|Home | About | Journals | Submit | Contact Us | Français|
An early event in the metastasis of epithelial ovarian carcinoma is shedding of cells from the primary tumor into the peritoneal cavity, followed by diffuse intra-peritoneal (i.p.) seeding of secondary lesions. Anchorage-independent metastatic cells are present as both single cells and multi-cellular aggregates (MCAs), the latter of which adhere to and disaggregate on human mesothelial cell monolayers, subsequently forming invasive foci. While this unique metastatic mechanism presents a distinct set of therapeutic challenges, factors that regulate MCA formation and dissemination have not been extensively evaluated. Proteolytic activity is important at multiple stages in i.p. metastasis, catalyzing migration through the mesothelial monolayer and invasion of the collagen-rich sub-mesothelial matrix to anchor secondary lesions, and acquisition of membrane type 1 matrix metalloproteinase (MT1-MMP; MMP-14) expression promotes a collagen-invasive phenotype in ovarian carcinoma. MT1-MMP is regulated post-translationally through multiple mechanisms including phosphorylation of its cytoplasmic tail, and the current data using ovarian cancer cells expressing wild type, phospho-mimetic (T567E-MT1-MMP) and phospho-defective (T567A-MT1-MMP) MT1-MMP show that MT1-MMP promotes MCA formation. Confluent T567E-MT1-MMP-expressing cells exhibit rapid detachment kinetics, spontaneous release as cell-cell adherent sheets concomitant with MT1-MMP-catalyzed α3 integrin ectodomain shedding, and robust MCA formation. Expansive growth within 3-dimensional collagen gels is also MT1-MMP dependent, with T567E-MT1-MMP-expressing cells exhibiting multiple collagen invasive foci. Analysis of human ovarian tumors demonstrates elevated MT1-MMP in metastases relative to paired primary tumors. These data suggest that MT1-MMP activity may be key to ovarian carcinoma metastatic success by promoting both formation and dissemination of MCAs.
Ovarian cancer is the leading cause of death from gynecologic malignancy, due primarily to complications of metastasis . Unlike most carcinomas that rely on the vasculature for metastasis, an early event in ovarian cancer dissemination is shedding of cells from the primary tumor into the peritoneal cavity, followed by diffuse “seeding” of the peritoneal cavity . This unique metastatic mechanism presents a distinct set of therapeutic challenges. Accumulation of malignant ascites is widely associated with advanced ovarian carcinoma  and it is hypothesized that ascites augments progression by facilitating the spread of cancer cells throughout the peritoneal cavity . Anchorage-independent metastatic cells are present as both single cells and multi-cellular aggregates (MCAs) that survive in suspension . The factors that regulate MCA formation have not been extensively evaluated; however evidence suggests that MCAs have a functional role in metastatic progression. MCAs adhere to and disaggregate on human mesothelial cell monolayers, subsequently forming invasive foci [6,7]. Transplanted ovarian cancer MCAs form murine xenografts with the same histopathology as the primary tumor and can be serially propagated in vivo [8,9]. Further, sphere-forming ovarian cancer initiating cells are significantly more tumorigenic in xenograft models, further demonstrating that the MCA population is a key target for anti-metastatic therapy .
Proteolytic activity is important at multiple stages in intraperitoneal metastasis, including localized proteinase-driven migration through the mesothelial monolayer and invasion of the collagen-rich sub-mesothelial matrix to anchor secondary lesions [11,12]. Invasion of collagenous matrices by ovarian cancer cells requires membrane type 1 matrix metalloproteinase (MT1-MMP, MMP-14) [13-15], a transmembrane collagenase that is not detected in normal ovarian surface epithelium or in benign ovarian tumors, but is widely expressed in ovarian carcinomas of all histotypes [15-20]. Acquisition of MT1-MMP expression promotes cell migration, extracellular matrix invasion, and growth within restricted three dimensional matrices [21-23].
