The type-1 inhibitor of plasminogen activator (PAI-1) is an important physiological regulator of extracellular matrix (ECM) homeostasis and cell motility. Various growth factors mediate temporal changes in the expression and/or focalization of PAI-1 and its protease target PAs, thereby influencing cell migration by barrier proteolysis and/or ECM adhesion modulation. TGF-β1, in particular, is an effective inducer of matrix deposition/turnover, cell locomotion and PAI-1 expression. Therefore, the relationship between motility and PAI-1 induction was assessed in TGF-β1-sensitive T2 renal epithelial cells. PAI-1 synthesis and its matrix deposition in response to TGF-β1 correlated with a significant increase in cell motility. PAI-1 expression was an important aspect in cellular movement as PAI-1-deficient cells had significantly impaired basal locomotion and were unresponsive to TGF-β1. However, the induced migratory response to this growth factor was complex. TGF-β1 concentrations of 1-2 ng/ml were significantly promigratory, whereas lower levels (0.2-0.6 ng/ml) were ineffective and final concentrations ≥5 ng/ml inhibited T2 cell motility. This same growth factor range progressively increased PAI-1 transcript levels in T2 cells consistent with a bifunctional role for PAI-1 in cell migration. TGF-β1 induced PAI-1 mRNA transcripts in quiescent T2 cells via an immediate-early response mechanism. Full TGF-β1-stimulated expression required tyrosine kinase activity and involved MAPK/ERK kinase (MEK). MEK appeared to be a major mediator of TGF-β1-dependent PAI-1 expression and T2 cell motility since PD98059 effectively attenuated both TGF-β1-induced ERK1/2 activation and PAI-1 transcription as well as basal and growth factor-stimulated planar migration. Since MEK activation in response to growth factors is adhesion-dependent, it was important to determine whether cellular adhesive state influenced TGF-β1-mediated PAI-1 expression in the T2 cell system. Cells maintained in suspension culture (i.e., over agarose underlays) in growth factor-free medium or treated with TGF-β1 in suspension expressed relatively low levels of PAI-1 transcripts compared with the significant induction of PAI-1 mRNA evident in T2 cells upon stimulation with TGF-β1 during adhesion to a fibronectin-coated substrate. Attachment to fibronectin alone (i.e., in the absence of added growth factor) was sufficient to initiate PAI-1 transcription, albeit at levels considerably lower than that induced by the combination of cell adhesion in the presence of TGF-β1. T2 cells allowed to attach to vitronectin-coated surfaces also expressed PAI-1 transcripts but to a significantly reduced extent relative to cells adherent to fibronectin. Moreover, newly vitronectin-attached cells did not exhibit a PAI-1 inductive response to TGF-β1, at least during the short 2 hour period of combined treatment. PAI-1 mRNA synthesis in response to substrate attachment, like TGF-β1-mediated induction in adherent cultures, also required MEK activity as fibronectin-stimulated PAI-1 expression was effectively attenuated by the MEK inhibitor PD98059. These data indicate that cellular adhesive state modulates TGF-β1 signaling to particular target genes (i.e., PAI-1) and that MEK is a critical mediator of the PAI-1+/promigratory phenotype switch induced by TGF-β1 in T2 cells.
INTRODUCTION
Genetic analysis and adaptation of physiologically-relevant in vitro models of wound repair have clarified basic mechanisms involved in the tissue response to injury (Garlick and Taichman, 1994; Romer et al., 1996; Creemers et al., 2000; Providence et al., 2000). Fundamental to this process is the conversion of normally sedentary cells to an actively migrating, invasive phenotype (Martin, 1997). However, stimulated cell movement and locomotion through the provisional extracellular matrix (ECM) requires cycles of adhesion-deadhesion and precise control of the pericellular proteolytic environment (Yamada and Clark, 1996; Greenwood and Murphy-Ullrich, 1998; Xie et al., 1998; Pilcher et al., 1999). Efficient wound re-epithelialization involves several protease systems with repair outcome highly dependent on the generation of plasmin by urokinase plasminogen activator (uPA). uPA activity is regulated, in turn, by its fast-acting type-1 inhibitor (PAI-1) (Andreasen et al., 1997; Lund et al., 1999; Zhou et al., 2000; Legrand et al., 2001). This cascade directly influences the overall injury site proteolytic balance and is a critical determinant in wound resolution (Mazzieri et al., 1997; Wysocki et al., 1999) as well as directed cell movement (Pepper et al., 1987; Pepper et al., 1992; Okedon et al., 1992; Providence et al., 2000).
