Search tips
Search criteria 


Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
Cancer Res. Author manuscript; available in PMC 2010 October 25.
Published in final edited form as:
PMCID: PMC2962982

TGF-β1 + EGF-Initiated Invasive Potential in Transformed Human Keratinocytes Is Coupled to a Plasmin/MMP-10/MMP-1–Dependent Collagen Remodeling Axis: Role for PAI-1


The phenotypic switching called epithelial-to-mesenchymal transition is frequently associated with epithelial tumor cell progression from a comparatively benign to an aggressive, invasive malignancy. Coincident with the emergence of such cellular plasticity is an altered response to transforming growth factor-β (TGF-β) as well as epidermal growth factor (EGF) receptor amplification. TGF-β in the tumor microenvironment promotes invasive traits largely through reprogramming gene expression, which paradoxically supports matrix-disruptive as well as stabilizing processes. ras-transformed HaCaT II-4 keratinocytes undergo phenotypic changes typical of epithelial-to-mesenchymal transition, acquire a collagenolytic phenotype, and effectively invade collagen type 1 gels as a consequence of TGF-β1 + EGF stimulation in a three-dimensional physiologically relevant model system that monitors collagen remodeling. Enhanced collagen degradation was coupled to a significant increase in matrix metalloproteinase (MMP)-10 expression and involved a proteolytic axis composed of plasmin, MMP-10, and MMP-1. Neutralization of any one component in this cascade inhibited collagen gel lysis. Similarly, addition of plasminogen activator inhibitor type 1 (SERPINE1) blocked collagen degradation as well as the conversion of both proMMP-10 and proMMP-1 to their catalytically active forms. This study therefore identifies an important mechanism in TGF-β1 + EGF-initiated collagen remodeling by transformed human keratinocytes and proposes a crucial upstream role for plasminogen activator inhibitor type 1–dependent regulation in this event.


Epithelial tumor progression, from a relatively indolent to a more aggressive phenotype, is frequently accompanied by acquisition of a plastic phenotype reminiscent of a developmental program called epithelial-to-mesenchymal transition (EMT; ref. 1). This process is typified by loss of normal epithelial properties (e.g., cell polarity and junctional complexes) and a gain in mesenchymal traits (expression of vimentin and smooth muscle actin and enhanced cell motility; ref. 2). Although essential during embryonic development, EMT is relatively limited in the adult organism, occurring during wound repair or, more atypically, in advanced pathologies largely in response to specific growth factors associated with tumor progression (35). Epidermal growth factor (EGF) receptor amplification and an altered cellular response to transforming growth factor-β (TGF-β), for example, accompany the progression of epithelial tumor cells from a benign phenotype to an aggressive, metastatic carcinoma (68). During this pathologic EMT, and despite increased autocrine/paracrine expression of TGF-β, cells become refractory to the normally growth-suppressive effects of TGF-β family members. Mouse models of multistage skin carcinogenesis support the concept that TGF-β functions as a tumor suppressor in the early stages of benign growth; in late stage tumors, however, TGF-β accelerates malignant conversion (6). Down-regulation of TGF-β receptors, alterations in TGF-β-dependent Smad signaling components, or a combination of both appear to contribute to this functional switch (8, 9).

TGF-β likely promotes tumor-invasive properties through expression of genes that encode stromal remodeling proteins, which paradoxically support matrix-disruptive as well as stabilizing processes. Structural extracellular matrix proteins such as fibronectin and collagen (10, 11) are up-regulated by TGF-β in conjunction with their proteolytic regulators, including plasminogen activator inhibitor type 1 (PAI-1; ref. 12) and matrix metalloproteinase (MMP)-1, -3, -9, -10, -11, and -13 (1315). Similar to TGF-β, EGF-dependent signaling contributes to up-regulation of several MMPs (1519) and is enhanced through increased receptor levels in various cancers (7).

Stringent temporal and spatial controls on MMP activation are essential for maintaining tissue homeostasis. In addition to transcriptional regulation, MMP-dependent activities are also modulated through proteolytic activation (20). Proteolytic cascades within the pericellular environment are largely initiated through conversion of matrix plasminogen to the broad-spectrum protease plasmin, which, in turn, directly converts several proMMPs to their active form and triggers a positive feedback mechanism for MMP activation (21). Indeed, regulation of plasminogen activation may substantially affect MMP-dependent remodeling processes and thereby cellular invasive traits. The ability of TGF-β and/or EGF stimulation to also up-regulate PAI-1 in several cell types (this study; refs. 12, 22) provides a potential mechanism for upstream negative regulation or titration of the MMP cascade.

The immortalized adult human keratinocyte cell line HaCaT (23) harbors genetic changes similar to those that accompany progression of a normal keratinocyte to an invasive squamous cell carcinoma (24). Stimulation of activated ras-expressing HaCaT II-4 cells with TGF-β alone or, more effectively, a combination of TGF-β1 and EGF promotes a highly plastic phenotype typified by loss of E-cadherin and de novo synthesis of N-cadherin and vimentin (4, 25). The ability of TGF-β1 and/or EGF to elicit EMT-related responses such as these in a more physiologically significant model, however, has not been explored. This article describes the use of a three-dimensional collagen gel system to evaluate proteolytic events associated with TGF-β1 + EGF-stimulated EMT and collagen invasion by HaCaT II-4 keratinocytes. The invasive potential of these keratinocytes was coupled to a plasmin/MMP-10/MMP-1–dependent collagen-remodeling axis, and a role for PAI-1 as a critical upstream regulator of this remodeling process was established.