Because MT1-MMP is central to a variety of biological processes, proteolytic activity is stringently controlled. MT1-MMP is internalized from the cell surface through a mechanism involving the cytoplasmic domain [24,25] and cytoplasmic tail truncation restricts MT1-MMP to the plasma membrane. The cytoplasmic domain of MT1-MMP has three potential phosphorylation sites: T567, Y573, and S577 and recent work indicates that MT1-MMP can be phosphorylated at T567 and Y573 [26-28]. T567 is localized within the sequence R563RHGT567PRRLLYCQRSLLDKV582 that has homology with the consensus sequence for protein kinase C (TXR) and ERK1/2 (XTP) . To examine the potential effect of T567 phosphorylation in the unique metastatic mechanism of ovarian carcinoma, the properties of cells expressing wild type MT1-MMP, a phospho-mimetic mutant (T567E-MT1-MMP) or a phospho-defective mutant (T-567A-MT1-MMP) were evaluated. Acquisition of MT1-MMP catalytic activity promotes rapid cell-matrix detachment kinetics concomitant with α3 integrin ectodomain shedding, enhanced MCA formation, and expansive growth in 3D collagen. This pro-metastatic phenotype was intensified in the phospho-mimetic mutant T567E-MT1-MMP, suggesting that phosphorylation of the MT1-MMP cytoplasmic tail may regulate intra-peritoneal metastatic dissemination.
DOV13 and OVCA433 cells were provided by Dr. R. Bast (Houston, TX). Anti-FLAG M2, anti-MT1-MMP (M3927), peroxidase conjugated secondary antibodies, and Protein G-Sepharose beads were from Sigma (St. Louis, MO). Super Signal-enhanced chemiluminescence (ECL) reagents were purchased from Pierce. TIMP-2 was provided by Dr. R. Fridman (Detroit, MI). Rat tail collagen type I, human type IV collagen and human fibronectin were purchased from BD Biosciences (San Diego, CA). Mouse anti-human integrin α3 (AMB1952Z and MAB2056) was purchased from Chemicon (Temecula, CA). Centriprep was purchased from Millipore (Temecula, CA).
The human MT1-MMP cDNA with C-terminal FLAG tag (DYKDDDDK) was provided by Dr. D. Pei (Minneapolis, MN). Subsequently, the T567A, T567E, and E240A point mutations were generated using quick-change (Stratagene, La Jolla, CA). Inserts were sequenced to verify mutation. Transfection of cells was done using FuGENE 6 (Roche, Germany) and stable cell lines were generated using G418 selection. Clones were routinely FACS-sorted (every 3-5 passage) using anti-Flag monoclonal M2 and pooled clones with similar expression levels were used for analyses. Untransfected parental OVCA433 cells (designated ‘control’) and OVCA433 cells transfected with empty vector (designated ‘vector’ or ‘vec’) behaved identically in all assays and have been used interchangably.
Cells were collected with lysis buffer [1% Triton X-100, 50mM Tris pH 7.5, 150 mM NaCl, 45 mM EDTA, 0.1% SDS, 20mM NaF, 10mM Na2P2O7] and protein concentration analyzed using the Bio-Rad DC detection kit (Hercules, CA). Cell lysates (50 μg) were electrophoresed on 9% SDS-PAGE, transferred to PVDF, and blocked with 3% BSA in 50 mM Trizma 300 mM NaCl, 0.2% Tween 20 (TBST). Membranes were incubated (1 h RT) with FLAG M2 monoclonal (1:1000) or anti-MMP-14 in 3% BSA/TBST. Immunoreactive bands were visualized with a peroxidase-conjugated anti-(rabbit-IgG) (1:4000 in 3% BSA/TBST) and ECL.
To detect integrin α3 in conditioned medium, 106 cells were serum starved overnight +/− the broad spectrum MMP inhibitor GM6001. Conditioned medium was collected, concentrated, then electrophoresed on 9% SDS-PAGE, transferred to PVDF, and blocked with 3% BSA/TBST. Membranes were incubated (1 h RT) with anti-integrin α3 (1:1000, MAB1952Z) in 3% BSA/TBST. Immunoreactive bands were visualized with a peroxidase-conjugated anti-(mouse-IgG) (1:4000 in 3% BSA/TBST) and ECL. For immunoprecipitation, 30ul Protein G beads were added to conditioned medium (6ml), and rotated for 2h at 4°C. Anti-integrin α3 antibody (5 ug, MAB2056) was added and samples were rotated overnight at 4°C followed by addition of Protein G beads (35 ul) for 2 h. Samples were centrifuged, washed with lysis buffer, and electrophoresed as described.
