KiSS-1 Represses 92-kDa Type IV Collagenase Expression by Down-regulating NF-κB Binding to the Promoter as a Consequence of IκBα-induced Block of p65/p50 Nuclear Translocation

The 92-kDa type IV collagenase (MMP-9) plays a critical role in tissue remodeling. We undertook a study to determine whether the KiSS-1 gene, previously shown to suppress cancer spread (metastases), negatively regulates MMP-9 expression. Six cell lines positive for MMP-9 mRNA were deficient in KiSS-1 mRNA. One of these cell lines, HT-1080, stably transfected with a KiSS-1 expression construct, demonstrated substantially lower MMP-9 enzyme activity/protein and in vitro invasiveness. The lower MMP-9 enzyme activity reflected reduced steady-state mRNA levels which, in turn, was due to attenuated transcription. Activation of ERKs and JNKs by phorbol 12-myristate 13-acetate and tumor necrosis factor α, respectively, leading to increased MMP-9 amounts was not antagonized by KiSS-1 expression, suggesting that MAPK pathways modulating MMP-9 synthesis are not the target ofKiSS-1. Although MMP-9 expression is regulated by AP-1, Sp1, and Ets transcription factors, KiSS-1 did not alter the binding of these factors to the MMP-9 promoter. However, NF-κB binding to the MMP-9 promoter required for expression of this collagenase was reduced by KiSS-1 expression. Diminished NF-κB binding reflected less p50/p65 in the nucleus secondary to increased IκBα levels in the cytosols of the KiSS-1transfectants. Thus, KiSS-1 diminishes MMP-9 expression by effecting reduced NF-κB binding to the promoter.

Tissue remodeling in physiological and pathological conditions such as trophoblast implantation, bone development, angiogenesis (1)(2)(3), and the spread of cancer (invasion/metastases) (4 -6) requires proteolytic action to degrade the surrounding extracellular matrix and to activate cytokines such as tumor growth factor ␤ and interleukin 1␤ (2,3). There is now compelling evidence implicating the type IV collagendegrading 92-kDa type IV collagenase (MMP-9) (4,7,8) in these processes. Thus, mice null for the MMP-9 gene exhibited an abnormal pattern of skeletal growth plate vascularization (1). Additionally, MMP-9 was shown to be required for human bronchial epithelial cell migration and spreading following injury (9). In cancer, MMP-9 mRNA/protein is produced in both cancer and normal cells (10 -12), and there is strong evidence implicating this type IV collagenase in the spread of the disease. Thus, Bernhard et al. (7) reported that the overexpression of this metalloproteinase in nonmetastatic rat embryo cells conferred a metastatic phenotype upon these cells. In contrast, inhibition of MMP-9 expression by a ribozyme blocked metastasis of rat sarcoma cells (6).
The MMP-9 gene, located on chromosome 20 (13), covers 13 exons spanning 7.7 kilobases. Transcription gives rise to a 2.5-kilobase mRNA (14,15). Translation of the message produces a 92-kDa precursor that is subsequently processed by the proteolytic removal of 73 amino acids at the amino terminus of the metalloproteinase (8, 16 -18). The activity of the enzyme is controlled by the levels of physiological inhibitors including tissue inhibitor of metalloproteinases 1 and 2, which form noncovalent bonds with the enzyme (19,20).
Regulation of MMP-9 protein levels has been ascribed to transcriptional activation of the gene (21) and to reduced mRNA turnover (22). The 5Ј flanking sequence contains putative binding sites for AP-1, NF-B, Sp1, and Ets transcription factors within the first 670 base pairs, and these have been implicated in the induction of MMP-9 gene expression by TNF-␣, 1 v-Src, c-Ha-Ras, Tat, and contact inhibition using transient transfection of cultured cells (15,21,(23)(24)(25)(26). In addition, studies with transgenic mice have demonstrated the requirement of regions Ϫ522/ϩ19 and Ϫ2722/Ϫ7745 for developmental regulation in mice (27) and for tissue-specific expression in osteoclasts and migrating keratinocytes, respectively (27,28).
Although  has been implicated in the invasive/metastatic phentoype of many cancers, how its expression is regulated is still poorly understood. In the last 5 years, a great deal of interest has been focused on a group of genes collectively referred to as metastases-associated genes, which modulate cancer metastases but not tumorigenesis (29,30). One of these is the KiSS-1 gene, which was discovered by subtractive hybridization and is reduced in its expression in metastatic cancer (31). Further, its enforced expression in MDA-MB-435 breast cancer cells and melanoma suppressed the metastatic potential of these cells by 95% without affecting their tumorigenicity (31,32). The KiSS-1 gene is comprised of 4 exons, the first two of which are not translated. The third exon contains 38 noncoding base pairs at the 5Ј end followed by 100 base pairs of translated sequence. The terminal exon contains 332 base pairs of translated sequence. The gene maps to chromosome 1q32 (33), and its sequence predicts a unique hydrophilic 145amino acid protein with a leader sequence and a predicted molecular mass of 15.4 kDa (34). However, the mechanism by which KiSS-1 represses the metastatic phenotype is unknown. Because MMP-9 has a well established role in tumor cell invasion and metastases, we undertook the present study to determine whether the expression of this collagenase is regulated by KiSS-1.

