The Transcription Factors Sp1, Sp3, and AP-2 Are Required for Constitutive Matrix Metalloproteinase-2 Gene Expression in Astroglioma Cells*

Matrix metalloproteinases (MMPs) are zinc-dependent endopeptidases that contribute to pathological conditions associated with angiogenesis and tumor invasion. MMP-2 is highly expressed in human astroglioma cells, and contributes to the invasiveness of these cells. The human MMP-2 promoter contains potential cis-acting regulatory elements including cAMP response element-binding protein, AP-1, AP-2, PEA3, C/EBP, and Sp1. Deletion and site-directed mutagenesis analysis of the MMP-2 promoter demonstrates that the Sp1 site at −91 to −84 base pairs and the AP-2 site at −61 to −53 base pairs are critical for constitutive activity of this gene in invasive astroglioma cell lines. Electrophoretic gel shift analysis demonstrates binding of specific DNA-protein complexes to the Sp1 and AP-2 sites: Sp1 and Sp3 bind to the Sp1 site, while the AP-2 transcription factor binds the AP-2 element. Co-transfection expression experiments inDrosophilia SL2 cells lacking endogenous Sp factors demonstrate that Sp1 and Sp3 function as activators of the MMP-2 promoter and synergize for enhanced MMP-2 activation. Overexpression of AP-2 in AP-2-deficient HepG2 cells enhances MMP-2 promoter activation. These findings document the functional importance of Sp1, Sp3, and AP-2 in regulating constitutive expression of MMP-2. Delineation of MMP-2 regulation may have implications for development of new therapeutic strategies to arrest glioma invasion.

The matrix metalloproteinases (MMPs) 1 are a family of structurally related zinc-dependent endopeptidases capable of degrading extracellular matrix components (for review, see Refs. [1][2][3]. The MMP family includes collagenases, gelatinases, stromelysins, membrane-type metalloproteinases, matrilysin, and metalloelastase. A hallmark of invasive tumors is their ability to degrade the surrounding extracellular matrix, resulting in a compromised matrix organization and disruption of tissue boundaries. A number of in vitro and in vivo studies have documented a direct correlation between high levels of expression of MMP and an increased invasive capacity of tumor cell lines (1, 4 -6). Glioblastoma multiforme is a highly malignant central nervous system tumor that is extremely refractory to therapy due to the rapid growth and local invasive potential of these tumors (for review, see Refs. 7 and 8). This rapid infiltrative growth prevents successful surgical resection of glioblastoma multiformes. The ability of glioma cells to invade the surrounding tissue has been attributed to their secretion of MMPs. In vitro, human glioma cell lines express a variety of MMPs, in particular the type IV collagenases MMP-2 and MMP-9 (9 -15). In vivo studies have also demonstrated the expression of MMP-2 and MMP-9 in human gliomas (11, 12, 16 -18); MMP-2 and MMP-9 expression was the highest in high grade gliomas (glioblastoma multiforme, anaplastic astroglioma) compared with non-invasive low grade astrogliomas and normal brain (11,18,19). Importantly, the invasiveness of glioma cells in vitro correlates with high levels of MMP-2 expression (9,12,14,15,20). A number of strategies have been utilized to modulate MMP-2 expression/activity, then assess subsequent changes in invasive potential. The MMP-2 proenzyme is activated by cell surface-associated membrane-type metalloproteinases (21,22). Transfection of U251.3 glioma cells with membrane-type metalloproteinases-1 leads to activation of the MMP-2 proenzyme and enhanced invasion as assessed by Matrigel assay (23), as well as remodeling of the extracellular matrix in vitro (24). We have recently demonstrated that two cytokines, tumor necrosis factor-␣ and interferon-␥, inhibit MMP-2 expression in glioma cells, which results in decreased invasiveness of these cells (15). Collectively, these results highlight the important role of MMP-2 in the invasive potential of astroglioma cells.
The activity of MMP-2 is regulated by several mechanisms, including gene expression, proenzyme activation by tissue inhibitor of metalloproteinase-2 and membrane-type metalloproteinases, and inhibition of enzyme activity by naturally occurring tissue inhibitor of metalloproteinases (for review, see Refs. [25][26][27]. Historically, the MMP-2 gene has been considered refractory to modulation, either inhibition or enhancement, due to a lack of well characterized regulatory elements in the MMP-2 promoter (for review, see Ref. 25). However, sequence analysis of the MMP-2 promoter has revealed a number of potential cis-acting regulatory elements including p53, AP-1, Ets-1, C/EBP, CREB, PEA3, Sp1, and AP-2 that could be involved in regulation of MMP-2 expression (see Fig. 2). The human MMP-2 promoter lacks a typical TATA box but has a relatively GC-rich region in the proximal region. Limited functional analysis of the MMP-2 promoter has been performed. An enhancer sequence of 42 bp located at Ϫ1635 bp relative to the transcription initiation start site in the human MMP-2 promoter has been identified in HT1080 cells (human fibrosarcoma line) (28). This enhancer region-2 (r2) was identified as an AP-2 binding sequence, allowing for several mismatches and gaps (28). It has been suggested that AP-2 is an important transcription factor for activation of the MMP-2 promoter, and that adenovirus EIA represses MMP-2 gene expression by targeting AP-2 (29). A silencer sequence at Ϫ1629 bp has also been functionally documented (28). Recently, a p53-binding site at Ϫ1649 to Ϫ1630 bp (20 bp) has been identified which specifically binds the p53 protein and plays an important role in constitutive MMP-2 gene expression in human sarcoma cell lines (30). Interestingly, the p53-binding site is localized within the enhancer region-2 (r2); however, Bian and Sun (30) did not detect AP-2 binding in this region, only that of p53. No information is available regarding any other potential functional cis-acting elements in the MMP-2 promoter.