Injury-induced cell motility is orchestrated by various autocrine/paracrine-acting growth factors (Martin, 1997). Most prominent are members of the transforming growth factor-β (TGF-β), fibroblast growth factor (FGF), and epidermal growth factor (EGF) families (Boland et al., 1996; Sato and Rifkin, 1988; Song et al., 2000; Ellis et al., 2001; Goke et al., 2001). TGF-β1 and activin A, in particular, integrate the complex processes of tissue repair and cell migration (Zambruno et al., 1995; Munz et al., 1999) largely through control of genes that encode matrix components (fibronectin, type I collagen), regulators of ECM homeostasis (e.g., uPA, PAI-1) and the cellular adhesive apparatus (e.g., PAI-1, integrin subunits) (Cajot et al., 1989; Cajot et al., 1990; Wrana et al., 1991; Munz et al., 1999; Lai et al., 2000; Providence et al., 2000). Therefore, growth factor-initiated changes in the expression, focalization and/or relative activity of uPA/PAI-1 may stimulate or inhibit cell migration via ECM barrier proteolysis or by altering cellular adhesive interactions with the ECM (Stefansson and Lawrence, 1996; Mignatti and Rifkin, 2000). Variances in PAI-1 synthesis (Providence et al., 2000) and/or site-localization (Kutz et al., 1997) would specifically impact on cellular migration by affecting uPA activity as well as uPAR/vitronectin- or integrin/vitronectin-dependent cell attachment (Ciambrone and McKeown-Longo, 1990; Deng et al., 1996; Chapman, 1997; Stefansson and Lawrence, 1996; Loskutoff et al., 1999).
Since TGF-β1 stimulates cell motility (Kutz et al., 2001), PAI-1 induction (Boehm et al., 1999) in response to TGF-β1 is probably critical to the motile process and the acquisition of epithelial cell ‘plasticity’ (Akiyoshi et al., 2001; Zavadil et al., 2001). This was confirmed in the present study using the PAI-1-deficient 4HH cell line (Providence et al., 2000) in a quantitative model of induced cell locomotion. Therefore, it was important to define mechanisms involved in TGF-β1-dependent PAI-1 gene expression. PAI-1 transcription in TGF-β1-responsive T2 epithelial cells used an immediate-early, tyrosine kinase-mediated, signaling pathway. Moreover, PAI-1 induction and basal as well as TGF-β1-stimulated T2 cell locomotion was MEK-dependent. The involvement of MEK in TGF-β1-initiated PAI-1 expression and the adhesion-dependency of MEK activation (Renshaw et al., 1997) suggested that substrate attachment may influence TGF-β1-induced PAI-1 gene regulation. TGF-β1, in fact, poorly induced PAI-1 transcription in cells maintained in suspension culture but significantly increased PAI-1 expression during attachment to fibronectin-coated surfaces. Cellular adherence to fibronectin alone (i.e., in the absence of TGF-β1), and to a lesser extent vitronectin, also stimulated PAI-1 mRNA synthesis indicating that adhesive state can modulate TGF-β1 signaling to particular target genes (i.e., PAI-1).
