2 itself cannot provide cell type specificity because it is found in a number of cell types. Rather, it is thought that endothelial cell-specific gene expression in these cases is achieved by cooperativity between GATA-2 and additional transcription factors. Several studies have shown cooperative interactions between GATA-2 and other transcription factors, including Sp1 (27) and c-fos and c-jun (32). In addition, a family of proteins that include "friend of gata," or FOG, FOG-2, and U-shaped (Ush) transcription factors are known to play critical roles as cofac-tors for GATA transcription factors in both erythroid (33) and cardiac muscle (34,35). It is possible that similar cooperative interactions can modulate the effectiveness of GATA-2 transactivation of the MMP-2 promoter.
Utilizing gel shift analysis, we observed two shifted bands corresponding to GATA-2 interactions with the MMP-2 promoter sequence (Ϫ1435/Ϫ1387). We think that both bands represent specific GATA-2 interactions with the MMP-2 promoter oligonucleotide, which has two putative GATA-2 binding sites. Both bands were competed with cold oligonucleotide. The addition of GATA-2-specific antibodies resulted in either 1) a supershifted band with a decrease in intensity of the slower Western blot showed an increase in GATA-2 protein in three-dimensional compared with two-dimensional nuclear extracts (A). The arrow indicates GATA-2 protein immunoreactivity at ϳ53 kDa. Electrophoretic mobility shift assay (B) using 32 P-labeled double stranded Ϫ1435/Ϫ1387 bp of the MMP-2 promoter demonstrated two shifted complexes (as seen when using in vitro synthesized GATA-2 protein) that were both much more prominent in three-dimensional compared with two-dimensional cultured cell extracts. Both bands were competed in the presence of increasing amounts of unlabelled oligonucleotide (-1437/Ϫ1387). C shows GATA-2 binding site activity in endothelial cells. The sequence Ϫ1435/Ϫ1387 of the double stranded MMP-2 promoter was subcloned into the pGL3 promoter and transiently transfected into microvascular endothelial cells. Luciferase activities of the Ϫ1435/Ϫ1387 promoter construct were measured after 2 days of culture as a monolayer (2D) or within threedimensional type I collagen matrix (3D) and normalized to the activity of the enhancerless pGL3 promoter. Data are presented as mean Ϯ S.E. for 3 independent experiments. The asterisk denotes significance (p Ͻ 0.05) compared with the two-dimensional culture condition. migrating protein-DNA complex and an increase in intensity of the faster migrating protein-DNA complex (Fig. 3D) or 2), when a greater amount of antibody was added, in the complete competition of both bands (data not shown). Despite this evidence, it is possible that the multiple gel shift bands result from related GATA protein interactions with the DNA probe, or from GATA-2 interactions with other proteins. Although GATA-2 was thought to be the predominant GATA factor found in endothelial cells (25), a more recent study has identified the mRNA for GATA-3 and GATA-6 in human umbilical vein endothelial cells and demonstrated that these factors can bind to and trans-activate the vascular cell adhesion molecule-1 promoter (36). We have shown that GATA-2 protein, by itself, is sufficient to trans-activate the MMP-2 promoter (Fig. 4B), but the possibility of enhanced MMP-2 transcription in the presence of additional GATA factors bears further investigation.
We previously found that the early growth response-1 (egr-1) transcription factor trans-activates the membrane type 1 (MT1)-MMP promoter sequence in response to three-dimensional type I collagen stimulation (17). Because MT1-MMP and MMP-2 gene products frequently appear to be co-regulated, we originally were surprised that the MMP-2 promoter lacks consensus binding sites for egr-1. The current study provides evidence for GATA-2 as a critical transcription factor in the regulation of the MMP-2 promoter under the pro-angiogenic condition of three-dimensional type I collagen stimulation. We have recently demonstrated a critical role for extracellular signal-regulated kinase-1/2 activation in the regulation of both MMP-2 and MT1-MMP mRNA in three-dimensional collagenstimulated endothelial cells. 2 Notably, both egr-1 and GATA-2 are downstream effectors of the extracellular signal-regulated kinase-1/2 signal pathway (19,37,38), implying that extracellular signal-regulated kinase-1/2 may be a critical component of a common signal pathway responsible for recruitment of both transcription factors. Further studies will elucidate the mechanisms by which extracellular matrix-driven signaling results in coordinated recruitment of diverse transcription factors and activation of the transcription of multiple genes critical for the process of angiogenesis.