Invasive matrix degradation at focal adhesions occurs via protease recruitment by a FAK–p130Cas complex

It is well known that invading tumor cells are surrounded by stroma and extracellular matrix (ECM) that is either displaced or degraded during the metastatic process. This degradative action is mediated by specialized organelles termed “invadopodia” that are centrally situated, dot-shaped actin-rich membrane protrusions that extend into the matrix-filled surroundings (Linder, 2007; Buccione et al., 2009; Caldieri et al., 2009). Active invadopodia are known to markedly potentiate invasion and are a hallmark of many types of cancer cells (Linder, 2007). Invadopodia perform both secretory and endocytic processes by mediating the targeted release of zinc-dependent matrix metalloproteinases (MMPs; Egeblad and Werb, 2002; Caldieri and Buccione, 2010) at the cell base and the subsequent internalization of the digested matrix for further processing by the lysosome (Coopman et al., 1996, 1998). Thus, cytoskeletal and membrane dynamics play an essential role in supporting invadopodia function (Gimona et al., 2008; Caldieri et al., 2009).

Both maturation and function of invadopodia are MMP dependent (Artym et al., 2006). To date, 25 MMPs have been identified in humans and are classified into either secreted MMPs or membrane-type MMPs (MT-MMPs). These two families share many structural elements except that MT-MMPs contain transmembrane or other membrane-tethering domains (Egeblad and Werb, 2002), and both have been identified at invadopodia. Among the known MMPs, MT1-MMP (MMP-14) is the best characterized and is believed to be the most prevalent form (Sabeh et al., 2009). MT1-MMP contains a single transmembrane domain, a 20-amino acid cytoplasmic tail, and is up-regulated in many types of cancer (Egeblad and Werb, 2002; Sato et al., 2005). Inhibition of MT1-MMP has been shown to significantly impair tumor cell invasion; however, the spatial and temporal regulation of MT1-MMP is not fully understood.

Although invadopodia are the primary site of action for MT1-MMP, it is also believed to function at other cellular locations such as lamellipodia where it is recruited via an interaction with CD44 (Mori et al., 2002). Moreover, Takino et al. (2006, 2007) have shown that in MT1-MMP–overexpressing Hela cells, ECM degradation takes place at the leading edge and in close proximity to focal adhesions (FAs). This is particularly interesting as FAs and invadopodia are related organelles that share many structural and regulatory components. In contrast to the dot-shaped invadopodia, FAs are streak-like structures at the basal membrane that bridge the actin cytoskeleton to the ECM, and the dynamic turnover of FAs is essential for the motility of both normal and tumor cells (Mitra et al., 2005; Gimona and Buccione, 2006). Despite distinct morphologies, FAs and invadopodia share many components including integrins, talin, paxillin, actin, cortactin, and dynamin, to name just a few (Linder and Aepfelbacher, 2003). A central component of FAs that provides both a scaffolding and signaling function is focal adhesion kinase (FAK). Although not observed at invadopodia (Bowden et al., 2006), FAK has been implicated in the regulation of invadopodia formation by modulating downstream signaling through Src and p130Cas (Hsia et al., 2003; Chan et al., 2009). Despite these molecular similarities, FAs are generally perceived to mediate cell migration but are not involved in matrix degradation.

Recently we have observed that a variety of cancer cell lines degrade matrix not only at invadopodia, but also at numerous peripheral sites that resemble FAs. Here we have confirmed that this peripheral ECM degradation represents bona fide FA sites, and have further revealed that this FA-type degradation is MT1-MMP dependent. Moreover, we find that this protease is targeted to FAs by a physical interaction between MT1-MMP and a FAK–p130Cas complex in response to the activation of Src. Formation of this FAK–p130Cas–MT1 complex is essential for the FA-centric ECM degradation as well as invasive migration of tumor cells through Matrigel.

