Regulation of DNA Methylation in Human Breast Cancer

Urokinase-type plasminogen activator (uPA) is a member of the serine protease family and can break down various components of the extracellular matrix to promote growth, invasion, and metastasis of several malignancies including breast cancer. In the current study we examined the role that the DNA methylation machinery might be playing in regulating differential uPA gene expression in breast cancer cell lines. uPA mRNA is expressed in the highly invasive, hormone-insensitive human breast cancer cell line MDA-MB-231 but not in hormone-responsive cell line MCF-7. Using methylation-sensitive PCR, we show that 90% of CpG dinucleotides in the uPA promoter are methylated in MCF-7 cells, whereas fully demethylated CpGs were detected in MDA-MB-231 cells. uPA promoter activity, which is directly regulated by the Ets-1 transcription factor, is inhibited by methylation as determined by uPA promoter-luciferase reporter assays. We then tested whether the state of expression and methylation of the uPA promoter correlates with the global level of DNA methyltransferase and demethylase activities in these cell lines. We show that maintenance DNA methyltransferase activity is significantly higher in MCF-7 cells than in MDA-MB-231 cells, whereas demethylase activity is higher in MDA-MB-231 cells. We suggest that the combination of increased DNA methyltransferase activity with reduced demethylase activity contributes to the methylation and silencing of uPA expression in MCF-7 cells. The converse is true in MDA-MB-231 cells, which represents a late stage highly invasive breast cancer. The histone deacetylase inhibitor, Trichostatin A, induces the expression of the uPA gene in MDA-MB-231 cells but not in MCF-7 cells. This supports the hypothesis that DNA methylation is the dominant mechanism involved in the silencing of uPA gene expression. Taken together, these results provide insight into the mechanism regulating the transcription of the uPA gene in the complex multistep process of breast cancer progression.

The breakdown of the extracellular matrix involves a variety of growth factors and proteases and is an important step in the process of tumor invasion and metastasis (1,2). Urokinase-type plasminogen activator (uPA) 1 and its cell surface glycophosphotidyl inositol-linked receptor (uPAR) play important roles in several malignancies (1). uPA produced by tumor cells and the surrounding stroma is intimately involved in tumor cell invasion, migration, and proliferation (1). The uPAR localizes the proteolytic effects of uPA within the tumor cell environment. Additionally, uPA enhances neovascularization of tumors thus further contributing to the process of tumor progression (1). Numerous data demonstrate the causal role of uPA in tumor growth and metastasis (1). First, expression of uPA or uPAR was shown to enhance tumor growth and metastasis (3,4). Second, increased uPA gene expression in various malignancies including breast cancer is closely related with disease stage (1,5). Third, inoculation of human breast cancer cells in mice lacking the uPA gene results in tumors of significantly smaller volume than tumors implanted in wild type mice (3). Fourth, we have demonstrated previously that a peptide derived from the non-receptor binding domain of uPA (Å6) decreases breast cancer invasion, growth, and metastases because of its pro-apoptotic and anti-angiogenic effects (6). Fifth, the active site inhibitor of uPA (B-428), alone or in combination with the anti-estrogen tamoxifen, blocks the growth and metastasis of prostate and breast cancers (7,8). Sixth, antibodies directed against uPA or uPAR are able to decrease tumor growth (9 -11).
DNA methylation marks inactive genes and can suppress gene expression directly by interfering with the binding of transcription factors or indirectly by attracting methylated DNA binding factors that recruit histone deacetylases and precipitate an inactive chromatin structure (12,13). Aberrant DNA methylation patterns are commonly observed in cancer (13). Neoplastic cells have the ability to simultaneously harbor widespread hypomethylation and regional hypermethylation that contribute to tumor progression (13,14). Whereas hypermethylation and silencing of tumor suppressor genes has attracted much attention recently (13,15,16), the molecular mechanisms underlying hypomethylation of tumor progression factors such as ras, myc, hox11, and xmrk that are up-regulated during cancer development are poorly described. This suggests that hypomethylation may also play an important role in regulating gene expression during tumorigenesis similar to hypermethylation (17)(18)(19)(20).
In the current study, we tested the hypothesis that the expression of uPA, a well defined marker of highly invasive tumor cells and activated at the late stages of breast cancer, can be regulated via the changes in the methylation status of its promoter region. As a follow-up to our previous studies where we demonstrated that the uPA gene is transcriptionally suppressed by DNA methylation (21), we have now focused on the examination of the role of DNA demethylation and DNA methylating enzymes such as DNA methyltransferase (DNMT) and demethylase (DMase) in regulating uPA expression during breast cancer progression.

