Induction of a MT1-MMP and MT2-MMP-dependent basement membrane transmigration program in cancer cells by Snail1

Abstract

The ability of carcinoma cells arising at primary sites to cross their underlying basement membrane (BM), a specialized form of extracellular matrix that subtends all epithelial cells, and to access the host vasculature are central features of the malignant phenotype. The initiation of the invasive phenotype has been linked to the aberrant expression of zinc-finger transcriptional repressors, like Snail1, which act by triggering an epithelial-mesenchymal cell-like transformation (EMT-like) via the regulation of largely undefined, downstream effectors. Herein, we find that Snail1 induces cancer cells to (i) degrade and perforate BM barriers, (ii) initiate angiogenesis, and (iii) and intravasate vascular networks in vivo via a matrix metalloproteinase (MMP)-dependent process. Unexpectedly, the complete Snail1 invasion program can be recapitulated by expressing directly either of the membrane-anchored metalloproteinases, MT1-MMP or MT2-MMP. The pro-invasive, angiogenic, and metastatic activities of MT1-MMP and MT2-MMP are unique relative to all other metalloproteinase family members and cannot be mimicked in vivo by the secreted MMPs, MMP-1, MMP-2, MMP-3, MMP-7, MMP-9, or MMP-13. Further, siRNA-specific silencing of MT1-MMP and MT2-MMP ablates completely the ability of Snail1 to drive cancer cell BM invasion, induce angiogenesis, or trigger intravasation. Taken together, these data demonstrate that MT1-MMP and MT2-MMP cooperatively function as direct-acting, pro-invasive factors that confer Snail1-triggered cells with the key activities most frequently linked to morbidity and mortality in cancer.

The transition of carcinoma in situ to a frankly carcinomatous lesion requires the cancer cells to acquire an ability to perforate and transmigrate the underlying basement membrane (BM), a specialized form of extracellular matrix (ECM) that subtends all epithelial cells (1, 2). Current evidence suggests that the induction of a BM-invasive phenotype may be linked to the expression of zinc-finger transcriptional repressors capable of promoting an epithelial-mesenchymal cell transition (EMT) which trigger epithelial cell-derived cancer cells to adopt a tissue-invasive, mesenchymal cell-like phenotype (2–4). Snail1 is a prototypical member of a family of EMT-inducing transcription factors, playing a required role in developmental programs, such as gastrulation, and capable of undergoing pathologic re-activation postnatally in neoplastic states (5). While Snail1 has been linked to cancer cell invasion programs, cancer recurrence, and the adoption of cancer stem cell-like properties (2, 5), the mechanisms by which the transcription factor induces BM degradation and invasion programs remain undefined.

Recent efforts to delineate normal or neoplastic cell interactions with the intact BM in an in vivo setting have been limited largely to model organisms where changes in BM structure during invasive processes can be evaluated directly by microscopic imaging (6, 7). In vertebrate systems, experimental models are unavailable where carcinoma cells can be situated atop linear, unbroken stretches of BM and invasion monitored in a fashion that lends itself to molecular characterization in vivo. Herein, we have adopted a live chick chorioallantoic membrane (CAM) model to analyze the cancer cell-BM interactions that underlie the earliest steps in the carcinoma invasion program (3, 4, 8, 9). These studies demonstrate that Snail1 induces cancer cells to transmigrate BM barriers by mobilizing the membrane-type matrix metalloproteinases (MT-MMPs), MT1-MMP and MT2-MMP (1). Working in tandem, these MT-MMP family members not only confer carcinoma cells with the ability to perforate BM structures in vivo, but also to trigger angiogenesis, cancer cell proliferation and dissemination of the transformed cells to distant sites through the host vasculature. These findings suggest that Snail1, and perhaps all EMT-inducing transcription factors, mobilize MT1-MMP and/or MT2-MMP as necessary co-factors during tumor progression.

Results

Snail1-Induced BM Degradation and Transmigration by Breast Carcinoma Cells.

