Dynamic Interactions of Cortactin and Membrane Type 1 Matrix Metalloproteinase at Invadopodia: Defining the Stages of Invadopodia Formation and Function

Proteolytic degradation of extracellular matrix (ECM) is a critical step in tumor cell invasion (1). Invadopodia are specialized membrane protrusions of invading tumor cells that extend into the ECM and mediate cellular digestion of, and invasion through, the ECM (2, 3). At the ultrastructural level, invadopodia are filament-like protrusions of the ventral cell membrane adherent to matrix that can range from 0.1 to 0.8 μm in diameter and reach >2 μm in length (2, 4). Recent studies based on correlative light electron microscopy show that invading tumor cells at sites of ECM degradation form complex superstructures: cell membrane invaginations 8 μm wide and 2 μm deep containing multiple slender invadopodial protrusions (5). Multiple proteins are associated with invadopodia and can be classified into four major functional groups (for review, see ref. 6). The first group is a minimal motility machine that includes actin and regulators of filamentous actin (F-actin) polymerization/branching (7). Among these are Arp2/3, N-WASP, Cdc42, Nck, cofilin, capping proteins, cortactin, and dynamin. The second group includes adhesion proteins, such as multiple integrins, that mediate invadopodia interaction with ECM. The third group comprises signaling proteins that regulate the actin cytoskeleton and membrane remodeling (tyrosine kinases and Ras-related GTPases). Membrane-associated proteases mediating ECM degradation at sites of invadopodia function constitute the fourth group. Among them are membrane type 1 matrix metalloproteinase (MT1-MMP), seprase, MMP-2, and the urokinase-type plasminogen activator (uPA)/uPA receptor proteolytic system (8–13).

It is generally accepted that the formation of cell protrusions is driven by actin polymerization at the leading edge. Indeed, it has been established that an N-WASP–Arp2/3–cortactin–dynamin complex is essential for invadopodia formation (4, 5, 14). Although invadopodia contain a variety of proteins, specific molecular markers unique to invadopodia have not been identified. Little is known about mechanisms of formation, function, and regulation of invadopodia. Therefore, to dissect the dynamic mechanisms that govern the formation and function of invadopodia, we have focused on two proteins, one of which allows the monitoring of cytoskeletal changes at invadopodia and another that yields information on the ECM-degrading function of invadopodia. These proteins are cortactin and MT1-MMP.

Cortactin, an actin-binding protein, has diverse cellular functions (for review, see refs. 15, 16). Its acidic NH₂-terminal domain mediates cortactin binding to the Arp2/3 complex, promoting actin filament branching and stabilization of branched networks (17). Internal to the acidic NH₂-terminal domain is the cortactin repeat region followed by a α-helical domain, a proline-rich domain with regulatory serine/threonine and tyrosine phosphorylation sites, and a Src homology 3 domain at the COOH terminus. The cortactin repeats mediate cortactin binding to F-actin. The cortactin Src homology 3 domain binds to a variety of targets providing a link between specific transmembrane receptors and the underlying actin cytoskeleton as well as recruiting particular binding proteins to regions of cortical actin and playing a role in a variety of vesicle transport pathways (16, 18). Cortactin is a target for phosphorylation by tyrosine and serine/threonine kinases (19). Differential regulation of cortactin by cellular kinases that effect cortical actin polymerization makes cortactin an important regulatory point. Moreover, the human gene EMS1 coding for cortactin is overexpressed in 13% of breast carcinomas and 29% of head and neck tumors, mainly due to gene amplification (20). Its overexpression in cells enhances cell migration, invasion, and tumor cell metastasis (21–24).

MT1-MMP is a membrane-bound metalloproteinase with broad substrate specificity and multiple cellular functions. MT1-MMP degrades multiple ECM components, including collagen types I, II, and III; fibronectin; and laminins 1 and 5 (for a review, see ref. 25). Furthermore, MT1-MMP increases the range of proteolytic activities by activating MMP-2 and possibly MMP-13. In addition, MT1-MMP may proteolytically modify CD44, α_V integrin, and transglutaminase, thereby exerting regulatory effects on cell migration (25). MT1-MMP activity is crucial for both physiologic processes and the progression of cancer (26, 27). It is highly expressed in different invasive cancers. MT1-MMP overexpression in cancer cells promotes cell migration, invasion, and metastasis in vitro and in vivo (25, 28, 29). In invasive tumor cell lines, MT1-MMP invadopodial localization requires membrane anchorage of the enzyme and its expression in cancer cells increases cell invasion by stimulating invadopodial activity (10).