MATERIALS AND METHODS
Cells and Cell Culture-All human breast cancer cell lines were obtained from the American Type Culture Collection (ATCC; Manassas, VA). MDA-MB-231 cells were maintained in L-15 medium (Invitrogen) with 2 mM L-glutamine, 10% fetal bovine serum (FBS), and 100 units/ml of penicillin-streptomycin sulfate (Invitrogen). MCF-7 cells was maintained in minimum Eagle's medium (Earle's salts; Invitrogen) with 2 mM L-glutamine, 10% FBS, 10 g/ml insulin, and 100 units/ml of penicillin-streptomycin sulfate. T47D and BT474 cells were maintained in RPMI 1640 (Invitrogen) with 2 mM L-glutamine, 10% FBS, and 100 units/ml of penicillin-streptomycin sulfate. T47D cells were supplemented with 10 g/ml insulin. HS578T and BT549 cells were maintained in Dulbecco's modified Eagle's medium (Invitrogen) with 2 mM L-glutamine, 5% FBS, and 100 units/ml of penicillin-streptomycin sulfate. All cells were incubated at 37°C in 5% CO 2 with the exception of MDA-MB-231 cells that are grown in the absence of CO 2 .
Northern Blot Analysis-MCF-7 cells were treated with or without 5-aza-2Ј-deoxycytidine (5-azaCdR) (Sigma) at a concentration of 20 M for 3 days and were cultured in the absence of the drug for an additional 24 h. MDA-MB-231 and MCF-7 cells were treated with Trichostatin A (TSA) (Sigma) at a concentration of 100 ng/ml for 2 days. RNA from these cells was isolated using the Trizol method following the manufacturer's instructions (Invitrogen), fractionated on a 1.5% agarose gel in MOPS buffer, and transferred onto nylon filters (Amersham Biosciences). The filters were hybridized with a 32 P-labeled human uPA cDNA at 65°C for 24 h, which was then stripped of the probe and re-hybridized with an 18 S RNA probe to normalize for the amount of RNA loaded in each lane. Autoradiography of the filters was carried out at Ϫ80°C using XAR film (Eastman Kodak Co., Rochester, NY). The intensity of hybridization to uPA mRNA and 18 S RNA was quantified by densitometric scanning. The normalized uPA mRNA expression was determined by dividing the intensity of the uPA mRNA signal by the intensity of the 18 S signal per each lane.
Boyden Chamber Matrigel Invasion Assay-Cell invasive capacity was determined using two compartment Boyden chamber Matrigel invasion assay as described previously (Transwell, Costar, Corning Corporation, Burlington, MA) (6, 7). The 8-m pore polycarbonate filters were coated with basement membrane Matrigel (50 g/filter). 5 ϫ 10 4 cells treated with or without 5-azaCdR in 0.1 ml of medium were added to the upper chamber and placed on top of a lower chamber pre-filled with 0.8 ml of serum-free medium supplemented with 25 g/ml fibronectin (Sigma) and then incubated at 37°C for 24 h. After the incubation, medium was removed, and the polycarbonate filters with invaded cells were fixed in 2% paraformaldehyde, 0.5% glutaraldehyde (Sigma) in 0.1 M phosphate buffer, pH 7.4, at room temperature for 30 min. They were then stained with 1.5% toluidine blue, washed, and mounted onto glass slides. Number of cells invaded was examined under a light microscope. Ten fields under 400ϫ magnification were randomly selected, and the mean cell number was calculated.
Southern Blot Analysis-Cellular genomic DNA isolated with DNA-ZOL (Invitrogen) was used for analyzing the methylation status of the CpG island of the uPA promoter by methylation-sensitive restriction enzyme digestion and Southern blotting. Briefly, 10 g of genomic DNA were extracted from MDA-MB-231, MCF-7, and MCF-7 treated with 20 M 5-azaCdR as recommended by the manufacturer. Extracted DNA was first digested with EcoRI or HindIII and was digested further with PstI, PstI/HpaII, or PstI/HhaI (8 units/g of total DNA), respectively, for 18 h at 37°C. The digests were fractionated on a 0.8% agarose gel and transferred to a nylon membrane (Amersham Biosciences) by capillary blotting. The filters were hybridized with a 32 P-labeled 778-bplong SmaI-AvrII fragment of the uPA promoter derived from the uPAchloramphenicol acetyltransferase (CAT) construct (gift from Dr. F. Blasi, Milan, Italy) at 42°C for 24 h. They were then successively washed in 1ϫ SSC (150 mM NaCl, 50 mM sodium citrate, pH 7.0), 1% SDS once for 15 min at room temperature, 0.5ϫ SSC, 0.58% SDS once for 15 min, 0.1ϫ SSC, 0.1% SDS twice for 15 min at room temperature, and at last once for 30 min at 55°C. Autoradiography of the filters was carried out at Ϫ80°C using XAR film (Kodak).