To define cancer cell-BM interactions in an in vivo setting, neoplastic cell populations were cultured atop the CAM, an extra-embryonic tissue consisting of a chorionic epithelium of ectodermal origin, an intermediate mesenchyme and an endodermal allantoic epithelium (Fig. 1A) (10). The upper chorionic epithelium is heavily vascularized by BM-encased capillaries and subtended by a continuous epithelial-derived BM that demarcates the epithelium from the underlying mesenchyme (Fig. 1A) (10). Immunohistochemical staining with chick type IV collagen-specific antibodies clearly identifies (i) capillary BMs interspersed within the chorionic epithelium, (ii) the upper chorionic and lower allantoic epithelial BMs, and (iii) the vascular BMs surrounding arterioles and venules that traverse the CAM mesenchyme (Fig. 1A).

Fig. 1.

Cancer cell-BM interactions in vivo. (A) CAM structure and matrix composition of the developing chick egg at day 11 as assessed in hematoxylin and eosin (H&E)-stained cross-section (left panel). Immunofluorescent micrographs of the CAM reveal type IV collagen staining of the ectodermal/endodermal BMs, intraepithelial capillaries (arrowheads) and interstitial blood vessels (right panel). (Scale bar, 10 μm.) (B) CAM invasion by human cancer cells. Human breast cancer MCF-7_MOCK, MCF-7_Snail1 or MDA-MB-231 cells [labeled with fluorescent nanospheres (green) or engineered to express GFP] were inoculated atop the CAM of 11-day-old chicks with or without doxycycline and cultured for 3 days in the absence or presence of the MMP inhibitors, BB-2516 or GM6001. Fluorescent micrographs of CAM cross-sections demonstrate the ability of MCF-7_Snail1 or MDA-MB-cells to perforate the CAM surface through the chorionic epithelial BM (arrow head). Red, type IV collagen; blue, DAPI staining for cell nuclei. Results are representative of four or more experiments performed. (Scale bar, 50 μm.)

When fluorescently labeled human MCF-7 breast carcinoma cells, a well-differentiated cancer cell line that displays minimal invasive activity in mouse model systems (3, 4), are cultured atop the live CAM surface, the underlying epithelial BM remains intact and no invasive activity can be discerned (Fig. 1). By contrast, when MCF-7 cells are induced to express Snail1 (i.e., MCF-7_Snail1 cells), the epithelial BM is destroyed and the CAM mesenchyme is widely infiltrated by the tumor cells (Fig. 1). Interestingly, Snail1-dependent MCF-7 cell invasion occurs in tandem with a burst in chick angiogenic activity as well as cancer cell intravasation of the chick vasculature (see below). Consistent with a major role for one or more MMP family members, BM degradation and invasion by MCF-7_Snail1 cells is abrogated completely in the presence of the pan-specific MMP inhibitors, BB-2516 or GM6001 (Fig. 1) (1). The ability of MCF-7_Snail1 cells to degrade and transmigrate the CAM BM is similar to that observed by MDA-MB-231 carcinoma cells, an undifferentiated and highly invasive human breast cancer cells whose invasive activity is similarly blocked by MMP inhibitors (Fig. 1) (11, 12).

MT-MMP Family Members Confer MCF-7 Cells with Pro-Invasive and Angiogenic Activities.