Here, we show that both cortactin and MT1-MMP are required for the assembly of functional invadopodia that degrade gelatin matrix in invasive MDA-MB-231 carcinoma cells. Cortactin is necessary for the maintenance of F-actin-enriched invadopodial core structures. In the presence of functional MT1-MMP, actin-cortactin aggregates of invadopodia support MT1-MMP accumulation at invadopodia required for initiation and sustenance of matrix degradation. Depletion of endogenous MT1-MMP levels results in the formation of actin-cortactin structural components of invadopodia that are functionally unable to mediate matrix proteolysis. After the initiation of MT1-MMP accumulation at invadopodia to initiate matrix degradation, cortactin dissociates from invadopodial regions leaving abundant MT1-MMP associated with foci of matrix degradation for continued matrix proteolysis. Based on our results, we propose a stepwise model of invadopodia formation and function.

Antibody. Anticortactin 4F11 monoclonal antibody (mAb) was from Upstate (Lake Placid, NY), anti-MT1-MMP AB815 from Chemicon (Temecula, CA), polyclonal anti-Src and anti-phospho-Src pY⁴¹⁸ antibody from BioSource (Camarillo, CA), and secondary antibody conjugated to Cy2 or Cy5 were from Jackson ImmunoResearch (West Grove, PA).

Cell lines and transfectants. Human breast carcinoma MDA-MB-231 and MCF-7 cell lines were obtained from Tissue Culture Shared Resource of the Lombardi Cancer Center. Cells were maintained in high glucose-DMEM (HyClone, Logan, UT) supplemented with 10% fetal bovine serum (HyClone), 2 mmol/L l-glutamine, 10 units/mL penicillin, and 10 μg/mL streptomycin.

Stable lines of MDA-MB-231 cells transfected with wild-type c-Src, Y527F constitutively active c-Src, and dominant-negative K295R c-Src were a generous gift from Dr. Toshiyuki Yoneda (Department of Medicine, University of Texas Health Science Center at San Antonio, San Antonio, TX; ref. 30).

cDNA vectors and small interfering RNA. Human MT1-MMP in the pEGFPN1 vector (BD Biosciences Clontech, Palo Alto, CA) was a kind gift from Drs. Henning Birkedal-Hansen and Sarah Netzel-Arnett (National Institute of Dental and Craniofacial Research, NIH, Bethesda, MD; ref. 31). Stable cell populations expressing MT1-MMP enhanced green fluorescent protein (MT1-MMP-eGFP) were generated from parental MDA-MB-231 and Y527F c-Src cells.

Human cortactin cDNA was purchased from Invitrogen (Carlsbad, CA). Twelve amino acids missing from the actin-binding domain compared with the original sequence (GenBank) was restored by site-directed mutagenesis. Complete cortactin cDNA was cloned into a modified monomeric dsRED pRFP1-C1 vector (BD Biosciences Clontech) that was a kind gift from Dr. Roger Tsien (Department of Pharmacology, University of California at San Diego, San Diego, CA). The cortactin monomeric dsRED [cortactin-mRFP(1)] vector was stably expressed in MDA-MB-231 cells stably expressing MT1-MMP-eGFP. Stable cell populations expressing MT1-MMP-eGFP as well as MT1-MMP-eGFP and cortactin-mRFP(1) were selected based on three consecutive sortings for eGFP and mRFP fluorescence with a Becton Dickinson FACStarPlus system (Becton Dickinson, Franklin Lakes, NJ).

Pools of four small interfering RNA (siRNA) duplexes each were designed to target human cortactin (accession no. NM_138565; Qiagen, Valencia, CA) or human MT1-MMP (accession no. NM_004995; Dharmacon, Lafayette, CO). As a negative control, cells were transfected with nonspecific siRNA pool (Dharmacon). All cDNA vectors and siRNA transfections were done with Lipofectamine 2000 (Invitrogen) according to the instructions of the manufacturer.

