Epidermal growth factor (EGF) receptor (EGFR) is frequently elevated in epithelial ovarian cancer, and E-cadherin expression is often reduced in advanced disease. In this study, we investigated a mechanism by which EGFR activation promotes disruption of adherens junctions through induction of matrix metalloproteinase 9 (MMP-9). We show that EGFR activation down-modulates E-cadherin, and broad spectrum MMP inhibition ameliorates EGF-stimulated junctional disruption and loss of E-cadherin protein. MMP-9 involvement in EGF-dependent down-regulation of E-cadherin was determined by siRNA specifically directed against MMP-9. Furthermore, treatment with recombinant MMP-9 or transient expression of MMP-9 is sufficient to reduce E-cadherin levels in differentiated ovarian tumor cells. Stable overexpression of MMP-9 led to a loss of E-cadherin and junctional integrity, and promoted a migratory and invasive phenotype. Thus, elevated MMP-9 protein expression is sufficient for junctional disruption and loss of E-cadherin in these cells. The associations between EGFR activation, MMP-9 expression, and E-cadherin were investigated in human ovarian tumors and paired peritoneal metastases wherein immunohistochemical staining for activated (phospho) EGFR and MMP-9 colocalized with regions of reduced E-cadherin. These data suggest that regulation of MMP-9 by EGFR may represent a novel mechanism for down-modulation of E-cadherin in ovarian cancer. [Cancer Res 2008;68(12):4606–13]
- epidermal growth factor receptor
- matrix metalloproteinases
Epithelial ovarian carcinoma is the leading cause of death from gynecologic malignancy in the United States, resulting in >15,000 deaths in 2006 ( 1). Because the disease is rarely detected when confined to the ovary, ∼75% of women are initially diagnosed with intra-abdominal disseminations resulting in a 5-year survival of <20%. Acquisition of the metastatic phenotype involves a complex series of interrelated cellular events including reversible modulation of cell to cell and cell to matrix adhesion and proteolytic activity, ultimately leading to dissociation (shedding) and dispersal of malignant cells. Modulation of adherens junctions and E-cadherin may promote metastatic progression of ovarian cancer.
The epidermal growth factor (EGF) receptor (EGFR) is often mutated or overexpressed in many tumor types including ovarian cancer (reviewed in ref. 2). Furthermore, EGFR is overexpressed in an estimated 35% to 70% of human epithelial ovarian cancer cases and is associated with more aggressive disease and poor clinical outcome ( 3– 12). Cellular responses to EGFR activation include those that play a critical role in tumor growth and survival, and EGFR overexpression in tumors is associated with metastatic progression ( 10, 11). We have shown that EGFR activation stimulates matrix metalloproteinase (MMP)-9 production and promotes migration and invasion in ovarian cancer cells ( 13– 17). Furthermore, EGFR regulates E-cadherin protein expression and complex formation ( 17, 18). We are interested in defining mechanisms by which EGFR activation modulates cell to cell interactions.
Aberrant epithelial differentiation is an early event in epithelial ovarian carcinogenesis; thus, in contrast to most carcinomas that lose E-cadherin expression with progression, E-cadherin is abundant in primary differentiated ovarian carcinomas (reviewed in ref. 19). In more advanced poorly differentiated carcinomas, both absent and persistent E-cadherin expression have been reported (reviewed in ref. 19). Although complete loss of E-cadherin expression is uncommon, reduced E-cadherin staining is often detected in late-stage carcinomas and in ascites-derived tumor cells. Furthermore, negative E-cadherin is predictive of poor overall survival (reviewed in ref. 19). Although mutations in the E-cadherin gene are rare in ovarian carcinomas ( 20), several mechanisms have been described for transient or sustained loss of E-cadherin function including E-cadherin cleavage and shedding ( 21– 24), epigenetic silencing ( 25, 26), and defective E-cadherin recycling and trafficking ( 27). Interestingly, although E-cadherin staining is significantly decreased in stage III/IV versus stage I/II tumors, a soluble ∼80-kDa E-cadherin ectodomain (sE-cad) is elevated in ascites from ovarian cancer patients ( 24). This observation is consistent with other reports demonstrating sE-cad ectodomain in peripheral blood, ascites, and cystic fluids from ovarian cancer patients ( 28– 30). Several proteinases have been implicated in E-cadherin cleavage including MMP-3 and MMP-7 ( 23, 31), a disintegrin and metalloproteinase-10, plasmin, membrane type I MMP ( 21, 32, 33), and MMP-9 ( 24). In this study, we report that EGFR activation leads to disruption of adherens junctions and down-regulation of E-cadherin protein. MMP-9 is a mediator of the EGFR-dependent decrease in E-cadherin protein, and MMP-9 expression alone is sufficient to cause E-cadherin and adherens junction loss. Furthermore, we find a relationship in human ovarian tumors between activated EGFR, MMP-9, and decreased E-cadherin levels. Together, our results suggest that regulation of MMP-9 by EGFR may represent a novel mechanism for down-modulation of E-cadherin in ovarian cancer.