The human MMP family is comprised of 23 members that include 15 secreted enzymes and eight membrane-anchored proteases (i.e., four type I transmembrane, one type II transmembrane, and two GPI-anchored enzymes) (1). While multiple MMPs have been characterized as expressing BM-degradative activity in various model systems in vitro (1), only 3 membrane-anchored MMPs, MT1-MMP, MT2-MMP, and MT3-MMP, have been shown to confer recipient cells with the ability to degrade or invade native BM barriers ex vivo (13). To first determine the ability of MT1-MMP, MT2-MMP, or MT3-MMP to confer MCF-7 cells with BM-degradative activity in vivo, the breast carcinoma cells were engineered to express the respective MT-MMPs and cultured atop live CAMs. As shown in Fig. 2, MT1-MMP- as well as MT2-MMP- expressing MCF-7 cells are not only able to degrade the underlying BM, but also to infiltrate the underlying mesenchyme while simultaneously triggering significant increases in angiogenesis as well as tumor cell proliferation via an MMP inhibitor-sensitive process (Fig. 2 A–D). Furthermore, although control MCF-7 cells are unable to intravasate the chick vasculature as determined by monitoring the lower CAM for human-specific Alu sequences (14), MT1-MMP- as well as MT2-MMP-expressing cells successfully access the host vascular bed (Fig.S1). As expected, the ability of MT1-MMP or MT2-MMP to trigger MCF-7 intravasation is likewise inhibited completely by BB-2516 (Fig. S1). Although recent studies have reported that MT1-MMP can modify cell behavior independently of its catalytic activity by initiating signaling cascades dependent upon the MT1-MMP cytosolic tail (15–18), MCF-7 cells expressing a catalytically-inactive MT1-MMP mutant lose all BM-degradative and invasive activities and no longer support increased rates of angiogenesis, proliferation or intravasation (Fig. 2 A–D and Fig. S1). Further, MCF-7 cells engineered to express a cytosolic tail-deleted form of MT1-MMP retain full BM-degradative, angiogenic, and proliferative activity in vivo while promoting intravasation as well (Fig. 2 A–D and Fig. S1). The ability of MT1-MMP or MT2-MMP to modulate MCF-7 cell behavior is, however, completely lost when the protease is expressed in an active, but secreted, form as a transmembrane-deleted mutant (Fig. 2 A–D). Finally, while MT3-MMP has been demonstrated previously to slowly remodel acellular BM sheets in vitro (13), the protease is unable to confer MCF-7 cells with BM remodeling activity in vivo, or any of the other ancillary activities associated with MT1-MMP or MT2-MMP activity despite similar levels of expression at the cell surface (Fig. 2 A–D).

Fig. 2.

MT1-MMP and MT2-MMP confer MCF-7 cells with tissue-invasive and pro-angiogenic activities. (A) MCF-7 cells transfected with control, wild-type or mutant MT1-MMP, MT2-MMP, or MT3-MMP constructs as indicated were labeled with fluorescent nanospheres (green), inoculated atop the CAM of 11-day-old chicks and cultured for 3 days in the absence or presence of BB-2516. Cross-sections were prepared and examined by fluorescent microscopy. Red, type IV collagen; blue, DAPI staining for cell nuclei. (Scale bar, 50 μm.) (B–D) MCF-7 cells transfected with wild-type or mutant MT1-MMP, MT2-MMP, or MT3-MMP constructs were inoculated atop the CAM of 11-day-old chicks and cultured as described above for 3 days. Invasion, angiogenic activity and MCF-7 cell proliferative responses were monitored as described in Methods. Results are expressed as the mean ± 1 SD of three or more experiments. (Scale bar, 50 μm.)

Secreted MMPs Fail to Support MCF-7 Cell Invasion, Proliferation, Angiogenesis, or Intravasation in Vivo.

