Epidermal Growth Factor Receptor–Mediated Membrane Type 1 Matrix Metalloproteinase Endocytosis Regulates the Transition between Invasive versus Expansive Growth of Ovarian Carcinoma Cells in Three-Dimensional Collagen

The epidermal growth factor receptor (EGFR) is overexpressed in ovarian carcinomas and promotes cellular responses that contribute to ovarian cancer pathobiology. In addition to modulation of mitogenic and motogenic behavior, emerging data identify EGFR activation as a novel mechanism for rapid modification of the cell surface proteome. The transmembrane collagenase membrane type 1 matrix metalloproteinase (MT1-MMP, MMP-14) is a major contributor to pericelluar proteolysis in the ovarian carcinoma microenvironment and is subjected to extensive posttranslational regulation. In the present study, the contribution of EGFR activation to control of MT1-MMP cell surface dynamics was investigated. Unstimulated ovarian cancer cells display caveolar colocalization of EGFR and MT1-MMP, whereas EGFR activation prompts internalization via distinct endocytic pathways. EGF treatment results in phosphorylation of the MT1-MMP cytoplasmic tail, and cells expressing a tyrosine mutated form of MT1-MMP (MT1-MMP-Y⁵⁷³F) exhibit defective MT1-MMP internalization. As a result of sustained cell surface MT1-MMP activity, a phenotypic epithelial-mesenchymal transition is observed, characterized by enhanced migration and collagen invasion, whereas growth within three-dimensional collagen gels is inhibited. These data support an EGFR-dependent mechanism for regulation of the transition between invasive and expansive growth of ovarian carcinoma cells via modulation of MT1-MMP cell surface dynamics. (Mol Cancer Res 2009;7(6):809–20)

The epidermal growth factor (EGF) receptor (EGFR) plays a prominent role in both physiologic and pathologic processes. In a ligand-dependent manner, the EGFR mediates the effects of EGF, transforming growth factor-α, and amphiregulin (1-4). The classic function of the EGFR is that of a receptor tyrosine kinase that catalyzes the phosphorylation of multiple substrates and downstream effector molecules at key tyrosine residues and thereby activates intracellular signaling networks related to cell growth and survival (4, 5), including those involving Src family kinases (6, 7), Ras/mitogen-activated protein kinase (8, 9), and Jak/Stat (10-12). The functional consequences of EGFR-mediated signal transduction include enhanced proliferation, migration, and cellular differentiation (6, 7, 13, 14).

The ability of the EGFR to promote cellular responses that contribute to cancer pathology underscores its overexpression in the vast majority of carcinomas (15-19). EGFR is expressed in 10% to 70% of ovarian cancer cases with an average reported expression of 48% (4, 20). In the normal ovary, the EGFR is preferentially expressed in actively proliferating cells, including the ovarian surface epithelium (4, 21-23). Tumors arising from the ovarian epithelium account for over 90% of ovarian malignancies. Although women diagnosed with early stage disease have a favorable outcome, detection after secondary metastasis beyond the ovaries results in poor patient prognosis. Evidence suggests that EGFR expression may be an early event in ovarian cancer development as well as in progression and metastasis (4, 24-26). Activated (phosphorylated) EGFR is found in 35% of primary ovarian cancers and is often elevated in peritoneal metastases relative to primary tumors from the same patient (27, 28). A recent meta-analysis revealed a relationship between EGFR status and decreased survival (29), supporting the conclusion that aberrant EGFR status is a factor in ovarian cancer outcome. With an estimated 15,000 mortalities expected each year (National Cancer Institute), ovarian cancer is the leading cause of death from gynecologic malignancy among women in the United States. As abrogation of EGFR signaling retards tumor progression (30, 31), a more detailed understanding of cellular events mitigated by this receptor is warranted.