The constitutive expression and regulation of the MMP-2 gene is cell-and stimulus-specific (15,30,31). Thus far, the molecular basis of MMP-2 gene expression in human astroglioma cells has not been addressed. In order to determine the sequence requirements for MMP-2 gene transcription, we utilized 5Ј-deletional MMP-2 reporter constructs in transcriptional activation studies. This approach, in conjunction with site-directed mutagenesis of the human MMP-2 promoter region, demonstrated that a Sp1 site spanning Ϫ91 to Ϫ84 bp and an AP-2 element at Ϫ61 to Ϫ53 bp are critical for constitutive expression of the MMP-2 gene in astroglioma cells. The p53/r2 elements identified as important for MMP-2 promoter activity in other cells are non-functional in astroglioma cells. We further defined DNA-protein complexes involved in MMP-2 transcription by electrophoretic mobility shift assays (EMSA). The transcription factors Sp1 and Sp3 both bind to the Sp1 element, while AP-2 binds to the AP-2 element. Furthermore, co-transfection experiments into mammalian cells lacking AP-2 and insect cells that lack Sp factors demonstrate that Sp1, Sp3, and AP-2 are all functionally important for constitutive expression of the MMP-2 gene.
Analysis of MMP Protein and mRNA Expression-Zymography was performed as described previously (15). In brief, astroglioma cell lines cells were incubated until ϳ80% confluent, then the medium was aspirated and fresh serum-free medium added to the dish. Supernatants were collected after a 48-h incubation and concentrated. Concentrated supernatants (750 l) were mixed with SDS sample buffer without reducing agent, and proteins subjected to SDS-polyacrylamide gel electrophoresis in 8% polyacrylamide gels that were copolymerized with 1 mg/ml gelatin. After electrophoresis, the gels were washed several times in 2.5% Triton X-100 for 1 h at room temperature to remove the SDS, and then incubated for 24 -48 h at 37°C in buffer containing 5 mM CaCl 2 and 1 M ZnCl 2 . The gels were stained with Coomassie Blue (0.25%) for 30 min, then destained for 1 h in a solution of acetic acid and methanol. The proteolytic activity was evidenced as clear bands (zones of gelatin degradation) against the blue background of stained gelatin.
The same supernatants were used in immunoblot analysis for MMP-2 protein. Proteins were detected using anti-MMP-2 antibody (5 g/ml) as described previously (15). Blots were washed four times in TBS with 0.01% Tween 20, and subsequently incubated in sheep anti-mouse peroxidase-conjugated antibody (1:3000) in antibody dilution buffer. ECL reagents were used for development. Total RNA was isolated from confluent monolayers of astroglioma cell lines that had been incubated in serum-free medium for 24 h (15). Human MMP-2 cDNA (gift of Dr. W. G. Stetler-Stevenson, National Cancer Institute, Bethesda, MD) was digested with SacI/PstI, and a 324-bp fragment corresponding to 1500 -1824 nucleotides was subcloned into the SacI/PstI polylinker site of the pGEM4Z vector (Promega, Madison, WI). The construct was linearized by EcoRI and used to generate a radiolabeled antisense RNA probe of 354 nucleotides with T7 RNA polymerase. A pAMP-1 vector containing a fragment of the human glyceraldehyde-3-phosphate dehydrogenase cDNA (corresponding to 43-531 nucleotides) was linearized with NcoI, and used to generate a radiolabeled antisense RNA probe of 290 nucleotides with T7 polymerase. Fifteen g of total RNA was hybridized with MMP-2 (50 ϫ 10 3 cpm) and glyceraldehyde-3-phosphate dehydrogenase (25 ϫ 10 3 cpm) riboprobes at 42°C overnight. The hybridized mixture was then treated with RNase A/T1 (1:200) at room temperature for 1 h, analyzed by 5% denaturing (8 M urea) polyacrylamide gel electrophoresis, and the gels were exposed to x-ray film. The protected fragments of the MMP-2 and glyceraldehyde-3-phosphate dehydrogenase riboprobes are 324 and 230 bp in length.
Human MMP-2 Promoter Constructs-A luciferase reporter plasmid driven by 1659 bp of the human MMP-2 promoter was used in this study, and is referred to as wild type (WT) (30). Serial deletion mutants were synthesized using polymerase chain reaction by Pfu DNA polymerase to delete out potential regulatory elements in the MMP-2 promoter, including one p53 site, two silencers, one AP-1 element, four PEA3 elements, two Sp1 elements, and one AP-2 site (see Fig. 2). They are named as follows: D1, which lacks the p53 site; D2, lacking the first silencer element (S1); D3, lacking both silencer elements (S1 and S2); D4, which lacks the AP-1 element; D5, lacking the Ets-1 and c-Myc/c-Myb sites; D6, lacking one PEA3 site and the C/EBP element; D7, lacking the second PEA3 site; D8, which lacks the third PEA3 site, the CREB site, and GCN-His region; D9, lacking the fourth PEA3 site; D10, lacking the first Sp1 element; D11, lacking both Sp1 elements; and D12, which lacks the AP-2 site (see Fig. 3, A and B). The polymerase chain reaction amplification protocol consisted of an initial 1-min melting step at 94°C, followed by 30 cycles with 40 s melting at 92°C, 40 s annealing at 60°C, and 1 min 30 s extension at 75°C, except for the last cycle which contained a 5-min extension step. The restriction enzyme EcoRI was used to check positive clones and AvaI to verify the correct orientation. The deletion mutants were inserted into the pGL2-Basic vector, which contains the gene for luciferase as reporter, using the KpnI/XhoI restriction site.