Fluorescent-gelatin degradation assay. Gelatin was conjugated with AlexaFluor (Molecular Probes, Eugene, OR) using the protocol of the manufacturer. Coverslips (18 mm) were precleaned with 20% nitric acid for 30 minutes followed by extensive washing. Ethanol-sterilized coverslips were coated with 50 μg/mL poly-l-lysine for 20 minutes at room temperature, washed with PBS, and fixed with 0.5% glutaraldehyde for 15 minutes followed by extensive washing. Then, the coverslips were inverted on an 80 μL drop of gelatin matrix (0.2% gelatin and AlexaFluor-gelatin at an 8:1 ratio) and incubated for 10 minutes at room temperature. After washing with PBS, the residual reactive groups in the gelatin matrix were quenched with sodium borohydride at 5 mg/mL for 15 minutes followed by extensive washing with PBS.

To assess the ability of cells to form invadopodia and degrade matrix, cells were plated on coverslips coated with fluorescent gelatin matrix in 12-well plates at 4 × 10³/mL and incubated at 37°C. Foci of degraded matrix were visible as dark areas ∼0.2 to 1.2 μm in diameter that lack fluorescence and appear as “holes” in the bright fluorescent gelatin matrix. A cell degrading at least one hole under the cell or near the cell edge was counted as a cell able to degrade matrix. For image presentation, MetaMorph (Molecular Devices, Sunnyvale, CA) median (2 × 2) filter was applied to fluorescent images of the matrix.

Immunocytochemistry. Cells were fixed-permeabilized with 3.7% formaldehyde/0.1% Triton X-100 for 10 minutes, washed with PBS, and labeled with 10 μg/mL primary antibodies for 1 hour followed by labeling with secondary fluorochrome-conjugated antibodies for 30 minutes. Actin filaments were visualized with Alexa-phalloidin (Molecular Probes).

Confocal fluorescence microscopy. Confocal images were collected using a Zeiss510-META laser scanning confocal microscope (Carl Zeiss, Thornwood, NY) with Zeiss Plan-Neofluar 100×/1.3 N.A. oil or Plan-Apochromat 63×/1.4 N.A. oil objectives. For optimal confocality, all images were collected with the pinhole photodetector aperture set equal or <1 Airy unit. Images were processed with AutoDeBlur (AutoQuant Imaging, Watervliet, NY) two-dimensional or three-dimensional blind deconvolution using a total of 30 iterations.

Live fluorescence microscopy. Live cells expressing eGFP and mRFP protein chimeras were imaged with a Zeiss Axiovert200M microscope equipped with an UltraVIEW-RS confocal live-cell imager (Perkin-Elmer, Boston, MA) with a Zeiss Plan-Apochromat 63×/1.4 N.A. oil objective. An environmental chamber mounted on the microscope maintained constant 37°C temperature, CO₂, and humidity. For live fluorescence microscopy, cells were seeded in microwell dishes (MatTek, Ashland, MA) with glass bottoms coated with fluorescent-gelatin matrix as described above. To suppress formation of cytotoxic free radicals, Oxyrase (Oxyrase, Mansfield, OH) was added at a 1:100 dilution to culture medium immediately before cell imaging.

Transmission electron microscopy. Cells at 5 × 10⁵/mL in complete culture medium were incubated in glass-bottom dishes coated with Gelatin-Alexa568 matrices for 3 hours. Cells were fixed with 2.5% glutaraldehyde/2% paraformaldehyde for 2 hours, postfixed in 1% osmium tetroxide for 1 hour, en bloc stained with uranyl acetate, and processed for conventional ultrathin section electron microscopy as described previously (4). Layers of plastic-embedded cells on matrix were removed from the coverslips, reembedded in plastic, and sections were cut perpendicular to the original plane of the coverslip. Sections were poststained with uranyl acetate and lead citrate, and photographed with a Hitachi H-7600 transmission electron microscope.

Immunoblotting. Cells cultured overnight at 2 × 10⁵/mL in six-well plates were extracted with radioimmunoprecipitation assay buffer, mixed with sample buffer containing β-mercaptoethanol, sonicated, and boiled for 5 minutes. SDS-PAGE was done using 4% to 12% Tris-glycine gels (Novex, Invitrogen). Semiquantitative analysis of protein levels was done using ScionImage software (Scion, Frederick, MD).