Emerging data support a link between EGFR activation and the regulation of matrix-degrading proteinases in ovarian carcinoma. Metastatic ovarian carcinoma cells require proteolytic activity for invasion of the mesothelial monolayer, i.p. anchoring in the submesothelial interstitial collagen (types I and III) stroma, and for tumor angiogenesis (32-37). Matrix metalloproteinases (MMP) facilitate matrix remodeling and play a key role in ovarian cancer pathobiology (38), as ovarian carcinomas display high levels of MT1-MMP (MMP-14), MMP-2, MMP-9, and tissue inhibitor of metalloproteinase (TIMP)-2 (27, 32, 33, 39-43). MT1-MMP is essential to matrix remodeling during physiologic processes (44), and key to acquisition of a metastatic phenotype in a variety of tumor cells (45-48). Acquisition of MT1-MMP expression promotes cell migration (49, 50), invasion of three-dimensional collagen (41, 48, 51-53), and three-dimensional growth (53), key cellular processes that drive ovarian pathology. Consequently, high levels of MT1-MMP correlate with poor survival of women with ovarian cancer (39). Given the central role of MT1-MMP in cancer progression and metastasis, multiple mechanisms have evolved for transcriptional and posttranslational regulation of MT1-MMP enzymatic activity. EGFR activation has been shown to induce MT1-MMP expression at both mRNA and protein levels, and lack of the EGFR is correlated with decreased MT1-MMP function (42, 54-56). Conversely, recent work shows a role for MT1-MMP in the trans-activation of the EGFR, suggesting potential mechanisms for reciprocal regulation (57).

Although there is an expansive body of literature outlining the central role of the EGFR in the architecture of intracellular signaling networks, less in known about its ability to regulate transmembrane protein trafficking. EGFR signaling may modify the molecular landscape of the cell surface by regulating the trafficking, and consequently, the function of specific cell surface proteins, highlighting an underappreciated outcome of EGFR activation. In the present study, the role of EGFR activation in MT1-MMP membrane trafficking was investigated. The current data support an EGFR-dependent mechanism for regulation of the transition between invasive and expansive growth of ovarian carcinoma cells via modulation of MT1-MMP cell surface dynamics.

Internalization of MT1-MMP is recognized as a mechanism to regulate surface expression and, as a result, pericellular proteolytic activity. This process has been shown to occur through both clathrin- and caveolae-dependent processes in response to temperature shift or stimulation with concanavilin A (58-62). Although physiologic mechanisms that regulate MT1-MMP trafficking have not been characterized, recent data show an EGF-induced change in MT1-MMP–dependent pericellular proteolytic activity (63). To evaluate the effect of EGF stimulation on MT1-MMP trafficking, OVCA 433 cells expressing GFP-tagged MT1-MMP were treated with 25 nm EGF and monitored via live cell imaging. Images were acquired at 1-minute intervals over 45 minutes. No evidence for enhanced MT1-MMP exocytosis was observed in this time frame. MT1-MMP is rapidly internalized from the cell surface after EGF stimulation (Supplementary Fig. S1). In unstimulated cells, MT1-MMP colocalization with EGFR is observed (Fig. 1A), whereas ligand stimulation results in dissociation of MT1-MMP and EGFR fluorescent signals (Fig. 1A). These data suggest that MT1-MMP is internalized via a pathway distinct from the EGFR. Previous studies have shown that before internalization, EGFRs are concentrated in caveolae (Supplementary Fig. S2A; refs. 64-67). After activation, EGFR dimers migrate to the bulk plasma membrane (Supplementary Fig. S2B; refs. 55, 66, 67) and subsequently localize to clathrin-coated pits, initiating endocytosis from the cell surface (1, 68, 69). In contrast, MT1-MMP immunoreactivity is colocalized with caveolin-1 both before and after EGF stimulation (Fig. 1B) and endocytosed MT1-MMP colocalizes with the caveosomal markers 58K and GRP94 but not Rab7 (Fig. 1C). This is supported by quantitative analysis of surface protein levels after EGFR activation using fluorescence-activated cell sorting (FACS). A rapid loss of cell surface EGFR is observed after ligand activation (Fig. 2A, closed circles). The kinetics of EGFR turnover are not altered by pretreatment of cells with methyl-β-cyclodextrin (Fig. 2A, open circles), which causes cholesterol depletion, and as a result, caveolar and lipid raft disruption (70, 71). Previous studies show that methyl-β-cyclodextrin treatment shifts caveolar EGFR to noncaveolar membrane (72). Thus, although cholesterol depletion affects the structure of caveolae, it does not influence the ability of EGFR to exit caveolae and undergo endocytosis in response to EGF treatment. In contrast to EGFR, endocytosis of MT1-MMP is dependent on caveolar integrity. Pretreatment of cells with methyl-β-cyclodextrin disrupts EGF-induced MT1-MMP endocytosis (Fig. 2B, open circles), resulting in accumulation of MT1-MMP on the cell surface. Together, these data support a caveolae-dependent mechanism for EGF-induced MT1-MMP endocytosis in ovarian cancer cells.