MATERIALS AND METHODS
Cell culture
T2 and EC-1 renal epithelial cells and the PAI-1-deficient 4HH line (Providence et al., 2000) were cultured in DMEM containing 10% fetal bovine serum (FBS) and antibiotics. Cells were washed twice with Hanks’ balanced salt solution (HBSS; 1.3 mM CaCl2, 5 mM KC1, 0.3 mM KH2PO4, 0.5 mM MgCl2-6H20, 0.4 mM MgSO4-7H20, 0.14 m NaCl, 4 mM NaHCO3, 0.3 mM Na2HPO4, 5.6 mM glucose) then incubated in serum-free DMEM for 3 days to initiate a quiescent state (Kutz et al., 1997). Cells were either maintained in quiescence medium or stimulated by the direct addition of FBS or TGF-β1 (to final concentrations of 20% and 0.2-10.0 ng/ml, respectively). Puromycin (100 μg/ml), actinomycin D (5 μg/ml), genistein (100 μM), PD98059 (5-50 nM) or wortmanin (50 nM) were added 30 minutes prior to stimulation with FBS or TGF-β1. For culture under non-adherent conditions, quiescent T2 cells were harvested by trypsinization, washed with soybean trypsin inhibitor, and plated on 1% agarose underlays in serum-free DMEM for 4 hours (Ryan et al., 1996). Agarose-suspended cells were maintained (for an additional 2 hours) under non-adherent conditions (in the presence or absence of TGF-β1) or seeded onto fibronectin-, vitronectin- or bovine serum albumin (BSA)-coated 100 mm plastic dishes (coating concentration 10 μg protein/ml) and allowed to re-attach for 2-6 hours (in the presence or absence of TGF-β1) prior to RNA isolation. Directional migration assays used methods developed previously (Providence et al., 2000). Three-day post-confluent T2 cell cultures were washed twice with HBSS and incubated for 3 days in serum-free DMEM. Monolayers were maintained under serum-free or TGF-β1-/serum-supplemented conditions prior to scrape-wounding using the small end of a 1000 μl pipette tip. Initial wound size was determined for each culture dish and extent of injury closure assessed 24 hours later with an inverted microscope fitted with a calibrated ocular grid.
Northern blot analysis
Total cellular RNA was isolated and denatured at 55°C for 15 minutes in 1× MOPS, 6.5% formaldehyde and 50% formamide prior to electrophoresis on agarose/formaldehyde gels (1.2% agarose, 1.1% formaldehyde, 1× MOPS, 50 mM sodium acetate, 1 mM EDTA, pH 8.0). RNA was transferred to Nytran membranes by capillary action in 10× SSC (3M NaCl, 0.3 M Na citrate, pH 7.0), UV crosslinked and incubated for 2 hours at 42°C in 50% formamide, 5× Denhardt’s solution, 1% SDS, 100 μg/ml sheared/heat-denatured salmon sperm DNA (ssDNA) and 5× SSC. RNA blots were hybridized with 32P-labeled cDNA probes for PAI-1 and A-50 (Ryan and Higgins, 1993) for 24 hours at 42°C in 50% formamide, 2.5× Denhardt’s solution, 1% SDS, 100 μg/ml ssDNA, 5× SSC and 10% dextran sulfate. Membranes were washed 3 times in 0.1× SSC/0.1% SDS for 15 minutes each at 42°C followed by 3 washes at 55°C prior to exposure to film.
Microscopy
Cells were washed twice with Ca2+/Mg2+ free phosphate-buffered saline (PBS-CMF; 2.7 mM KC1, 1.2 mM KH2PO4, 0.14 M NaC1, 8.1 mM Na2HPO4-7H2O) and fixed in 10% formalin/PBS-CMF for 10 minutes. Following permeabilization with cold 0.5% Triton X-100/PBS-CMF (for PAI-1 immunolocalization) or 1% NP-40/PBS-CMF (for phalloidin-actin binding) for 10 minutes at 4°C, cells were washed 3 times (5 minutes each) with PBS-CMF then overlayed with rabbit antibodies to PAI-1 (Kutz et al., 1997) in BSA (3 mg/ml)/PBS-CMF. After three PBS-CMF washes, cells were incubated with fluorescein-conjugated goat anti-rabbit IgG (1:20 in BSA/PBS-CMF) for 30 minutes at 37°C, washed, and coverslips mounted with 100 mM n-propylgalate in 50% glycerol/PBS-CMF. Rhodamine-phalloidin was used to visualize actin microfilament structures (Ryan and Higgins, 1993).