Matrix degradation by a wide variety of different tumor cell lines appears to occur at linear, peripherally located FA-like sites as well as the well-characterized, centrally situated and dot-shaped invadopodia. To determine if this pattern represents the action of FAs, human fibrosarcoma cells (HT-1080) were plated on fluorescent gelatin-coated coverslips for 6 h, and stained for the FA marker vinculin. As shown in Fig. 1 a, many of the peripheral, linear degradation sites colocalized with vinculin. The enlarged images reveal that mature FAs (Fig. 1 a’) generated a solid, dark pattern of degradation whereas nascent FAs (Fig. 1 a’’) appeared to be in the process of degrading matrix. A similar result was obtained from several cell lines (Fig. S1). As many actin stress fibers terminate in FAs, we colocalized actin with a FA marker, and sites of matrix degradation as just one feature of many used to distinguish FAs from other adhesive structures (Fig. S1, c and d). It is well known that cells exert physical tension on the ECM at FA sites that can result in matrix tearing. To distinguish enzymatic activity from physical effects, HT-1080 cells were treated with 10 µM of the MMP inhibitor BB94 (Davies et al., 1993) as the cells were plated on gelatin. As shown in Fig. S2 (a–a’), overall degradation including the ones at FAs were completely inhibited, while no significant change in FA morphology was observed. From these observations we conclude that ECM degradation at FAs occurs by the proteolytic activity of MMPs.

Because the FA-like degradation was sharply focused rather than diffuse in nature, we hypothesized that a membrane-type MMP, not a secreted form, was responsible for degradation at FAs. MT1-MMP is the most prominent member of MT-MMPs and has been shown by many others to support ECM degradation (Sato et al., 2005; Itoh, 2006; Strongin, 2006). As shown in Fig. 1, c–c’, knocking down MT1 in HT-1080 cells (dashed line) significantly impaired ECM degradation, including the ones at FAs (Fig. 1 d). To further test the role of MT1-MMP in ECM degradation, we used a pancreatic cancer cell line (PANC-1) that does not express endogenous MT1 (Fig. 1 e). Although gelatin degradation was not observed by these cells even 24 h after plating (Fig. S2 b-b’), some FA-type degradations were observed in cells transfected with MT1-GFP just 2 h after plating on gelatin (Fig. 1 f). The same findings were obtained by cells expressing either an internal mCherry-tagged (see Fig. 5 h) or a nontagged MT1, indicating that this protease is essential for FA-mediated degradation.

As MT1-MMP mediates ECM degradation at FAs, it is likely that a FA component acts to bind and recruit this protease to these sites. To test this possibility, MT1-GFP was overexpressed in PANC-1 cells, then immunoprecipitated (IP) with a GFP antibody and blotted for several core FA proteins, including FAK, paxillin, vinculin, talin, and β1-integrin. Of all these proteins, FAK was by far the most prominent interacting partner (Fig. 2 a). As FAK localizes to FAs and is generally not considered an invadopodial component (Bowden et al., 2006; Chan et al., 2009), this FAK–MT1 interaction was predicted to occur at FAs. To define this novel interaction, we generated a series of FAK truncation mutants as illustrated in Fig. 2 b. PANC-1 cells expressing MT1-GFP were lysed and incubated with different GST-tagged FAK domains to test for interaction. As shown in Fig. 2, c and d, the proline-rich region (PRR) of FAK is sufficient to pull down MT1. To confirm this finding, MT1-GFP and SFB-tagged (SFB: S protein, FLAG, and streptavidin-binding peptide triple tagged) FAK WT or a PRR mutant were expressed in 293T cells, followed by a FAK IP with S protein beads. As predicted, FAKΔPRR did not interact with MT1 (Fig. 2 e, lane 3). Within the FAK PRR domain the most characteristic features are two conserved PXXP motifs (second and third in the full-length protein; Mitra et al., 2005; Wu et al., 2005). Mutations of these two motifs (FAK P2/3A) did not fully abolish, but significantly impaired FAK’s interaction with MT1-MMP (Fig. 2 e, lane 4). Taken together, these data suggest that two PXXP motifs within the PPR domain of FAK are essential to interact with MT1-MMP.

To test if FAK–MT1 association is important for matrix degradation at FA sites, HT-1080 cells were treated with FAK siRNA to knock down endogenous FAK (Fig. 2 f), then were plated on gelatin-coated coverslips for 16 h before IF analysis. Interestingly, degradation at FAs was significantly compromised in FAK knockdown cells (Fig. 2 g). To further confirm the role of FAK, FAK knockdown cells were rescued with siRNA-resistant rat-FAK. Surprisingly, WT FAK did not significantly increase degradation at FAs; however, expression of a FAK Y397F mutant rescued degradation to control levels (Fig. 2 h). As discussed later, this result is likely due to the fact that expression of WT FAK promotes FA turnover while the YF mutant is known to stabilize this structure (Hamadi et al., 2005; Wang et al., 2011). Accordingly, expression of a FAK double mutant, Y397F carrying the P2/3A mutation that reduces MT1 binding, did not rescue matrix degradation at FAs (Fig. 2 i). Taken together, we conclude that FAK interacts with MT1-MMP through the PRR domain and this interaction is important for FA-mediated degradation.