Biotinylation and immunoprecipitation. Cell surface protein biotinylation and immunoprecipitation to identify surface MT1-MMP was done as described (32).

Gelatin zymography. Cells at 3 × 10⁵ per well in six-well plates were incubated overnight. For culture medium samples, cells were cultured in serum-free DMEM for 48 hours. Cells were lysed with 1% NP40 in 10 mmol/L Tris (pH 7.4), 150 mmol/L NaCl, 1 mmol/L CaCl₂, and 1 mmol/L MgCl₂. To detect MMP-2 activity, cell culture medium and cell lysates were concentrated six times and twice, respectively, by filtration on Microcon filter devices (Millipore, Billerica, MA). Samples were mixed with sample buffer without reducing agent, incubated for 10 minutes at room temperature, and separated on 10% polyacrylamide/0.1% gelatin gels. Then, gels were extensively washed in 2.5% Triton X-100 and incubated in 50 mmol/L Tris-HCl (pH 7.5), containing 0.2 mol/L NaCl, 5 mmol/L CaCl₂, 1 μmol/L ZnCl₂, 0.02% Brij 35, and 1 mmol/L p-aminophenylmercuric acetate at 37°C for 36 hours. The gel was stained with 0.5% Coomasie blue R-250. Gel loadings were normalized to total protein measured with a micro-BCA kit (Pierce, Rockford, IL).

Real-time PCR. MMP expression levels were measured by quantitative real-time PCR (qRT-PCR). Cells plated in six-well plates at 2 × 10⁵/mL were cultured overnight, then mRNA was collected using the RNAaqueous 4-PCR kit (Ambion, Austin, TX). cDNA was synthesized using a GeneAmp reverse transcription reaction kit (Applied Biosystems, Foster City, CA). qRT-PCR was carried out with SYBR green PCR master mix (Applied Biosystems) using an ABI Prism7700 Sequence detector (Perkin-Elmer). Thermal cycling variables for the qRT-PCR were 50°C for 2 minutes, 95°C for 10 minutes, 40 cycles of denaturation at 95°C for 15 seconds, and annealing/extension at 65°C for 30 seconds. The primers for human MMPs and internal controls used for RT-PCR are provided in Supplementary Fig. S1A. The results were presented as fold change in gene expression.

Statistical analysis. The data are represented graphically as a mean value with the corresponding SD, calculated after pooling three independent trials. For comparison between groups, t test was used. For analysis of live-cell imaging data, the sign and Wilcoxon signed-rank (two-sided) test were applied for pairwise comparisons between groups.

Elevated levels of c-Src stimulate ECM degradation via increased invadopodia formation. To study invadopodial dynamics, we searched for a cell line with a high percentage of cells capable of degrading matrix with significant numbers of invadopodia per cell. Because c-Src increases cell invasive potential (30, 33–37), we investigated the effect of c-Src expression on gelatin degradation by the breast carcinoma cell line MDA-MB-231 as a cell model. Degradation assays involved culturing cells on coverslips coated with thin, fluorochrome-conjugated gelatin matrices that were not cross-linked with glutaraldehyde (see Materials and Methods). Local proteolytic activity was identified by appearance of dark areas lacking fluorescence in the bright fluorescent matrix. According to electron microscopy, the thickness of our matrices is 48.9 ± 2.9 nm (Fig. 1D , inset). Unlike widely used glutaraldehyde cross-linked fluorescent gelatin films, the fluorescent matrices that we developed provide better sensitivity, allowing precise detection of local cell proteolytic activity after short incubation periods.

Incubation of parental carcinoma cells and cells expressing wild-type c-Src as well as mutant Y527F c-Src and K295R c-Src on fluorescent gelatin matrixes for different periods of time detected differences in numbers of cells degrading matrix (Fig. 1A). The percentage of parental carcinoma cells able to degrade the matrix gradually increased with time and reached ∼20% upon 3 hours of incubation. Elevated levels of wild-type c-Src or Y527F c-Src in these cells stimulated matrix degradation as detected by increases in the percentage of cells degrading matrix over time (Fig. 1A). Increased cellular Src conferred ability to ∼40% of the cells to degrade the gelatin matrix by 3 hours, whereas expression of Y527F c-Src increased the percentage to 55% (Fig. 1A).