Site-directed Mutagenesis of the Human MMP-2 Promoter-Site-directed mutagenesis was utilized to mutate potential transcription elements in the proximal sequence (Ϫ139 to Ϫ7) of the human MMP-2 promoter. 15-bp lengths of DNA in the D3 and D9 reporter constructs were serially mutated to the EcoRV and XbaI restriction enzyme sequence by polymerase chain reaction using QuickChange TM Site-Directed Mutagenesis Kit (Stratagene, La Jolla, CA) (see Fig. 4). The introduction of mutations was verified by the restriction enzymes EcoRV and XbaI and DNA sequencing. After 18 cycles of polymerase chain reaction by Pfu polymerase at 95°C for 30 s, 55°C for 1 min, and 68°C for 20 min, DpnI was used to digest the parental supercoiled double-stranded DNA. Transformations were performed in Escherichia coli supercompetent cells using DpnI-treated DNA. A total of three site-directed mutants were generated using the D3 and D9 promoter constructs as templates (Fig. 4), and were confirmed by sequencing. They are mSp1A, mSp1B, and mAP-2. As well, constructs containing mutations in both the Sp1A and AP-2 elements were created.
Transient Transfection and Luciferase Assay-Ten g of the MMP-2 promoter constructs (both MMP-2 deletion and mutation constructs) were co-transfected into human astroglioma cells with 1 g of the pCMV-␤-galactosidase construct into 3 ϫ 10 6 cells by electroporation with a Bio-Rad Gene Pulser set at 250 V, 960 microfarads, as described previously (15). After transfection, cells were allowed to recover for 18 h and cultured in 1% FBS/Dulbecco's modified Eagle's/F-12 medium for 24 h. Cells were washed with phosphate-buffered saline and lysed with 180 l of lysis buffer containing 25 mM trisphosphate (pH 7.8), 2 mM DTT, 2 mM diaminocyclohexane tetraacetic acid, 10% glycerol, and 1% Triton X-100. Extracts were assayed in triplicate for luciferase activity in a volume of 130 l containing 30 l of cell extract, 20 mM Tricine, 0.1 mM EDTA, 1 mM magnesium carbonate, 2.67 MgSO 4 , 33.3 mM DTT, 0.27 mM coenzyme A, 0.47 mM luciferin, and 0.53 mM ATP, and light intensity was measured using a luminometer (Promega, Madison, WI). Luciferase activity was integrated over a 10-s time period. Extracts were also assayed in triplicate for ␤-galactosidase enzyme activity as described previously (15). The luciferase activity of each sample was normalized to ␤-galactosidase activity to calculate relative luciferase activity (RLA) before calculating the fold activation value.
For transfection with HepG2 cells, 3 ϫ 10 6 cells were electroporated with 10 g of the indicated MMP-2 luciferase promoter constructs, 1 g of the pCMV-␤-galactosidase construct, and 1.0 g of either the pSX-AP-2 expression vector or pSX control vector (34). After transfection, cells were allowed to recover for 18 h, changed to serum-free medium for 24 h, and luciferase and ␤-galactosidase activities assayed as described above.
For transfection of SL2 cells, 1 day prior to transfection, cells were plated onto 60-mm 2 dishes at a density of 4 ϫ 10 6 cells/plate. Cells were transfected by the calcium phosphate method as described previously (35). Each plate received up to 20 g of DNA including 10 g of the indicated MMP-2 promoter constructs, and variable amounts of the expression plasmids pPacSp1 and/or pPacUSp3 (36). Variable amounts of the expression plasmids were adjusted with the control plasmid pPac. The medium was changed 24 h after addition of DNA, and 24 h later the cells were harvested for luciferase activity. Luciferase values were normalized against total protein concentrations determined by the Bio-Rad protein assay.