Reverse Transcriptase-PCR-Total RNA was extracted from by a single-step method using Trizol reagent. The reaction template, denatured for 10 min at 70°C, consisted of 2 g of total RNA in 10 l of diethylpyrocarbonate H 2 O. To the reaction mixture, 1ϫ PCR buffer (MBI Fermentas, Burlington, ON, Canada), dNTP mix 0.2 mM each (MBI Fermentas), 10 units of RNasin inhibitor (Amersham Biosciences) and of M-MuLV Rtase (Invitrogen) were added. The RNA was reversedtranscribed for 1 h at 42°C. Using 2.5 units of Taq DNA polymerase (Invitrogen), uPA was amplified by PCR for 25 cycles with the following primers: uPA-forward, TGCGTCCTGGTCGTGAGCGA; UPA-reverse, CAAGCGTGTCAGCGCTGTAG. Glyceraldehyde-3-phosphate dehydrogenase primers were used as a control. The amplified PCR products was fractionated on a 1.1% agarose gel and visualized by ethidium bromide staining.
Methylation-specific PCR-Genomic DNA (20 g) was digested with restriction enzyme DraI and denatured in 0.3 M NaOH for 20 min at 37°C in a volume of 100 l. The purified genomic DNA (2 g) was treated with 2.2 M sodium bisulfite and 1 mM hydroquinone, pH 5.0, at 55°C for 14 h, purified using a Qiagen PCR purification column, denatured in 0.3 M NaOH for 20 min at 37°C, and neutralized with 3 M ammonium acetate, pH 7.0. Following ethanol purification, an aliquot of treated DNA was amplified with the following PCR primers modified for nested methylation-specific PCR (MSP) that correspond to the uPA gene: 5Ј outer primer Ϫ532 to Ϫ512, TATAGAGGGAGTTTTTATAGG; 3Ј outer primer ϩ287 to ϩ306, ATAACCAAACTCCCCAACTA; 5Ј inner primer Ϫ421 to Ϫ402, TTTATAGTTTTATTTAGTTG; and 3Ј inner primer ϩ35 to ϩ54, ACAAAAACAAATAAACCCTA (22,23). PCR was carried out for 35 cycles at the amplification conditions that consist of 1-min denaturation at 95°C, 45-s annealing at 56°C (outer primers) or 58°C (inner primers) followed by 1-min extension at 72°C. A 10-min extension step at 72°C was added at the end of the PCR cycles. All reactions were carried out in 100 l of total reaction buffer containing 1ϫ PCR buffer, 1.5 mM MgCl 2 , 0.25 mM dNTP, 20 pmol of primers, and 2.5 units of Taq DNA polymerase (MBI, Flamborough, ON, Canada) with GeneAmp PCR system 9600 (PerkinElmer Life Sciences). The PCR products were run on a 1.2% agarose gel and were extracted and purified from the gel. They were then subcloned into Topo-PCR TA cloning vectors according to the manufacturer's instruction (Clonetics, San Diego, CA) for sequencing (Bio S&T, Montreal, Quebec, Canada).