Matrix degradation by cancer cells is mediated by invadopodia. Calculation of the mean number of invadopodia per cell showed substantial increases in formation of invadopodia with elevations in wild-type c-Src and dominant-active c-Src levels (Fig. 1A and D). Specifically, the mean number of invadopodia in our cell lines quadrupled after augmentation of c-Src levels compared with parental carcinoma cells, and it increased nearly 7-fold with expression of Y527F c-Src. The expression of K295R c-Src decreased the mean number of invadopodia per cell by ∼65% compared with parental cell line. Because Src induced a high percentage of cells capable of cleaving gelatin matrix by forming abundant invadopodia, we chose wild-type c-Src and Y527F c-Src MDA-MB-231 cells as convenient models to study the mechanisms of invadopodia formation and function in depth.

We also examined the effect of Y527F c-Src on expression and activity of several MMPs capable of gelatin degradation. Quantitative RT-PCR detected increases in MT1-MMP and MMP-9 mRNA expression levels of 1.4- and 16-fold, respectively, in carcinoma cells expressing Y527F c-Src compared with parental (Fig. 1B). No effects were observed on the mRNA expression levels of MMP-2, MMP-13, MMP-15, or MMP-16. The increase in MT1-MMP mRNA expression level correlated with the increase in amount of cell surface MT1-MMP in Y527F c-Src cells (Fig. 1B). Gelatin zymography of cell culture medium and cell lysates detected elevated levels of pro-MMP-9 (92 kDa) in Y527F c-Src–expressing cells (Fig. 1C), consistent with qRT-PCR data. The active 86 kDa form of MMP-9 appeared only in cell lysates and was elevated in Y527F c-Src cells. Gelatin zymography detected only pro-MMP-2 (72 kDa) in both cell lysates and culture medium with no difference in pro-MMP-2 levels in Y527F c-Src cells compared with parental cells (Fig. 1C).

Cortactin is indispensable for invadopodia formation. Early studies on invadopodia revealed that microinjection of anticortactin polyclonal antibodies decreased the number of matrix-degrading cells (4). To assess the role of cortactin in the dynamics of formation and function of invadopodia, we implemented a siRNA-targeted protein depletion approach. Based on dose dependence analysis using parental MDA-MB-231 to obtain ∼90% suppression of cortactin levels (Supplementary Fig. S1B), we treated parental, wild-type c-Src, and Y527F c-Src MDA-MB-231 cells with 80 nmol/L cortactin-specific pools of four siRNAs for 72 hours (Fig. 2A).

Cortactin depletion decreased cell migration on gelatin matrices (Supplementary Fig. S1C) and inhibited matrix degradation (Fig. 2B). To evaluate the ability of cells to form invadopodia under conditions of depleted cortactin, the cells were immunolabeled for actin and cortactin (Fig. 2C). In the control cells, the majority of actin-cortactin aggregates were associated with sites of degraded matrix after 3 hours of incubation on the gelatin matrix, indicating that functional invadopodia were formed. Depletion of cellular cortactin inhibited formation of the actin-cortactin aggregates and was accompanied by strong inhibition of matrix degradation (Fig. 2B and C). These results indicate that cortactin is required for formation of invadopodia and they suggest that the inhibition of matrix proteolysis by cells with depleted cortactin is caused by the impairment of invadopodia formation.