Nuclear Extracts and Electrophoretic Mobility Shift Assay (EMSA)-Nuclear extracts were prepared as described previously (37). Cells were grown in 100-mm dishes, allowed to adhere overnight, and then were incubated in serum-free medium for 24 h. Cells were then washed with cold phosphate-buffered saline, harvested by scraping, and pelleted. Cells were resuspended in 1 ml of buffer A (10 mM KCl, 20 mM HEPES, 1 mM MgCl 2, 1 mM DTT, 0.4 mM phenylmethylsulfonyl fluoride, 1 mM sodium fluoride, 1 mM Na 3 VO 4 ), incubated on ice for 10 min, and pelleted at 1000 ϫ g for 10 min. Pellets were resuspended in 0.5 ml of buffer A plus 0.1% Nonidet P-40, incubated on ice for 10 min, and centrifuged at 3,000 ϫ g for 10 min. The nuclear pellet was resuspended in 1 ml of buffer B (10 mM HEPES, 400 mM NaCl, 0.1 mM EDTA, 1 mM MgCl 2, 1 mM DTT, 0.4 mM phenylmethylsulfonyl fluoride, 15% glycerol, 1 mM sodium fluoride, 1 mM Na 3 VO 4 ) and incubated for 30 min at 4°C with constant gentle mixing. Nuclei were then pelleted at 40,000 ϫ g for 30 min, and extracts were dialyzed for 2 h at 4°C against 1 liter of buffer C (20 mM HEPES, 200 mM KCl, 1 mM MgCl 2, 0.1 mM EDTA, 1 mM DTT, 0.4 mM phenylmethylsulfonyl fluoride, 15% glycerol, 1 mM sodium fluoride, 1 mM Na 3 VO 4 ). Extracts were cleared by centrifugation at 14,000 ϫ g for 15 min at 4°C. Protein concentrations were determined using a Bio-Rad protein assay.
EMSA was performed using the following oligonucleotides as probes and/or competitors: the oligonucleotide Sp1A has the sequence 5Ј-CA-GAGAGGGGCGGGCCCGAGTG-3Ј, corresponding to the human MMP-2 promoter sequence Ϫ98 to Ϫ76, and the AP-2 oligonucleotide has the sequence 5Ј-CCCCAGCCCCGCTCTGCCAGCT-3Ј, and corresponds to the human MMP-2 promoter sequence Ϫ66 to Ϫ44. The mutant Sp1 oligonucleotide (mSp1A) has the sequence 5Ј-CAGAtAtcta-gatGatatcGTG-3Ј, and the mutant AP-2 oligonucleotide (mAP-2) is 5Ј-CCgatatCatctagaatCAGCT-3Ј. Mutations are indicated by lowercase letters. 0.2 ng of 32 P-labeled oligonucleotide (20,000 cpm) were incubated for 30 min at room temperature with 10 g of nuclear extract in a volume of 20 l containing 50 mM KCl, 2.5 mM MgCl 2, 1 mM EDTA, 1 mM DTT, 10 mM Tris-Cl (pH 7.5), 10% glycerol, 1 g of salmon sperm DNA, and 1 g of poly(dI-dC). For supershift analysis, 1 l of antibody was incubated with the nuclear extracts at 4°C for 30 min in binding buffer, followed by an additional incubation for 30 min at room temperature with labeled oligonucleotide. For competitions, unlabeled DNA was incubated with the nuclear extracts at 4°C for 20 min before addition of labeled probe. Bound and free DNA were resolved by electrophoresis through a 6% polyacrylamide gel at 250 V in 0.25 ϫ TBE (50 mM Tris-Cl, 2 mM EDTA). Dried gels were exposed to Kodak XAR-5 film at Ϫ70°C with intensifying screens. Four different preparations of nuclear extracts were tested by EMSA.
Statistical Analysis-Levels of significance for comparisons between samples were determined using Student's t test distribution.

Identification of the Transcription Elements Controlling
Constitutive MMP-2 Gene Expression-High levels of MMP-2 expression correlate with malignant progression of astrogliomas in vivo (11,12,14,16,18). We have recently determined that a variety of human astroglioma cell lines constitutively express high levels of MMP-2, which correlates with the in vitro invasive capacity of these cells (15). As illustrated in Fig. 1, the MMP-2 gene is constitutively expressed in U251-MG astroglioma cells. The 72-kDa MMP-2 protein can be detected by zymography (lane 1) and immunoblotting (lane 2). MMP-2 mRNA is also constitutively expressed (Fig. 1, lane 3). To assess the molecular basis of constitutive MMP-2 gene expression in astroglioma cells, we utilized a MMP-2 construct in which 1659 bp of the 5Ј region was inserted into the pGL2-Basic reporter construct to define promoter elements involved in this response. Fig. 2 depicts potential regulatory elements that have been identified through the Mat Inspector program (38). Deletion constructs were generated from the 5Ј end of the MMP-2 promoter, as described under "Experimental Procedures" (see Fig. 3, A and B). Each of the constructs were transiently transfected into U251-MG cells with a pCMV-␤galactosidase construct to monitor transfection efficiency, the cells were cultured for 24 h in 1% Dulbecco's modified Eagle's/ F-12 medium, and then RLA was determined. The RLA from the vector control was set at 1 to calculate fold activation, then the fold activation value for the WT construct was plotted as 100%. As illustrated in Fig. 3A, the WT MMP-2 promoter construct is constitutively active in U251-MG cells. Deletion of the p53 element (D1, Ϫ1629 bp) had no effect on constitutive MMP-2 promoter activity in astroglioma cells, which is different from that observed in sarcoma cell lines (28,30). MMP-2 promoter activity increased ϳ1.6-fold after the first silencer was deleted (D2, Ϫ1611 bp), and further increased ϳ2.2-fold after deletion of the second silencer (D3, Ϫ1591 bp), indicating that the region containing the two silencers negatively affects MMP-2 promoter activity. Deletion of the region that contains a potential AP-1 site (D4, Ϫ1259 bp) did not significantly affect promoter activity compared with the D3 construct. In construct D5 (Ϫ1050 bp), deletion of the potential Ets-1 and c-Myc/c-Myb elements did not affect MMP-2 promoter activity compared with the D4 deletion construct. The first distal PEA3 element and the C/EBP element were deleted in construct D6 (Ϫ562 bp); luciferase activity of this deletion construct was not significantly different compared with the D5 deletion construct. The luciferase activity of construct D7 in which the second distal PEA3 element was deleted is comparable to that of the D6 construct. The luciferase activity of construct D8 (Ϫ161 bp) which deletes the third PEA3 element, and the CREB and GCN-His elements, was not significantly different than that of D7 construct. Deletion of the fourth PEA3 element (D9, Ϫ139 bp), also had no effect on MMP-2 promoter activity compared with the D8 construct. It should be noted that the promoter activity of constructs D4-D8 are significantly higher than that of the WT construct. The identical pattern of MMP-2 promoter activity was observed in two other astroglioma lines, U373-MG and CH235-MG (data not shown), which also constitutively express MMP-2 (15). These data collectively indicate that the two distal silencer elements negatively affect constitutive MMP-2 promoter activity, and other potential elements (p53, AP-1, Ets-1, c-Myc/c-Myb, C/EBP, CREB, GCN-His, and the four PEA3 elements) do not contribute to regulation of constitutive human MMP-2 promoter activity in astroglioma cells.