MT1-MMP is a transmembrane protein with a 20-amino acid cytoplasmic tail (CP; Fig. 3 a) that is known to interact with a variety of cellular proteins such as MTCBP-1 and AP2 (Uekita et al., 2004). To test if FAK also interacts with this domain, we generated a MT1 mutant missing the cytoplasmic tail (MT1ΔCP) and analyzed its interaction with FAK via GST pull-down assay. As predicted, removal of the CP domain completely abolished the FAK–MT1 interaction (Fig. 3 b), thus this mutant is likely defective in mediating degradation at FAs. Indeed, in contrast to MT1 WT (Fig. 1 f and Fig. S4 c), MT1 ΔCP expressing PANC-1 cells failed to generate any gelatin degradation at FAs (Fig. 3c). As this effect could result from defects in either MT1 targeting or enzymatic activity, we generated a chimeric protein by replacing the MT1 cytoplasmic tail with the focal adhesion targeting domain of FAK (FAT). Thus, the cytoplasmic tail truncated protease could be constitutively targeted to FAs. As shown in Fig. 3 d, PANC-1 cells expressing MT1ΔCP-FAT generated excessive degradation at FA sites (Fig. 3 e), indicating that MT1ΔCP remains enzymatically active but cannot be targeted to FAs.

Interaction between MT1-MMP and FAK is essential for FA-centric matrix degradation; however, whether this interaction is direct was unclear. To test this, we purified His-MT1 protein from Escherichia coli, and performed direct binding analysis with GST-PRR of FAK. No direct interaction was detected (unpublished data), suggesting the participation of an adaptor. One candidate known to bind directly to both MT1 (Gingras et al., 2008; Gonzalo et al., 2010) and to the second and third PXXP motifs of FAK is p130Cas (Harte et al., 1996; Mitra et al., 2005). However, whether these three proteins form a functional complex is not clear. To test this possibility, we transfected 293T cells with either MT1 alone or together with p130Cas and then performed pull-down assays with the purified GST-PRR of FAK. As shown in Fig. 4 a, p130Cas strongly binds to GST-PRR, and overexpression of p130Cas significantly enhances the amount of MT1-MMP in the pulldown, suggesting that p130Cas facilitates MT1–FAK interaction. To test if these three proteins interact endogenously, FAK was immunoprecipitated from HT-1080 cells, and both p130Cas and MT1 were detected in the Co-IP (Fig. 4 b). Importantly, the FAK–MT1 interaction was attenuated in p130Cas knockdown cells (Fig. 4 b), suggesting that p130Cas mediates the interaction between MT1-MMP and FAK.

Although it is known that p130Cas binds to FAK via the N-terminal SH3 domain (Harte et al., 1996), and FAK recruits p130Cas to FAs (Fig. S3, a–b’), it remained unclear how p130Cas might interact with MT1. To address this question, a series of SFB-p130Cas truncation mutants were generated (Fig. 4 c) and coexpressed with MT1-GFP in 293T cells followed by Co-IP analysis. As shown in Fig. 4 d, the substrate domain (SD) of p130 Cas was sufficient to bind to MT1, and the deletion of this domain abolished the interaction (Fig. 4 e). Although p130Cas is an established signaling molecule at FA sites (O’Neill et al., 2000), its function during matrix degradation remains elusive. To test this premise, HT-1080 cells were treated with p130Cas siRNA for 72 h to knock down the endogenous protein (Fig. 4 f), and then were plated on gelatin-coated coverslips. As shown in Fig. 4 g, p130Cas knockdown significantly reduced FA-mediated degradation, while FA morphology remained unchanged (Fig. S3, c–c’). To further test p130Cas’s effect, p130Cas WT or ΔSD (Fig. 4, h and i) were reexpressed and tested for matrix degradation. Importantly, p130Cas WT but not the MT1-binding mutant (ΔSD) rescued ECM degradation at FAs. These data suggest that ECM degradation at FAs is regulated by a novel mechanism involving a FAK–p130Cas–MT1 complex.