MT1-MMP is required for invadopodia function but not for invadopodia initiation. Several proteases have been shown to be associated with invadopodia of cancer cells. In our cell model, Y527F c-Src expression stimulated MT1-MMP and MMP-9 expression. To establish which proteases are responsible for gelatin matrix degradation in MDA-MB-231 carcinoma cells, we tested the effect of different protease inhibitors on matrix degradation (Fig. 3). Only tissue inhibitor of metalloproteinase (TIMP)-2, an inhibitor of soluble and membrane-bound MMPs, including MT1-MMP, was able to suppress matrix degradation at 0.3 μg/mL (Fig. 3A and D) and completely inhibited at 3 μg/mL (data not shown). In contrast, TIMP-1, which inhibits soluble MMPs but not membrane-type MMPs, did not reduce gelatin matrix degradation. Nevertheless, immunostaining of the TIMP-2-treated cells still revealed association of actin-cortactin aggregates with the cell membrane adherent to nondegraded matrix. These aggregates were identical in appearance and location to those in invadopodia of control cells. In wild-type c-Src and Y527F c-Src cells treated with TIMP-2, the total numbers of actin-cortactin aggregates were ∼40% and ∼30% less than those of the control, respectively (Fig. 3A).

Similar results were obtained upon carcinoma cell transfection with siRNA targeting MT1-MMP gene expression (Fig. 3B , C, and D). By suppressing MT1-MMP expression, we observed strong inhibition of matrix degradation by parental, wild-type c-Src, and Y527F c-Src–expressing cells. However, the formation of actin-cortactin structures was only modestly reduced (60-70% remained). Thus, we conclude that onset of invadopodia formation is independent of MT1-MMP, whereas the matrix-degrading capacity of invadopodia requires availability of catalytically functional MT1-MMP. Furthermore, matrix degradation of fluorescent type I collagen matrices by Y527F c-Src cells was observed at invadopodia (data not shown), further implicating MT1-MMP in invadopodia function.

Cortactin and MT1-MMP association at invadopodia is sequential. Given the conclusion that cortactin and MT1-MMP are essential components of functional invadopodia in MDA-MB-231 carcinoma cells capable of gelatin matrix degradation, we closely examined cells immunostained for cortactin and MT1-MMP (Fig. 4A). The cells revealed four distinct patterns of protein accumulation and matrix degradation: accumulations of only cortactin at the cell adherence membrane, cortactin and MT1-MMP aggregation overlying intact matrix, cortactin and MT1-MMP accumulation localized to foci of degraded matrix (single holes), and MT1-MMP association with foci of degraded matrix. These four types of combinations of cortactin and MT1-MMP over intact or degraded matrix were classified as invadopodia stages I, II, III, and IV, respectively.

Next, we examined cortactin and MT1-MMP temporal sequences relative to matrix degradation (i.e., appearance of matrix holes). Y527F c-Src cells stably expressing MT1-MMP-eGFP chimeras were plated on fluorescent matrix and allowed to degrade it for 45 minutes to 3 hours (Fig. 4B). After fixation, cells were immunolabeled for cortactin. Figure 4B shows the percentage of cell invadopodia at each stage after specific periods of incubation. After 45 minutes of incubation on matrix, ∼60% of the invadopodia formed by cells were at stage I, indicating that the initial step of invadopodia formation is linked to cortactin accumulation at the ventral cell membrane adherent to matrix. With time, the number of invadopodia at stage I decreased to ∼50% after 1 hour and then 10% after 2 hours of cell incubation on matrix, although it rebounded slightly to ∼20% after 3 hours of incubation. The percentage of invadopodia at stage II remained relatively constant (∼35%) after short periods of cell incubation on matrix (45 minutes and 1 hour), but then decreased to ∼10% with further incubation time. This time course indicates that invadopodial stage II (accumulation of cortactin and MT1-MMP above intact matrix) is relatively short-lived. The number of invadopodia at stages III and IV increased progressively with longer cell incubation on matrix, and after 3 hours, stage IV became dominant. This pattern obviously suggests that upon MT1-MMP accumulation at sites of cortactin aggregates, matrix degradation takes place. Subsequently, cortactin dissociates from the invadopodia, leaving behind MT1-MMP associated with the foci of degraded matrix.

Cortactin is an F-actin-binding protein that associates with the centers of active cortical actin remodeling. To assess the state of F-actin at each of the four invadopodial stages, we did multispectral microscopy of invadopodia (Fig. 4C). Y527F c-Src MDA-MB-231 cells stably expressing MT1-MMP-eGFP were plated on fluorescent gelatin matrices for different periods of time. The cells were then fixed and labeled for cortactin and F-actin. Representative micrographs of invadopodia at each stage are shown in Fig. 4C. Cortactin was always associated with F-actin accumulation at invadopodia in stages I, II, and III. The dissociation of cortactin from the invadopodia enriched in MT1-MMP-eGFP at stage IV was accompanied by disappearance of F-actin from that site.