EXPERIMENTAL PROCEDURES
Cell Culture and Stable Transfections-HT-1080 cells were maintained in McCoys 5A medium supplemented with 10% fetal bovine serum. For stable transfections, cells were transfected at ϳ60% confluence using poly-L-ornithine as described previously (35). Essentially, the cells were incubated with 20 g of DNA and 9.3 g of poly-Lornithine (Sigma) for 6 h and then shocked for 4 min with 25% glycerol. Cells were incubated for 2 days and then selected with 600 g/ml of G418 (active concentration). Clones were isolated, expanded, and screened for KiSS-1 cDNA integration by PCR.
Invasion Assays-These were carried out as described previously but with minor modifications (36). Briefly, Matrigel® was diluted in cold serum-free medium, and 25 g of the preparation was added to the porous filters (pore size, 8 m) and allowed to gel at 37°C. Cells are dispersed with Ca 2ϩ /Mg 2ϩ -free medium containing 3 mM EDTA, and 250,000 cells dispensed into the Transwell®. The Transwell® was subsequently inserted into a well, with the latter also containing culture medium. The cells were incubated at 37°C for 5 h. After this time, cells on both the upper and lower aspects of the membrane were stained with Diff-Quik (Baxter Scientific Products, McGaw Park, IL) and enumerated.
Northern Blotting-The level of steady state mRNAs was determined by Northern blot analysis (37). Total cellular RNA was extracted from 90% confluent cultures using 5.0 M guanidinium isothiocyanate and purified on a cesium chloride cushion (5.7 M) by centrifugation at 150,000 ϫ g for 20 h. Purified RNA was electrophoresed in a 1.5% agarose-formaldehyde gel and transferred to Nytran-modified nylon by capillary action using 10ϫ SSC. The Northern blot was probed at 42°C with random-primed radiolabeled cDNAs (38). The KiSS-1 cDNA coding sequence was generated by restriction digestion of the KiSS-1 expression plasmid with XhoI/BamHI, yielding a 0.7-kilobase fragment. The blots were then washed at 65°C using 0.25ϫ SSC in the presence of 0.75% SDS. Loading efficiencies were checked by reprobing the blot with a radioactive GAPDH cDNA.
Mobility Shift Assays (EMSA)-Nuclear extract was prepared essentially as described elsewhere (41). Cultured cells were collected by centrifugation, washed, and suspended in a buffer containing 10 mM Hepes, pH 7.9, 10 mM KCl, 0.1 mM EDTA, 0.1 mM EGTA, 1 mM DTT, and 0.5 mM PMSF. After 15 min on ice, the cells were vortexed in the presence of 0.5% Nonidet P-40. The nuclear pellet was then collected by centrifugation and extracted in a buffer containing 20 mM Hepes, pH 7.9, 0.4 M NaCl, 1 mM EDTA, 1 mM EGTA, 1 mM DTT, and 1 mM PMSF for 15 min at 4°C.
Nuclear extract (10 -20 g) was preincubated at 4°C for 30 min with 1 g of the indicated antibody or a 100-fold excess of an unlabeled oligonucleotide spanning the MMP-9 cis element of interest. The sequences were as follows: AP-1, CTGACCCCTGAGTCAGCACTT; NF-B, CAGTGGAATTCCCCAGCC; Ets, GGAGAGGAAGCT; and Sp-1, TCCTTCCGCCCCCA. After this time, the reaction mixture was incubated at 4°C for 20 min in a buffer (25 mM Hepes buffer, pH 7.9, 0.5 mM EDTA, 0.5 mM DTT, 0.05 M NaCl, and 2.5% glycerol) with 2 g of poly(dI⅐dC) and 5 fmol (2 ϫ 10 4 cpm) of a Klenow end-labeled ([ 32 P]ATP) oligonucleotide that spans the DNA binding site in the MMP-9 promoter. The reaction mixture was electrophoresed at 4°C in a 6% polyacrylamide gel using TBE (89 mM Tris, 89 mM boric acid, and 1 mM EDTA) running buffer. The gel was rinsed with water, dried, and exposed to x-ray film overnight.
Zymography-Zymography was performed as described previously (24). Culture supernatants were denatured in the absence of reducing agent and were electrophoresed in a 7.5% polyacrylamide gel containing 0.1% (w/v) gelatin. The gel was incubated at room temperature for 2 h in the presence of 2.5% Triton X-100 and subsequently at 37°C overnight in a buffer containing 10 mM CaCl 2 , 0.15 M NaCl, and 50 mM Tris, pH 7.5. The gel was then stained for protein with 0.25% Coomassie and photographed on a light box. Proteolysis was detected as a white zone in a dark field.
Western Blotting-For analysis of p65 and p50 NF-B proteins and IB (42), 2-3 ϫ 10 6 cells were washed with cold phosphate-buffered saline and were suspended in lysis buffer (10 mM Hepes, pH 7.9, 10 mM KCl, 0.1 mM EDTA, 0.1 mM EGTA, 1 mM DTT, 0.5 mM PMSF, and proteinase inhibitors). The cells were allowed to swell on ice for 10 min, after which 20 l of 10% Nonidet P-40 was added. The samples were then vortexed and centrifuged for 30 s at 12,000 ϫ g. Supernatants were saved as cytosol extract, whereas the pellets were resuspended in icecold nuclear extraction buffer (20 mM Hepes, pH 7.9, 0.4 M NaCl, 1 mM EDTA, 1 mM EGTA, 1 mM DTT, 1 mM PMSF, and proteinase inhibitors). The nuclear pellets were incubated on ice for 30 min followed by centrifugation for 5 min at 12,000 ϫ g at 4°C. The supernatants were retained as nuclear extract.