Materials and Methods
Materials. EGF was purchased from Biomedical Technologies, Inc. Matrigel was purchased from Becton Dickinson. Leupeptin, phenylmethylsulfonyl fluoride, and pepstatin were obtained from Sigma. FITC-conjugated anti-rabbit secondary antibody was purchased from Chemicon. Cell culture reagents were obtained from Sigma.
Cell culture and generation of stable cell lines. The epithelial ovarian carcinoma cell lines OVCA433 and OVCA 429 (and all of the stable cell lines generated from the OVCA 433 cells) were grown in Minimal Essential Medium Eagle (MEME) supplemented with 10% fetal bovine serum, 0.2 mmol/L l-glutamine, 1 mmol/L sodium pyruvate, 50 units per mL penicillin, and 50 mg/mL streptomycin. OVCA 433 and OVCA 429 cells were a kind gift from Robert Bast Jr. (M. D. Anderson Cancer Center, Houston, TX). A mouse MMP-9 cDNA image clone was obtained from American Type Culture Collection. OVCA 433 cells were transfected with 10 μg of MMP-9/pCDNA3 or pCDNA3 alone according to manufacturer's instructions using Superfect (Qiagen). Stable cell lines for further analysis were generated by selection with 2 mg/mL G418 and maintained in 0.75 mg/mL G418. Ten MMP-9–expressing cell lines were generated. Stable cell lines were selected based on their MMP-9 mRNA expression and gelatinase activity in a zymography assay.
Zymogram analysis. For all zymography analysis, cells were serum deprived for 24 h and zymography was performed on conditioned medium as described previously ( 13, 14). Results shown are representative of a minimum of three independent experiments. Densitometry was performed on Western blots using the Kodak Image Station using Molecular Imaging Software version 4.0.
Immunofluorescence and microscopy on OVCA 433 cells. Phase-contrast microscopy and immunofluorescence microscopy were performed using an Olympus BH-2 inverted microscope or an Olympus 1X70 fluorescence microscope, respectively. Images were collected using an Olympus America camera and Magnafire 2.1 software.
OVCA 433 cells were treated as described above, fixed with freshly prepared 3.7% (w/v) formaldehyde in PBS [137 mmol/L NaCl, 2.7 mmol/L KCl, 8.1 mmol/L Na2HPO4, and 1.5 mmol/L KH2PO4 (pH 7.4)] containing 0.8 mmol/L MgCl2 and 0.18 mmol/L CaCl2 for 10 min at room temperature, permeabilized with 0.1% Triton X-100 in PBS for 5 min at room temperature, and blocked with 3% bovine serum albumin (BSA)/PBS for 1 h at 37°C. Fixed cells were incubated with mouse anti–E-cadherin (5 μg/mL; Zymed) or β-catenin (1:200; Chemicon) for 1 h at 37°C. After washing thrice with PBS, samples were incubated with FITC-conjugated anti-mouse IgG (Chemicon). Confocal images were acquired at room temperature using a Zeiss LSM510 system equipped with argon lasers for excitation at 488 nm (green). Samples were viewed with the 63 × 1.4 oil immersion objective lens. Immunofluorescence on OVCA 429 cells was performed as described in Symowicz and colleagues ( 24).