A series of secreted MMPs, including MMP-1 (collagenase-1), MMP-2 (gelatinase A), MMP-3 (stromelysin-1), MMP-7 (matrilysin), and MMP-9 (gelatinase B) have been postulated to alter the invasive properties of cancer cells by virtue of their ability to either digest extracellular matrix components, cleave cell surface-associated targets, hydrolyze soluble substrates or induce EMT programs (1, 19). However, in direct contrast with MT1-MMP or MT2-MMP, MMP-1, MMP-2, MMP-9, and MMP-7 do not affect MCF-7 cell invasion, angiogenic activity, or proliferation in vivo (Fig. 3 A–C). Similarly, although MMP-3 has been reported to induce EMT-like changes in breast carcinoma cells via a Snail1-dependent process (20, 21), the protease is unable to induce BM degradative, invasive, proliferative or pro-angiogenic activity in MCF-7 cells (Fig. 3 A–C). Because each of these MMPs are synthesized and secreted as zymogens, and the enzymatic activities of the proteases cannot be readily monitored in the complex, in vivo setting, the functional status of the secreted proteases remains at issue. Nevertheless, when MMP-2, -9, -1, -3, or -7 are purposely expressed in MCF-7 cells as RXKR chimeras (R = Arg, X = any amino acid, K = Lys) that undergo efficient intracellular processing to their active forms within the trans-Golgi network by proprotein convertase family members before their extracellular secretion (13, 22), the engineered tumor cells continue to express the indolent activity characteristic of the parental cell population (Fig. S2). Interestingly, MMP-13, expressed either as the wild-type or RXKR mutant, is able to trigger significant BM degradation (Fig. S2). MMP-13-dependent BM remodeling did not, however, induce MCF-7 invasion, proliferation or angiogenesis (Fig. S2). Apparently, BM degradation alone is not sufficient to trigger the entire repertoire of activities associated with MT1-MMP or MT2-MMP activity as the tumor cells must also traffic through the interstitial collagen-rich CAM mesenchyme (11, 23).

Fig. 3.

MT1-MMP and MT2-MMP mediate MDA-MB-231 cell BM invasion and intravasation in vivo. (A) CAM invasion by MDA-MB-231 cells electroporated with a scrambled siRNA control alone or siRNAs directed against MT1-MMP, MT2-MMP, MMP-1, or MMP-9. In selected experiments, MDA-MB-231 cells electroporated with siRNAs directed against both MT1-MMP and MT2-MMP were transfected with mouse MT1-MMP or MT2-MMP expression vectors. Cancer cell invasion was visualized by fluorescent microscopy of CAM cross-sections after a 3-day incubation period. The upper CAM BM is marked by arrowheads. Red, type IV collagen; blue, DAPI staining for cell nuclei. (Scale bar, 50 μm.) (B) RT-PCR analysis of MT1-MMP and MT2-MMP expression in MDA-MB-231 cells following electroporation with scrambled, MT1-MMP- or MT2-MMP- specific siRNAs. (C–E) MDA-MB-231 cell invasion (C), angiogenic activity (D), and proliferative (E) indices were determined in CAM cross-sections. Results are expressed as the mean ± 1 SD of three or more experiments.

Tandem Expression of MT1-MMP and MT2-MMP Regulates the MDA-MB-231 Invasive Phenotype.

MDA-MB-231 cells express tissue-invasive activity in vivo that is largely dependent on Snail1 (4, 24). Consistent with the potential existence of a Snail1-MT-MMP axis, both the BM-invasive and pro-angiogenic activities of MDA-MB-231 cells correlate with their ability to express MT1-MMP as well as MT2-MMP (Fig. 3 A and B). To determine the degree to which MDA-MB-231 phenotype is linked to endogenous MT1-MMP or MT2-MMP activity, each of the proteases was silenced alone or in combination. Following knockdown of either MT1-MMP or MT2-MMP alone, MDA-MB-231 cells continue to degrade the underlying BM although invasive activity is attenuated significantly (Fig. 3 A and B). Despite the decrease in MDA-MB-231 invasion into the chick mesenchyme, pro-angiogenic activity and carcinoma cell proliferation continue in an unabated fashion (Fig. 3 C and D). By contrast, when MT1-MMP and MT2-MMP are both silenced, the BM remains intact, and invasion is inhibited significantly in tandem with comparable decreases in angiogenesis, proliferation and intravasation (Fig. 3 A–E and Fig. S3). Consistent with the absence of any siRNA-dependent off-target effects on cell function, the ability of MT1-MMP/MT2-MMP-silenced MDA-MB-231 cells to degrade the underlying BM, invade the CAM mesenchyme, proliferate, trigger angiogenic activity and intravasate is rescued following the expression of mouse orthologues of MT1-MMP or MT2-MMP that escape siRNA-mediated silencing (Fig. 3 A–E and Fig. S3). While MDA-MB-231 cells also express MMP-1 and MMP-9, secreted MMPs whose expression has previously been linked indirectly with carcinoma cell invasion or pro-angiogenic activities (25, 26), neither MMP-1 nor MMP-9 silencing affects BM transmigration, intravasation, or angiogenic responses (Fig. 3 A–E and Fig. S3).