Cells (106) were treated with either monoclonal anti-Flag for MT1-MMP detection (1:100), or monoclonal anti-integrin α3 (MAB1952Z) (1:100) for 30m at 4°C. Cells were stained with secondary antibody conjugated to Alexa Fluor 488 (1:200) for 30m at 4°C, washed twice with PBS and re-suspended in 0.7 ml of PBS for analysis with Summit Software 4.3 on a Beckman Coulter FACS. Results are expressed as averages of mean fluorescence units of experiments conducted in triplicate.
MCAs were generated using a modification of the hanging drop method as previously described . MCA formation was monitored after 24-48h.
RNA was extracted from 106 cells using the SV Total RNA Isolation System (Promega). cDNA was synthesized from 5-10 μg of total RNA using iScript cDNA Synthesis Kit (BIO-RAD). Real time PCR used SYBR Green chemistry and the 7500 ABI Prizm (Applied Biosystems). Reaction setup, normalization, and primers were used as previously described and fold-changes were quantified as 2-(ΔCt sample - ΔCt control) .
Cells (106) were seeded onto 60mm dishes and allowed to adhere overnight. Adherent cells were subjected to controlled trypsinization (0.25% Trypsin/2.21 mM EDTA in HBSS without sodium bicarbonate, calcium & magnesium) and both adherent and detached cells were quantified. Experiments were repeated 3-5 times.
Chambers were coated with 10ug/ml type I collagen, type IV collagen, or fibronectin in PBS for 4h at 37°, blocked with 3% BSA in MEM for 1h at 37°C, washed with PBS, and air dried. Cells were seeded at 105/well, allowed to adhere for 75 min (determined empirically based on a time course of 0-2h), washed with PBS to remove non-adherent cells, fixed, stained using Diff-Quick, and enumerated. Assays were performed in triplicate.
3D cultures were prepared as described  by diluting type I rat tail collagen with 10 × MEM to 1.5 mg/ml. Cells (104) were added to the collagen mixture prior to solidification +/− TIMP-2 (5ug/ml). In additional controls, cells were added atop solidified gels.
Immunohistochemical analysis was done retrospectively on tumor tissue microarrays (TMA) prepared with IRB approval by the Pathology Core Facility, Northwestern University. TMAs included 17 paired primary and metastatic ovarian cancer tissues obtained during the same surgical procedure from patients who were not treated against stage III,IV ovarian cancer before the operation (15 serous, 2 endometroid). Immunohistochemical staining used antibody to MT1-MMP (clone RB1544B 1:100 dilution, Neomarkers, Kalamazoo, MI)  with breast carcinoma tissue as a positive control. Scoring was assigned according to the average overall intensity of the staining: 0, no staining; 1, fine granular staining; 2, somewhat coarse staining, but less than positive control tissue (breast carcinoma); 3, very coarse staining, similar or greater than positive control tissue. Staining <10% of tumor cells, regardless of intensity, was considered negative. Staining of between 10-75% of tumor cells was considered focal positive, and staining of greater than 75% of tumor cells was considered diffuse positive.
Based on previous studies showing that phosphorylation of cytoplasmic tail residues in MT1-MMP can modify cell behavior [26-28], MT1-MMP mutants were generated in which T567 was mutated to glutamic acid (T567E) to mimic constitutive phosphorylation, or alanine (T567A) to represent a phospho-deficient mutation. Constructs were transfected into OVCA433 cells, chosen because of the lack of endogenous MT1-MMP expression. Analysis of MT1-MMP in whole cell lysates obtained from wild type and phospho-mutant cell lines showed equivalent expression of wild type and mutant MT1-MMP (suppl. Fig. 1A). Relative to cells expressing wild type MT1-MMP, gelatin zymography indicated no significant change in proMMP-2 activation by T567A-MT1-MMP or T567E-MT1-MMP mutant cell lines [16, 21, 32-35] (suppl. Fig. 1B). The cytoplasmic tail alters the surface presentation of active MT1-MMP by regulating its internalization from the cell surface [24,25], the functional consequences of which may impact net proteolytic activity. Because the specific sequence(s) in the cytoplasmic tail responsible for modulating internalization have yet to be delineated, the effect of T567 mutation on MT1-MMP surface presentation was evaluated. Relative to cells expressing wild type MT1-MMP, T567A- and T567E-MT1-MMP-expressing cells exhibit similar basal levels of surface expression (suppl. Fig. 1C). Further, expression levels of exogenous MT1-MMP in OVCA433 stable transfectants are similar to endogenous MT1-MMP levels in DOV13 cells (suppl. 1D,E). Potential changes in the relative distribution of MT1-MMP mutants within specific membrane microdomains or in rates of internalization in response to specific stimuli were not evaluated.