Flow cytometry. Cell lines were washed twice with PBS and then harvested with Typsin-EDTA. Cells (106) were blocked with 3% BSA/PBS for 10 min at 4°C and incubated with primary antibody mouse anti–E-cadherin 5 μg/mL (Zymed) for 40 min at 4°C and then washed twice with PBS. Cells were aliquoted for E-cadherin staining, no antibody control, and IgG control, and resuspended in 0.5 mL PBS. Flow cytometric analysis was performed on a Becton Dickinson FACScan flow cytometer (Immunocytometry Systems). Mean fluorescence intensity for three independent experiments was shown, and error bars represent ± SD. Student's t test was used to calculate significance.
E-cadherin ectodomain generation and immunoprecipitation. OVCA429 cells were plated in 6-well plates at 60% to 70% confluence, preincubated in serum-free MEME and 50 μmol/L GM6001 (Chemicon) or equivalent DMSO overnight, and then treated with 20 ng/mL EGF. After 8 h, the total volume of conditioned medium (1 mL) was collected and centrifuged for 5 min. E-cadherin ectodomain immunoprecipitation was conducted as described in Symowicz and colleagues ( 24). The results of three independent experiments are included in the quantification with Image J. P values were calculated using t test (two sample-unequal variance type and one-tailed distribution). The first asterisk compares untreated cells and cells treated with EGF. The second asterisk compares cells treated with EGF to cells treated with both EGF and GM6001.
Transient transfections. At 80% confluence, OVCA 433 cells were transfected with pCDNA or pCDNA/MMP9. Two micrograms of plasmid DNA was transfected into the cells using Superfect (Qiagen) according to manufacturer's instructions. For transient transfections, cells were subcultured 24 h after transfection, serum starved for 24 h, followed by 24 h of EGF treatment before collection of cell lysates.
SiRNA analysis. SiRNA against MMP-9 was purchased from Ambion. Two different siRNA oligonucleotides were tested, and the following two were used in the described knockdown experiments: Si#1 and Si#2. Silencer glyceraldehyde-3-phosphate dehydrogenase (GAPDH) siRNA controls (which includes GAPDH siRNA and a negative control siRNA) were used as additional controls for the siRNA technique. At 50% confluency, OVCA 433 cells were mock transfected on 10-cm plates or transfected with 50 nmol/L MMP-9 siRNA, 50 nmol/L GAPDH siRNA, or 50 nmol/L negative control siRNA using Superfect (Qiagen). Forty-eight hours after transfection, cells were serum-starved for 24 h and then treated with 20 nmol/L EGF for 24 h before cell lysate collection. Error bars represent SE. Student's t tests were performed to determine statistical significance.
Western blotting. Cells were lysed in radioimmunoprecipitation assay lysis buffer [50 mmol/L Tris (pH 7.5), 150 mmol/L NaCl, 1% triton X-100, and 5 mmol/L EDTA]. Protein concentration was determined using the BCA kit (Pierce). Thirty micrograms of protein were diluted in sample dilution buffer with 2-mercaptoethanol and boiled for 5 min. Samples were separated by SDS-PAGE and subsequently transferred to nitrocellulose membrane. Blots were probed with the following antibodies: anti–E-cadherin from Dako (1:1,000), β-catenin (Chemicon), β-tubulin from Santa Cruz (1:1,000), and β-actin obtained from Sigma. Densitometry was performed on Western blots using the Kodak Image Station using Molecular Imaging Software version 4.0.
Activation of recombinant proMMP-9. Purified recombinant proMMP-9 was generously provided by Dr. Rafael Fridman (Wayne State University, Detroit, MI). It was activated by 1 mmol/L 4-aminophenylmercuric acetate (Sigma Aldrich) at 37°C for 1 to 3 h and then dialyzed in collagenase buffer [50 mmol/L Tris, 150 mmol/L NaCl, 5 mmol/L CaCl2, 0.02% Brij-35 (pH 7.5)] at 4°C for 8 h to overnight. The concentration of recombinant MMP-9 was determined using the Bio-Rad Protein Assay. The presence of the active form of MMP-9 was determined using gelatin zymography.