MAP kinase assays
Cells were extracted for 30 minutes in cold lysis buffer (0.5% deoxycholate, 50 mM Hepes [pH 7.5], 1% Triton X-100, 1% NP-40, 150 mM NaCl, 50 mM NaF, 1 mM Na-orthovanadate, 0.1% aprotinin, 4 μg/ml pepstatin A, 10 μg/ml leupeptin, 1 mM PMSF) and lysates clarified by centrifugation at 14,000 g for 15 minutes at 4°C. For immunoprecipitation, aliquots containing 500 μg protein were incubated with 2 μg ERK1/2 antibody for 2 hours with gentle rocking. Protein A/G Plus-agarose (30 μl) was added for 2 hours, immune complexes collected by centrifugation, washed twice with lysis buffer and twice with 100 mM NaCl in 50 mM Hepes (pH 8.0) and then incubated at 37°C for 15 minutes in kinase reaction buffer (10 μCi 32P-ATP, 50 μM ATP, 20 mM Hepes (pH 8.0), 10 mM MgCl2, 1 mM DTT, 1 mM benzamidine, 0.3 mg/ml myelin basic protein (MBP)). Electrophoresis buffer (50 mM Tris (pH 6.0), 10% glycerol, 1% SDS, 1% β-mercaptoethanol) was added, samples boiled for 10 minutes and 15 μl aliquots separated on SDS/15% polyacrylamide gels. Proteins were transferred to nitrocellulose in 25 mM Tris, 190 mM glycine, 20% methanol and membranes exposed to film for visualization of phosphorylated MBP. Western blotting for ERK2 and total MBP detection by Ponceau S staining were used to confirm equivalent loading per lane. For detection of phosphorylated ERK1/2, membranes were washed for 10 minutes in 0.05% Triton X-100/PBS-CMF followed by 2 hours in wash solution containing 3% milk. Phospho-ERK monoclonal antibody (1:1000) was added for an overnight incubation in blocking solution at room temperature. Following 3 washes for 20 minutes, horseradish peroxidase (HRP)-labeled anti-mouse secondary antibody (1:3000 in blocking solution) was added and incubated for an additional 1 hour. Membranes were washed 5 times for 10 minutes each in wash solution, incubated with ECL substrate solution (Amersham, Piscataway, NJ) for 2 minutes with gentle rocking and exposed to film. Membranes were stripped for 90 minutes at room temperature using the Western Stripper Kit (Bioworld, Dublin, Ohio), neutralized then incubated in a ERK1/2 primary antibody mixture (each diluted 1:3000 in blocking solution) followed by HRP-anti-rabbit secondary antibody and ECL reagent as described above.
RESULTS
TGF-β1 stimulation of PAI-1 expression and directional motility is MEK-dependent
PAI-1 synthesis and accumulation in the ventral undersurface region in response to TGF-β1 (Fig. 1) correlated with a significant increase in T2 cell motility (relative to the basal rate of movement in the scrape-wound assay) approximating that of serum-stimulated cells (Fig. 2). This rather dramatic effect of TGF-β1 on cell migration and cytoarchitecture is consistent with involvement of TGF-β1 target genes (e.g., PAI-1) in cellular ‘plasticity’ and invasive behavior (Zavadil et al., 2001). Antibodies to PAI-1, in fact, inhibit cell attachment (Higgins et al., 1991), promote substrate detachment (Rheinwald et al., 1987) and attenuate cell activation (Kutz et al., 1997). Furthermore, recent findings have highlighted the functional linkage between induced PAI-1 synthesis and cellular motility (Mignatti and Rifkin, 2000; Providence et al., 2000; Kutz et al., 2001). To directly assess the role of PAI-1 in growth factor-stimulated cell locomotion, TGF-β1-dependent wound closure was compared in EC-1 and 4HH cells. 4HH is a derivative of EC-1 in which PAI-1 synthesis is specifically ablated by stable constitutive expression of PAI-1 antisense transcripts under control of a strong CMV promoter (Higgins et al., 1997). These cells do not produce detectable PAI-1 protein under growth factor-supplemented culture conditions (Providence et al., 2000), thus, providing a tool to assess the relationship between PAI-1 expression and cell motility. The 4HH basal migration rate (i.e., locomotion in serum-/growth factor-free medium) was ≤30% that of parental controls. Moreover, relative to EC-1 cells in which wound closure is significantly enhanced by TGF-β1, 4HH cells were unresponsive and failed to increase substantially their rate of movement in TGF-β1-supplemented medium (Fig. 2).