Further analysis of the MMP-2 promoter revealed that dele-tion of the first proximal Sp1 element (D10, Ϫ85 bp) lead to a significant decrease in promoter activity compared with WT, and deletion of the second proximal Sp1 element (D11, Ϫ64 bp) also reduced promoter activity (Fig. 3B). In the last construct tested (D12, in which the proximal AP-2 element was deleted), promoter activity was almost completely abolished (ϳ14% of the WT MMP-2 promoter) (Fig. 3B). These data indicate that the proximal Ϫ139 bp region positively controls the constitutive activity of the MMP-2 promoter in U251-MG cells. Comparable results were obtained using U373-MG and CH235-MG astroglioma cells (data not shown). These results demonstrate that the constitutive activation of the MMP-2 gene in astroglioma cells is dependent on elements contained within a 139-bp region of the human MMP-2 promoter; this region contains two Sp1 elements and one AP-2 element.

Contribution of the Proximal Sp1 and AP-2 Elements to the Constitutive Activity of the Human MMP-2
Promoter-To further define which specific element(s) in the proximal 139-bp region of the MMP-2 promoter contribute to constitutive MMP-2 expression in astroglioma cells, we generated a series of MMP-2 promoter constructs with mutations in the two Sp1 elements and the AP-2 element (see Fig. 4). Because the D9 deletion construct (Ϫ139 bp) was the shortest construct that had high constitutive luciferase activity compared with the WT promoter, we first used the D9 deletion construct as a template for mutagenesis. As illustrated in Fig. 4, mutation of the first Sp1 element (mSp1A) resulted in a ϳ70% reduction in constitutive MMP-2 promoter activity in U251-MG cells; mutation of the second Sp1 element (mSp1B) had no effect on MMP-2 promoter activity; and mutation of the AP-2 element (mAP-2) reduced promoter activity by ϳ50%. These results demonstrate that the first Sp1 element (Sp1A) and the AP-2 element are critical for constitutive MMP-2 gene expression. To further assess the importance of these two elements, a construct was generated in which both the Sp1A and AP-2 elements were mutated. The mSp1A ϩ mAP-2 construct was minimally active in U251-MG astroglioma cells (83% inhibition compared with the D9 construct) (Fig. 4). As a further confirmation of the results obtained using the D9 template, we used the D3 deletion construct (Ϫ1591 bp) as a template to study the function of the proximal Sp1 and AP-2 elements because D3 has the highest constitutive luciferase activity. The D3/mSp1A construct resulted in a ϳ50% reduction of constitutive MMP-2 promoter activity compared with the D3 construct; the activity of the D3/mSp1B construct was not affected, and mutation of the AP-2 element (D3/mAP-2) resulted in a ϳ44% reduction in promoter activity (Fig. 4). As well, the combined mutation of the Sp1A and AP-2 elements (mSp1A ϩ mAP-2) resulted in a ϳ75% reduction compared with the D3 construct (Fig. 4). Thus, comparable results were obtained using two different templates (D9 and D3) for mutagenesis. Additional site-specific constructs were generated in both the D3 and D9 templates in which only the 6-bp Sp1A element (GGGCGG) was mutated; similar results as those for D9/mSp1A and D3/mSp1A were obtained (data not shown). Identical results were obtained using two other astroglioma lines, U373-MG and CH235-MG

FIG. 3. Effect of deletion mutations on human MMP-2 promoter activity.
Putative cis-acting elements identified within the MMP-2 promoter are indicated. U251-MG cells were co-transfected with 10 g of the WT or truncated MMP-2 promoter constructs and 1 g of the pCMV-␤-galactosidase construct as indicated under "Experimental Procedures." Cells were allowed to recover for 18 h, then were treated with serum-free medium for 24 h. Luciferase (LUC) and ␤-galactosidase activity was determined in triplicate, and the luciferase activity of each sample was normalized to ␤-galactosidase activity to calculate RLA. Fold activity was calculated by dividing the RLA of the MMP-2 promoter samples by the RLA of the pGL2-Basic vector sample. The actual fold activity of the WT construct is 63.4 Ϯ 3.3. The activity of the WT construct is represented as 100%, and the activities of the deletion constructs compared with WT. In A, the promoter activity of the WT and first nine deletion constructs are shown. In B, the promoter activity of the WT and last three deletion constructs are shown. Data are the mean Ϯ S.D. from at least four experiments. Statistical analysis was performed comparing the activity of the deletion constructs to that of WT. *, p Ͻ 0.05, **, p Ͻ 0.01; ***, p Ͻ 0.001.