Materials and Methods


Vitrogen (Cohesion Technologies) or PureCol (Inamed; Advanced BioMatrix) provided sources of bovine collagen type 1. Both products yielded comparable results and were used interchangeably. Where indicated, FITC-labeled collagen type 1 (Sigma-Aldrich) or DQ FITC-labeled collagen type 1 (Molecular Probes/Invitrogen) were added to monitor gel degradation. Recombinant human TGF-β1 (R&D Systems) was used at 1 ng/mL and recombinant human EGF (Upstate/Millipore) at 10 ng/mL. Plasminogen, aprotinin, E-64, amiloride, o-phenylenediamine dihydrochloride, hydrogen peroxide, and phosphate-citrate buffer were from Sigma-Aldrich. Recombinant human PAI-1 protein and α2-antiplasmin were from Calbiochem. GM6001 was obtained from Chemicon/Millipore and AG1478 was obtained from Biosource International/Invitrogen. Immunofluorescence antibodies included tubulin (clone DM1A; Sigma), vimentin (LN6 Ab-1; Calbiochem), E-cadherin (clone 36), and N-cadherin (clone 32) from BD Biosciences; MMP-10 from Santa Cruz Biotechnology; and 4′,6-diamidino-2-phenylindole stain, phalloidin-594, and Cell Tracker CMTPX from Molecular Probes/Invitrogen. For Western blotting, antibodies to MMP-1 and MMP-10 or biotinylated antibodies to MMP-1 and MMP-10 were obtained from R&D Systems; antibodies to actin or extracellular signal-regulated kinase 1/2 were from Santa Cruz Biotechnology. Antibodies against human plasminogen and PAI-1 (neutralizing) were from American Diagnostica. Neutralizing antibodies to MMP-1 and MMP-10 were obtained from R&D Systems. A polyclonal antibody to PAI-1 was used for ELISA and immunofluorescence. Unless otherwise indicated, Alexa Fluor 488 (green) or 594 (red) conjugates (Molecular Probes/Invitrogen) were used for immunocytochemistry detection; horseradish peroxidase conjugates (Pierce Biotechnology) were used for Western blot and ELISA analyses.

Cell culture

HaCaT II-4 keratinocytes were maintained in low-glucose DMEM supplemented with 10% fetal bovine serum (Life Technologies/Invitrogen). Cells were harvested with trypsin/EDTA, washed with PBS, and seeded onto collagen in serum-free Advanced DMEM overnight before stimulation with TGF-β1 and/or EGF. Phenol red-free medium (Life Technologies/Invitrogen) was used in fluorescence assays.

Collagen gel-based studies

Collagen type 1 was neutralized according to the manufacturer’s instructions using 10× PBS and 0.1 N NaOH and then diluted to 1.8 mg/mL (unless stated otherwise) with DMEM. Gels were polymerized in 48-well tissue culture plates (150 μL), OptiCell Chambers (1.0 mL; USA Scientific), MatTek glass bottom dishes (200 μL; MatTek), or onto cell culture inserts (20 μL, 700 μg/mL) at 37°C for 2 to 3 h. For OptiCell-based invasion assays, 2 × 105 cells were added in 1 mL Advanced DMEM to gels polymerized vertically within the chamber. Cells were viewed on an inverted Olympus IX70 by laying the chamber on its side and images were captured with Image-Pro Plus software. For Transwell invasion assays, 1 × 105 cells were seeded onto thin collagen gels in Advanced DMEM. Invading cells were visualized with 4′,6-diamidino-2-phenylindole and counted in four random fields. For collagen gel dissolution assays, 5 × 104 cells were seeded onto polymerized gels in Advanced DMEM; plasminogen (5–20 μg/mL) was added 24 to 48 h post-growth factor stimulation for up to 24 h. Cells were pretreated with inhibitors or neutralizing antibodies as indicated. In experiments with anti-PAI-1 neutralizing antibody, time (not shown) and/or plasminogen concentration were reduced to capture differences in the state of dissolution with increasing antibody concentration. Cells on intact gels were fixed in 3% paraformaldehyde before viewing on an inverted Olympus IX70 microscope. To quantify collagen degradation, FITC-labeled collagen (25 μg/mL) was incorporated into polymerized gels. Cells were stimulated in phenol red-free DMEM and incubated with plasminogen for 7.5 h and 100 μL conditioned medium was removed for fluorescence spectroscopy using Synergy HT microplate reader equipped with KC4 software (BioTek Instruments).


HaCaT II-4 cells were seeded onto collagen-coated coverslips (50 μg/mL) or onto collagen gels polymerized in MatTek glass-bottomed dishes. Following treatment, cells were fixed in 3% paraformaldehyde, permeabilized, blocked, and incubated with primary and secondary antibodies for 1 h each. Coverslips were mounted using ProLong Gold with 4′,6-diamidino-2-phenylindole and viewed on an Olympus BX61 microscope with Image-Pro Lab software version 3.6.5. or an Olympus IX70 inverted scope with Image-Pro Plus software. For visualization of collagen digestion, cells were seeded onto coverslips coated with collagen type 1 (50 μg/mL) + DQ FITC-labeled collagen type 1 (25 μg/mL).