The cytoplasmic tail of MT1-MMP undergoes reversible Src-dependent phosphorylation at tyrosine 573 (Y⁵⁷³) in response to the platelet-derived chemoattractant sphingosine-1-phosphate (73, 74). As EGFR downstream signaling can also activate the nonreceptor tyrosine kinase Src, the contribution of Y⁵⁷³ phosphorylation to MT1-MMP trafficking was evaluated. Using site-directed mutagenesis, Y⁵⁷³ in wild-type GFP-tagged MT1-MMP was converted to the nonphosphorylatable homologue Phe to generate GFP-tagged MT1-MMP-Y⁵⁷³F. After transfection and G418 selection, stable cell populations of OvCa433 cells expressing either wild-type-Y⁵⁷³F or MT1-MMP-Y⁵⁷³F were isolated by FACS and shown to have equivalent MT1-MMP total and surface expression under basal conditions (Fig. 3A). Whereas treatment of serum-starved cells with EGF induced phosphorylation of wild-type MT1-MMP, phospho-Tyr signal was not detected in cells expressing MT1-MMP-Y⁵⁷³F (Fig. 3B). Altered trafficking of MT1-MMP-Y⁵⁷³F in response to EGF was also observed. Whereas ∼30% of wild-type MT1-MMP is internalized in response to EGF (Fig. 2B), no internalization of the mutant was detected after EGF treatment in either control or methyl-β-cyclodextrin–treated cells (Fig. 2C). Persistence of cell surface localization of MT1-MMP-Y⁵⁷³F was also shown by confocal microscopy, as after 30 minutes of EGFR stimulation, little detectable loss of surface MT1-MMP-Y⁵⁷³F staining was observed (Fig. 3C). Lack of significant changes in colocalization with endocytic markers after EGF stimulation (Fig. 3D) further underscores the role of Y⁵⁷³ phosphorylation in mediating internalization of MT1-MMP in response to EGF treatment.

Ovarian cancer metastasis is characterized by adhesion to and invasion of the type I collagen-rich submesothelial membrane, followed by proliferation within the interstitial collagenous matrix to anchor secondary lesions (35, 75, 76). Both in vivo and in vitro studies indicate that expression of MT1-MMP facilitates tumor cell growth within a three-dimensional matrix (45, 52, 53). In control experiments, analysis of cell growth atop of collagen surfaces illustrated similar growth rates for cells expressing either wild-type-Y⁵⁷³F or MT1-MMP-Y⁵⁷³F (Fig. 4A-C). To determine whether the Y⁵⁷³F modification exerts an effect on MT1-MMP–mediated three-dimensional growth, cells were seeded within three-dimensional type I collagen gels and cultured for 6 days before quantitation of cell proliferation. As previously reported (53), proliferation in three-dimensional collagen gels is largely attenuated in the absence of MT1-MMP expression or in the presence of MMP inhibitors. However, cells expressing wild-type MT1-MMP exhibited reproducibly enhanced proliferation within a three-dimensional matrix relative to MT1-MMP-Y⁵⁷³F cells (Fig. 5A and C) relative to vector-transfected cells or cells in matrices containing the metalloproteinase inhibitor TIMP-2 (data not shown). These cells grew as expansive, well-organized multicellular aggregate structures with readily apparent cell-cell interactions (Fig. 5A, left; Fig. 5C). However, in contrast, striking morphologic differences were exhibited by MT1-MMP-Y⁵⁷³F cells, which survived in three-dimensional collagen gels, but grew as elongated, single-cell entities with a mesenchymal phenotype (Fig. 5A, right; Fig. 5B). Enhanced invasion of three-dimensional collagen gels (Fig. 5D) and chemotactic migration (data not shown) was evident with MT1-MMP-Y⁵⁷³F–expressing cells, consistent with the observed morphologic alterations characteristic of a migratory, invasive phenotype.