Src kinase is a central signaling molecule known to promote ECM degradation (Yeatman, 2004; Courtneidge et al., 2005; Kelley et al., 2010) by phosphorylating a variety of substrates including Tyr573 on the cytoplasmic tail of MT1-MMP (Nyalendo et al., 2007). Consistent with this premise, we observed a marked tyrosine phosphorylation of MT1-MMP when coexpressed with active Src (Y530F; Fig. 5 a). As expression of an MT1-Y573F mutant did not eliminate the phospho-tyrosine signal (unpublished data), we assume that other tyrosine substrates exist on MT1-MMP. Importantly, however, phosphorylation of Y573 appears to facilitate the MT1–p130Cas interaction as the MT1 Y573F mutant exhibited a significantly reduced affinity to p130Cas compared with the MT1 WT in a Co-IP assay (Fig. 5 b). To confirm a role for Src, HT-1080 cells were treated with the Src family kinase inhibitor PP2 (10 µM, 3 h) before IP with p130Cas antibody. As shown in Fig. 5 c, PP2 treatment significantly reduced p130Cas–MT1 interaction compared with the DMSO control. Because p130Cas adapts MT1-MMP to FAK, we predicted that a Y573F mutant should also reduce the MT1–FAK interaction. This was indeed the case, as confirmed by both Co-IP and GST pull-down assay (Fig. S4, a and b). To test if Src activity is important for ECM degradation at FAs, HT-1080 cells were plated on gelatin in the presence of 10 µM PP2. As expected, PP2 treatment significantly inhibited ECM degradation at FAs (Fig. 5, e–e’). Consistent with this result, PANC-1 cells expressing MT1 Y573F showed a reduced FA-type degradation compared with MT1 WT–expressing cells (Fig. S4, c–d’). Taken together, these data suggest that Src regulates matrix degradation at FAs by phosphorylating MT1-MMP on Y573 that promotes MT1’s interaction with the FAK–p130Cas complex.

The observations described above suggest that Src-mediated formation of a FAK–p130Cas–MT1 complex regulates ECM degradation at FAs. It is intriguing to test if invadopodia are regulated by the same mechanism. It is well established that Src kinase is a central regulator of invadopodia (Courtneidge et al., 2005). Surprisingly the numerous, centrally located degradation sites generated by the HT-1080 cells were not attenuated by PP2 treatment (Fig. 5 e). Importantly, these central sites did not co-stain for invadopodial markers such as actin and cortactin (Fig. S5, a and c), suggesting that these are distinct structures. In contrast, BxPC-3 cells, which do form numerous classical invadopodia (Fig. S5, b and d), exhibited a marked inhibition in the formation of centrally located ECM degradation sites by PP2 treatment (Fig. 6, d and e). Therefore, we tested if a disruption of the FAK–p130Cas–MT1 complex in BxPC-3 cells affects ECM degradation at invadopodia as it did at FAs (Figs. 2 and 4). BxPC3 cells were treated with siRNAs against MT1, FAK, or p130Cas, then plated on gelatin for 16 h, and stained for the invadopodia marker cortactin (Fig. 6, a–c). As expected, the MT1 knockdown affected invadopodia-type degradation due to an inhibition of invadopodia formation (Fig. 6 a). In contrast, FAK or p130Cas knockdown did not affect invadopodia function (Fig. 6, b and c), suggesting that, unlike that observed at FAs, MT1-MMP’s function at invadopodia is independent of the FAK–p130Cas complex.

The observations described above have implicated a novel role for a FAK–p130Cas–MT1 complex in the degradation of ECM at FAs. To test if this FA-centric protein complex contributes to tumor cell invasion, an in vitro Boyden chamber invasion assay was implemented. As PANC-1 cells do not express endogenous MT1-MMP, cells were transfected to express MT1 WT, Y573F, or MT1ΔCP-FAT (Fig. 7 a). Cells were suspended in serum-free medium and plated in the upper chamber 24 h after transfection, then challenged to invade through a Matrigel-coated filter into the lower chamber containing full serum (10% FBS) medium. After a 36-h period, cells were fixed and the number of invaded cells per field was quantified. Importantly, the protein components that facilitate ECM degradation at FAs (MT1 WT and MT1ΔCP-FAT, Fig. 3 e) also significantly enhanced invasion through the filter compared with MT1 Y573F or GFP vector (Fig. 7 d). As a reciprocal approach, HT-1080 cells were transfected twice with siRNA against FAK or p130Cas, and subsequently rescued with either WT or mutant constructs (Fig. 7, b and c). Interestingly, cells with reduced levels of FAK or p130Cas exhibited significantly reduced invasion compared with siRNA control cells, which is consistent with previous reports (Ueda et al., 2003; Chan et al., 2009; Cunningham-Edmondson and Hanks, 2009). Moreover, reexpression of the respective WT proteins rescued invasion whereas expression of the mutants FAK P2/3A and CasΔSD that perturb formation of the trimeric complex did not (Fig. 7 e). Taken together, these data suggest that the targeting of MT1-MMP to FAs by the FAK–p130Cas complex is important for cancer cell invasion.