Protein analysis

Collagen gels were digested with collagenase D; cells were separated from digested collagen by centrifugation and lysed in a 50 mmol/L HEPES containing 150 mmol/L NaCl, 1% Triton X-100, 0.5% deoxycholate, 1% NP-40, 10 mmol/L NaF, 1 mmol/L orthovanadate, and protease inhibitors; and extracts were probed for N-cadherin and E-cadherin. Western blot analysis of MMP-1 and MMP-10 used conditioned medium from cells stimulated with TGF-β1 + EGF followed by incubation with plasminogen ± inhibitors (as indicated). The Human MMP Antibody Array 1.1 from RayBio (RayBiotech) was used to detect changes in MMP protein levels in conditioned medium. For measurement of PAI-1 levels by ELISA, 1.2 × 105 cells seeded on collagen type 1-coated, BSA-blocked wells were maintained under serum-free conditions for 6 h, pretreated with AG1478, as indicated, and then stimulated with TGF-β1 and/or EGF overnight. Cells were fixed with 3% paraformaldehyde, permeabilized, blocked, and incubated with PAI-1 polyclonal antibodies for 1 h followed by a horseradish peroxidase-conjugated secondary antibody. Cell layer PAI-1 was detected by colorimetric assay using an o-phenylenediamine dihydrochloride substrate and measured by spectrophotometer at 492 nm. Results were normalized to cell number by measuring the level of cell-associated crystal violet staining.

Statistical analysis

The Student’s t test for two samples, assuming unequal variance, was used to compare conditions within a group. Two-tailed values with P ≤ 0.05 were considered significant.


HaCaT II-4 keratinocytes stimulated with TGF-β1 + EGF undergo EMT and invade collagen gels in a MMP-dependent manner

To recapitulate events associated with cutaneous EMT in a relevant context, p53 mutant, Ha-ras-expressing human keratinocytes (HaCaT II-4 cells; ref. 23) were cultured on a collagen coat (Fig. 1A) or onto a more physiologically related three-dimensional collagen gel (Fig. 1B and C) and simultaneously treated with TGF-β1 and EGF to mimic the increased TGF-β expression/EGF receptor signaling characteristics of late-stage tumors. Under these conditions, EGF stimulation was mitogenic, whereas TGF-β1 maintained its growth-suppressive activity even in the presence of EGF (Supplementary Fig. S1). Whereas HaCaT II-4 colonies cultured on a three-dimensional collagen gel appeared more compact than cells cultured on a collagen coat, dually stimulated cells displayed traits typical of an EMT (2) on both substrates including increased scattering (Fig. 1A and B; tubulin), de novo vimentin, and N-cadherin expression (Fig. 1A–C) as well as loss of E-cadherin at cell-cell junctions (Fig. 1A–C). To our knowledge, these observations represent the first evidence that EMT-related events take place in human keratinocytes cultured in a three-dimensional environment.

Figure 1
TGF-β1 + EGF costimulation induces EMT-like plasticity in HaCaT II-4 cells cultured on a three-dimensional collagen matrix. A and B, immunocytochemical detection of EMT-associated processes in TGF-β1 and/or EGF stimulated HaCaT II-4 cells ...

Because the phenotypic plasticity characteristic of EMT may promote tumor metastasis (5), it was important to evaluate the invasive capacities of TGF-β1 + EGF-treated HaCaT II-4 cells. TGF-β1 + EGF enhanced cell invasion and migration into a collagen matrix, as assessed in both OptiCell and Transwell three-dimensional systems (Fig. 2A and B), and was effectively attenuated by the broad-spectrum MMP inhibitor GM6001 (Fig. 2C). Confocal imagery of cells seeded onto thin collagen gels (Supplementary Fig. S2) also supported the observation that TGF-β1 + EGF promoted HaCaT II-4 collagen gel invasion. Enhanced collagen type 1 degradation following TGF-β1 + EGF treatment was confirmed by using collagen matrices prepared from a quenched FITC-labeled collagen type 1 substrate that fluoresces on cleavage (Fig. 2D). Together, these data indicate that TGF-β1 and EGF play integral roles in modulating HaCaT II-4-based collagenase activity, effectively supporting collagen gel invasion.

Figure 2
TGF-β1 + EGF stimulation enhances MMP-dependent collagen gel remodeling and invasion by HaCaT II-4 cells. A, OptiCell tissue culture chambers were used to visualize collagen gel invasion 6 d post-stimulation with TGF-β1 and/or EGF. The ...