As both MT1-MMP and the EGFR are overexpressed in epithelial ovarian carcinoma (4, 27, 51), elucidating the role of EGFR signal transduction in modulating MT1-MMP activity bears potential clinical significance. A growing body of work shows that acquisition of MT1-MMP expression alone promotes enhanced cell migration (49, 50), invasion of the basement membrane or three-dimensional collagen gels (41, 51-53, 77), and three-dimensional expansive growth (53). Studies done in multiple cell lines indicate that activation of the EGFR induces MT1-MMP protein expression (54-56). Although the ability to augment protein levels offers an effective means of influencing molecular activity, altering the surface presentation of the transmembrane proteinase MT1-MMP represents an alternative efficient regulatory mechanism for rapid changes in pericellular proteolytic potential.

Results in the present study show that activation of the EGFR regulates MT1-MMP cell surface topology, leading to internalization of MT1-MMP. Although the EGFR and MT1-MMP colocalize with caveolae before activation, stimulation with EGF disrupts this association. Although the exact function of caveolae in cell signaling is not well-defined, emerging lines of evidence indicate that the caveolar structure functions in the compartmentalization of receptors and second messengers involved in signal transduction (66, 78-81). Caveolin-1 polymerizes to form a scaffold within these compartments (78), promoting direct interaction with many of the receptors and signaling proteins that reside in caveolae (82). Caveolar localization may thereby facilitate signaling by placing the appropriate receptors and effectors at optimal proximity to elicit efficient intermolecular action. This is underscored by the observation that although caveolae comprise only 5% to 10% of the plasma membrane, 40% to 60% of unstimulated EGFR is concentrated in these domains (66). In addition to the EGFR (66), nerve growth factor receptors (83), vascular EGFR (84), several components of the MAP kinase signaling pathway, including extracellular signal-regulated kinase and Ras (85, 86), protein kinase C (81, 87), and Src kinases (81, 87) are present in caveolae. Although this is not a comprehensive list, the compartmentalization of entire signaling units supports a role for caveolae as mediators in cell signaling.

Although cytoplasmic levels of both MT1-MMP and the EGFR increase after EGF treatment, intracellular colocalization was not detected. After activation, punctuate staining of the EGFR is observed with clathrin and the recycling endosome Rab7 (67), both markers of clathrin-dependent internalization. This is in agreement with previous data showing that activated EGFR dimers migrate to the bulk plasma membrane (5, 66) and subsequently localize to clathrin-coated pits, initiating endocytosis from the cell surface (1, 68, 69). Intracellular vesicles are eventually subject to either lysosomal degradation or recycling to the cell surface (69). In contrast, internalized MT1-MMP did not colocalize with either clathrin or Rab7, but was detected in association with GRP94 and 58K, both markers of caveolae-dependent trafficking. These findings indicate that intracellular trafficking of the EGFR and MT1-MMP occurs through two distinct mechanisms.

To our knowledge, analysis of MT1-MMP internalization in response to a biological stimulus such as EGF has not been reported. However, previous studies have evaluated MT1-MMP endocytosis in response to temperature shift or concanavilin A treatment. It has been shown that MT1-MMP internalization from the surface of HT1080 fibrosarcoma cells requires dynamin (58), a GTPase required for vesicle formation from the plasma membrane during clathrin-mediated endocytosis (88). Abrogation of MT1-MMP internalization in cells expressing a dominant-negative form of dynamin (dynK44A) supported a role for clathrin-mediated endocytosis in MT1-MMP trafficking induced by treatment with concanavalin A (48). Conversely, caveolae-mediated endocytosis was also implicated as a distinct, clathrin-independent mechanism of MT1-MMP internalization in response to temperature shift (89). A physical association between MT1-MMP and caveolae has previously been shown, as MT1-MMP preferentially localizes to caveolae at the surface of HT1080 and glioblastoma cells (60, 61). Furthermore, using the mutant EpsE_95 to specifically block clathrin-mediated endocytosis, internalization of MT1-MMP was retained in HT1080 cells (62). It is unknown whether mutation of Tyr573 would affect MT1-MMP trafficking by these diverse mechanisms. However, the involvement of separate mechanisms for endocytosis in different cell types suggests that internalized MT1-MMP may be subject to a differential fate that is dependent on the mode of internalization.