Initial dissemination of ovarian cancer involves exfoliation of cells from the primary ovarian tumor into the peritoneal cavity as matrix-detached single cells and MCAs [5,6]. Malignant ascites accumulates in advanced ovarian carcinoma and contains a population of non-adherent MCAs ranging in size from 30-200 um . These cells survive anoikis  and proliferate as a free-floating population of highly neoplastic cells ; however molecular mechanisms regulating the genesis of MCAs have not been extensively evaluated. Exploratory cDNA microarray analysis of changes in gene expression induced by MCA culture in DOV13 cells, which express endogenous MT1-MMP [13,15], identified MT1-MMP as one of <100 genes significantly upregulated in MCA culture (M. Barbolina and S. Stack, unpublished results). Validation using qPCR and western blotting confirmed these results, showing a 2.1-fold increase in MT1-MMP in MCAs relative to 2-D cultures (Fig. 1A, B). To evaluate the effect of acquisition of MT1-MMP expression on MCA formation, OVCA433 cells expressing wild-type or mutant MT1-MMP were evaluated relative to DOV13 using the hanging drop method  to generate MCAs in vitro (Fig. 1C). Cells expressing MT1-MMP formed larger diameter MCAs relative to untransfected control cells (Fig. 1D). Interestingly, smaller MCAs were formed by cells expressing the phospho-defective T567A-MT1-MMP mutant relative to those expressing wild type or T567E phospho-mimetic mutant MT1-MMP, suggesting that phosphorylation at this site may modulate spheroid size and overall growth.
Although MCAs are prevalent in ascites obtained from women with ovarian carcinoma [6,7,36,39], it is not known whether cellular aggregation occurs prior to or following detachment of metastatic cells from the primary tumor. Microscopic examination of cell cultures expressing T567E-MT1-MMP revealed a striking phenotype, showing spontaneous detachment as cell-cell adherent MCAs (Fig. 2A). To further characterize this phenotype, relative adhesive strength was evaluated using a controlled trypsinization assay to monitor the kinetics of cell-matrix dissociation. In sub-confluent cultures, detachment kinetics were similar in cells expressing wild type-, T567A- or T567E-MT1-MMP (Fig. 2B). Upon confluence, cells expressing T567E-MT1-MMP exhibit a more complete and accelerated detachment relative to cells expressing wild type- or T567A-MT1-MMP (Fig. 2C). Cells lacking MT1-MMP expression (vector controls) are strongly adherent and detach with extended kinetics (>20 min) under these conditions (not shown). DOV13 cells do not exhibit spontaneous detachment; however it should be noted that these cells also express high levels of the endogenous MMP inhibitor TIMP-2 [13,15]. To determine whether MT1-MMP catalytic activity was required for aggregate detachment, the active site mutation E240A was introduced into the T567E background to generate T567E/E240A-MT1-MMP-expressing cells. Detachment kinetics of T567E/E240A-MT1-MMP cells were significantly attenuated relative to cells expressing the catalytically active phospho-mimetic mutant (Fig. 2D). Together these results suggest that cell-cell communication regulates cell-matrix adhesion via the activity of MT1-MMP.