TGF-β1 + EGF treatment enhances collagen degradation via a plasmin/MMP dependent mechanism

Physiologic control of pericellular proteolysis occurs primarily through the regulation of plasminogen activation at the cell surface, which, in turn, contributes to downstream extracellular MMP activity (Fig. 3A). To explore the mechanisms associated with plasmin-based proteolysis in a cutaneous model, exogenous plasminogen was added to HaCaT II-4 cultures stimulated with TGF-β1 and/or EGF, as HaCaT II-4 cells secrete only low levels of plasminogen (Fig. 3A). Dissolution of the supporting collagen matrix accompanied TGF-β1 + EGF + plasminogen treatment (Fig. 3B), a process significantly reduced by the plasmin inhibitors aprotinin and α2-antiplasmin, but not with the cysteine protease inhibitor E-64 (Fig. 3C and D), which affects cellular cathepsins. Inhibition of MMP activity with GM6001 also blocked plasmin-initiated collagen degradation, confirming a role for MMPs in the remodeling process (Figs. 2D and 3C and D). Control experiments revealed no evidence of adverse effects arising from treatment with these inhibitors (data not shown). Cells that were detached as a result of collagen gel dissolution, in fact, reattached to tissue culture wells within 24 h (Supplementary Fig. S3).

Figure 3
Plasmin inhibition blocks MMP-dependent collagen gel dissolution. A, Western blot of endogenous plasminogen from HaCaT II-4 and HepG2 cells cultured on collagen (serum-free) overnight and then stimulated ± TGF-β1 + EGF for 8 h. Two samples ...

Plasmin-dependent collagen degradation was quantified through the release of digested FITC-labeled collagen type 1 (Fig. 3D). Stimulation with either TGF-β1 or EGF independently significantly increased collagen type 1 proteolysis within 7.5 h of plasminogen addition. This by itself, however, was insufficient to trigger a distinguishable loss in the fibrillar network even at later time points (Fig. 3B). Stimulation with the combination of TGF-β1 + EGF clearly evoked a more substantial proteolytic response (Fig. 3D) that resulted in dissolution of the polymerized gel within 20 h (Fig. 3B), suggesting that the collagenolytic activity promoted by combining these two growth factors in the presence of plasminogen surpassed any threshold limitations.

TGF-β1 + EGF-stimulated collagen gel dissolution occurs via a plasmin/MMP-10/MMP-1–dependent axis

Because plasmin-dependent collagen degradation has been linked to MMP-13 up-regulation in mouse keratinocytes (26), it was necessary to assess whether MMP-13 might be involved in TGF-β1 + EGF-initiated collagenolysis. Similar to what has been reported previously for other HaCaT variants (13, 27), TGF-β1 stimulation significantly increased MMP-13 levels in HaCaT II-4 cells (Fig. 4A). Only a modest elevation in MMP-13 was evident, however, following coincubation with TGF-β1 + EGF (Fig. 4A). MMP-10, in contrast, was substantially increased under these conditions (Fig. 4A and B).

Figure 4
TGF-β1 + EGF stimulation increases MMP-10 levels in HaCaT II-4 cells cultured on collagen. A, evaluation of downstream MMP targets by protein microarray analysis of conditioned medium from HaCaT II-4 cells cultured on collagen gels and stimulated ...

ProMMP-10 is a plasmin substrate (21), and whereas active MMP-10 does not cleave collagen type 1 directly, it does activate the collagenase MMP-1 (20). Subsequent to TGF-β1 + EGF stimulation, MMP-10 activation was evident by 4 h post-plasminogen addition and complete by 24 h (Fig. 5A, top), whereas the kinetics of MMP-1 activation closely followed the conversion of MMP-10 to a catalytic form (Fig. 5A, bottom). To confirm the role of MMP-10 in collagen matrix degradation, plasminogen was added to TGF-β1 + EGF-stimulated HaCaT II-4 cultures in the presence of increasing concentrations of neutralizing antibodies to either MMP-1 or MMP-10. MMP-1 inhibition prevented plasminogen-dependent collagen dissolution (Fig. 5B, top). Importantly, neutralization of MMP-10 activity also blocked plasminogen-initiated collagen degradation (Fig. 5B, bottom), supporting a plasmin/MMP-10/MMP-1–dependent axis in matrix remodeling. A notable decrease in the level of active MMP-1 was also consistently evident following MMP-10 neutralization (Fig. 5C). Despite residual levels of active, likely plasmin-generated MMP-1, this activity by itself was insufficient, however, to trigger gel dissolution on MMP-10 inhibition (Fig. 5B).

Figure 5
MMP-10 and MMP-1 direct plasmin-dependent collagen gel dissolution. A, Western blot analysis to detect levels of active MMP-10 or MMP-1 (bottom band of each doublet) in conditioned medium from HaCaT II-4 cells cultured on collagen ± TGF-β1 ...

PAI-1 functions as an upstream regulator of a MMP-10–initiated collagenolytic phenotype

Similar to MMP-10, PAI-1 expression in HaCaT II-4 cells is increased in response to TGF-β1 stimulation (12), whereas the combination of TGF-β1 + EGF synergistically enhanced PAI-1 protein levels (Fig. 6A; Supplementary Fig. S4A). Despite the inability of EGF alone to increase PAI-1 levels in this system, enhanced PAI-1 synthesis resulting from TGF-β1 + EGF stimulation, as well as from TGF-β1 alone, was attenuated by inhibition of EGF receptor signaling with AG1478 (Fig. 6A). Similar results were evident in human dermal fibroblasts (Supplementary Fig. S4B) and kidney epithelial cells (22), emphasizing the generality of EGF receptor involvement in TGF-β1–dependent PAI-1 production.