FIG. 2. Potential regulatory elements in the human MMP-2 pro-
moter. S1 is silencer 1, and S2 is silencer 2. (Table I). These results demonstrate that both the proximal Sp1A and AP-2 elements play functionally important roles in constitutive MMP-2 promoter activity.
The Transcription Factors Sp1 and Sp3 Bind to the MMP-2 Proximal Sp1A Element-To determine the DNA-protein complexes forming over the proximal functional Sp1A element of the MMP-2 promoter, nuclear extracts from unstimulated astroglioma cells were analyzed by EMSA using labeled oligonucleotides spanning the Sp1A element. As illustrated in Fig. 5A, three DNA-protein complexes were detected using the labeled Sp1A oligonucleotide: complex 1, complex 2, and complex 3 (lane 1). To determine the specificity of the three DNA-protein complexes, an excess of unlabeled Sp1A oligonucleotide was used as a competitor, and all three complexes competed away (Fig. 5A, lanes 2 and 3). As well, inclusion of unlabeled proximal mSp1A oligonucleotide as a competitor did not affect formation of the three complexes (Fig. 5A, lanes 4 and 5). Using the labeled mSp1A oligonucleotide as probe, no DNA-protein complexes were observed (Fig. 5A, lane 6). HeLa cells were used as a control (Fig. 5A, lane 7); the same three DNA-protein complexes were formed. Recently, a number of GC and/or GT box-binding proteins homologous to Sp1 have been identified; Sp3 and Sp4 are the most homologous to Sp1, contain similar functional domains, and recognize GC/GT boxes with similar specificity and affinity as Sp1 (for review, see Ref. 39). To further investigate the identities of complexes 1-3, we performed supershift experiments using antibodies against Sp family members (Sp1 and Sp3). As shown in Fig. 5B, complex 1 was supershifted in the presence of Sp1 antisera, and complex 2 was partially supershifted (lane 2). Complex 3 was supershifted in the presence of Sp3 antisera, and complex 2 was partially supershifted (lane 3). All three complexes were supershifted in the presence of both Sp1 and Sp3 antisera (Fig. 5B,  lane 4). These results indicate that complex 1 contains Sp1, complex 2 is composed of both Sp1 and Sp3, and complex 3 contains Sp3. The inclusion of AP-2 antisera or normal rabbit serum (NRS) did not affect binding of the three complexes (Fig.  5B, lanes 5 and 6). Using nuclear extracts prepared from unstimulated HeLa cells as a control, we observed the same pattern of reactivity with Sp1 and Sp3 antisera (Fig. 5B, lanes  7-10). These results demonstrate that the transcription factors Sp1 and Sp3 bind to the functional MMP-2 Sp1A element. It should be noted that we did not detect Sp1 or Sp3 binding to the nonfunctional Sp1B element (data not shown).

Sp1 and Sp3 Function as Activators of the MMP-2 Promoter-
To determine whether Sp1, Sp3, or both could modulate MMP-2 promoter activity, SL2 cells, which are deficient in Sp-related proteins (40) were utilized. Initially, SL2 cells were co-transfected with the D3 MMP-2 construct, and increasing concentrations of pPacSp1 or pPacUSp3 expression vectors. As shown in Fig. 6A, both pPacSp1 and pPacUSp3 stimulated D3 MMP-2 promoter activity in a dose-dependent manner, with pPacSp1 being the most potent activator. Optimal responses by Sp1 and Sp3 were observed using 1 and 5 g, respectively. Interestingly, co-transfection of pPacSp1 and pPacUSp3 (0.5 g) lead to a synergistic activation of the MMP-2 promoter (Fig.  6B). To evaluate the specificity of Sp1/Sp3 activity, the MMP-2 Sp1A mutant, D3/mSp1A, was tested in SL2 cells. Sp1-mediated activation of the D3/mSp1A construct was reduced ϳ85% compared with the WT D3 construct, Sp3-mediated activation inhibited by ϳ93%, and the synergistic response by both Sp1 and Sp3 reduced by ϳ83% (Fig. 6B). These results indicate that the proximal Sp1A element is responsible for mediating the transactivating effect of both Sp1 and Sp3 on MMP-2 promoter activity. FIG. 4. Effect of site-specific mutations on human MMP-2 promoter activity. U251-MG cells were co-transfected with 10 g of the WT templates (either the D9 or D3 construct) or mutated MMP-2 promoter constructs and 1 g of the pCMV-␤-galactosidase construct as indicated under "Experimental Procedures." Cells were allowed to recover for 18 h, then were treated with serum-free medium for 24 h. Luciferase (LUC) and ␤-galactosidase activity was determined in triplicate, and the luciferase activity of each sample was normalized to ␤-galactosidase activity to calculate RLA. The activity of the appropriate WT construct is represented as 100%, and the activities of the mutated constructs compared with WT. D9 construct fold activity is 84.5 Ϯ 19.2, and D3 construct fold activity is 144.0 Ϯ 11.3. Data are the mean Ϯ S.D. from at least four experiments. a U373-MG cells or CH235-MG cells were co-transfected with 10 g of the WT template (either the D9 or D3 construct) or mutated MMP-2 promoter constructs and 1 g of the pCMV-␤-galactosidase construct. Cells were allowed to recover for 18 h, then were treated with serumfree medium for 24 h. Cells were harvested for the determination of luciferase and ␤-galactosidase activity.
b The activity of the appropriate WT construct is represented as 100%, and the activity of the mutated constructs compared to WT. Data are the mean Ϯ S.D. from three experiments.