Ascites samples. Ascites were collected from women undergoing surgical procedures at Prentice Women's Hospital (Chicago, IL) for gynecologic indications with Institutional Review Board (IRB)-approved consent. Preoperative and intraoperative ascites were collected under sterile conditions and frozen at −20°C. ELISA was performed for MMP-9 on ascites specimens. Ascites samples from 35 ovarian cancer patients and women with benign ovarian cysts were diluted 1:100 in PBS analyzed using a human MMP-9 (total) Immunoassay (R&D) according to manufacturer's specifications.
Migration and invasion assays. Cell migration assays were performed on modified Boyden chambers as described in Cowden Dahl and colleagues ( 13). To quantify invasive cells, three independent fields of migratory or invasive cells per well were photographed under phase-contrast microscopy. The number of cells per field was counted, and an average of the three determinations was obtained for each chamber. Each migration or invasion assay was performed a minimum of thrice. Error bars represent SE. Student's t tests were performed to determine statistical significance.
Immunohistochemistry. Immunohistochemical analysis was performed retrospectively on tumor tissue microarrays prepared by the Pathology Core Facility of the Robert H. Lurie Comprehensive Cancer Center at Northwestern University assembled from tissue originally taken for routine diagnostic purposes under an IRB-approved protocol. The microarray tissue specimens included 146 primary human ovarian carcinomas (74 serous, 45 endometroid, 6 mucinous, and 18 clear cell) or 17 paired samples of primary ovarian carcinoma (serous) and paired peritoneal metastatic tissue obtained from the same patient. All cores were 1 mm in diameter. Samples were cut 3- to 4-μm thick and deparaffinized. Antigen retrieval was accomplished by heat induction at 99°C for 45 min. Immunohistochemical staining with antibodies to phospho-EGFR (1:400; Zymed), MMP-9 (1:200 dilution; Lab Vision), and E-cadherin (1:200 dilution; NCH-38; Dako Cytomation) was conducted according to standard procedures. Tissue sections were analyzed by light microscopy by a pathologist (B.P.A.) without prior knowledge of the clinical variables. Scoring was assigned according to the intensity of the staining using a four-tiered scale and graded 0 (negative, with <25% cells positive), 1+ (weak, 25–50% positive cells), 2+ (moderate, 51–75% positive cells), or 3+ (strong, >75% positive cells). Spearman's rank correlation analysis was performed using Systat 11 (Systat, Inc.).
EGFR activation disrupts adherens junctions and reduces E-cadherin protein levels. Because the EGFR is overexpressed in ovarian cancer ( 3– 9), we wanted to determine how activation of the EGFR influences adherens junctions in a differentiated ovarian tumor cell line. EGF treatment disrupted adherens junctions as detected by decreased E-cadherin at sites of cell to cell contact and accumulation of nuclear β-catenin ( Fig. 1A ). Immunoblotting performed on lysates collected from OVCA 433 and OVCA 429 cells treated with EGF for 0, 2, 4, 8, 24, and 48 hours showed little decrease in total E-cadherin protein until 8 to 24 hours of treatment ( Fig. 1B). E-cadherin down-regulation in response to EGF continued over an extended time course ( Fig. 1B). Due to the observation that E-cadherin was decreased at cell borders, we examined cell surface–localized E-cadherin levels by flow cytometry. Loss of E-cadherin from the cell surface was evident within 4 hours after EGF treatment, with maximal E-cadherin loss between 8 and 24 hours ( Fig. 1C). Therefore, EGFR activation decreased the stability of adherens junctions, ultimately leading to a decrease in total E-cadherin protein.