Figure 6
PAI-1 regulates MMP-10/MMP-1–dependent collagen gel dissolution by HaCaT II-4 cells. A, cell-based ELISA for PAI-1 in TGF-β1 and/or EGF-stimulated HaCaT II-4 cells cultured on collagen-coated tissue culture plastic (50 μg/mL). ...

PAI-1, through its inhibition of urokinase-type plasminogen activator, is critical for regulating the generation of pericellular plasmin. It was important, therefore, to assess the effect of PAI-1 on collagen gel dissolution. Blocking urokinase-type plasminogen activator activity with the inhibitor amiloride, or by adding a stable recombinant form of PAI-1 protein (N150H, K154T, Q319L, and M345I; ref. 28), completely inhibited collagen gel dissolution (Fig. 6B). Addition of PAI-1 protein also effectively blocked conversion of MMP-10 and MMP-1 to their active forms (Fig. 6C). In contrast, neutralization of endogenous PAI-1 with function-blocking antibodies accelerated both collagenolysis (Fig. 6B) and activation of MMP-10 and MMP-1 (Fig. 6C). Collectively, these results indicate that, in the physiologically relevant setting of a complex three-dimensional collagen environment, PAI-1 regulates MMP-10–initiated collagenolytic activity (Fig. 6D). A key factor in this model is the ability of active MMP-10 to superactivate MMP-1, creating a plasmin/MMP-10/MMP-1 proteolytic axis that enhances collagen type 1 degradation and facilitates collagen gel invasion.


MMPs are integral components of a complex stromal remodeling program designed to modulate matrix integrity, release bioactive fragments, growth factors, and cytokines from matrix constituents, and enhance cell motility (29). Amplified MMP expression appears linked to increased tumor aggressiveness, metastasis, and poor patient survival (30). Not surprisingly, recently, studies also implicate several MMPs, including MMP-3, -7, -9, and -28, in directly triggering EMT-related processes (30). The combination of TGF-β and EGF, which effectively promotes EMT, also up-regulates certain MMPs synergistically including MMP-1, -3, -9, -10, and -14 (17, 18), posing some interesting questions regarding potential mechanisms that support amplification of EMT-associated events.

An acute collagenolytic phenotype linked to plasmin-dependent activation of stromelysin-2 (MMP-10) emerged in response to costimulation of HaCaT II-4 keratinocytes with TGF-β1 and EGF and was coincident with collagen gel invasion. MMP-10, which is generally limited to epithelial cells (15, 19), has broad substrate specificity, targeting proMMP-1, -7, -8, -9, and -13 as well as collagen types III, IV, and V, gelatin elastin, fibronectin, proteoglycans, and laminin (20, 21). Rigorous control over MMP-10 levels and activation are likely critical, therefore, for normal cutaneous homeostasis. MMP-10, in fact, is not evident in intact skin but is expressed during cutaneous injury repair where it localizes to migrating keratinocytes at the wound edge, suggesting that its presence may facilitate invasive behavior (31).

Similar to what has been observed in other systems (15, 18, 19, 27), MMP-10 and MMP-1 were up-regulated in HaCaT II-4 cells seeded onto collagen gels and stimulated with TGF-β1 and/or EGF. Both proenzymes are plasmin substrates; however, following MMP-10 inhibition, the residual level of active, likely plasmin-generated MMP-1 (Fig. 5C) was insufficient by itself to trigger collagen gel dissolution (Fig. 5B). These data likely reflect the established ability of MMP-10 to “superactivate” MMP-1 and enhance its collagenase activity 7- to 10-fold over that observed with plasmin alone (19). Given these parameters, neutralization of MMP-10 activity would have quenched this hypercollagenase activity and, as observed, impeded collagen degradation. Similar results regarding MMP-10–dependent superactivation of MMP-1, -8, and -13 in an arthritis-based model have been reported but not linked to plasminogen activation (32). Consequently, this article is the first to show a plasmin/MMP-10/MMP-1–dependent collagen remodeling axis and establish its relevance in a keratinocyte-based three-dimensional model.

Clearly, MMP-10 activity can have significant stromal consequences, particularly in a cutaneous environment, irrespective of its level of up-regulation. Amplified MMP-10 expression does, however, accompany the progression of several epithelial cancers, including squamous cell carcinomas of the head and neck and esophagus (33, 34). Increased MMP-10 expression also occurs in colorectal carcinoma (35), breast cancer (36), prostate cancer (37), and lymphoma (38). It localizes to cells at the invasive front of renal cell carcinomas and signals a lower survival rate compared with patients with MMP-10–negative tumors (39). Similarly, tumor MMP-10 levels predict poor survival in non-small cell lung cancer (40).

Like MMPs, lysosomal proteinase cathepsins are also associated with tumor cell invasion, particularly the cysteine proteinases cathepsins L and B, which degrade collagen type 1 and activate MMP-1, respectively (41). Inhibition of cysteine cathepsins had no effect, however, on collagen gel dissolution in the HaCaT II-4 model, whereas serine proteinase blockade effectively attenuated collagen degradation. These data emphasize a critical role for active plasmin, and not cathepsins, in the initiation of collagen degradation by TGF-β1 + EGF-stimulated cells and are consistent with observations regarding the ability of TGF-β to down-regulate cathepsins (42).