Proteins (60 g) or conditioned medium were resolved in polyacrylamide gels and then transferred to nitrocellulose membrane. The blot was blocked in 5% milk solution and incubated in primary antibody solution at 4°C overnight. The antibodies to phosphorylated JNKs and ERKs was purchased from New England Biolabs (Beverley, MA) (catalog number 9251S) and Promega (Madison, WI) (catalog number V8031), respectively. A mouse monoclonal antibody (catalog number 1M10L) (Oncogene Research Products, Cambridge, MA) was used for the detection of MMP-9. All the other antibodies were purchased from Santa Cruz Biotechnology. After being washed with TBS (10 mM Tris-HCl, pH 8.0, 150 mM NaCl) buffer, the blot was incubated with an horseradish peroxidase-conjugated secondary antibody, and proteins were visualized with ECL reagents (PerkinElmer Life Sciences) according to the manufacturer's recommendation.
PCR-To verify the insertion of KiSS-1 cDNA into the genome, PCR was performed using genomic DNA as the template. To prepare the genomic DNA, the cells were lysed in a solution containing 10 mM Tris-HCl, pH 8.0, 100 mM EDTA, pH 8.0, 20 g/ml RNase A, and 0.5% SDS at 37°C for 1 h. The cell lysates were then treated with 100 g/ml proteinase K at 50°C overnight and extracted twice with phenol/chloroform (1:1). DNA was precipitated with ammonium acetate/ethanol.
The PCR reaction mixture contained 50 ng of genomic DNA, 1.5 mM MgCl 2 , 0.2 mM each of dATP, dTTP, dGTP, and dCTP, 0.5 M of each of the primers, and 1ϫ Q solution (Qiagen). The upstream primer (5Ј-CCACTGCTTACTGGCTTATCG-3Ј) was complementary to the cytomegalovirus promoter sequence in the plasmid pcDNA3, whereas the downstream primer (5Ј-CAGTAGCAGCAGGCTTCCTC-3Ј) was complementary to the KiSS-1 cDNA sequence (nucleotides 376 to 395). After the template genomic DNA was denatured at 95°C, the PCR reaction was initiated by adding 1 unit of Taq DNA polymerase (Qiagen). The template was amplified for 32 cycles. Each PCR cycle consisted of denaturation at 95°C, annealing at 65°C, and extension at 72°C (1 min, 1 min, and 40 s, respectively). The PCR cycles were terminated by extending at 72°C for 7 min. The PCR products were resolved in a 2% agarose gel.
For amplification of KiSS-1 cDNA, the reaction mixture contained 0.5 M KiSS-1 primers and 0.05 M ␤-actin primer, whereas the cycle number was 30. For MMP-9 amplification, the concentration of ␤-actin primers was 0.025 M, and the cycle number was 34. For MMP-2 amplification, the concentration of ␤-actin primers was 0.05 M, and the cycle number was 32. The PCR products were resolved on 2% agarose gel.
Chromatin Immunoprecipitation Assays-Cells (2-3 ϫ 10 6 ) were treated with 1% formaldehyde in fresh medium at 37°C for 10 min followed by the addition of 0.125 M glycine. After rinsing with ice-cold phosphate-buffered saline, the cells were resuspended in 200 l of SDS lysis buffer (1% SDS, 10 mM EDTA, 50 mM Tris-HCl, pH 8.1) and incubated for 10 min on ice. The cell lysates were then sonicated with a Sonic Dismembrator (Fisher) at 30% maximum power for six 20-s pulses on ice. After removing cell debris by centrifugation, the cell lysates were diluted 10-fold in IP buffer (0.01% SDS, 1.1% Triton X-100, 1.2 mM EDTA, 16.7 mM Tris-HCl, pH 8.1, 167 mM NaCl) supplemented with 1 mM PMSF and protease inhibitor mixture (Roche Molecular Biochemicals). At this step, 1% of the diluted chromatin solutions was saved to control (input DNA) for the total DNA amounts of each sample. The diluted chromatin solutions were then precleared with 80 l of a 50% protein A-agarose slurry containing 20 g of ssDNA and 1 mg/ml bovine serum albumin (Upstate Biotechnology, Lake Placid, NY) for 30 min at 4°C with agitation to reduce nonspecific background. Anti-p65 antibody (2 g; Santa Cruz Biotechnology) was then added to 1 ml of chromatin solution and rotated overnight at 4°C. The immune complexes were then mixed with 80 l of a 50% protein A-agarose slurry containing 20 g of ssDNA and 1 mg/ml bovine serum albumin. After centrifugation, the agarose beads were washed once in 1 ml of Solution 1 (0.1% SDS, 1% Triton X-100, 2 mM EDTA, 20 mM Tris-HCl, pH 8.1, 150 mM NaCl) followed by two additional washes in 1 ml of Solution 2 (0.1% SDS, 1% Triton X-100, 2 mM EDTA, 20 mM Tris-HCl, pH 8.1, 500 mM NaCl) and 1 ml of Solution 3 (0.25 M LiCl, 1% Nonidet P-40, 1% sodium deoxycholate, 1 mM EDTA, 10 mM Tris-HCl, pH 8.1), respectively. The beads were finally washed twice in TE buffer (10 mM Tris-HCl, 1 mM EDTA, pH 8.0). The immune complexes were eluted by adding 250 l of 1% SDS in 0.1 M NaHCO 3 to the pelleted beads and incubated at room temperature for 15 min. NaCl (20 l of a 5 M solution) was then added to the eluted solution and incubated at 65°C for 4 h. After being treated with proteinase K, the DNAs in the samples were recovered by phenol/chloroform extraction and ethanol precipitation using 20 g of glycogen as carrier. The precipitated DNAs were then dissolved in 20 l of TE and blotted into nylon membrane. The slot blot was hybridized with [ 32 P]-labeled MMP-9 promoter (ϩ52 to Ϫ670) fragments.