Live-cell imaging reveals the dynamics of stepwise formation of invadopodia and uncovers two types of cortactin association: long lived and short lived. To examine cortactin and MT1-MMP dynamic interactions at invadopodia more extensively and to confirm our proposed four stages of invadopodia formation and function, we turned to live-cell imaging. Three-channel live-cell imaging was used to track the kinetics of matrix degradation by parental carcinoma cells stably expressing MT1-MMP-eGFP and cortactin-mRFP(1) chimeras (Fig. 5). Upon cell adhesion to the fluorescent matrix, dynamic accumulations of cortactin near the ventral cell membrane were observed at multiple sites (data not shown). Some of these cortactin aggregates dispersed whereas others became invadopodia. Invadopodia formation first involved the arrival of a cortactin aggregate to the cell membrane adherent to the matrix, followed by the accumulation of MT1-MMP at the cortactin-rich regions (Fig. 5A and B; Movies 1 and 2 in Supplementary Data). Accumulation of MT1-MMP at invadopodia was accompanied by high motility of the aggregates of membrane-associated MT1-MMP in and adjacent to invadopodia. The average time for onset of MT1-MMP accumulation at cortactin-rich invadopodia after cortactin initially appeared was 2.25 minutes for migrating cells and 1.5 minutes for nonmigrating cells (Table 1). Accumulation of MT1-MMP at cortactin-containing invadopodial structures resulted in subsequent degradation of matrix underlying the invadopodia. The onset of matrix degradation compared with cortactin arrival at invadopodia was 6 minutes for nonmigrating carcinoma cells and 4.5 minutes for migrating cells (Table 1).

Live-cell imaging also revealed different patterns of cortactin association with invadopodia. In nonmigrating cells, cortactin remained associated with invadopodia for extensive periods of time. Imaging of invadopodia for 90 minutes showed continual presence of cortactin. However, the intensity of this long-lived invadopodial cortactin changed over the 90-minute period (Fig. 5A,, kymograph). These fluctuations may indicate dynamic actin remodeling at invadopodia. In striking contrast, imaging of carcinoma cells undergoing migration showed short-lived association of cortactin with invadopodia (Fig. 5B). In migrating cells, cortactin remained associated with invadopodia for only an average of 10 minutes (median time for cortactin loss from invadopodia is 10.13 minutes; interquartile range is 7.50-20.25 minutes). During this short period of time, invadopodia assembled and matrix degradation was initiated. Cortactin disappearance from the invadopodia resulted in MT1-MMP-enriched but cortactin-free invadopodia that were still able to continue matrix degradation.

Finally, we characterized and compared the relative fluorescence intensities of fluorescently tagged MT1-MMP, cortactin, actin, and matrix to assess relative amounts of each constituent and the extent of matrix proteolysis at each invadopodial stage (Fig. 6A). By semiquantitative determinations of pixel intensity integrated over the area of individual invadopodia, the levels of cortactin and actin were confirmed to be initially elevated at stage I. Cortactin and actin levels then rose moderately further at stages II and III, then dropped to zero at stage IV. In contrast, MT1-MMP was absent at stage I and rose sharply to a plateau at stages II to III, and remained high or slightly increased at stage IV. Matrix staining was the opposite: It was highest at stages I and II, then dropped as holes were formed at stage III and IV.

Invadopodia are specialized membrane protrusions of invasive tumor cells that mediate focal pericellular proteolysis of ECM and promote tumor cell invasion and metastasis (6). Consequently, understanding the mechanisms governing the formation and function of invadopodia should provide insights into the biology, regulation, and potential therapeutic approaches to cancer metastasis.