Although the observed effect of EGFR on MT1-MMP surface presentation has not been previously reported, EGFR activation has been shown to regulate the surface receptor integrin α2, which also localizes at caveolae, in a similar fashion (67). The ability of EGFR activation to affect the trafficking of multiple receptors that occupy caveolae suggests that the EGFR may regulate internalization of the cellular compartment itself, and as a result, alter the surface profiles of proteins occupying these domains. Although the molecular mechanism underlying caveolae function has yet to be determined, several lines of data indicate that the EGFR may function in cell signaling before exiting caveolae. The first phases of EGFR signal transduction occur in caveolae, suggesting that the entire signaling pathway is preorganized in these compartments (80). It is clear that after activation, the EGFR exits caveolae through a process that requires the activity of both the activated EGFR and Src kinases (66, 81). Although endocytosis of EGFR does not occur through caveolae, it has been shown that tyrosine kinase activity mediated by Src drives caveolae internalization (81). Caveolin-1 may act as a negative regulator of endocytosis mediated through this pathway (90, 91). Caveolin-1 functions as a structural support for highly immobile caveolae (92). The ability of caveolin-1 to inhibit caveolae budding may be overcome by reversible phosphorylation (93), potentially initiated by EGFR-induced Src activation (94). It is interesting to speculate that this functions as a mechanism to initiate endocytosis of caveolar proteins including MT1-MMP. The ability of the EGFR to modify caveolin-1 via reversible tyrosine phosphorylation (93, 94) also raises the possibility that the cytoplasmic domain of MT1-MMP can serve as a substrate for kinase activity. It has been previously reported that MT1-MMP cytoplasmic tail residue Y⁵⁷³ undergoes reversible phosphorylation in response to sphingosine-1-phosphate (95). This modification is catalyzed by Src kinase, which can be activated downstream of EGFR signaling, raising the hypothesis that EGFR can stimulate, either directly or indirectly, tyrosine phosphorylation of MT1-MMP. This is supported by data in the current study showing EGF-induced phopshorylation of wild-type but not Y⁵⁷³F-mutated MT1-MMP as well as defective EGF-stimulated internalization kinetics in cells expressing MT1-MMP-Y⁵⁷³F.

To determine the functional consequences of prolonged surface presentation, ovarian cancer cells expressing MT1-MMP-Y⁵⁷³F were evaluated for the ability to invade through and proliferate within three-dimensional collagen gels. After shedding from the surface of the ovary, metastatic cells anchor, migrate, and invade the mesothelium lining the organs of the peritoneum. Subsequently, ovarian cancer cells encounter the submesothelial matrix, which is rich in type I collagen, and proliferate within this matrix to anchor secondary lesions (35, 41, 51, 75). MT1-MMP-Y⁵⁷³F–expressing cells exhibited enhanced invasion through type I collagen, and superior chemotactic migration relative to cells expressing wild-type MT1-MMP. Similar results were obtained using MT1-MMP-Y⁵⁷³F–transfected MDA-MB-231 mammary carcinoma cells.⁵

As localization of MT1-MMP at invadopodia and cellular constriction rings is required for efficient cell invasion (96-100), it may be proposed that tyrosine phosphorylation of the MT1-MMP cytoplasmic tail abrogates its ability to properly localize within the migratory apparatus, resulting in impaired migration and invasion. However, these results conflict with those presented in a recent report (74) using HT1080 fibrosarcoma cells transfected with MT1-MMP-Y⁵⁷³F, in which collagen invasion was impaired in cells expressing the mutant proteinase. It should be noted that in addition to posttranslational regulation of MT1-MMP trafficking, EGFR activation has also been shown to increase total MT1-MMP expression in ovarian cancers (42) as well as in squamous cell carcinoma and glioma (52, 53). In addition, after siRNA-induced down-regulation of MT1-MMP expression in ovarian cancer cells, residual invasive activity is observed, suggesting that MT1-MMP is not the only contributor to invasion (42). Alternatively, the apparent discrepancy between the current study and Nyalendo et al. (74) may reflect a distinction between the responses of epithelial versus mesenchymal cells to altered MT1-MMP dynamics. In this regard, it is interesting to note that ovarian cancer cells expressing MT1-MMP-Y⁵⁷³F undergo a striking transition to a mesenchymal phenotype, particularly apparent in three-dimensional collagen culture. Cells expressing wild-type MT1-MMP form large three-dimensional multicellular aggregates within collagen gels with a long axis of 200 to 700 μm. In contrast, cells expressing MT1-MMP-Y⁵⁷³F display minimal cell-cell contact and, alternatively, survive in collagen gels as strikingly elongated single cells or 2 to 5 cell clusters with individual cells approaching 300 to 500 μm in length. Expression of the epithelial cadherin E-cadherin is lost, whereas cells acquire expression of the mesenchymal markers N-cadherin and vimentin. A unique feature of epithelial ovarian carcinoma progression is that E-cadherin becomes more abundant in primary differentiated ovarian carcinomas, relative to the lack of E-cadherin expression by normal ovarian epithelium (75). Thus, ovarian carcinomas seem to follow a mesenchymal to epithelial transition during early tumorigenesis. In advanced ovarian tumors, expression of E-cadherin is generally reduced, whereas N-cadherin is increased, suggestive of an epithelial-mesenchymal transition (EMT) late in tumor progression (75). Although the role of dysregulated MT1-MMP activity in promotion of mesenchymal characteristics requires further investigation, previous studies have implicated MMPs in EMT via a mechanism involving induction of a Rac1 isoform that promotes snail expression and subsequent EMT (101). However, in the current study, culture of MT1-MMP-Y⁵⁷³F cells in the presence of the broad spectrum MMP inhibitor GM6001 for 6 days resulted in an incomplete reversal of EMT, characterized by a partial gain of E-cadherin expression (to ∼50% wild-type levels), while N-cadherin expression was retained (data not shown), suggesting alternative mechanisms of EMT regulation.