Expression of KiSS-1 in HT-1080 Cells Decreases MMP-9
Protein/mRNA-Although KiSS-1 has been shown to dramatically attenuate the metastatic capacity of cancer cells, the mechanism as to how this is achieved is not known. We hypothesized that this may be due, at least in part, to a suppression of the expression of MMP-9, which degrades the extracellular matrix. To test this hypothesis, MMP-9-producing HT-1080 cells, which demonstrate an invasive phenotype (43) and which have undetectable KiSS-1 expression, were stably transfected with an expression construct encoding the KiSS-1 coding sequence. G418-resistant colonies were cloned and expanded, and genomic DNA was screened by PCR for the integration of the KiSS-1 sequence into the DNA of the recipient cells. Several clones (numbers 4, 5, 7, and 10) positive for the exogenous KiSS-1 sequence were detected (Fig. 1A). The 470-base pair amplified sequence was not due to the detection of the endogenous KiSS-1 gene, because the 5Ј PCR primer employed is unique to the cytomegalovirus promoter of the pcDNA3 expression plasmid.
To verify the expression of the integrated KiSS-1 cDNA, we performed RT-PCR. RNA from HT-1080 cells transfected with the empty vector or the KiSS-1 expression construct was purified and subjected to RT-PCR. HT-1080 cells transfected with the empty vector (clones 1 and 3) showed no amplified product using primers corresponding to the KiSS-1 coding sequence (Fig. 1B). These data indicate that the expression of KiSS-1 mRNA (endogenous) in HT-1080 cells transfected with the pcDNA3 vector is below the detection limit of RT-PCR. In contrast, an amplified product (of the predicted size, 389 base pairs) corresponding to the KiSS-1 mRNA was readily detected in HT-1080 clones 4, 5, 7, and 10, which were positive for genomic integration of the exogenous KiSS-1 plasmid.
To corroborate the earlier findings (31,32) that KiSS-1 functions as a metastases suppressor gene, in vitro invasion assays were performed. In these assays, cells are plated on an extracellular matrix-coated porous filter, and invasion is defined as the percent of cells penetrating this impervious barrier. Expectedly, HT-1080 cells bearing the vector alone (clone 3-V3) (Fig. 1C) penetrated the extracellular matrix-coated porous filter (32 Ϯ 3% invaded). In contrast, a representative KiSS-1expressing clone (clone 7-K7) showed over a 65% reduction in invasiveness (Fig. 1C). This reduced invasiveness was not due to diminished adhesion. Thus, KiSS-1 clone 7 demonstrated an adhesion rate (percent of cells attached 1 h post-plating) of 44 Ϯ 2 compared with 46 Ϯ 4% for the vector clone 3. Thus, these data illustrate that in HT-1080 cells, KiSS-1 represses invasion, which is a crucial component of the metastatic phenotype.
Because the ability of cells to demonstrate an invasive/metastatic phenotype requires proteolysis, we hypothesized that KiSS-1 negatively regulates MMP-9 expression. To answer this question, conditioned medium from clones positive for KiSS-1 expression or the vector only were assayed for MMP-9 enzyme activity by zymography ( Fig. 2A). Conditioned medium from HT-1080 cells harboring the empty vector (pcDNA3) contained a collagenase activity that was indistinguishable in size (92 kDa) from MMP-9. The amount of this enzyme was greatly reduced in the KiSS-1 transfectants and below the detection limit of this assay for clones 4 and 7. In contrast, the activity of a metalloproteinase, whose size was identical to that of the 72-kDa type IV collagenase MMP-2 (44,45), was not reduced in the KiSS-1 transfectants.
Because the reduced MMP-9 enzyme activity evident by zymography could arguably be due to the increased presence of a physiological inhibitor such as tissue inhibitor of metalloproteinases family members (20), we also assayed the conditioned medium for MMP-9 protein amounts by Western blotting (Fig.  2B). Similar to the zymography data, immunoblotting of conditioned medium from the KiSS-1-transfected HT-1080 clones showed a marked reduction in the amount of this metalloproteinase when compared with the empty vector (pcDNA3) clones. Thus, it is likely that the reduced MMP-9 enzyme activity is due to a diminished amount of the protein.
If KiSS-1 is indeed a negative regulator of MMP-9 protein levels, we would predict that other cell lines or resected cancer that secrete this collagenase would be deficient in KiSS-1 expression. To address this question, 5 squamous cell carcinoma cell lines from the oral cavity (UM-SCC-1, TU-159, Tu167, Tu177, and Tu684) and two resected tumors, which all secrete MMP-9, were analyzed for KiSS-1 mRNA levels by RT-PCR (Fig. 2C). The level of KiSS-1 mRNA was below the detection limit of RT-PCR, whereas amplified fragments corresponding to the ␤-actin and MMP-9 transcripts were evident with these 5 different oral cancer cell lines and the resected tumors. KiSS-1 Expression Reduces the Transcription of the MMP-9 Gene-To determine whether the reduced amount of MMP-9 activity was a consequence of less mRNA encoding this collagenase, we then compared MMP-9 mRNA levels in the HT-1080 clones expressing the KiSS-1 or the empty vector. An amplified fragment of the predicted size (120 base pairs) was readily detected in HT-1080 cells and two clones bearing the empty pcDNA3 vector (Fig. 3) by RT-PCR. However, in all four of the KiSS-1-expressing clones examined, the amount of MMP-9 mRNA determined by RT-PCR was substantially less than that evident for the untransfected HT-1080 or HT-1080 cells harboring the pcDNA3 vector (Fig. 3). In contrast to MMP-9, KiSS-1 expression in HT-1080 cells did not affect the levels of MMP-2 mRNA, a metalloproteinase encoded by a separate gene (45). Thus, although a fragment of the correct size (447 base pairs) was easily detected in HT-1080 cells, the amount of this amplified product was unchanged in HT-1080 cells stably expressing the KiSS-1 cDNA (data not shown).