In this study we provide evidence for a novel, unconventional function of focal adhesions that entails the regulated degradation of the surrounding matrix (Fig. 1 and Fig. S1). This process is observed in a wide variety of human tumor cell types such as the fibrosarcoma cell line HT1080 and several different pancreatic tumor cell lines including BxPC3, HuPT3, CFPAC-1, and HPAF. It is mediated by, and dependent upon, MT1-MMP as cells without this protease do not degrade matrix at these sites (Fig. 1 and Fig. S2). In addition, we have found that the MT1-MMP is targeted to FAs through a novel interaction with a FAK–p130Cas complex. Disruption of these interactions results in a significant reduction of ECM degradation at FAs (Figs. 2–4), but not at invadopodia (Fig. 6), suggesting a FA-specific mechanism. Finally, the formation of this trimeric complex is regulated by Src, and is essential for tumor cell invasion in vitro.

Two of the interesting central concepts addressed in this study focus on the mechanics of MT1-MMP recruitment to FAs and the functional equilibrium of matrix degradation occurring at FAs and invadopodia during cancer invasion. Although degradation predominates at invadopodia in most tumor cells, FA-based degradation does occur and is not from mechanical tearing of the ECM, as inhibiting MMP activity with the drug BB94 abolished FA-centric degradation. This is further supported by the fact that MT1-MMP–deficient cells (PANC-1) or HT-1080 cells with MT1 knockdown failed to degrade ECM at FAs. These data suggest that matrix degradation at FAs, like that at invadopodia, is a proteolytic event, and the essential role of a membrane-type MMP (MT1-MMP) is consistent with the confined degradation pattern at FAs.

Key evidence supporting MT1-MMP’s function at FAs is an interaction with FAK. Although FAs and invadopodia share a substantial number of components, FAK is generally believed to be FA specific. Accordingly, we hypothesized that MT1-MMP may be selectively recruited to FAs by FAK. Further analysis revealed that FAK–MT1 interaction is mediated by the PRR domain of FAK and the cytoplasmic tail of MT1-MMP (Figs. 2 and 3). Functionally, FAK knockdown in HT-1080 cells resulted in a reduction of FA-mediated degradation, and consistent with a previous report (Chan et al., 2009) invadopodia-type degradation was not affected (Fig. 6 b), again supporting FAK’s distinct roles between these two structures. However, this defect could not be rescued by reexpression of WT FAK. We believe this is due to the fact that FAK overexpression promotes FA turnover (Ezratty et al., 2005; Mitra et al., 2005) and thus reduces the time allotted for matrix degradation to occur. This concept is supported by a subsequent finding that FAK Y397F, a mutant known to inhibit FA turnover (Hamadi et al., 2005), exhibited a significant rescue of FA-based degradation (Fig. 2), and importantly, no rescue was observed by FAK Y397F carrying the two PXXP mutations (P2/3A) that reduce MT1 binding. Moreover, cells expressing a MT1-MMP mutant that cannot interact with FAK (MT1ΔCP) were defective in degrading ECM at FAs (Fig. 3). These data strongly suggest that the FAK–MT1 interaction is important for FA-based degradation, but is dispensable for degradation by invadopodia. As a direct interaction between FAK and MT1 was not detected, we had searched for an adaptor protein that could act as a functional linker. Based on previous reports (Harte et al., 1996; Gingras et al., 2008) we pursued, and subsequently found, that the FAK–MT1 connection is largely mediated by p130Cas (Fig. 4).