These functional data emphasize and resolve two distinct outcomes available to cells expressing MT1-MMP that encounter collagen-rich matrices: invasion versus expansion. Numerous studies have supported a requirement for MT1-MMP in collagen invasion (45, 52, 53, 102-104). MT1-MMP–catalyzed processing of collagen fibers results in realignment of fibers in a forward movement direction, and these collagen degradation tracks are permissive for the formation of invasive cell strands (100). MT1-MMP has also been recognized as a key component for expansive cell growth in three-dimensional matrices (52, 53, 74, 100). In the context of serum-derived mitogens, MT1-MMP is necessary to remove physical constraints imposed by the collagenous matrix that prevent cytoskeletal alterations necessary to drive an effective proliferative response (53). In fibroblasts, MT1-MMP drives collagen phagocytosis during tissue remodeling, and MT1-MMP is localized to sites of collagen cleavage intracellularly as well as in the pericellular space (102), suggesting that MT1-MMP internalization is necessary for tissue remodeling. Furthermore, cooperation between MT1-MMP and the endocytic receptor uPARAP/Endo180 enables endocytosis and further intracellular processing of large collagen fragments (105, 106), facilitating large-scale turnover of collagen sufficient to allow aggregate proliferation. Thus, it is interesting to speculate that MT1-MMP–expressing cells capable of responding to EGF signaling can effectively internalize MT1-MMP, as well as pericellular collagen, establishing an environment permissive for expansive growth. In contrast, lack of or inability to respond to EGFR signaling may target MT1-MMP–expressing cells to an invasion pathway favored by mesenchymal cells. Although the contribution of EGFR activation to regulation of MT1-MMP interaction with collagen endocytic receptors has not been explored, the current data support the hypothesis that cellular response to growth factor signals may shift tissue dynamics of MT1-MMP–expressing cells between invasive and expansive growth. As the vast majority of women with ovarian cancer succumb due to complications of metastasis, a more detailed understanding of the factors that differentially regulate localized invasion, i.p. anchoring and uncontrolled proliferation of secondary lesions is warranted.

The ovarian carcinoma cell line OVCA 433 was provided by Dr. Robert Bast, Jr., M. D. Anderson Cancer Center, Houston, TX. Cells were maintained in minimal essential medium containing 10% fetal bovine serum. Cell culture media and reagents were purchased from Mediatech. For experiments involving EGF (Chemicon), OVCA 433 cell lines were serum starved for 24 h before growth factor treatment. The human MT1-MMP cDNA with COOH-terminal FLAG tag (DYKDDDDK) was kindly provided by Dr. Duanqing Pei (University of Minnesota, Minneapolis, MN). The open reading frame sequences of MT1-MMP cDNA were TA subcloned into eukaryotic expression vector pCR3.1-Uni (Invitrogen) to obtain the wild-type construct. The wild-type construct was FLAG tagged in the stalk region as described (107). Subsequently, the Y573F point mutation was generated on the wild-type cDNA and on the tagged constructs using quick-change (Stratagene) to create MT1-MMP-Y⁵⁷³F. The wild-type MT1-MMP coding sequence was also subcloned in a pEGFP-N1 vector (Clontech) to generate a fluorescent protein tag at the COOH terminal using Nhe I site from pCR3.1 vector and a COOH terminal PCR-generated Sac II site. Transfection of cells was done using FuGENE 6 (Roche) according to manufacturer's instructions, and stable cell lines were generated using G418 selection. Cell populations were FACS-sorted using an antibody against the Flag tag (M2) in the Flow Cytometry Facility, Northwestern University Feinberg School of Medicine. To determine the effect of caveolae disruption on EGFR and MT1-MMP surface dynamics, cells were treated for 1 h with 10 mmol/L methyl-β-cyclodextrin diluted in serum free 10× MEM before EGF treatment and processing for confocal microscopy or FACS analysis as described below.