Reciprocal interplay between integrin and cadherin function has been reported [40,41] and the results described above suggest that cadherin-engaged, confluent cells lose integrin function. Relative to phospho-deficient T567A cells, expression of both wild-type- and T567E-MT1-MMP significantly impaired cell adhesion to both type I collagen (Fig. 3A) and type IV collagen (suppl. Fig. 2A). In contrast, adhesion to fibronectin was unaffected by MT1-MMP expression (suppl. Fig. 2B), suggesting a specific defect in the expression and/or function of collagen-binding integrins. As detachment of cell-cell adherent sheets of T567E-MT1-MMP-expressing cells occurs upon confluence (Fig. 2A), integrin profiles were compared under these conditions. A significant reduction in levels of cell surface α3 integrin was observed in confluent cultures of T567E-MT1-MMP-expressing cells (Fig. 3B, inset, red trace) relative to cells expressing wild type (Fig. 3B inset black trace) and T567A-MT1-MMP (Fig. 3B inset, blue trace). Vector control cells maintain α3 expression at levels seen in subconfluent cells (not shown). To evaluate a potential role for MT1-MMP activity in this process, cells were cultured with GM6001 to block MT1-MMP catalytic activity. Inhibition of MMP catalytic activity rescues α3 integrin surface expression in confluent cells expressing wild type- or T567E-MT1-MMP (Fig. 3B). In control experiments, surface levels of α3 integrin were similar in subconfluent cells expressing wild type-, T567A-, or T567E-MT1-MMP (suppl. Fig. 3A). Additional controls demonstrate no loss of surface levels of α2 integrin (suppl. Fig. 3B), β1 integrin (suppl. Fig. 3C), E-cadherin (suppl. Fig. 3D), transferrin receptor or integrins αv, α5 and α6 (not shown) in subconfluent or confluent cells in the presence or absence of GM6001, suggesting that loss of α3 integrin expression does not represent non-specific loss of cell surface protein.
The ability of GM6001 to rescue surface expression of integrin α3 indicates that the observed reduction in surface presentation may be the result of an MT1-MMP-catalyzed cleavage event. To explore this possibility, conditioned medium from confluent T567E-MT1-MMP-expressing cells was concentrated and analyzed for the presence of polypeptides that cross-react with anti-α3 integrin antibodies. Western blot analysis of concentrated conditioned medium identified a protein of approximately 70.8 kDa and this species was undetectable in the medium of cells cultured overnight with GM6001 (Fig. 3C, compare lanes 2 and 3). Similar results were obtained following immunoprecipitation of non-concentrated conditioned medium with anti-α3 integrin antibodies followed by western blotting for α3 integrin (Fig. 3C, lanes 4,5). Impaired collagen adhesion, spontaneous detachment, decreased surface α3 integrin levels and α3 integrin ectodomain shedding were not observed in cells expressing the catalytically inactive T567E/E240A-MT1-MMP (Fig. 3D and data not shown). Together these data support a mechanism for loss of adhesion subsequent to MT1-MMP-catalyzed α3 integrin ectodomain shedding, particularly evident in phospho-mimetic T567E-MT1-MMP-expressing cells.
Successful intra-peritoneal metastasis requires proliferation within the confines of the interstitial collagen-rich submesothelial matrix to establish collagen-anchored intra-peritoneal secondary lesions [11,42,43]. MT1-MMP is necessary for cell proliferation in a 3D collagen network, functioning to remove cytoskeletal constraints necessary to drive a proliferative response . To examine the contribution of MT1-MMP activity to growth of ovarian cancer cells in 3D collagen, cells were seeded at single cell density in 3D collagen gels and were qualitatively assessed for growth after 6 days . Large multi-cellular aggregate structures were generated by all MT1-MMP-expressing cells (Fig. 4A and suppl. Table 1). Proliferation was significantly attenuated in vector-transfected control cells (not shown) or by copolymerization of tissue inhibitor of metalloproteinases-2 (TIMP-2) within collagen gels (Fig. 4B), verifying the dependence of growth in 3D collagen on MT1-MMP activity. DOV13 cells were also highly proliferative within 3D collagen gels, forming clusters of cells averaging 2.43 +/− 0.65 × 105 cells/cluster. Although overall structures generated by all MT1-MMP-expressing cells were similar in size (suppl. Table 1), multi-cellular aggregates generated by T567A-MT1-MMP-expressing cells were more spherical (length/width ratio of 1.47) and lacked distinct foci