An interesting component of this study is the Src-dependent regulation of protease–adaptor interactions. Consistent with the findings of others (Gingras et al., 2008; Gonzalo et al., 2010), we observed that phosphorylation of a specific tyrosine (Tyr 573) on the cytoplasmic tail of MT1 by Src facilitates interactions with p130Cas (Fig. 5) and FAK (Fig. S4). It is well known that Src signaling promotes ECM degradation (Yeatman, 2004; Courtneidge et al., 2005; Kelley et al., 2010), yet it appears that Src regulates FAs and invadopodia through distinct mechanisms. For example, Src-mediated phosphorylation of cortactin, AMAP1, paxillin, and p190 RhoGAP (Stylli et al., 2008) are essential for invadopodia-mediated degradation, presumably by regulating the assembly of this organelle. In contrast, Src phosphorylation of MT1-MMP showed no obvious effects on FA structure but did enhance MT1-MMP’s recruitment to FAs (Fig. 5, I and j). This is both significant and consistent with our biochemical observations that Src facilitates MT1 interactions with FAK–p130Cas (Fig. 5 and Fig. S4). Moreover, FAK or p130Cas knockdown (Figs. 2 and 4) significantly compromised degradation at FAs, but not at invadopodia (Fig. 6, b and c). From these contrasting findings we believe that matrix degradations at these two structures are related yet distinct.

Despite the fact that FAs and invadopodia share many structural components, the relative contributions of these two organelles to cell migration and invasion remain unclear. One interesting study (Oikawa et al., 2008) has suggested that FAs could be transformed into podosomes, an invadopodia homologue, in a Src-dependent manner. As many signaling molecules, including Src, are elevated or activated in cancer cells, it is attractive to predict that FAs could acquire some invadopodia-like properties during cancer progression, while largely maintaining its FA morphology and function. The Boyden chamber–based invasion assay (Fig. 7) suggests that FA-centric degradation is relevant for cancer invasion. As demonstrated in Figs. 2–4, disruption of the FAK–p130Cas–MT1 complex significantly impaired ECM degradation at FAs, as well as invasion through Matrigel (Fig. 7). Admittedly, due to the multifunctional nature of these proteins, we cannot rule out the possibility that the mutant FAK, p130Cas, and MT1-MMP proteins affect invasion through other mechanisms.

Finally, it is attractive to speculate that FA-based degradation may play a central role during cancer invasion in vivo, in which cells acquire a compact spindle-like morphology (Cukierman et al., 2001, 2002) while FAs become smaller and predominate at the pseudopodia–ECM contact at the leading edge (Friedl and Wolf, 2009; Wolf and Friedl, 2009; Kubow and Horwitz, 2011). From this organization it is likely that in situ FAs act to macerate, degrade, and reorganize matrix at the front edge and allow the cell body to migrate through. Although the biology of FAs in situ remains elusive and a topic of debate (Fraley et al., 2010; Kubow and Horwitz, 2011), it has been reported that the leading edge localization of FAK–p130Cas is important for invasion (Hsia et al., 2003), and proteolytic activity has been observed at the pseudopodia–ECM contacts (Wolf and Friedl, 2009). Moreover, specific targeting of an enzymatic defective chimeric MT1 to FAs significantly hinders cancer invasion (Takino et al., 2007). All of these related findings suggest that FA-mediated ECM degradation could play a significant role in the invasive process. It will be of great interest to define how FAs maintain a balance between adhesive and degradative properties.

MT1-MMP antibody (MAB3328) was purchased from Millipore. The antibodies against FAK, paxillin, and p130Cas were from BD. Anti-Src antibody (sc-18) and anti-GST (sc-138) were purchased from Santa Cruz Biotechnology, Inc. The phospho-Src family antibody pY416 was from Cell Signaling Technology, the monoclonal vinculin antibody was from Sigma-Aldrich, anti-Flag antibody from Cell Signaling Technology, and anti-GFP antibody from Roche. Goat anti–rabbit or goat anti–mouse secondary antibodies conjugated to either Alexa 488 or 594 were from Invitrogen. BB-94 was purchased from Tocris Bioscience. All the other chemicals and reagents, unless otherwise stated, were from Sigma-Aldrich.

Pancreatic cancer cell lines BxPC3, Panc 04.03 and PANC-1 cells were purchased from American Type Culture Collection (Manassas, VA). HT1080 cells were a gift from Dr. Zhenkun Lou (Mayo Clinic, Rochester, MN). BxPC-3 and Panc04.03 cells were maintained in RPMI plus 10% fetal bovine serum, and PANC-1 and HT1080 cells were maintained in DME plus 10% fetal bovine serum (Invitrogen), and all cell lines had 100 U/ml penicillin and 100 µg/ml streptomycin in 5% CO₂ at 37°C. Cells were cultured in T-75 flasks (Thermo Fisher Scientific).