Snail1 Triggers an MT1-MMP/MT2-MMP-Dependent Invasion Program.

To characterize directly the role of MT1-MMP and/or MT2-MMP in regulating MCF-7_Snail1 behavior, the expression of each of the membrane-anchored MMPs was examined in vivo following doxycycline induction of the invasive phenotype. Coincident with the ability of MCF-7_Snail1 cells to degrade and transmigrate the BM as well as trigger angiogenesis and tumor cell intravasation, both MT1-MMP and MT2-MMP expression are upregulated in response to Snail1 induction (Fig. 4 A–E). Following silencing of either MT1-MMP or MT2-MMP in MCF-7_Snail1 cells, invasion as well as angiogenesis is inhibited significantly, but not completely (Fig. 4 A–C). However, following the tandem knockdown of MT1-MMP and MT2-MMP, the ability of MCF-7_Snail1 cells to degrade the chick BM, mount an invasive response, stimulate angiogenesis or intravasate the host vasculature is blocked (Fig. 4 A–E). Under these conditions, changes in apoptosis are not observed (Fig. S4). As expected, the inability of MT1/MT2-MMP-silenced MCF-7_Snail1 cells to mount a BM-invasive program is rescued following expression of either mouse MT1-MMP or MT2-MMP (Fig. 4 A–E). Hence, the ability of Snail1 to induce an aggressive carcinoma cell-like phenotype characterized by BM invasion, pro-angiogenic activity, and intravasation is mediated by the combined expression of MT1-MMP and MT2-MMP.

Fig. 4.

Snail1 induces an MT1-MMP/MT2-MMP-dependent BM invasion program. (A) MCF-7_Snail1 cells were electroporated with a scrambled siRNA control or siRNAs directed against MT1-MMP, MT2-MMP, or both MT1-MMP and MT2-MMP. In selected experiments, cells co-electroporated with MT1-MMP and MT2-MMP siRNAs were transfected with mouse MT1-MMP or MT2-MMP expression vectors. (Scale bar, 50 μm.) Results are representative of three or more experiments performed. (B) RT-PCR analysis of MT1-MMP and MT2-MMP mRNA levels in non-induced and Snail1-induced MCF-7 cells cultured atop the CAM surface for 3 days as well as in induced cells wherein MT1-MMP, MT2-MMP, or both MT1-MMP and MT2-MMP were targeted with specific siRNAs. (C and D) CAM invasion (C) and angiogenic activity (D) were quantified as described. Results are the mean ± 1 SD of three or more experiments performed. (E) Intravasation by MCF-7_Snail1 cells was monitored by Alu-sequence PCR as described following silencing of MT1-MMP and MT2-MMP in tandem or after rescue with mouse MT1-MMP or MT2-MMP. Results are representative of three or more experiments performed.

Discussion

The first structural barrier encountered by invasive carcinoma cells is the BM, a 100- to 300-nm thick ECM composite dominated by an interwoven network of type IV collagen fibrils (1, 27). During EMT, normal as well as neoplastic cells exhibit an ability to perforate the underlying BM by largely uncharacterized mechanisms (1, 13). Whereas well-differentiated carcinoma cell lines that retain epithelial characteristics, such as MCF-7 cells, do not display BM-invasive activity, Snail1 induces these cells to assume an aggressive phenotype characterized by BM remodeling, transmigration, angiogenesis and intravasation. Furthermore, MCF-7_Snail1 cells display carcinomatous properties similar, if not identical, to those exhibited by MDA-MB-231 cells, a de-differentiated breast cancer cell line known to express high levels of Snail1 as well as other EMT-inducing transcription factors, including ZEB1 (4, 24, 28, 29). Importantly, both MCF-7_Snail1 and MDA-MB-231 cells degrade and invade the chick BM by a process that was inhibited completely by synthetic MMP inhibitors, raising the possibility that cancer- or chick-cell-derived MMPs, acting alone or in combination, drive tissue-invasive programs.