Although the RT-PCR indicated that steady-state MMP-9 mRNA levels were lower in the KiSS-1-transfected HT-1080 cells, it was not clear whether this was because of decreased transcription or increased mRNA degradation. Therefore, to determine the contribution of reduced MMP-9 mRNA synthesis to the diminished levels of mRNA evident in the KiSS-1-trans-fectants, nuclear run-on experiments were carried out. Nuclei from KiSS-1 transfectants, verified by Northern blotting to be overexpressing KiSS-1 (Fig. 4A), and from vector-only clones were isolated. Nuclei were then incubated with radioactive dUTP, and RNA was isolated and hybridized to various cDNAs. It is apparent from Fig. 4B that the KiSS-1-expresssing HT-1080 clones (numbers 4 and 7) have a lower rate of MMP-9 mRNA synthesis when compared with the pcDNA3 vector pool. Thus, for the pcDNA3 vector pool, the ratio of MMP-9/GADPH mRNA synthesis was 0.98, whereas for representative KiSS-1expressing clones (numbers 4 and 7), it was reduced by 75% to 0.24 and 0.23. In contrast, the nuclear run-on experiments indicated that the mRNA for MMP-2 was transcribed essentially at the same rate in HT-1080 cells expressing the vector or the KiSS-1. These findings suggest that the decreased steadystate MMP-9 mRNA observed in the KiSS-1 transfectants is due largely to a reduced rate of synthesis. Further, the attenuated MMP-9 transcription is not a consequence of a generalized shut-down of gene transcription.
Because MMP-9 gene expression is known to be growthassociated (46 -48), we considered the possibility that the diminished metalloproteinase synthesis evident in the KiSS-1 transfectants was secondary to a slower proliferation rate. To address this possibility, HT-1080 cells expressing either the KiSS-1 or the vector were assayed for proliferation. The growth rates of HT-1080 KiSS-1 clones 4, 5, 7, and 10, once established

KiSS-1 Represses MMP-9 Transcription via Reduced NF-B Binding
in logarithmic phase, were unchanged relative to HT-1080 vector clones 1 and 3 (data not shown). Thus, the reduced expression of MMP-9 evident in KiSS-1-transfected HT-1080 cells cannot be due to a slower proliferative rate.

Stimulation of MMP-9 Activity by Agents That Increase ERK
and JNK Activity Is Unaffected by KiSS-1 Expression-MMP-9 expression is regulated partly by ERK-and JNK-dependent signaling pathways (49,50). We therefore determined whether stimulation of these MAPK signaling cascades leading to increased MMP-9 expression was countered by KiSS-1 expression. To address this question, we employed the phorbol ester, PMA, and TNF-␣ for activating the ERKs (51-53) and JNKs (54), respectively. HT-1080 cells expressing either the empty vector pcDNA3 or KiSS-1 were treated with PMA, and conditioned medium was collected and analyzed for metalloproteinase activity (Fig. 5A). Expectedly, PMA treatment strongly elevated MMP-9 (but not MMP-2) activity in HT-1080 cells harboring the vector only (clones 1 and 3). However, although the basal MMP-9 activity was reduced in all KiSS-1 clones, these clones all responded to the phorbol ester with a strong increase in the activity of this collagenase. Expectedly, PMA increased the amount of phospho-ERK in HT-1080 cells (Fig. 5B) bearing either the empty vector (pcDNA3) or the KiSS-1 expression construct, confirming that the phorbol ester was indeed activating the ERK FIG. 2. Reduced MMP-9 activity/protein in HT-1080 cells made to express KiSS-1. At 80% confluence, the indicated HT-1080 clones were changed to serum-free medium and incubated for 2 days. After this time, conditioned medium was collected and clarified by centrifugation, and cells were counted. Aliquots of conditioned medium, corrected for differences in cell number, was subjected to zymography (A) using an acrylamide gel containing gelatin or to immunoblotting for MMP-9 protein (B). The data are representative of at least duplicate experiments. C, RNA (2 g) extracted from the various cell lines or resected tumors was reverse transcribed and subjected to multiplex RT-PCR using primers to amplify the MMP-9, ␤-actin, and KiSS-1 transcripts. The KiSS-1 and MMP-9 plasmids used as positive controls corresponded to 50 ng of DNA and were amplified without the reverse transcription step.
FIG. 3. Diminished MMP-9 mRNA levels in HT-1080 clones stably expressing an exogenous KiSS-1. At 90% confluence, HT-1080 cells expressing pcDNA3 or the KiSS-1 construct were harvested, and RNA was extracted. After reverse transcription, RNA was subjected to multiplex PCR using primers specific for MMP-9 and ␤-actin. As a positive control, the MMP-9 plasmid (50 ng) was amplified. For the negative control, RT-PCR was performed using buffer only.

FIG. 4. Reduced MMP-9 transcription in HT-1080 cells expressing KiSS-1.