MT1-MMP-GFP and MT1-mCherry constructs were gifts from Drs. Roberto Buccione (Consorzio Mario Negri Sud, Chieti, Italy) and Philippe Chavrier (Institut Curie, Paris, France). FAK was amplified from a rat brain cDNA library and inserted into pEGFP-C1 or SFB vector (pIRES2-EGFP with S peptide, Flag tag, and streptavidin-binding peptide) as described previously (Wang et al., 2011). p130Cas constructs were subcloned from pEBG-p130cas (Addgene). Constructs were transfected with Lipofectamine 2000 (Invitrogen) according to the manufacturer’s instructions. siRNAs targeting MT1-MMP (D-004145-02), FAK (M-003146-02), and p130Cas (M-020465-01) were purchased from Thermo Fisher Scientific and transfected with Lipofectamine RNAiMAX (Invitrogen) following the standard protocol. Knockdown and reexpression experiments were performed as follows: two rounds of siRNA transfection in a 24-h interval, and 24 h after the second transfection, cells were rescued with indicated constructs.

Cells were lysed with NETN buffer (0.5% NP-40, 150 mM NaCl, 50 mM Tris, and 1 mM EDTA) at 4°C. Cell debris was removed by centrifugation, and the supernatant was incubated with 5 µg of the appropriate antibody and protein A beads at 4°C for 4 h. For the S protein IP, cell lysate was incubated with S protein Agarose (EMD) at 4°C for 2 h. The pellet was washed with NETN buffer three times, eluted in Laemmli sample buffer, and analyzed by Western blot as described previously (Wang et al., 2011).

Cells were grown on 22-mm coverslips for transfections and immunocytochemistry. Cells were fixed in formaldehyde and processed as described previously (Wang et al., 2011). Images were taken with an AcioObserver microscope (Carl Zeiss) and Orca II camera (Hamamatsu Photonics), analyzed with iVision software (BioVision Technologies) and processed with Adobe Photoshop (Wang et al., 2011).

Acid-washed coverslips were first coated with 50 mg/ml poly-l-lysine for 20 min at room temperature, and then fixed with 0.5% glutaraldehyde for 15 min. Gelatin matrix was prepared by mixing 0.2% gelatin and Oregon Green gelatin (Invitrogen) at an 8:1 ratio. After coating for 10 min, coverslips were washed with PBS and quenched with 5 mg/ml sodium borohydride for 15 min followed by extensive wash. For degradation assay, cell suspensions were plated on top of these coated coverslips for designated amounts of time, and subjected to IF analysis as described above. Cells that degraded ECM at FA sites were scored as positive, and more than 600 random selected cells were quantified (n > 200, three independent experiments). The percentage of cells with FA-mediated degradation was plotted and statistical significance was analyzed by two-sample t test.

This assay was performed in a Boyden chamber (#BW200L; Neuro Probe) with #PFA 8 filter (Neuro Probe). The filters were coated with 1 mg/ml growth factor–reduced Matrigel as described previously (Hauck et al., 2002). The lower chamber was filled with full serum medium, and cells suspended in serum-free medium were plated on the upper chamber. Cells were allowed to invade for 24 or 36 h at 37°C before fixing and IF analysis.

Fig. S1 shows ECM degradation at FAs in multiple cell lines. Fig. S2 shows that deficiency or inhibition of MT1-MMP abolishes ECM degradation at FAs. Fig. S3 shows that FA localization of p130Cas is FAK dependent, and that p130Cas knockdown does not affect FA integrity. Fig. S4 shows that phosphorylation on MT1 Y573 facilitates MT1–FAK interaction, as well as ECM degradation at FAs. Fig. S5 shows cell model–specific FA and invadopodia matrix degradation patterns.

Thanks to Robbin Eppinga and Kristi Simmons for reading the manuscript and helpful comments. Special thanks to Shaun Weller and Eugene Krueger for providing images as well as preparation of the manuscript.

This study was supported by National Cancer Institute grant CA104125 to M.A. McNiven and the Optical Morphology Core of the Mayo Clinic Center for Cell Signaling in Gastroenterology (P30DK84567).

The authors declare no conflicts of interest.