The Transcription Factor AP-2 Binds to the MMP-2 AP-2
Element-To determine the DNA-protein complexes forming over the functional AP-2 element of the MMP-2 promoter, nuclear extracts from unstimulated astroglioma cells were analyzed by EMSA using labeled oligonucleotides spanning the proximal AP-2 element. Two DNA-protein complexes were detected: complex I and complex II (Fig. 7, lane 1). We next investigated the nature of the protein(s) binding to the AP-2 element. The two DNA-protein complexes were competed away using an excess of unlabeled proximal AP-2 oligonucleotide (Fig. 7, lanes 2 and 3), but were not affected by unlabeled proximal mAP-2 oligonucleotide (lanes 4 and 5). As well, no DNA-protein complexes were formed with labeled mAP-2 oligonucleotide as the probe (Fig. 7, lane 6). The identities of complex I and II were analyzed by supershift experiments using antibodies against AP-2, Sp1, and Sp3. Inclusion of AP-2 antisera caused a supershift in complex I, but did not affect complex II (Fig. 7, lane 7). Complex I or II binding was not influenced by Sp1 or Sp3 antisera or NRS (Fig. 7, lanes 8 -10).
These results indicate that complex I contains the AP-2 transcription factor, while the composition of complex II is unknown thus far.
The AP-2 Site Is Functional in the MMP-2 Promoter-We utilized AP-2-deficient HepG2 cells (41) to investigate the functional relevance of AP-2 for MMP-2 expression. HepG2 cells were co-transfected with the D3 MMP-2 construct and either the pSX-AP-2 expression vector or pSX as a control vector, then MMP-2 promoter activity was assessed. The MMP-2 promoter is constitutively active in HepG2 cells, and expression of AP-2 enhances MMP-2 promoter activity (3.9-fold enhancement) (Fig. 8). The specificity of this response was determined using the mutant MMP-2 construct D3/mAP-2 that contains a mutant AP-2 site. This construct is weakly activated in HepG2 cells, and the expression of AP-2 has no effect on promoter activity (Fig. 8), demonstrating the specificity of the transactivating effect of AP-2. DISCUSSION MMP-2-deficient mice exhibit reduced tumor progression (42), illustrating the importance of this molecule. Importantly, a strong correlation has been observed between astroglioma invasion and high levels of MMP-2 expression (11,12,14,15,18). Astroglioma cells constitutively express high levels of MMP-2 mRNA, protein, and bioactivity as assessed by ribonuclease protection assay, immunoblotting, and zymography assay, respectively (15). In addition, the 62-kDa active form of MMP-2 can be detected in the plasma membrane of astroglioma cells (15). We have also determined that the MMP-2 promoter is constitutively active in these cells (15). Given the prominent role of MMP-2 in tumor invasion, it is critical to understand the molecular basis of MMP-2 gene expression. In the present study, we identified the cis-acting elements and transcription factors involved in constitutive MMP-2 expression in astroglioma cells.
Functional analysis of the human MMP-2 promoter was initially performed utilizing 5Ј-deletion constructs in three human astroglioma cell lines, U251-MG, U373-MG, and CH235-MG, which all express high levels of MMP-2 constitutively. The distal p53 element does not contribute to MMP-2 promoter activity in astroglioma cells since deletion of that element (construct D1, Fig. 3A) had no effect compared with the WT promoter. This results differs from what has been observed in human sarcoma cells; deletion of the p53 element decreased MMP-2 promoter activity by 2-4-fold (30). It was also demonstrated that wild-type p53 protein, but not p53 mutant proteins commonly found in human cancers, could activate MMP-2 promoter activity via the p53 element (30). The p53 element does not contribute to MMP-2 activation in astroglioma cells; this may be because the majority of astrogliomas express mutant p53 protein (43). Deletion of the two silencer elements (S1, S2) immediately adjacent to the p53 site resulted in a 2.2-fold enhancement of MMP-2 promoter activity (constructs D2 and D3, Fig. 3A), indicating that these two elements negatively regulate constitutive MMP-2 expression. This two-element transcriptional silencer has been previously described to function efficiently in MCF-7 mammary carcinoma cells, but not in HT1080 cells, suggesting that cell-type specific silencing may contribute to regulation of MMP-2 gene expression (28). Our findings indicate that the silencer elements are indeed functional in human astroglioma cells, although the mechanism of transcriptional repression has not been elucidated. Further deletion of the AP-1, Ets-1, c-Myc/c-Myb, C/EBP, CREB, PEA3, and GCN-His elements did not significantly affect promoter activity as compared with the D3 construct, which exhibits maximal MMP-2 promoter activity (Fig. 3A). These results indicate that those elements are not functionally important for constitutive MMP-2 promoter activity in astroglioma cells.