Previously, intermediates other than MMP-10, including MMP-13, have been associated with linking keratinocyte-based collagen type 1 dissolution and plasminogen activation (26, 43). Studies suggest, however, that contrary to observations in primary human keratinocytes, MMP-13 expression is, in fact, correlated with transformation of human keratinocytes and enhanced in these cells, including HaCaT derivatives, following stimulation with TGF-β1 (13, 27). Our data in a three-dimensional system also indicate that TGF-β1 stimulation alone increases the level of MMP-13; however, the combination of TGF-β1 + EGF did not produce this effect. Costimulated cells instead exhibited a robust induction of MMP-10. This disparity may be due, in part, to differences among the ras-HaCaT variants in MMP expression programs (44), TGF-β/EGF-related receptor cross-talk promoting EGF-dependent down-regulation of TGF-β1–enhanced MMP-13 levels (18, 45), or culture in two-dimensional versus a more complex three-dimensional stromal-equivalent system. The potential for EGF to counteract a TGF-β1–dependent increase in MMP-13 production reinforces the complexity of this process and presents some intriguing questions for future investigation regarding the regulation of these matrix-modifying enzymes in cancer progression.

Consistent with recent observations regarding receptor crosstalk and synergy (17, 18), PAI-1 levels increased synergistically following TGF-β1 + EGF treatment of HaCaT II-4 cells on a collagen substrate (Fig. 6A). During tumor progression, synergistic amplification of PAI-1 would, in effect, inhibit a disproportionate level of stromal degradation and, in doing so, facilitate cell migration preserving stromal architecture as well as by interacting with low-density lipoprotein-related receptor (46). Up-regulation of MMP-10 and PAI-1 in conjunction with a TGF-β1 + EGF-stimulated EMT, therefore, promotes an environment conducive to the acqui sition of an invasive phenotype. Indeed, like MMP-10, PAI expression is up-regulated in various cancers where its presence is associated with poor patient outcome (47, 48). The incidence of stromal PAI-1 is an important factor in determining progression, reflecting its capacity to stabilize the microenvironment and promote tumor vascularization (49, 50).

Temporal and spatial balance between extracellular components that reorganize tissue architecture is a significant aspect tumor progression. This study has identified an important proteolytic axis for regulating collagen type 1 degradation in a three-dimensional environment. The data provided are consistent with a model (Fig. 6D) in which transformed keratinocytes, in response to TGF-β1 + EGF in the microenvironment, up-regulate proMMP-10, which is converted to its active form in the presence of plasmin and can subsequently superactivate the catalytic activity of MMP-1. This process results in a plasmin/MMP-10/MMP-1–dependent proteolytic axis that effectively enhances collagen type 1 degradation and facilitates collagen gel invasion. PAI-1 plays a crucial role in this paradigm through its ability to counter excessive collagen degradation and maintain stromal integrity for cell migration. Identification of critical components involved in managing the rate and level of collagen type 1 degradation may have far reaching implications for therapeutic targeting of cutaneous pathologies.

Supplementary Material






Grant support: NIH grant GM57242 (P.J. Higgins) and NIH training grant T32-HL07194 (C.E. Wilkins-Port).


Note: Supplementary data for this article are available at Cancer Research Online (

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.