of collagen invasion (Fig. 4A left panels; suppl. Table 1). Cells expressing wild type MT1-MMP produced more elongated structures with small invasive foci (Fig. 4A middle panels; suppl. Table 1). In contrast, cells expressing the phospho-mimetic mutant T567E-MT1-MMP grew as multi-cellular prolate ellipsoid aggregates (length/width ratio of 2.16) with multiple large invasive foci (Fig. 4A right panels; suppl. Table 1). Quantative analysis of invasive projections indicates that both projection number and projection length were significantly increased relative to cells expressing wild type MT1-MM (suppl. Table 1). These results support the hypothesis that the phosphorylation status of MT1-MMP cytoplasmic residue T567 may regulate invasive growth within the collagen-rich microenvironment of the sub-mesothelial matrix. Furthermore, the data indicate that the ability of ovarian cancer cells to survive long-term and proliferate in 3D collagen is enhanced by MT1-MMP activity, as proliferation is blocked by inclusion of TIMP-2 within the 3D collagen gels (Fig. 4B). This is supported by an immunohistochemical analysis of MT1-MMP expression in sets of paired primary ovarian tumors and peritoneal metastatic lesions derived from the same patient (n=17; suppl. Table 2, Fig. 5A-C). Although the dataset is not sufficiently powered for rigorous statistical analysis, the data support a trend wherein 76% (13/17) show sustained or increased high level (2+,3+) MT1-MMP expression in metastases. A further 12% (2/17) maintain low level (1+) expression in metastases, while only 12% (2/17) show decreased expression relative to the primary tumor. It is interesting to note that MT1-MMP is expressed in 100% of the metastatic lesions overall (suppl. Table 2).
Reversible phosphorylation is widely recognized as a key post-translational modification that regulates protein function and a growing body of work suggests that MT1-MMP action may be altered through phosphorylation of cytoplasmic tail residues [26-28]. For example, src- or EGF-dependent phosphorylation of cytoplasmic residue Y573 has been shown to modulate cell migration and invasion [26,28], suggesting that the intracellular domain of MT1-MMP may be vital to “inside-out” signaling processes. The observation that several cell surface proteins undergo phosphorylation at multiple sites underscores our observations implicating T567 as a second site for post-translational modification . While regulatory mechanisms for the control of T567 phosphorylation have yet to be elucidated, results in the present study highlight a pivotal role for the cytoplasmic tail in regulating MT1-MMP function and associated cellular phenotypes relevant to metastasis of ovarian carcinoma.
The majority of women with ovarian cancer are initially diagnosed with disseminated intra-abdominal disease [2,3], indicating that a more detailed understanding of the cellular and biological factors that promote successful metastatic dissemination can ultimately improve patient survival. Unlike other solid tumors, hematogenous dissemination of ovarian cancer cells is uncommon, as metastasis proceeds primarily through exfoliation of cells from the primary tumor into the peritoneal cavity as free floating cells and MCAs. This imposes challenges to tumor cell survival because these metastatic cells must escape anoikis, and it has been speculated that MCA formation may function to promote anchorage-independent growth . Although originally considered a non-adhesive subset of ovarian tumor cells, recent data demonstrate that human ovarian cancer ascites-derived MCAs adhere to and invade mesothelial monolayers and can thereby contribute to intra-peritoneal implantation and metastasis [6,7,12]. Further, MCAs that survive in ascites may generate a subpopulation of highly neoplastic cells . This is supported by recent data demonstrating that cells isolated from murine ascites are more aggressive than parental cells when re-injected in vivo . Gene expression profiling shows clustering of MCA expression profiles with tumor xenograft patterns, rather than with monolayer cells [9,44], suggesting that MCAs represent a more advanced stage of malignancy. Self-renewing spheroid forming cells isolated from ovarian primary tumors show increased tumorigenicity and are associated with metastasis to omentum and colon . Together these data support the hypothesis that the MCA population in human ovarian ascites may be a primary source of intra-peritoneal metastases and thereby represents a key target for anti-metastatic therapy.