Luciferase Reporter Assay-The uPA promoter region (Ϫ745 to ϩ30) was cut from uPA-chloramphenicol acetyltransferase reporter vector (gift from Dr. F Blasi, Milan, Italy) at AvrII and SmaI sites and then inserted into a luciferase reporter vector pGL-3 basic (Promega, Madison, WI) digested with NheI and SmaI to generate uPA-luc plasmid. Unmethylated uPA promoter construct was obtained by treating uPAluc plasmid with the methyl donor S-adenosylmethionine (AdoMet) in the absence of bacterial CpG methylases. The uPA-luc plasmid was methylated at different CpG sites in vitro using methylases (mSssI, mHpaII, and mHhaI) and the methyl donor as recommended by the manufacturer (New England Biolabs, Mississauga, ON, Canada). Complete methylation was confirmed by resistance to HpaII and HhaI restriction enzymes. PGL-3 basic was used as a negative control. The different treated plasmids were transiently transfected into MDA-MB-231 cells using LipofectAMINE as a carrier according to the manufacturer's protocol (Invitrogen). pEVRF0 or pEVRF-Ets-1 were gifts from B. J. Graves at University of Utah School of Medicine, Salt Lake City, UT. PSV-␤-gal (Promega) containing the ␤-galactosidase gene under the control of the constitutively active SV40 promoter and enhancer was co-transfected at a concentration of 0.5 g/sample to normalize for transfection efficiency. 48 h after transfection, cells were scraped in 1ϫ reporter lysis buffer (Promega) followed by centrifugation at 14,000 rpm. Luciferase activity in the supernatants was then analyzed by . The reaction was stopped by adding 0.5 ml of 1 M Na 2 CO 2 , and the absorbance was measured at 420 nm in a V max plate reader (Molecular Devices). Activity was determined by comparison to a standard curve. Luciferase reporter activity in relative luminescence units was normalized to ␤-galactosidase activity as described (24).
Electrophoretic Mobility Shift Assay-PEVRF-Ets-1 was transiently transfected into MDA-MB-231 cells using LipofectAMINE according to the manufacturer's protocol (Invitrogen). Nuclear extracts were isolated as described and used for electrophoretic mobility shift assay. Doublestranded synthetic oligonucleotides (ODN) corresponding to methylated or unmethylated Ets binding sites were generated using the following oligonucleotides: and unmethylated antisense, ACACTCGCAACGCCTTCGTGCGCCCCAGGC (Sheldon Biotechnology, Montreal, QC). Subsequently, the ODN were end-labeled with [␥-32 P]dATP using T4 polynucleotide kinase (New England Biolabs, Mississauga, ON). For each assay, 25,000 counts per minute of labeled ODN were incubated with 5 g of nuclear extract in a final volume of 20 l containing 1 g of poly(dI-dC), 10 mM Tris-HCl, pH 7.5, 50 mM NaCl, 0.5 mM EDTA, 1 Mm MgCl 2 , 4% glycerol, and 1 mM dithiothreitol and incubated on ice for 20 min. Competition experiments was done by co-incubation with 10 -50-fold excess of unlabeled ODN in the DNAprotein binding reaction. Complexes were resolved on 6% polyacrylamide non denaturing gels in 0.25ϫ TBE buffer at 7.5 V/cm. Gels were vacuum-dried and autoradiographed. These results are representative of three independent experiments (25).
DNA Methyltransferase Activity Assay-5 g of the nuclear extract obtained were incubated with the methyl donor [ 3 H]-AdoMet (S-adenosyl-L-[methyl-3 H]-methionine; Amersham Biosciences) and hemimethylated double stranded oligos (poly(mdC-dG⅐dC-dG); Sheldon Biotechnology Center) to measure maintenance methyltransferase activity or unmethylated double stranded oligos (poly(dI-dC⅐dI-dC); Amersham Biosciences) to measure de novo methyltransferase activity. After incubating for 3 h at 37°C, the level of incorporation of 3 H-labeled methyl groups into DNA substrates was determined by 10% trichloroacetic acid precipitation and filtration through GF/C filters (Whatman Ltd., Maidston, England), followed by liquid scintillation counting. Reaction mixtures containing only substrate oligos but no nuclear extract were used as negative controls. Total radioactivity retained by GF/C filters from reaction mixtures of respective cell lines in the absence of substrate oligos but with nuclear extract was subtracted to exclude DNA-independent methylation activities in the extracts. Results are expressed as the mean disintegration/min of [ 3 H]-CH 3 incorporated into substrate oligonucleotides per g of nuclear protein Ϯ S.E. of triplicate cell extracts. Each nuclear extract was assayed in triplicate (26).
DNA Demethylase Activity Assay-Total nuclear extracts (6 g) obtained from MDA-MB-231 and MCF-7 cells were loaded onto DEAE-Sephadex A 50 columns. Following elution with a continuously increasing gradient salt buffer (from 0.2-5 N NaCl), eluted fractions (500 l ϫ 10) from the total nuclear extract were collected. A 20-l sample of each eluted fraction was incubated with [ 3 H]-methyl-DNA substrate in an open microfuge placed in a sealed scintillation vial containing 2 ml of scintillation mixture overnight at 37°C. Demethylation of DNA results in release of the methyl moiety as a volatile methyl residue ( 3 H-labeled methanol) that is trapped in the scintillation mixture (27)(28)(29). To determine the level of released methyl groups, the vials were counted in a scintillation counter. Results are expressed as the mean disintegration per min per sample Ϯ S.E. per the peak fraction of each of the cell extracts, which were assayed in triplicate.