1. Thiery JP. Epithelial-mesenchymal transitions in development and pathologies. Curr Opin Cell Biol. 2003;15:740–6. [PubMed]
2. Boyer B, Vallés AM, Edme N. Induction and regulation of epithelial-mesenchymal transitions. Biochem Pharmacol. 2000;60:1091–9. [PubMed]
3. Ackland ML, Newgreen DF, Fridman M, et al. Epidermal growth factor-induced epithelio-mesenchymal transition in human breast carcinoma cells. Lab Invest. 2003;83:435–48. [PubMed]
4. Zavadil J, Bitzer M, Liang D, et al. Genetic programs of epithelial cell plasticity directed by transforming growth factor-β Proc Natl Acad Sci U S A. 2001;98:6686–91. [PubMed]
5. Thiery JP. Epithelial-mesenchymal transitions in tumour progression. Nat Rev Cancer. 2002;2:442–54. [PubMed]
6. Cui W, Fowlis DJ, Bryson S, et al. TGFβ1 inhibits the formation of benign skin tumors, but enhances progression to invasive spindle carcinomas in transgenic mice. Cell. 1996;86:531–42. [PubMed]
7. Rho O, Beltran LM, Gimenez-Conti IB, DiGiovanni J. Altered expression of the epidermal growth factor receptor and transforming growth factor-α during multistage skin carcinogenesis in SENCAR mice. Mol Carcinog. 1994;11:19–28. [PubMed]
8. Derynck R, Akhurst RJ, Balmain A. TGF-β signaling in tumor suppression and cancer progression. Nat Genet. 2001;29:117–29. [PubMed]
9. Han G, Lu SL, Li AG, et al. Distinct mechanisms of TGF-β1-mediated epithelial-to-mesenchymal transition and metastasis during skin carcinogenesis. J Clin Invest. 2005;115:1714–23. [PMC free article] [PubMed]
10. Roberts AB, Heine UI, Flanders KC, Sporn MB. Transforming growth factor-β. Major role in regulation of extracellular matrix. Ann NY Acad Sci. 1990;580:225–32. [PubMed]
11. Vollberg TM, Sr, George MD, Jetten AM. Induction of extracellular matrix gene expression in normal human keratinocytes by transforming growth factor β is altered by cellular differentiation. Exp Cell Res. 1991;193:93–100. [PubMed]
12. Akiyoshi S, Ishii M, Nemoto N, Kawabata M, Aburatani H, Miyazono K. Targets of transcriptional regulation by transforming growth factor-β: expression profile analysis using oligonucleotide arrays. Jpn J Cancer Res. 2001;92:257–68. [PubMed]
13. Johansson N, Westermarck J, Leppä S, et al. Collagenase 3 (matrix metalloproteinase 13) gene expression by HaCaT keratinocytes is enhanced by tumor necrosis factor a and transforming growth factor β1. Cell Growth Differ. 1997;8:243–50. [PubMed]
14. Kim HS, Shang T, Chen Z, Pflugfelder SC, Li DQ. TGF-β1 stimulates production of gelatinase (MMP-9), collagenases (MMP-1, -13) and stromelysins (MMP-3, -10, -11) by human corneal epithelial cells. Exp Eye Res. 2004;79:263–74. [PubMed]
15. Madlener M, Mauch C, Conca W, Brauchle M, Parks WC, Werner S. Regulation of the expression of stromelysin-2 by growth factors in keratinocytes: implications for normal and impaired wound healing. Biochem J. 1996;320:659–64. [PubMed]
16. Sudbeck BD, Baumann P, Ryan GJ, et al. Selective loss of PMA-stimulated expression of matrix metalloproteinase 1 in HaCaT keratinocytes is correlated with the inability to induce mitogen-activated protein family kinases. Biochem J. 1999;339:167–75. [PubMed]
17. Tian YC, Chen YC, Chang CT, et al. Epidermal growth factor and transforming growth factor-β1 enhance HK-2 cell migration through a synergistic increase of matrix metalloproteinase and sustained activation of ERK signaling pathway. Exp Cell Res. 2007;313:2367–77. [PubMed]
18. Uttamsingh S, Bao X, Nguyen KT, et al. Synergistic effect between EGF and TGF-β1 in inducing oncogenic properties of intestinal epithelial cells. Oncogene. 2007;27:2626–34. [PubMed]
19. Windsor LJ, Grenett H, Birkedal-Hansen B, Bodden MK, Engler JA, Birkedal-Hansen H. Cell type-specific regulation of SL-1 and SL-2 genes. Induction of the SL-2 gene but not the SL-1 gene by human keratinocytes in response to cytokines and phorbolesters. J Biol Chem. 1993;268:17341–7. [PubMed]
20. Chakraborti S, Mandal M, Das S, Mandal A, Chakraborti T. Regulation of matrix metalloproteinases. An overview. Mol Cell Biochem. 2003;253:269–85. [PubMed]
21. Lijnen HR. Matrix metalloproteinases and cellular fibrinolytic activity. Biochemistry (Moscow) 2002;67:92–8. [PubMed]
22. Kutz SM, Higgins CE, Samarakoon R, et al. TGF-β1-induced PAI-1 expression is E box/USF-dependent and requires EGFR signaling. Exp Cell Res. 2006;312:1093–105. [PubMed]
23. Fusenig NE, Boukamp P. Multiple stages and genetic alterations in immortalization, malignant transformation, and tumor progression of human skin keratinocytes. Mol Carcinog. 1998;23:144–58. [PubMed]
24. Boukamp P. UV-induced skin cancer: similarities-variations. JDDG J German Soc Dermatol. 2005;3:493–503. [PubMed]
25. Davies M, Robinson M, Smith E, Huntley S, Prime S, Paterson I. Induction of an epithelial to mesenchymal transition in human immortal and malignant keratinocytes by TGF-β1 involves MAPK, Smad and AP-1 signalling pathways. J Cell Biochem. 2005;95:918–31. [PubMed]
26. Netzel-Arnett S, Mitola DJ, Yamada SS, et al. Collagen dissolution by keratinocytes requires cell surface plasminogen activation and matrix metalloproteinase activity. J Biol Chem. 2002;277:45154–61. [PubMed]