Anti-FLAG monoclonal M2 antibody, monoclonal anti-58K, anti–MMP-14 (hinge region), monoclonal antivimentin (clone VIM-13.2), and rabbit polyclonal anti-Rab7 were purchased from Sigma. The mouse monoclonal EGFR, rabbit polyclonal antibodies against the extracellular domain of EGFR, ERK1 antibodies (sc-93), the mouse monoclonal anti-GRP 94, and mouse monoclonal anti caveolin-1 antibodies were purchased from Santa Cruz Biotechnology. The mouse monoclonal anticlathrin antibody was purchased from AbCam, Inc. Antiphosphotyrosine (clone PY20) was purchased from Millipore, and monoclonal antibodies directed againse N-cadherin (clone 3B9) and E-cadherin (clone HECD-1) were obtained from Zymed. The mouse monoclonal antibody against CD71/Transferrin Receptor was purchased from BD Pharmingen. The Alexa Fluor 555 cholera toxin subunit B conjugate was purchased from Molecular Probes. Alexa Fluor 546 goat anti-mouse IgG, Alexa Fluor 488 goat anti-mouse IgG, Alexa Fluor 546 goat anti-rabbit IgG, and Alexa Fluor 488 goat anti-rabbit IgG were purchased from Molecular Probes. EGF labeled with Alexa Fluor 546 was purchased from Molecular Probes.

To acquire live cell images, cells expressing GFP-tagged MT1-MMP were seeded onto 35-mm Becton glass bottom dishes at a density of 10⁶ and serum starved overnight. Cells were treated with EGF (25 nmol/L) and images were acquired using a Nikon TE2000 inverted microscope (Nikon, Inc.), and analyzed using Metamorph Software (Molecular Devices). Images were taken in 1-min increments for 60 m with a ×60 oil immersion objective with numerical aperture 1.40. The experiment was repeated in triplicate.

After treatment with 25 nmol/L EGF, OVCA 433 cells were fixed with 4% paraformaldehyde in PBS containing MgCl₂ and CaCl₂, for 10 min at room temperature, rinsed with PBS, permeabilized with 0.1% triton X-100 for 5 min, and blocked with 10% bovine serum albumin (BSA)/PBS for 1 h at 37°C. For dual staining, fixed GFP-tagged MT1-MMP– or MT1-MMP-Y⁵⁷³F–expressing OVCA 433 cells grown on glass coverslips were stained with primary antibodies directed against EGFR, transferrin receptor, clathrin, GRP-94, 58K, caveolin-1 or Rab7 for 30 min at 37°C. After 3 washes with PBS, samples were incubated with Alexa Fluor 546 goat anti-mouse or Alexa Fluor 546 goat anti-rabbit for 30 min at 37°C. For dual staining for EGFR and other antigens, samples were treated with anti-EGFR (rabbit or mouse), and anti-CD71/transferin receptor, anti-clathrin, anti–caveolin-1, anti-GRP 94, anti-58K, or anti-Rab7. Confocal microscopy was conducted at the Cell Imaging Facility, Northwestern University Feinberg School of Medicine. Images were attained using a Zeiss LSM510 system equipped with HeNe lasers for excitation at 488 nm (green) and 543 nm (red) and analyzed using Metamorph. All images were viewed using the 63 × 1.4 oil immersion objective. All experiments examining MT1-MMP or EGFR dynamics were done a minimum of thrice.