Evaluation of uPA mRNA Expression and Its Effect on Tumor Cell
Invasion-To test the hypothesis that uPA might play a role in breast cancer cell invasion, we first examined the expression of uPA mRNA in the estrogen receptor (ER)-positive, hormone-sensitive human breast cancer cell line MCF-7 and in highly invasive ER-negative, hormone-insensitive MDA-MB-231 cells using Northern blot analysis. Additionally, mRNA expression was also evaluated in MCF-7 cells following treatment with the DNA methyltransferase inhibitor 5-aza-CdR. As seen in Fig. 1A, uPA mRNA is detected in MDA-MB-231 cells and in 5-azaCdR treated MCF-7 cells indicating that this potent demethylating agent is capable of inducing the re-activation of uPA gene expression in MCF-7 cells. To examine the effect of uPA on the invasive capacity of these cells, we carried out a modified Boyden chamber Matrigel invasion assay. These studies show that MDA-MB-231 cells expressing high levels of uPA are able to invade through the Matrigel, whereas MCF-7 cells that do not express uPA are unable to invade. Furthermore, the non-invasive nature of MCF-7 cells was reversed, and a significant number of tumor cells invaded through the Matrigel, following 5-azaCdR treatment of MCF-7 cells (Fig. 1B). This increase in tumor cell invasive capacity following demethylation correlated with induction of uPA  (lanes a1, b1, and c1) or with methylationsensitive enzymes HpaII (lanes a2, b2, and c2) and HhaI (lanes a3, b3,  and c3). DNA digests were resolved on 0.8% agarose gel, and all blots were probed with a 778-bp human uPA promoter fragment (C). Results are representative of at least three different experiments. Data of invasion assay (B) are expressed as mean Ϯ S.E. of values from three independent experiments. Significant difference in the number of invading from control is represented by an asterisk (p Ͻ 0.05).
mRNA. These results are consistent with previous studies by us and others showing that uPA expression is directly related to the invasive capacity of the tumor cell (30).
Analysis of uPA Gene Methylation in Breast Cancer-We then determined whether this switch in expression of uPA is controlled epigenetically. The change in uPA gene methylation status was examined as a potential molecular mechanism regulating the differential expression of uPA in tumor cells representing early (MCF-7) and late (MDA-MB-231) stage human breast cancer. Genomic DNA was isolated from these cell lines and digested first with the methylation-insensitive endonuclease PstI (Fig. 1C, lane a) and then with the methylationsensitive (HpaII, HhaI) endonucleases (Fig. 1C, lanes b and c). These samples were subjected to a Southern blot analysis and hybridized with a probe recognizing the uPA promoter region. Because this region of the gene bears multiple CCGG and CGCG sites, it can be cleaved extensively by HpaII and HhaI restriction enzymes when it is unmethylated, as is the case with DNA prepared from MDA-MB-231 cells expressing abundant amounts of uPA (Fig. 1C, lanes b1 and c1). In contrast, in MCF-7 cells, both HpaII and HhaI failed to cleave this region resulting in an identical pattern to the one observed after PstI digestion alone (Fig. 1C, lanes b2 and c2). To further confirm that this differential digestion with HpaII and HhaI is because of DNA methylation, MCF-7 cells were treated with 5-azaCdR and subjected to a similar Southern blot analysis. The pattern of cleavage of the uPA promoter was identical to the one observed in MDA-MB-231 cells (Fig. 1C, lanes b3 and c3).
Characterization of Human Breast Cancer Cell Lines for uPA Gene Expression, Cell Invasive Capacity, and Methylation Status within the uPA Promoter Region-To determine whether the correlation among uPA expression, cell invasiveness, and uPA promoter methylation is representative of all breast cancer cell lines, we characterized several additional cell lines representing different stages of breast cancer. Using reverse transcriptase-PCR, we found that, like MCF-7 cells, T47D and BT474 cells that are known to be ER-positive do not express uPA. (Fig. 2A). On the other hand, HS578T and BT549, which are ER-negative, were found to express uPA, as does MDA-MB-231 ( Fig. 2A). To identify whether a correlation between uPA expression and cell invasive capacity exists in all of these cell lines, we carried out a modified Boyden chamber Matrigel invasion assay. These studies show that HS578T and BT549 cells, which express uPA, act in a similar fashion to MDA-MB-231 and are able to invade through the Matrigel. T47D and BT474, in addition to MCF-7, are found to have low invasive capacity (Fig. 2B).