27. Johansson N, Ala-aho R, Uitto VJ, et al. Expression of collagenase-3 (MMP-13) and collagenase-1 (MMP-1) by transformed keratinocytes is dependent on the activity of p38 mitogen-activated protein kinase. J Cell Sci. 2000;113:227–35. [PubMed]
28. Berkenpas MB, Lawrence DA, Ginsburg D. Molecular evolution of plasminogen activator inhibitor-1 functional stability. EMBO J. 1995;14:2969–77. [PubMed]
29. Mott JD, Werb Z. Regulation of matrix biology by matrix metalloproteinases. Curr Opin Cell Biol. 2004;16:558–64. [PMC free article] [PubMed]
30. Orlichenko LS, Radisky DC. Matrix metalloproteinases stimulate epithelial-mesenchymal transition during tumor development. Clin Exp Metastasis. 2008;25:593–600. [PubMed]
31. Krampert M, Bloch W, Sasaki T, et al. Activities of the matrix metalloproteinase stromelysin-2 (MMP-10) in matrix degradation and keratinocyte organization in wounded skin. Mol Biol Cell. 2004;15:5242–54. [PMC free article] [PubMed]
32. Barksby HE, Milner JM, Patterson AM, et al. Matrix metalloproteinase 10 promotion of collagenolysis via procollagenase activation: implications for cartilage degradation in arthritis. Arthritis Rheum. 2006;54:3244–53. [PubMed]
33. O-charoenrat P, Rhys-Evans PH, Eccles SA. Expression of matrix metalloproteinases and their inhibitors correlates with invasion and metastasis in squamous cell carcinoma of the head and neck. Arch Otolaryngol Head Neck Surg. 2001;127:813–20. [PubMed]
34. Mathew R, Khanna R, Kumar R, Mathur M, Shukla NK, Ralhan R. Stromelysin-2 overexpression in human esophageal squamous cell carcinoma: potential clinical implications. Cancer Detect Prev. 2002;26:222–8. [PubMed]
35. Asano T, Tada M, Cheng S, et al. Prognostic values of matrix metalloproteinase family expression in human colorectal carcinoma. J Surg Res. 2008;146:32–42. [PubMed]
36. McGowan PM, Duffy MJ. Matrix metalloproteinase expression and outcome in patients with breast cancer: analysis of a published database. Ann Oncol. 2008;19:1566–72. [PubMed]
37. Riddick ACP, Shukla CJ, Pennington CJ, et al. Identification of degradome components associated with prostate cancer progression by expression analysis of human prostatic tissues. Br J Cancer. 2005;92:2171–80. [PMC free article] [PubMed]
38. Van Themsche C, Alain T, Kossakowska AE, Urbanski S, Potworowski E, Pierre Y. Stromelysin-2 (matrix metalloproteinase 10) is inducible in lymphoma cells and accelerates the growth of lymphoid tumors in vivo. J Immunol. 2004;173:3605–11. [PubMed]
39. Miyata Y, Iwata T, Maruta S, et al. Expression of matrix metalloproteinase-10 in renal cell carcinoma and its prognostic role. Eur Urol. 2007;52:791–7. [PubMed]
40. Frederick LA, Matthews JA, Jamieson L, et al. Matrix metalloproteinase-10 is a critical effector of protein kinase Cτ-Par6α-mediated lung cancer. Oncogene. 2008;27:4841–53. [PMC free article] [PubMed]
41. Song F, Wisithphrom K, Zhou J, Windsor LJ. Matrix metalloproteinase dependent and independent collagen degradation. Front Biosci. 2006;11:3100–20. [PubMed]
42. Gerber A, Wille A, Welte T, Ansorge S, Bühling F. Interleukin-6 and transforming growth factor-β1 control expression of cathepsins B and L in human lung epithelial cells. J Interferon Cytokine Res. 2001;21:11–9. [PubMed]
43. Birkedal-Hansen H, Lin HY, Birkedal-Hansen B, Windsor LJ, Pierson MC. Degradation of collagen fibrils by live cells: role of expression and activation of procollagenase. Matrix Suppl. 1992;1:368–74. [PubMed]
44. Meade-Tollin LC, Boukamp P, Fusenig NE, Bowen CPR, Tsang TC, Bowden GT. Differential expression of matrix metalloproteinases in activated c-ras(Ha)-transfected immortalized human keratinocytes. Br J Cancer. 1998;77:724–30. [PMC free article] [PubMed]
45. Joo CK, Kim HS, Park JY, Seomun Y, Son MJ, Kim JT. Ligand release-independent transactivation of epidermal growth factor receptor by transforming growth factor-β involves multiple signaling pathways. Oncogene. 2008;27:614–28. [PubMed]
46. Degryse B, Neels JG, Czekay RP, Aertgeerts K, Kamikubo Yi, Loskutoff DJ. The low density lipoprotein receptor-related protein is a motogenic receptor for plasminogen activator inhibitor-1. J Biol Chem. 2004;279:22595–604. [PubMed]
47. Pedersen H, Brunner N, Francis D, et al. Prognostic impact of urokinase, urokinase receptor, and type 1 plasminogen activator inhibitor in squamous and large cell lung cancer tissue. Cancer Res. 1994;54:4671–5. [PubMed]
48. Sternlicht MD, Dunning AM, Moore DH, et al. Prognostic value of PAI1 in invasive breast cancer: evidence that tumor-specific factors are more important than genetic variation in regulating PAI1 expression. Cancer Epidemiol Biomarkers Prev. 2006;15:2107–14. [PMC free article] [PubMed]
49. Bajou K, Noël A, Gerard RD, et al. Absence of host plasminogen activator inhibitor 1 prevents cancer invasion and vascularization. Nat Med. 1998;4:923–8. [PubMed]
50. Maillard C, Jost M, Romer MU, et al. Host plasminogen activator inhibitor-1 promotes human skin carcinoma progression in a stage-dependent manner. Neoplasia. 2005;7:57–66. [PMC free article] [PubMed]