Multiple MMP family members are thought to participate in BM remodeling, but the ability of these enzymes to confer cancer cells with BM-degradative or BM-invasive activity in an in vivo setting remains controversial and has not been examined previously (1). As MT1-MMP, MT2-MMP, and MT3-MMP have recently been shown to degrade acellular sheets of intact BM recovered from animal or human tissues (13), we first sought to determine the impact of MT-MMP expression on MCF-7-BM interactions in vivo. Remarkably, MT1-MMP- or MT2-MMP-expressing MCF-7 cells not only acquire the ability to degrade the underlying BM, but to also mount a potent invasion program that coincides with the induction of an angiogenic response, increased cancer cell proliferation and tumor cell intravasation. The broad impact of MT1-MMP and MT2-MMP expression on cancer cell phenotype is further confirmed by silencing endogenous levels of MT1-MMP and MT2-MMP in MDA-MB-231 cells. In the absence of these two proteases, the ability of MDA-MB-231 cells to remodel or transmigrate the BM, to initiate angiogenesis, accelerate proliferation or intravasate is blocked completely, and neither cancer cells nor chick host cells are able to activate alternate pathways to re-engage the cancer invasion program. We considered the possibility that MT1-MMP or MT2-MMP might, in and of themselves, initiate EMT programs (30), but MT1-MMP- or MT2-MMP-expressing MCF-7 cells cultured in vitro maintain epithelial phenotypes and do not upregulate Snail1 levels (Fig. S5). More likely, in an in vivo environment, MCF-7 cells may be triggered to undergo EMT-mimetic programs by growth factors, chemokines or ECM degradation products generated as a consequence of MT1-MMP or MT2-MMP activity (Fig. S5) (1). Although earlier studies have suggested that MT3-MMP might display activity similar to MT1-MMP or MT2-MMP (13), MT3-MMP plays neither sufficient nor necessary roles in BM transmigration in vivo. Apparently in the dynamic system of the live chick CAM, either MT3-MMP-dependent BM degradation did not outstrip the ability of the CAM epithelium to maintain ECM architecture, or alternatively, MT3-MMP activity was muted by chick-derived inhibitors that preferentially target the protease.

The ability of MT1-MMP or MT2-MMP to trigger cancer cell invasion, angiogenesis, proliferation, and intravasation correlates closely with the ability of these proteases to degrade the underlying BM. However, the secreted MMP, MMP-13 (collagenase-3), likewise induced BM degradation, but was unable to support MCF-7 cell invasion into the interstitial matrix or any of the other sequelae associated with MT1-MMP or MT2-MMP expression. In vitro studies have failed to identify MMP-13 as a BM-degradative protease (13), raising the possibility that MMP-13 acts in collaboration with host cells or other proteinases expressed in vivo to remodel the BM. Nevertheless, while BM degradation has been suggested as a process sufficient to elicit an EMT-linked invasion program (31, 32), MMP-13 alone is unable to initiate this more complex response. As the BM interfaces with a dense, underlying ECM comprised largely of types I and III collagen, and only MT1-MMP and MT2-MMP can support invasive activity in collagen-rich interstitial tissues (8, 11, 23, 33), we posit that invasive cancer cells must not only be able to degrade the BM, but also to infiltrate type I/III collagen-rich tissues to access the host vasculature or induce angiogenesis. Although other secreted MMPs such as MMP-1, -2, -3, -7, or -9, have been linked indirectly to cancer cell invasion programs or cancer cell-induced neovascularization (19, 34), none of these proteinases played required roles in either BM degradation, transmigration, angiogenesis, or intravasation in our in vivo system.