Complete characterization of the methylation status of the uPA promoter region in human breast cancer cell lines was carried out by MSP. For these studies, we selected PCR primers that could amplify a 475-bp DNA fragment that included part of the promoter and of exon 1 within the uPA gene. This area is rich in CpG dinucleotides (Fig. 2C), which can serve as sites for methylation modification. The analysis of this region was also carried out by a TESS (transcription element search system) DNA analysis program showing the presence of several important DNA transcription factor binding sites in this region, which may play a significant role in regulating uPA expression (Fig. 2C). MSP analysis revealed that greater than 80% of cytosines of CpG dinucleotides are methylated in MCF-7 cells that do not express uPA. Similarly, non-uPA expressing and low invasive T47D and BT474 cells also have a high degree of methylation in the uPA promoter at 21.2 and 18.2%, respectively. By contrast, in uPA expressing MDA-MB-231, HS578T, and BT549 cells, the uPA promoter was completely unmethyl-ated (Fig. 2C). In Fig. 3 we have provided the complete DNA sequence from the results of MSP analysis of two of the most commonly described highly invasive (MDA-MB-231) and low invasive (MCF-7) breast cancer cell lines. Overall, these results are in agreement with methylation of the uPA gene in MCF-7 cells resulting in silencing of uPA gene expression and provide a clear correlation among uPA expression, cell invasive capacity, and uPA promoter methylation.
Analysis of the Impact of Methylation on uPA Gene Promoter Activity in Human Breast Cancer Cells-After we established a correlation between promoter methylation and uPA gene transcription, we examined the effect of changes in promoter methylation on gene transcription. For these studies, the uPA promoter reporter construct (Ϫ745 to ϩ30 bp) uPA-luc was methylated by mHpaII, mHhaI, or mSssI methylases or mock- treated in vitro and was transiently transfected into the human breast cancer cells MDA-MB-231. Luciferase reporter activity was determined 48 h after transfection. As a control we used a pGL-3 basic luciferase construct lacking the promoter. The mock-treated unmethylated uPA-luc plasmid exhibited at least 4-fold higher luciferase activity as compared with cells transfected with the promoterless control plasmid as expected. Methylation of the different plasmids with different methylases reduced the promoter activity significantly (Fig. 4A).
Transcription factor Ets-1 activates uPA promoter and induces its gene expression (31,32). As shown in Fig. 2C, Ets-1 binding site coincides with methylation sites and is therefore a candidate to be affected by DNA methylation. We then determined whether the inducible effect of Ets-1 could be blocked by methylation. Methylated or unmethylated uPA-luc was cotransfected with either pEVRF0 (empty vector for pEVRF-Ets-1; lane 2) or pEVRF-Ets-1 (lane 3) encoding Ets-1 as shown in Fig. 4B. Although pEVRF-Ets-1 activates unmethylated uPA-luc more than 5-fold, it fails to stimulate the transcription activity of all methylated uPA-luc plasmids (Fig. 4B). This marked suppression is consistent with methylation regulating the uPA promoter. Although there are several methylated cytosines in regions surrounding Sp1 binding sequences, transfection of Sp1 plasmid failed to cause any significant change in the uPA-mediated luciferase activity, as was observed following Ets-1 plasmid transfection (data not shown).
Electrophoretic Mobility Shift Assay of Methylated Versus Unmethylated ETS Binding Site-To further evaluate the involvement of the Ets-1 transcription site in the regulation of the uPA promoter, we next examined the ability of the Ets-1 protein to bind to both the methylated and unmethylated Ets-1 binding sites by electrophoretic mobility shift assay. Methyl-ated and unmethylated oligonucleotides corresponding to the Ets-1 binding site within the uPA promoter were synthesized, labeled, and incubated with nuclear extracts from MDA-MB-231 cells where the Ets-1-binding protein was overexpressed. The resulting DNA-protein complex indicates that there is a greater amount of binding with the unmethylated ODN than with the methylated ODN (Fig. 5). Addition of 10 -50-fold excess of unlabeled methylated oligonucleotide did not effect the Ets-1 binding activity. By contrast, addition of a 10 -50-fold excess of unlabeled unmethylated ODN resulted in a dose-dependent displacement in binding of Ets-1-binding protein to the unmethylated oligonucleotide (Fig. 5). These results confirm that the Ets-1-binding protein selectively binds to the unmethylated, and not the methylated, Ets-1 transcription factor site within the uPA promoter.