Published data show that pericellular proteinases can catalyze shedding of the E-cadherin ectodomain ( 21, 23, 31, 33), and detection of sE-cad is significantly increased the ascites of ovarian cancer patients ( 24). To evaluate whether EGFR activation leads to generation of sE-cad, conditioned medium were collected from OVCA 429 cells exposed to 20 nmol/L EGF for 8 hours in the presence or absence of the broad-spectrum MMP inhibitor GM6001 followed by immunoprecipitation using an antibody against the extracellular portion of E-cadherin. The 8-hour time point was chosen to specifically determine if sE-cad shedding occurred before the complete breakdown in adherens junctions seen at 24 hours of EGF treatment ( Fig. 2 ). sE-cad was increased in the conditioned medium of EGF-treated cells, and ectodomain shedding was reduced in EGF-treated cells concurrently exposed to GM6001 ( Fig. 1D), suggesting that a metalloproteinase activity contributes to EGF-dependent junction disruption and sE-cad generation.
EGFR activation induces expression and activity of MMP-9 in many ovarian cancer cells ( 15) and induces MMP-9 and MMP-14 in OVCA 433 cells; however, EGF does not increase expression of MMP-1, MMP-7, or MMP-13 in these cells ( 13). We next chose to examine the effects of 24 hours of EGF treatment on both E-cadherin protein and cellular morphology because although E-cadherin is decreased after 8 hours of EGF treatment, the cellular morphology is largely still epithelial. Importantly, EGF treatment does not decrease total N-cadherin (by immunoblot analysis) or surface N-cadherin (by immunofluorescence; data not shown). In two different ovarian cancer cell lines (OVCA 429 and OVCA 433), total E-cadherin protein was decreased after EGF treatment for 24 hours, but E-cadherin protein was retained in EGF-treated cells coincubated with GM6001 ( Fig. 2A). Similarly, EGF treatment led to loss of E-cadherin staining at cell to cell borders, and GM6001 largely protected junctional integrity from EGF-dependent disruption ( Fig. 2B).
To establish whether MMP-9 is important for the observed EGF-stimulated decrease in E-cadherin protein, we used a siRNA approach. Two independent siRNA oligos significantly knocked down MMP-9 mRNA levels (by 53% and 83% compared with mock-transfected cells treated with EGF) and MMP-9 protein (by 70% and 64% compared with mock-transfected cells treated with EGF; Fig. 2C). Importantly, the EGF-dependent decrease in E-cadherin was minimal in cells transfected with these MMP-9–directed siRNAs ( Fig. 2D). Taken together, these data suggest EGF-stimulated down-regulation of E-cadherin is largely mediated by MMP-9.
Elevated MMP-9 disrupts adherens junctions and reduces E-cadherin levels. MMP-9 activity has been reported previously in human ovarian cancer ascites using gelatin zymography ( 24, 34– 36). Quantification of MMP-9 levels as detected by ELISA assay in nonmalignant ascites versus ascites collected from patients with stage III/IV ovarian cancer revealed that total MMP-9 expression was significantly higher in carcinomas, indicating that MMP-9 is elevated in the extracellular ovarian cancer microenvironment ( Fig. 3A ). In addition, the ratio of sE-cad ectodomain ( 24) to MMP-9 was significantly higher in stage III/IV ovarian cancer ascites compared with nonmalignant samples ( Fig. 3B). Because the loss of E-cadherin in cells treated with EGF is associated with MMP-9 activity ( 24), we evaluated whether MMP-9 is sufficient to disrupt adherens junctions and down-regulate E-cadherin protein levels. Cells were treated with exogenous activated MMP-9 at a concentration representing the average level found in human ovarian cancer ascites ( Fig. 3A; ref. 24). Exposure to exogenous MMP-9 decreased E-cadherin border staining ( Fig. 3C). This was confirmed by transient expression of MMP-9, as immunoblot analysis showed a decrease in E-cadherin protein as a consequence of elevated MMP-9 expression in the absence of EGF stimulation ( Fig. 3D). As further confirmation, we examined the effect of constitutive MMP-9 expression on adherens junctions. Clones of cells transfected with an empty vector (pCDNA3) or MMP-9 (pCDNA3/MMP-9) were analyzed. Twelve MMP-9–overexpressing clones were isolated and a representative example (M9) is shown in Fig. 4 . The MMP-9 transfected clones expressed elevated gelatinase activity compared with vector control cells (V3), and elevated MMP-9 mRNA ( Fig. 4A) was confirmed by Northern blot (data not shown). M9 cells do not form epithelial-like sheets ( Fig. 4C) and have no detectable E-cadherin mRNA (data not shown). M9 cells exhibit a loss of E-cadherin protein ( Fig. 4B), loss of E-cadherrn staining at of cell to cell borders ( Fig. 4C), and morphology reminiscent of that observed in EGF-stimulated cells ( 16, 17). N-cadherin expression is detected by both immunoblot and immunofluorescence in M9 cells (data not shown). These findings indicate that elevated MMP-9 in the ovarian tumor environment is likely to have a deleterious effect on adherens junction integrity via modulation of E-cadherin levels.