Further deletion of the MMP-2 promoter revealed that the proximal Ϫ139 bp region is essential for constitutive MMP-2 promoter activity in astroglioma cells (Fig. 3B). Within this region are two Sp1 elements (designated as Sp1A and Sp1B) and one AP-2 element. To define the contribution of these elements to MMP-2 promoter activity, site-directed mutagenesis was performed. Mutation of the Sp1A element resulted in a ϳ41-70% reduction in MMP-2 transcription, compared with the WT construct ( Fig. 4 and Table I). Mutation of the Sp1B element had no effect on promoter activity, while mutation of the AP-2 element resulted in a ϳ36 -60% loss of MMP-2 transcription compared with WT ( Fig. 4 and Table I). Indeed, double mutation of both the Sp1A and AP-2 elements caused a ϳ74 -94% inhibition in MMP-2 promoter activity, indicating the functional importance of these two elements ( Fig. 4 and Table I).
Sp1 is a well characterized sequence-specific DNA-binding protein that is important in the transcription of many cellular and viral genes that contain GC boxes in their promoter (44,45). Additional transcription factors (Sp2, Sp3, and Sp4) have been cloned that are similar in structural and transcriptional properties to Sp1, thus forming an Sp1 multigene family (for review, see Ref. 39). We analyzed the factor(s) binding to the functional Sp1A element by EMSA and supershift analysis. Three constitutively expressed DNA-protein complexes were detected using labeled Sp1A oligonucleotides (Fig. 5). Analysis with antibodies against Sp1 and Sp3 revealed that complex 1 contained Sp1, complex 2 was composed of both Sp1 and Sp3, and complex 3 contains Sp3 exclusively (Fig. 5). While Sp1 stimulates transcriptional activity, the function of Sp3 is less clear. Sp3 has been shown to act as a dual-function regulator whose activity is dependent upon context of DNA-binding sites in a promoter. Sp3 functions as a repressor when it is bound to a promoter through multiple DNA-binding sites, and functions as an activator when targeted to a promoter through a single DNA-binding site (35,36,46,47). Sp3 has the ability to repress Sp1-mediated transcriptional activation of the human alcohol dehydrogenase 5 gene, possibly by competing with Sp1 for binding (47). Regarding the MMP-2 gene, we have determined that Sp1 potently activates the MMP-2 promoter in SL2 cells, while Sp3 also functions as a transcriptional activator. Of most HepG2 cells were co-transfected with 10 g of the MMP-2 D3 deletion construct or the MMP-2 D3/mAP-2 mutant construct, 1 g of the pCMV-␤-galactosidase construct, and 1.0 g of the expression plasmid for AP-2 (pSX-AP-2) or control vector pSX. Cells were allowed to recover for 18 h, then were incubated with serum-free medium for 24 h. Luciferase and ␤-galactosidase activity were determined in triplicate, and the luciferase activity of each sample was normalized to ␤-galactosidase activity to calculate RLA. Data are the mean Ϯ S.D. from at least four experiments.
interest is the observation that Sp1 and Sp3 function in a synergistic manner for MMP-2 activity (Fig. 6). Sp3 contains two glutamine-rich regions that have strong activation functions, however, these activation modules are silenced by the presence of an inhibitory domain (36). The molecular basis underlying the activation function of Sp3 for the MMP-2 promoter, as well as the synergistic activity of Sp1 and Sp3 is unknown at this time, but may involve alleviation of the inhibition mediated by the inhibitory domain of Sp3.
The transcription factor AP-2 is a 52-kDa protein that binds as a dimer to a palindromic sequence 5Ј-GCCNNNGGC-3Ј (41), although some AP-2-binding sites deviate from this consensus sequence. Mutagenesis of the AP-2 element at Ϫ91 to Ϫ84 bp revealed that this element is functionally important for constitutive MMP-2 promoter activity (Fig. 4). By EMSA, we demonstrated that two DNA-protein complexes were competed away by unlabeled AP-2 oligonucleotide (Fig. 7). However, only the slower migrating species (complex I) was supershifted with anti-AP-2 antiserum (Fig. 7). At present, we do not know the composition of complex II. The functional significance of AP-2 was demonstrated using HepG2 cells, which are deficient in AP-2 (41). The MMP-2 promoter is constitutively active in HepG2 cells, likely due to Sp1-related factors, and introduction of an AP-2 expression vector enhances MMP-2 promoter activity in these cells (Fig. 8). The enhancing effect of AP-2 was eliminated using a MMP-2 promoter construct with a mutation in the AP-2 element. AP-2 has been implicated in the regulation of a number of genes, including transforming growth factor-␣ (34), apolipoprotein E (48), vascular permeability factor/ vascular endothelial growth factor (49), and metallothionein IIa (50). We can now include MMP-2 as a gene regulated by AP-2.
In summary, we have demonstrated, for the first time, the importance of elements within the proximal region of the human MMP-2 promoter for constitutive MMP-2 expression. This analysis also provides novel information on how the MMP-2 gene is regulated in human astroglioma cells. Given the importance of the Sp1 and AP-2 elements for constitutive MMP-2 expression in astroglioma cells, it will be of interest to determine if the transcription factors that bind those elements (Sp1, Sp3, and AP-2) are aberrantly expressed in invasive, high grade gliomas, compared with non-invasive low grade astrogliomas or normal astrocytes. A further understanding of the molecular basis of MMP-2 regulation will have functional implications for suppressing glioma invasion and ultimately lead to design of new therapeutics.