Since PAI-1 appears to be a critical element in cellular migration, it was important to clarify pathways by which TGF-β1 influenced PAI-1 gene expression and the motile phenotype. TGF-β1-mediated PAI-1 transcription, similar to induction by serum, was an immediate-early response (i.e., resistant to protein synthesis inhibitors) and significantly reduced by prior exposure to genistein (Fig. 3). The MEK-specific compound PD98059 effectively attenuated (at the 5-20 nM range) (Fig. 3A) and completely ablated (at 50 nM) (Fig. 3B) TGF-β1-induced PAI-1 expression. In agreement with previous studies (Hartsough and Mulder, 1995; Yonekura et al., 1999), and the PD98059-sensitivity of PAI-1 induction (Fig. 3), addition of TGF-β1 to quiescent T2 cells stimulated ERK1/2 activity. However, the time course of ERK activation by TGF-β1 (as assessed by the ability of ERK1/2 to phosphorylate the target substrate MBP in a linked immunoprecipitation-in vitro kinase assay), was delayed (by approximately 60 minutes) compared with the rapid induction (within 15 minutes) typical of serum-treated cells (Fig. 4). Similarly, increases in phospho-ERK1/2 levels (sixfold those of unstimulated cells) were not evident until 1 hour after addition of TGF-β1 to quiescent cultures (not shown). Consistent with the metabolic requirements for PAI-1 expression in response to TGF-β1 (Fig. 3), genistein as well as PD98059 effectively blocked TGF-β1-mediated ERK1/2 activation in coupled immunoprecipitation/MBP-phosphorylation assays (Fig. 4).
Although MEK was an important intermediate in TGF-β1-dependent PAI-1 transcription, the role of this signaling pathway in TGF-β1-stimulated T2 cell motility was unclear (Kutz et al., 2001). Since the migration-promoting effects of TGF-β1 are often concentration, as well as context-dependent, it was necessary to define the optimal level of growth factor required to maximally stimulate directional T2 cell locomotion. Dose-assessments indicated that the migratory response to TGF-β1 was complex. Motility rates in cultures supplemented with low (0.2-0.6 ng/ml) as well as high (5 ng/ml) TGF-β1 concentrations were not significantly different from control values compared with the obvious promigratory effect associated with exposure to 1 and 2 ng/ml (Fig. 5A). TGF-β1 levels >5 μg/ml actually inhibited wound-induced T2 cell locomotion (when used alone or in the presence of serum) (not shown). PAI-1 transcripts progressively increased over the same growth factor concentration range (0.5-5.0 ng/ml) (Fig. 6). Collectively, these data are consistent with the suggestion that PAI-1 is a bifunctional regulator of cellular motility with positive effects likely restricted to a relatively narrow expression level ‘window’ (Mignatti and Rifkin, 2000). Using the determined optimal TGF-β1 concentration (1-2 ng/ml), as well as a subeffective 5 ng/ml dose, MEK activity appeared to be important to both TGF-β1-stimulated PAI-1 gene expression and planar motility, as PD98059 effectively attenuated both responses (Fig. 3; Fig. 5B).