Next, we wanted to ascertain the relative importance of MMP-9 overexpression on cellular invasion because MMP-9 expression in either ovarian epithelial tumor cells or the adjacent malignant stroma is indicative of poor prognosis ( 37– 39). Cells constitutively expressing MMP-9 displayed a significant increase in migration (2.4-fold; P = 0.0034) compared with control cells ( Fig. 4D). Although MMP-9 overexpressing cells can migrate and invade in <48 hours, we wanted to allow sufficient time for untreated V3 cells to exhibit any invasive potential. In vitro invasion of cells through an artificial basement membrane gave similar results; MMP-9–expressing cells were 5.5-fold more invasive than the untreated vector control cells (P = 0.0004; Fig. 4D). These data suggest that elevated MMP-9 is capable of promoting ovarian cancer cell migration and invasion.
Expression of activated EGFR, MMP-9, and E-cadherin in patient samples. Although MMP-9 is commonly expressed by stromal elements in tumors ( 39), in situ hybridization analyses have provided definitive evidence for robust expression of MMP-9 by malignant ovarian epithelium ( 37, 40, 41). This was supported by immunohistochemical analysis of 141 primary ovarian tumors, which showed MMP-9 expression in 74% of specimens ( 24). Furthermore, analysis of MMP-9 levels in ovarian cancer patient ascites revealed that MMP-9 is prevalent in the ovarian tumor microenvironment ( Fig. 3A and B).
To determine whether EGFR activation is correlated with MMP-9 expression and E-cadherin loss in human tumors, primary ovarian tumors were examined for activated (phospho-)EGFR, MMP-9, and E-cadherin immunoreactivity. We reported previously that 86% of ovarian tumors displayed zones of positive E-cadherin immunoreactivity, whereas 74% exhibited positive MMP-9 staining ( 24). Similar results were obtained in the current analysis, with the majority of ovarian tumors displaying positive E-cadherin (76%) and MMP-9 (81%) reactivity. EGFR activation (phospho-EGFR) was evident in one third of the specimens (35%). EGFR activation was statistically positively correlated with MMP-9 expression (Spearman's rho, 0.429; P < 0001). Although both the Spearman's rank correlation analysis and McNemar's test indicated a negative relationship between phospho-EGFR and E-cadherin, results were not statistically significant. However, examination of serial tumor sections revealed numerous areas of colocalized staining for activated EGFR and MMP-9 that coincided with reduced levels of E-cadherin. ( Fig. 5A–D ). This is further supported by immunohistochemical analysis of a panel of paired primary tumor and peritoneal metastases obtained from the same patient. Approximately one third (35%) of metastases exhibited elevated EGFR activation (phospho-EGFR staining) relative to the paired primary tumor ( Fig. 6A and B ). MMP-9 expression was high in all (100%) phospho-EGFR–positive metastases, whereas 83% concomitantly exhibited decreased E-cadherin staining relative to the paired primary tumor ( Fig. 6A and B). Together, these in vivo data strongly support a relationship between EGFR activation, elevated MMP-9 expression, and decreased E-cadherin.