Maintenance and de Novo DNMT and DMase Enzyme Activities in Human Breast Cancer Cells-We then examined whether differences in the state of methylation of the uPA promoter between MDA-MB-231 and MCF-7 cells reflect a global change in the DNA methylation machinery. We first examined the levels of de novo and maintenance DNA DNMT activities in the two cell lines by quantitation of the total amount of [ 3 H]-methyl group that has been catalyzed by nuclear extracts prepared from these cells onto either hemimethylated (maintenance DNMT activity) or unmethylated DNA substrates (de novo DNMT activity). As shown in Fig. 6A with MCF-7 cells (lane 5). Thus, the maintenance but not the de novo DNMT activity correlates with the state of methylation of the uPA gene.
We then determined whether there are differences in global DNA DMase activity in MCF-7 cells in comparison with MDA-MB-231 cells. It has been suggested previously that tumor cells have high levels of DMase activity that may be responsible in part for the hypomethylation observed in these cells. Active removal of the [ 3 H]-CH 3 moiety from methylated cytosines in in vitro methylated DNA has been shown previously to result in a release of a volatile residue that was identified as methanol (28). Using a volatile assay we quantitated DMase activity in these breast cancer cells (26 -28). DMase activity was 3-4-fold higher in MDA-MB-231 cells where the uPA promoter is hypomethylated, as compared with MCF-7 cells, where the promoter is methylated (Fig. 6C). Thus, global DMase activity correlates with the state of methylation of the uPA promoter in breast cancer cells.

TSA Treatment Increases uPA Expression in MDA-MB-231
But Not in MCF-7 Cells-Gene silencing is frequently associated with DNA methylation and histone deacetylation, and conversely, gene expression is associated with DNA demethylation and histone acetylation. Together, these two regulatory mechanisms play a critical role in the regulation of gene transcription during tissue development, tumor transformation, and tumor progression. It has been shown that DNMT1, and several methyl-CpG-binding proteins such as MeCP2, MBD2, and MBD3, are associated with histone deacetylases that are involved intimately in gene silencing (33)(34)(35). These observations have provided key links between DNA methylation (demethylation) and histone deacetylation (acetylation). TSA is a well described chemical deacetylase inhibitor that has been shown to induce DNA acetylation resulting in gene activation. To examine the role of histone acetylation in uPA gene transcription, MDA-MB-231 and MCF-7 cells were treated with TSA for 2 days followed by determination of uPA mRNA expression by Northern blot analysis. These studies showed that uPA mRNA was increased markedly following treatment of MDA-MB-231 cells with TSA. In contrast, MCF-7 cells treated with TSA failed to exhibit any significant change in uPA mRNA expression. uPA mRNA continued to be undetectable in these breast cancer cells, which are non-invasive and show hypermethylation of the uPA promoter (Fig. 7). These results demonstrate that DNA methylation and not histone deacetylation is the dominant mechanism suppressing uPA expression in MCF-7 cells. DISCUSSION uPA is now believed to play an important role in several cancers where increased uPA production is associated with late stages of cancer (1,36). It is therefore important to understand the mechanisms responsible for regulation of uPA expression during tumor progression. In this paper, we tested the hypothesis that DNA methylation is involved in the differential reg- ulation of uPA during tumor progression. After fully characterizing several human breast cancer cell lines representing various stages of breast cancer, we focused on 2 cell lines, MCF-7, a hormone-sensitive and low invasive cell line that represents early stage human breast cancer, and MDA-MB-231 cells, which are hormone-insensitive and represent late stage breast cancer. Animals inoculated with MDA-MB-231 develop large tumors that can metastasize to several sites in vivo (6). In these cells, levels of uPA expression correlate with tumor cell invasive capacity and provide a link between uPA expression and tumor stage (5,37). Using this model we examined whether the methylation status of the uPA gene plays a role in regulating the differential expression of this gene through the multistep process of tumor progression. Our data show that the methylation status of the uPA promoter correlates with its state of expression as demonstrated by methylation-sensitive endonuclease Southern blot analysis and sodium bisulfate mapping. DNA methylation plays a causal role in controlling the uPA expression, because the DNMT inhibitor 5-azaCdR induces the expression of uPA in MCF-7 cells and increases their invasive capacity. Further support for the causal role of DNA methylation in regulating uPA gene expression is drawn A 50 columns. Following elution with a continuously increasing gradient salt buffer, eluted fractions (500 l ϫ 10) from the total nuclear extract were collected. A 20-l sample of each eluted fraction was incubated with [ 3 H]-methyl-DNA substrate overnight at 37°C. Amount of generated volatized [ 3 H]-CH 3 OH was counted for each cell line (C). Results are expressed as the mean disintegration per min per sample Ϯ S.E. of cell extracts, each of which was assayed in triplicate.  