A, total RNA was extracted and purified from 90% confluent HT-1080 cells or clones harboring pcDNA3 or KiSS-1. Purified RNA (20 g) was resolved by electrophoresis and then transferred to a nylon membrane. The filter was subsequently probed with multiprime-labeled cDNAs specific for KiSS-1 or GAPDH. B, nuclei from 90% confluent cells were isolated and incubated with [␣ 32 P]-labeled UTP. Radioactive mRNA was subsequently purified and hybridized with nylon filter-immobilized cDNAs specific for MMP-9, MMP-2, or GAPDH. The intensity of the MMP-9 and GADPH signals was measured using QuantityOne® software. The data are typical of duplicate experiments.

KiSS-1 Represses MMP-9 Transcription via Reduced NF-B Binding
pathway. Equally important, KiSS-1 did not decrease the amount of activated ERKs in the untreated HT-1080 cells.
Similarly, TNF-␣ was an effective stimulant for MMP-9 activity and JNK activation (Fig. 5, C and D) in both the vector clone and the KiSS-1 transfectants. Furthermore, the amount of activated JNKs (phosphorylated p54 and p46) in untreated HT-1080 cells was undetectable by immunoblotting. Taken together, these data argue against the possibility that KiSS-1 represses MMP-9 expression by interfering with either ERK or JNK activation.
KiSS-1 Does Not Affect AP-1 Binding to the MMP-9 Promoter-Several studies from ours and other laboratories (15,21,24,48) have reported on the trans-acting factors regulating the expression of MMP-9. These include transcription factors in the AP-1, NF-B, Sp1, and Ets families (15,21,24,48). We therefore speculated that KiSS-1 reduces MMP-9 expression by modulating the binding of one, or more, of these DNA-binding proteins to the MMP-9 promoter. Toward this end, we first investigated the possibility that the ability of KiSS-1 to downregulate MMP-9 expression was due to reduced binding of AP-1 factors, because these transcription factors have been shown to be a converging point for many regulators of MMP-9 synthesis including v-Src, c-Ha-Ras, and TNF-␣. Nuclear extracts were generated from HT-1080 cells stably expressing the pcDNA3 vector, KiSS-1, or no exogenous construct and were added to an oligonucleotide spanning the proximal AP-1 motif at Ϫ79 of the MMP-9 promoter, which has been implicated in the regulation of expression of this collagenase by diverse stimuli (15,21,24,48). Nuclear extract from HT-1080 cells (parental) resulted in a shifted band in EMSA (data now shown) that was specific for AP-1, because it could be competed by an excess of an unlabeled AP-1 consensus sequence but not by an oligonucleotide containing an NF-B binding motif. However, a comparison of nuclear extracts generated from HT-1080 cells stably expressing either the empty vector or the KiSS-1 coding sequence revealed that the intensity of the shifted band was indistinguishable (data not shown). These data indicate that it is unlikely that KiSS-1 down-regulates MMP-9 transcription by reducing the amount of AP-1 transcription factor(s) bound to the Ϫ79 recognition motif in the MMP-9 promoter. Similarly, KiSS-1 expression did not effect any change in the binding of transcription factors to the Ets or Sp1 motifs in the MMP-9 promoter (data not shown).
KiSS-1 Reduces NF-B Binding to the MMP-9 Promoter Subsequent to Cytosolic "Trapping" of p65/p50 by IB-We next tested the possibility that KiSS-1 effects a reduction in the binding of NF-B to the MMP-9 promoter sequence, because (a) FIG. 5. Stimulation of MMP-9 enzyme activity by phorbol ester or TNF-␣ is unaffected by KiSS-1 expression. A and C, at 80% confluence, the indicated clones were changed to serum-free medium supplemented with or without varying amounts of PMA or TNF-␣. After 24 h, conditioned medium was harvested and clarified, and cells were counted. Aliquots of conditioned medium normalized for any differences in cell number were subjected to zymography. B and D, the indicated clones were treated for varying times with PMA (100 nM) or TNF-␣. The resulting cellular extracts (equal protein) were subjected to Western blotting using antibodies specific for the phosphorylated MAPKs or that recognized the total (phosphorylated and unphosphorylated) amounts. The experiment was performed at least twice.

KiSS-1 Represses MMP-9 Transcription via Reduced NF-B Binding
constitutive expression of this metalloproteinase in HT-1080 cells is achieved partly via trans-activation of an NF-B motif (at Ϫ600) in the MMP-9 promoter (21), and (b) our own studies have shown that mutation of this NF-B motif in the MMP-9 promoter results in up to a 75% reduction in promoter activity. Nuclear extract from untransfected HT-1080 cells bound to an oligonucleotide spanning the NF-B motif at Ϫ600, as evident by a shifted band (parenthesis) in EMSA (Fig. 6A) that could be competed by an excess of an oligonucleotide bearing a consensus NF-B motif (lane 3) but not an AP-1 consensus site (lane 4). Interestingly, the amount of the shifted band was greatly reduced (Fig. 6A), with nuclear extract generated from representative KiSS-1 clones (numbers 4 and 7) (lanes 7 and 8) when compared with nuclear extract (equal protein) from the pcDNA3 vector clones (numbers 1 and 3) (lanes 5 and 6). Note, however, that the amount of a nonspecific band (marked with an asterisk) bound to the oligonucleotide was not decreased in nuclear extract from the KiSS-1 transfectants.