In a current view, the induction of invadopodia—F-actin-rich membrane protrusions with matrix degradation activity—is associated with activation of integrin-mediated or growth factor–stimulated signaling events (3, 8, 14, 38). Requirements for cortical actin regulators N-WASP and Arp2/3 and their upstream regulators Nck1, Cdc42, and WIP have been shown for epidermal growth factor–stimulated formation of invadopodia in adenocarcinoma cells (14, 39). In addition, the regulators of actin remodeling cofilin, cortactin, Arf6, and dynamin2, which also play roles in membrane trafficking, remodeling, and recycling, localize to invadopodia, and their functional activities are required for invadopodia-mediated matrix degradation (4, 5, 14, 40, 41). Nonetheless, current understanding of molecular mechanisms that govern invadopodia formation has been limited to the early events of actin polymerization induction through the Arp2/3 complex that drives invadopodia initiation (14). In this study, based on an approach that combines the monitoring of the cytoskeletal core of invadopodia with the monitoring of invadopodia function, we substantially extend existing knowledge about the organization and proteolytic function of invadopodia.

While establishing an in vitro cell model for invadopodia studies, we found that the expression of wild-type c-Src or a dominant-active c-Src stimulates the formation of invadopodia in breast carcinoma cells. The application of our thin and non-cross-linked fluorescent gelatin matrixes, which provide rapid and sensitive detection of proteolytic activity associated with invadopodia, allowed identification of distinct structural and functional features of invadopodial dynamics. The structural aspect includes the initiation, organization, and stabilization of the F-actin-rich core of invadopodia and invadopodial protrusion extension, all dependent on cortactin, whereas functional activity was associated with the recruitment and organization of proteolytic enzymes at invadopodia that were required for ECM degradation. We found that in MDA-MB-231 carcinoma cells, invadopodial function in gelatin matrix proteolysis required MT1-MMP activity. Down-regulation of MT1-MMP activity blocked matrix degradation but only moderately affected actin/cortactin aggregation of the invadopodia, which otherwise appeared intact, suggesting that dynamic cell-matrix interactions initiate invadopodial actin core formation and dictate protease accumulation at invadopodia. During live-cell imaging, we also observed that only cortactin aggregates at the cell adherence membrane that are destined to become invadopodia initiate MT1-MMP accumulation, indicating the existence of regulatory signaling events yet to be established.

Several studies showed MT1-MMP localization to the migration front of cells in parallel with actin cytoskeleton remodeling when cells are stimulated to migrate. Such reorganization of MT1-MMP has been accompanied by enhancement of MMP-2 activation, showing that MT1-MMP is in an active state at the leading edge (42–44). In this study, we revealed that MT1-MMP, although present over the entire plasma membrane, is specifically concentrated at the invadopodia of carcinoma cells and promotes matrix degradation at these sites. MDA-MB-231 cells plated on gelatin matrix and labeled for FAK or paxillin, in addition to invadopodia, show the formation of focal adhesions that do not overlap with invadopodial structures or degradation sites (data not shown). We speculate that focal adhesions, by providing the main sites of cell adhesion to ECM and sites of signal transduction into the cell interior, might support local membrane attachment to permit initiation and extension of invadopodia into the matrix. The proteolytic activity accumulated at invadopodial tips degrades the ECM and promotes invasion by weakening matrix resistance.

Some cell types engaged in matrix remodeling and tissue invasion form podosomes, actin-rich adhesion structures that share many molecular components with invadopodia (45, 46). The classic podosome structure is a circular F-actin core surrounded by a ring of vinculin. In contrast, ∼95% of the invadopodia of c-Src-expressing carcinoma cells plated on gelatin matrix and labeled for actin and vinculin lack vinculin-rich rings (data not shown). Thus, invadopodia of MDA-MB-231 cells are distinct from classic podosomes.

In summary, based on the results of examination of cells at different time points, live-cell imaging, and fluorescence intensity analysis, we have identified a stepwise mechanism of invadopodia formation and function (Fig. 6B). Invadopodia start as an invadopodial initiation site by accumulation of actin and cortactin at areas of cell membrane adherence to ECM. The onset of MT1-MMP accumulation defines the full-fledged preinvadopodia stage. This stage involves ongoing increases in the amounts of actin and cortactin along with the increase in MT1-MMP. The beginning of matrix degradation defines the formation of mature invadopodia, which are enriched in actin, cortactin, and MT1-MMP. The dispersal of F-actin and cortactin from mature invadopodia results in the formation of late invadopodia, which remain enriched in MT1-MMP capable of continuing focal degradation.

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.