Given the ability of MT1-MMP or MT2-MMP to phenocopy MCF-7_Snail1 cell behavior, we examined directly the role of these membrane-anchored MMPs during the Snail1-dependent EMT program in vivo. In accordance with our predictions, Snail1 induced the expression of MT1-MMP and MT2-MMP not only in tandem with BM degradation and invasion, but also angiogenesis and intravasation. At first glance, the ability of Snail1 to drive intravasation within a 72-h culture period in vivo is surprising. Yet, recent studies suggest that BM invasion and intravasation may occur at the earliest steps of carcinoma formation (35, 36). Indeed, circulating neoplastic cells have been shown to express either MT1-MMP or MT2-MMP in mouse models as well as human cancer (35, 37). Most remarkably, despite the ability of Snail1 to alter the expression of hundreds of gene products associated with motility, invasion, cell cycle control, apoptosis and angiogenesis (38–41), the ability of MCF-7_Snail1 cells to degrade and transmigrate the BM, invade local tissues, trigger an angiogenic response and intravasate were all inhibited following MT1-MMP and MT2-MMP silencing. Likewise, we have previously reported that Snail1 silencing inhibits MDA-MB-231 cell invasion in the CAM system although the downstream proteolytic effectors were not identified (4). While MDA-MB-231 cells also express ZEB1 (24, 28, 29), the tandem knockdown of MT1-MMP and MT2-MMP expression similarly abrogated the carcinomatous phenotype.

The mechanisms by which MT1-MMP or MT2-MMP can exert such global effects on cancer cell function remain to be established, but it is important to stress the fact that the substrate repertoire of these enzymes can extend far beyond the ECM itself to include hundreds of potential targets (42). While MT1-MMP can specifically regulate cancer cell proliferation within the context of the 3-D ECM by exerting collagenolytic activity (22), the growth promoting effects observed atop the CAM surface more likely occurs as a consequence of neovascularization. In this regard, MT1-MMP activity (like Snail1) has been linked to pro-angiogenic responses through largely uncharacterized processes that may involve VEGF, semaphorin secretion, or the generation of bioactive ECM products (38). MT2-MMP has not previously been shown to affect angiogenesis, intravasation, or cancer cell proliferation, but our results provide evidence that a second MT-MMP family member can induce an aggressive carcinoma-like phenotype in expressing cells. Finally, independent of the ability of MT1-MMP or MT2-MMP to hydrolyze substrates in their surrounding milieu, it is important to note that MT1-MMP/MT2-MMP-dependent proteolysis may indirectly exert complex effects on gene expression in the cancer cells themselves by altering local ECM rigidity as well as cell shape (1, 38). With an increasing body of evidence supporting the early expression of EMT-inducing transcription factors as well as MT1-MMP or MT2-MMP at the cancer cell invasion front, we posit that these two proteolytic enzymes, working alone or in concert, play critical roles in orchestrating carcinoma cell behavior.

Materials and Methods

Cell Culture and Transfection.

MCF-7 and MDA-MB231 cells were routinely maintained in complete DMEM supplemented with 10% FBS (FBS). Tet-on-Snail1 MCF-7 cells were provided by E. Fearon.

CAM Invasion and Intravasation Assays.

In vivo cancer cell invasion and intravasation assays were conducted using 11-day-old chick embryos wherein MCF-7, MDA-MB-231, or the transfected cells (10⁵ cells labeled with Fluoresbrite carboxylated polystyrene nanospheres of 45-nm diameter; Polysciences) (8) were seeded atop the CAM and incubated for 72 h as described previously (8, 18). The CAM was dropped without damaging the epithelial BM by applying gentle negative pressure at the air sac, and an opening of approximately 1 cm² was cut in the shell above the CAM with an electric drill. Tumor cell intravasation was detected as human-specific Alu-sequences by PCR on DNA extracted from the lower CAM after a 3-day culture period (Fig. S1). Chick GAPDH (chGADPH) serves as the loading control (14).

Additional Details.

For details of expression plasmids, siRNA electroporation and RT-PCR analysis, and histology and microscopy, see SI Materials and Methods.