To confirm the identity of NF-B factors bound to the MMP-9 oligonucleotide, the following experiments were carried out. First, we tested the ability of an oligonucleotide bearing a mutated NF-B binding site to compete for the shifted band in EMSA using HT-1080 nuclear extract (Fig. 6B). The oligonucleotide harboring nucleotide substitutions at the NF-B motif, rendering it incapable of binding NF-B, failed to compete for the shifted band in gel retardation assays (Fig. 6B, lane 4). Second, the identity of the transcription factors in the slowermigrating band (marked with a parenthesis) was determined using antibodies specific for p65 or p50 NF-B proteins. Addition of the anti-p65 antibody to the HT-1080 nuclear extract resulted in a "supershifted" band (indicated with a bracket) (Fig. 6B, lane 5). Similarly, an antibody to the p50 NF-B protein also retarded the mobility of the shifted band (Fig. 6B, lane 6) with a concomitant reduction in the intensity of the shifted band (marked with a parenthesis). In contrast, a control antibody directed at the Ets-1 transcription factor had no effect (lane 7) on the migration profile. Taken together, these data indicate that in KiSS-1-tranfected cells NF-B, which trans-regulates MMP-9 expression, is bound in reduced amount to this promoter.
To corroborate the data generated by the EMSA, which is an in vitro assay that does not address the contribution of the in vivo environment of the promoter (DNA wrapped around a histone protein core) to transcription factor binding, we performed chromatin immunoprecipitation assays. Cultured HT-1080 cells bearing the pcDNA3 vector (clone 3) or clones harboring the KiSS-1 expression construct (clones 4 and 7) were treated with formaldehyde to cross-link transcription factors to the DNA in vivo. Subsequently, cells were lysed, and DNA was fragmented by sonication and subjected to immunoprecipitation with the anti-p65 antibody. The immunoprecipitated DNA, after reversal of cross-links, was purified and hybridized with a MMP-9 promoter-specific probe (Fig. 7A). For the empty vector clone (number 3), the signal ratio (p65 precipitable/input amount) was 0.69 as determined using QuantityOne® software (Bio-Rad). This ratio was reduced (to 0.34 and 0.13) with chromatin derived from the two KiSS-1-expressing clones.
Western blotting revealed that the decreased amount of NF-B proteins bound to the MMP-9 promoter in the KiSS-1 transfectants is due, in part, to lower amounts of these proteins in the nuclear compartment with an accompanying increase in the cytosolic subcellular fraction (Fig. 7B). Because the ankyrin repeat-containing IB proteins mask the nuclear localization signal of the NF-B proteins located at the carboxyl terminus of the Rel homology domain (55), we speculated that the reduced presence of p65/p50 in the nuclear fractions of the KiSS-1 transfectants was a consequence of higher amounts of IB. Indeed, immunoblotting (Fig. 7C) revealed larger amounts of IB␣ in the cytosols of the KiSS-1-transfected HT-1080 cells. Thus, this observation would suggest that the diminished amounts of p65/p50 in the nuclear compartment of the KiSS-1 transfectants is due to an inhibition of nuclear translocation of the NF-B proteins by IB␣. Thus, for the KiSS-1 transfectants, the MMP-9 promoter is bound with less NF-B as a result of blockade of p65/p50 nuclear migration because of the increased IB␣ cytosolic amounts.
FIG. 6. KiSS-1 reduces the amount of NF-B bound to the MMP-9 promoter. A, nuclear extract (10 g) from the indicated cells was reacted with a radioactive oligonucleotide (CAGTGGAAT-TCCCCAGCC) spanning the MMP-9 promoter NF-B motif at Ϫ600 without or with an excess of unlabeled oligonucleotides. Competitors included oligonucleotides spanning the NF-B binding site (-600) and the AP-1 motif at Ϫ79. Bound complexes were resolved by electrophoresis. wt, wild-type. B, EMSA was carried out as described for A with the exception that an oligonucleotide corresponding to the mutated (mt) NF-B motif at Ϫ600 was included, where indicated, as a competitor (CAGTttAATTCCCCAGCC). Also, antibodies to Ets-1 and the p65 and p50 NF-B family members were included in the reaction mixture where shown. wt, wild type DISCUSSION The MMP-9 collagenase plays a critical role in tissue remodeling in both physiology and pathology. Thus, this metalloproteinase is a key regulator of growth plate angiogenesis (1), and cell migration required for respiratory epithelium repair has been attributed to this enzyme (9). In cancer, receptor-bound MMP-9 is required for tumor invasion and angiogenesis (2). We undertook the current study to determine whether KiSS-1, a gene that suppresses the invasive/metastatic phenotype of divergent tumors (31,32), regulates MMP-9 expression. Our finding that MMP-9 expression is down-regulated by KiSS-1 provides the first explanation as to how this metastases suppressor gene diminishes the spread of cancer. It should be emphasized, however, that the reduction in MMP-9 expression is probably one of multiple mechanisms responsible for diminished metastases achieved with KiSS-1. On the other hand, it is clear that KiSS-1 expression does not have a global effect on collagenase synthesis. Thus, the expression of a separate metalloproteinase (the 72-kDA type IV collagenase, MMP-2) was not reduced by this metastases suppressor gene, and this observation might partly account for the incomplete inhibition of in vitro invasion by KiSS-1 expression.