FACS analysis was used to quantify surface dynamics of MT1-MMP or EGFR in response to EGF treatment. OVCA 433 cells expressing flag-tagged wild-type MT1-MMP or Y⁵⁷³F mutant MT1-MMP (10⁶ cells) were seeded onto tissue culture dishes for each experimental time point. Before analysis, cells were serum starved for 24 h, then treated with EGF (25 nmol/L). At each time point, cells were trypsinized and incubated on ice with anti-flag monoclonal M2 primary antibody (to detect MT1-MMP; 1:100 dilution) or with monoclonal anti-EGFR antibody (1:100) for 30 min, washed thrice, and incubated on ice with Alexa Fluor 488 goat anti mouse IgG (1:200 dilution; 30 min). Cells were washed thrice and analyzed using a Becton Dickinson FACScan flow cytometer. Assays were done with three to six replicates. Data are normalized such that surface expression in the absence of EGF treatment is designated as 100%.

For Western blotting of whole cell lysates, cells were lysed using 50 mmol/L Tris (pH 7.5), 150 mmol/L NaCl, 1% Triton X-100, and the protein concentration of lysates was analyzed using the Bio-Rad DC detection kit and bovine albumin standards. Cell lysates (50 μg) were electrophoresed on 9% SDS-polyacrylamide gels, transferred to polyvinylidene difluoride membrane, and blocked with 3% BSA in 50 mmol/L Trizma (Tris base; pH 7.5), 300 mmol/L NaCl, 0.2% Tween 20 (TBST). Membranes were incubated for 1 h at room temperature with a 1:1,000 dilution of antibodies directed against the FLAG epitope tag (FLAG M2 monoclonal antibody), MT1-MMP hinge, E-cadherin, N-cadherin, vimentin, or glyceraldehyde-3-phosphate dehydrogenase in 3% BSA/TBST. Immunoreactive bands were visualized with a peroxidase-conjugated anti–rabbit-IgG (1:4,000 in 3% BSA/TBST) and enhanced chemiluminescence. For immunoprecipitation analyses, cells were serum starved in the presence of the broad spectrum MMP inhibitor GM6001 (Chemicon), treated with EGF (25 nmol/L) for 30 min, collected with lysis buffer (above) and subjected to immunoprecipiation using anti–MT1-MMP (hinge antibody, 1:4,000 dilution) and protein G beads. Immunoprecipitates were electrophoresed on 9% polyacrylamide gels and subjected to Western blotting using anti–MT1-MMP catalytic domain (1:4,000) or anti–phospho-tyrosine antibodies (1:1,000) followed by enhanced chemiluminescence detection as described above.

Three-dimensional cultures were prepared by diluting type I rat tail collagen (BD Biosciences) with 10 X MEM (Life Technologies) to a final concentration of 1.5 mg/mL. Cells (5 × 10⁴) were added to the collagen mixture before solidification. In parallel control experiments, the MMP inhibitor TIMP-2 was included at a final concentration of 5 μg/mL (53). After growth for 6 d in three-dimensional collagen, gels were dissolved using bacterial collagenase (2 mg/mL; Worthington) and cell number was evaluated by hemocytometry as described (41). Experiments were repeated in triplicate.

Type I collagen was dissolved in 0.5 mol/L acetic acid at a concentration of 2 mg/mL. For invasion experiments, the collagen stock was neutralized with 100 mmol/L Na2CO3 (pH 9.6) to a final concentration of 0.4 mg/mL. Transwell inserts (0.8 μm; Becton Dickinson) were coated on the underside with 500 μL of collagen diluted to a concentration of 100 μg/mL at 37°C for 1 h. Collagen gels were prepared in the inner well by adding 50 μL of collagen (20 μg) at room temperature and allowing gels to air dry overnight. Collagen-coated inserts were then washed with MEM thrice to remove salts and used immediately. Cells were trypsinized, washed with serum-free medium, and 1 × 10⁵ cells were added to the inner invasion chamber in a volume of 200 μL. The outer wells contained 400 μL of culture medium. Cells were allowed to invade for 24 h; noninvading cells were removed from inner wells using a cotton swab, and invading cells adherent to the bottom of membrane were fixed and stained using a Diff-Quick staining kit (DADE AG). Invading cells were enumerated by dividing membranes into 4 quadrants and counting the number of cells in 3 distinct areas for each quadrant under a ×10 objective using an ocular micrometer. Assays were done in triplicate.

No potential conflicts of interest were disclosed.