5 in A and B). CTL represents the reaction mixture containing no substrate oligos or nuclear extract. Total radioactivity retained by GF/C filters from reaction mixtures of respective cell lines in the absence of substrate oligos are shown by Ϫ signs (lanes 2 and 4 in A and B). Results are expressed as the mean disintegration/min of [ 3 H]-CH 3 incorporated into substrate oligonucleotide per g of nuclear protein Ϯ S.E. of triplicate cell extracts, each of which was assayed in triplicate. Total nuclear extracts (6 g) obtained from MDA-MB-231 and MCF-7 cells were loaded onto DEAE-Sephadex from experiments that show that in vitro methylation inhibits uPA promoter activity and its transactivation by the transcription factor Ets-1. Ets-1 is required for the expression of uPA in a number of tissues (31,32,38) and interacts with sequences that are close to CpG dinucleotides as shown in Fig. 2. Methylation might inhibit uPA promoter activity by either inhibiting the interaction of Ets-1 with methylated CpGs in its recognition sequence or by recruiting methylated DNA-binding proteins that results in an inactive chromatin structure (12,39,40). These results demonstrate that the uPA promoter region involved in these effects contains an Ets-1 binding site; however, further experiments are required to determine the exact role of the Ets-1 site.
Aberrations in the DNA methylation machinery resulting in global hypomethylation and regional hypermethylation are well documented in cancer. It has been suggested that induction of both DNMT and DMase activities might play a role in the complex changes in DNA methylation observed in cancer cells (15). However, most of the attention in the field has been directed to the hypermethylation of tumor suppressor genes in cancer cells and its potential role in tumorigenesis (41). A large number of studies have demonstrated that inhibition of DNMT reverses tumor growth, and in addition, antisense DNMT1 inhibitors are currently in clinical trials (42)(43)(44). Our data, however, suggests that the involvement of methylation in tumorigenesis is more complex and that hypomethylation of certain genes might play a critical role in tumor progression. Therefore, activities that are responsible for demethylating genes required for tumor invasion might be important anticancer targets.
Our data are consistent with a model in which DNMT and DMase activities are expressed differentially and play distinct roles at different stages of tumor progression (Fig. 8). Early stage cancer cells such as MCF-7 show higher maintenance DNMT activity that may be required for maintaining the transformed state and involved in the silencing of tumor suppressor genes. However, in the later stages of tumor progression, increased DMase activity is vital to induce the expression of genes that are silenced by DNA methylation but are critical for tumor invasion such as uPA. We show that in later stage MDA-MB-231 cells, maintenance DNMT activity is reduced whereas DMase activity is increased. It has been shown previously that ectopic expression of the ras oncogene can lead to increased DMase activity, and an active DMase was purified recently from human lung carcinoma cells (29,45). It is not yet clear which of the DMases is specifically induced in MCF-7 cells, and further experiments are required to characterize this DMase. Nevertheless our results demonstrate that DNA methylation activities do undergo distinct changes during tumor progression and might play different roles at specific stages. FIG. 8. Schematic representation of the potential mechanisms that control uPA gene expression through DNA methylation during breast cancer progression. During early stage breast cancer, uPA expression is undetectable, and the uPA promoter is hypermethylated and maintained in this hypermethylated form after DNA replication. A high level of maintenance DNMT activity catalyzes this reaction. During cancer progression, hypermethylation of the uPA promoter is switched to hypomethylation because of increased DMase activity in coordination with gradually decreased maintenance DNMT activity. This is achieved either directly (via demethylation) or indirectly (via hemimethylation) during DNA replication. Overexpression of various tumor progression factors in coordination with the suppression of tumor suppressor genes are able to break the endogenous balance of DNMTs and DMases to promote hypomethylation of the uPA gene at late stage breast cancer progression.