For KiSS-1, the down-regulation of MMP-9 enzyme activity is largely due to altered transcription as occurs with v-Src and c-Ha-Ras (21,24). Further, our data suggest that the diminished synthesis of MMP-9 in the KiSS-1 transfectants can be partly accounted for by decreased promoter binding of NF-B. trans-Activation of gene expression by NF-B has been ascribed to multiple mechanisms including the degradation of the IBs, increased DNA binding affinity, and increased trans-acting potential (55)(56)(57)(58). The findings, herein, point to a decreased binding of the p65 and p50 NF-B proteins to the MMP-9 promoter, although reduced trans-acting potential of the bound transcription factor, as previously reported for c-Rel or p65, cannot be ruled out at this stage (56). The attenuated binding of the p65/p50 evident in our study could potentially be due to a lower affinity for its MMP-9 binding site, reduced production of the transcription factor themselves, or a consequence of increased IB synthesis as occurs in interleukin 10-treated cells (57). However, we found no evidence that the total amount of p65/p50 (i.e. whole cell lysates) differed between the vector control and the KiSS-1 transfections. On the other hand, the observation of higher amounts of IB␣ in the KiSS-1 transfectants is consistent with the contention of increased synthesis (or decreased degradation), thereby preventing the nuclear localization of p65/p50 and thus diminishing activation of the MMP-9 expression by these NF-B proteins.
How KiSS-1 regulates IB amounts thereby sequestering the p65/p50 NF-B proteins in the cytosolic compartment is currently unknown. The KiSS-1 sequence predicts a 15.4-kDa protein (34) lacking any signature transcriptional domains arguing against the possibility that it acts directly at the level of MMP-9 gene transcription. On the other hand, the presence of a putative signal peptide is consistent with the notion that the molecule is secreted. If KiSS-1 is indeed secreted, it could potentially interfere with a variety of growth factors/cytokines previously shown to up-regulate MMP-9 expression (47,49,59) or, alternatively, generate a suppressive signal via an autocrine mechanism. However, preliminary findings (data not shown) have failed to show a repression of MMP-9 mRNA levels in untransfected HT-1080 cells incubated with conditioned medium from KiSS-1-transfected HT-1080 cells arguing against the above-mentioned scenario.
Another possibility is that KiSS-1 interferes with MAPKs that connect the cell surface to the nuclear transcriptional machinery. However, our observations again would argue against this possibility. First, KiSS-1 expression did not decrease the amount of the phosphorylated ERKs, and the activity of the JNKs was undetectable in untransfected HT-1080 cells. Second, the ability of the phorbol ester, PMA (activating the ERKs), and TNF-␣ (which activates the JNKs) to stimulate MMP-9 activity in the HT-1080 cells was not blunted by KiSS-1 expression. Thus, it is unlikely that the ability of KiSS-1 to repress MMP-9 synthesis is by an antagonism of the signaling events culminating in the activation of either the ERKs or the JNKs. However, these observations do not rule out the possibility that KiSS-1 interferes with these signaling cascades downstream of ERK or JNK activation. Alternatively, other signaling cascades, not examined in the current study, may themselves be sensitive to KiSS-1 expression. For example, activation of NF-B by the interaction of atypical protein kinase C (60, 61) with p62 and RIP, culminating in the activation of IB kinase ␤ and phosphorylation of IB, may represent a KiSS-1 target.
Interestingly, although the regulation of MMP-9 expression by the KiSS-1 metastasis suppressor is unique to date, this collagenase has recently been shown to physically interact with the metastases-promoting CD44 gene product (62). In pioneering work, CD44, which encodes a membrane glycoprotein, was reported to act as a receptor for proteolytically active MMP-9 thereby promoting tumor invasion and angiogenesis possibly via activation of transforming growth factor ␤. (2). Thus, for CD44 the interaction with MMP-9 is a physical one and obviously separate from that of KiSS-1, which regulates MMP-9 synthesis by interfering with the trans-activation of MMP-9 gene expression by NF-B. Another gene modulating the FIG. 7. Reduced in vivo binding of NF-B to the MMP-9 promoter is associated with higher IB amounts and reduced nuclear localization of p65/p50. A, cell lysates were sonicated and chromatin incubated without or with an anti-p65 antibody and protein A-agarose. DNA in immunocomplexes was recovered by proteinase K treatment and phenol/chloroform extraction and was hybridized to a radioactive MMP-9 promoter probe. As a control, a fixed portion (1%) of the total cell lysate was also hybridized to the radioactive MMP-9 promoter probe (Input). Quantitation of the bands was accomplished using QuantityOne® software. The background detected in the absence of an immunoprecipitating antibody (No Ab) was subtracted from all quantitations. B, subcellular fractions (equal protein) from the indicated clones was subjected to Western blotting for the p50 and p65 NF-B proteins. C, the cytosolic fraction was analyzed for IB␣ protein by Western blotting (Santa Cruz Biotechnology antibody sc-371). The blot was reprobed with an antibody to ␤-actin. spread of cancer is KAI-1 (29) which, like KiSS-1, is a potent suppressor of metastasis (29,63). However, the mechanism by which KAI-1 suppresses tumor cell invasion is also distinct from that of KiSS-1 and is related to its effects on cell-cell aggregation and binding to fibronectin substratum (64) rather than any change in the activity and/or amount of any collagenase (or, for that matter, protease).
Thus, in conclusion, we report for the first time that KiSS-1 down-regulates MMP-9 expression, a finding that explains, at least partly, how KiSS-1 attenuates the invasive/metastatic phenotype of different cancers (31,32). Equally important, we have also demonstrated that the suppression of MMP-9 synthesis by KiSS-1 can be partly accounted for by reduced binding of NF-B to the promoter secondary to the cytosolic sequestration of the p50/p65 NF-B proteins by IB␣.