Elevated EGFR expression in ovarian cancer is associated with poor patient outcome ( 8, 10– 12). In addition to the effect of EGFR on tumor cell growth and survival, EGFR also influences tumor metastasis. EGFR is elevated in advanced ovarian tumors ( 42), and E-cadherin expression is often reduced in advanced disease ( 19), suggesting that both EGFR activity and E-cadherin status promote ovarian cancer progression. EGFR activation in vitro alters cellular morphology and adhesion leading to tumor cell invasion ( 13– 17). Disruption of cell to cell contacts is an important event that fosters cell dissociation from the primary tumor and tumor cell invasion.
Numerous activators of EGFR exist in the ovarian tumor microenvironment ( 43) including EGF, heparin-binding EGF, lysophosphatidic acid, transforming growth factor α, and amphiregulin ( 43, 44). EGFR engagement by ligand induces dimerization, and receptor and intracellular substrate phosphorylation, resulting in activation of signaling cascades ( 45, 46). One consequence of EGFR activation is induction of MMP-9 ( 13– 15). We find MMP-9 in ascites isolated from ovarian cancer patients ( Fig. 3), and interestingly, newly isolated ascites cells secrete MMP-9 in culture ( 34). However, these cells lose MMP-9 expression with successive passages ( 34), suggesting microenvironmental factors are necessary for MMP-9 production. In this study, serial sections of ovarian tumor microarrays support a localized relationship between activated EGFR, increased MMP-9, and diminished E-cadherin ( Fig. 5), and this relationship is particularly evident in metastatic lesions ( Fig. 6). Although there are multiple mechanisms that lead to loss of E-cadherin, including transcriptional repression, trafficking, and methylation ( 47, 48), our data suggest a novel mechanism for down-modulation of E-cadherin in ovarian cancer involving EGFR-dependent regulation of MMP-9 expression.
E-cadherin cleavage by MMPs such as MMP-3, MMP-7, and MMP-9 has been reported in different contexts ( 21, 24, 49, 50). In ovarian cancer, MMP-9 is expressed in both the stroma and the ovarian epithelial tumor cells ( 37, 40, 41), and MMP-9 expression in either compartment is indicative of poor prognosis ( 37– 39). Thus, stromal MMP-9 sources such as tumor-associated neutrophils or fibroblasts in addition to tumor cell–generated MMP-9 are likely to lead to the observed elevation of MMP-9 in ovarian ascites ( Fig. 3; ref. 34). Of note, in our studies, MMP-9 is more prevalent than activated EGFR in human tumors (81% versus 35%), suggesting that ovarian tumors have alternative MMP-9 sources and mechanisms of regulation. In this study, we provide several lines of evidence that elevated MMP-9 down-regulates E-cadherin. EGFR activation induces MMP-9, and both EGFR and MMP-9 activation lead to adherens junction disruption in ovarian tumor cells ( Figs. 1– 4, respectively). Inhibition of MMP-9 by siRNA reduces EGFR-dependent E-cadherin loss ( Fig. 2), whereas elevated MMP-9 by multiple mechanisms is sufficient to induce junctional disruption and decrease of E-cadherin protein ( Figs. 3 and 4). Furthermore, we find a relationship between activated EGFR, elevated MMP-9, and decreased E-cadherin in tumor samples ( Figs. 5 and 6). These data support our conclusion that MMP-9 is a mediator of junctional disruption by EGF and provide the first example of MMP-9–dependent down-regulation of E-cadherin in response to EGF signaling.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Grant support: Department of Health and Human Services grant from the U. S. NIH R01 CA90492 (L.G. Hudson) and CA109545 (M.S. Stack), a postdoctoral fellowship PF-06-036-01-GMC from the American Cancer Society and support by the NRSA F32 award 1F32CA119729-01 (K.D. Cowden Dahl), University of New Mexico Cancer Research and Treatment Center NIH P20 CA888070 and the University of New Mexico National Institute of Environmental Health Sciences Center P30 ES-012072, and partial support by an NIH Institutional National Research Service Award (T32HL076595; E. Guttierrez).
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Received August 22, 2007.
- Revision received February 28, 2008.
- Accepted March 5, 2008.
- ©2008 American Association for Cancer Research.