Although recent studies highlight the importance of MCAs in ovarian cancer pathobiology, the process of self-assembly of ovarian tumor cells into microtissues such as MCAs has not been extensively evaluated. Further, it is unknown whether the temporal sequence of events in MCA formation involves shedding of individual tumor cells which then aggregate in ascites versus exfoliation of multi-cellular tumor cell sheets that subsequently reorganize into MCAs. The current data suggest that acquisition of MT1-MMP activity would promote MCA formation by either mechanism. Cells expressing endogenous MT1-MMP (DOV13) readily aggregate from single cells into MCAs, and qPCR and western blotting analyses confirm elevated expression of MT1-MMP in MCA cultures. Alternatively, OVCA433 cells transfected with wild-type- or T567E-MT1-MMP exhibit rapid detachment kinetics and sheet-like exfoliation as cell-cell adherent aggregates. Interestingly, this striking detachment phenotype is not manifested in sub-confluent cultures, suggesting the potential contribution of cell-cell adhesion molecules to the regulation of cell-matrix adhesion. It is interesting to speculate that acquisition of MT1-MMP expression by primary tumor cells may promote metastasis via enhanced tumor cell shedding. In support of this hypothesis, previous immunohistochemical and in situ hybridization analyses show MT1-MMP expression in 78-100% of primary ovarian tumors [15,18]. MT1-MMP expression is also detected in cancer cells obtained from malignant effusions  and in 88-98% of peritoneal metastases, wherein expression is correlated with poor outcome [current study, 17].
Detachment of cell-cell adherent sheets bears similarities to cohort migration . The ability of protease inhibitors to impede this process reinforces the role for MMP activity in establishing this phenotype. This is further supported by results from the current study showing that the catalytically inactive mutant T567E/E240A-MT1-MMP did not induce cell detachment and sheet-like exfoliation. Our data suggest that MT1-MMP catalytic activity contributes mechanistically to cell detachment via catalysis of α3 integrin ectodomain shedding, as cell adhesion was restored and soluble integrin ectodomain was not detectable in the presence of a broad spectrum MMP inhibitor or with the catalytically inactive mutant. It is plausible that shedding of cells as multi-cellular masses may represent a more efficient means of dissemination. In addition, the ability of cells to function collectively may allow the mass to produce high levels of matrix proteases that promote migration and invasion at secondary sites .
Metastasizing ovarian cancer cells encounter an interstitial collagen-rich microenvironment, as the sub-mesothelial matrix is comprised primarily of types I and III collagen [12,46-50]. Acquisition of MT1-MMP collagenolytic activity may be key to metastatic success, as MT1-MMP activity is required for invasion of 3D collagen gels by ovarian cancer cells . Furthermore, MT1-MMP collagenolysis is necessary to remove matrix barriers to allow for the cytoskeletal reorganization necessary to drive cellular proliferation . This is supported by results of the current study showing lack of proliferation within 3D collagen gels in the absence of MT1-MMP expression or in gels containing TIMP-2. While all MT1-MMP-expressing cultures formed proliferative colonies in 3D collagen, cells expressing the phosphomimetic T567E-MT1-MMP construct exhibited more invasive patterns of growth. Thus, it is interesting to speculate that both the presence and the phosphorylation status of MT1-MMP may control metastatic success. Metastatic disease is the main cause of death for women with epithelial ovarian carcinoma, as disseminated tumor cells attach to abdominal surfaces, anchor, and grow multiple secondary lesions that disrupt the function of essential organs [1,2]. A molecular-level understanding of metastasis is necessary for the development of therapies to inhibit intra-peritoneal spread and thereby improve the long-term survival of thousands of women with ovarian cancer. To this end, a more detailed understanding of what regulates the transition from primary tumor to free-floating MCA to life-threatening peritoneally-anchored metastatic lesion may provide novel insight necessary to target intra-peritoneal therapies to appropriate multi-cellular populations.
This work was supported by NIH/NCI Grants CA86984 (M.S.S.), CA86984-S1 (N.M.M.) and CA109545 (M.S.S.), and the Illinois Department of Public Health Penny Severns Breast, Cervical and Ovarian Cancer Research Fund (M.V.B.). The authors would like to thank Dr. Brian Adley, Advocate Lutheran General Hospital, Chicago, IL, for scoring of immunohistochemical stains.