Adhesive controls on TGF-β1-induced PAI-1 expression in T2 cells
PD98059 blockade implicated MEK as a critical intermediate in the TGF-β1-initiated pathway of PAI-1 gene expression (Fig. 3). Growth factor- and integrin-activated signaling pathways are interdependent, often converging on the MAPK cascade (Zhu and Assoian, 1995; Lin et al., 1997; Roovers et al., 1999), and MEK activation in response to growth factors requires substrate adhesion (Renshaw et al., 1997). Therefore, experiments were designed to assess whether cellular adhesive state modulated TGF-β1-induced PAI-1 expression. Quiescent T2 cells were trypsinized and plated over agarose underlays in serum-/TGF-β1-free DMEM where they remained in suspension as single cells. After 3 hours, agarose-cultured cells were maintained in non-supplemented medium, stimulated with TGF-β1 (1 ng/ml) in suspension, transferred to fibronectin-coated dishes in serum-/TGF-β1-free medium, or stimulated with TGF-β1 during adhesion to fibronectin (all treatments were for 2 hours). Cells in agarose culture under supplement-free conditions or treated with TGF-β1 in suspension expressed relatively low levels of PAI-1 mRNA compared with the robust expression evident upon stimulation with TGF-β1 during attachment to fibronectin (Fig. 7A). Adhesion to fibronectin matrices alone, in the absence of added TGF-β1, was sufficient to initiate modest PAI-1 transcription (Fig. 7A,B). This adhesion-dependent induction reflected a similarly conservative increase (i.e. threefold) in phospho-ERK1/2 levels (not shown) and, like TGF-β1-mediated expression in normally anchored cells (Fig. 3), also required MEK activity as it was effectively inhibited by PD98059 (Fig. 7B). To further assess if this adhesion-related PAI-1 induction (in either TGF-β1-stimulated or unstimulated cells) was dependent on the nature of the ‘matrix’ encountered, T2 cells were plated onto dishes coated with 10 μg/ml fibronectin, vitronectin or BSA. TGF-β1 did not induce PAI-1 under non-adherent conditions (i.e., culture on BSA-coated surfaces). Attachment to fibronectin for 4 hours (in the absence of TGF-β1) was an effective inducer of PAI-1 transcripts (by three to fivefold) relative to adhesion to vitronectin (Fig. 8). Moreover, preliminary kinetic determinations indicated that PAI-1 mRNA levels increased as a function of time of attachment suggesting that subsequent cell spreading may be a factor in expression control. This was not unique to T2 cells as a similar response was evident in human dermal fibroblasts and microvessel endothelial cells (data not shown). However, no difference in either the attachment and/or spreading of T2 cells plated on fibronectin- compared with vitronectin-coated surfaces was evident within the same time (Fig. 9). More importantly, TGF-β1 significantly enhanced PAI-1 expression only in T2 cells during attachment to fibronectin (Fig. 8). PAI-1 mRNA levels in cells seeded on vitronectin in the presence of TGF-β1 were not different from that expressed during adhesion to vitronectin alone.
DISCUSSION
Several important aspects of in vivo injury repair (i.e. regional uPA/PAI-1 expression, spatial/temporal distinctions between motile and proliferative phenotypes) (Reidy et al., 1995; Romer et al., 1991; Romer et al., 1994) are recapitulated during cell migration into the denuded areas of a scrape-injured monolayer (Pepper et al., 1987; Pepper et al., 1992; Garlick and Taichman, 1994; Zahm et al., 1997; Providence et al., 2000). PAI-1 is rapidly synthesized by cells immediately adjacent to experimentally-created wounds (Pepper et al., 1992; Pawar et al., 1995; Providence et al., 2000). PAI-1 synthesis and deposition into cellular migration tracks are characteristics of a mobile cohort (Seebacher et al., 1992; Pepper et al., 1992) and an essential component of the migratory program (Providence et al., 2000; Kutz et al., 2001). The in situ distribution of this protein is consistent with a function in cell locomotion. De novo synthesized PAI-1 protein accumulates in the cellular undersurface region, probably in a complex with matrix vitronectin (Seiffert et al., 1994; Lawrence et al., 1997) although it appears that PAI-1 may also associate with fibronectin and/or laminin deposits in migration tracks (Seebacher et al., 1992). Therefore, this SERPIN is well-positioned to modulate integrin-ECM or uPA/uPAR-ECM interactions as well as ECM barrier proteolysis. In vitro studies suggest that PAI-1 may dissociate bound vitronectin from the uPAR, detaching cells that use this receptor as a vitronectin anchor (Deng et al., 1996; Kjoller et al., 1997; Loskutoff et al., 1999). Alternatively, PAI-1 may directly inhibit αv integrin-mediated attachment to vitronectin by blocking accessibility to the RGD sequence located proximal to the uPAR binding site (Stefansson and Lawrence, 1996; Loskutoff et al., 1999), although this inhibition is subject to spatial-temporal constraints (Germer et al., 1998). Furthermore, uPAR-associated uPA/PAI-1 complexes are internalized by endocytosis, which promotes uPA receptor recycling (Andreasen et al., 1997) and thereby vitronectin-dependent cell movement. However, transgenic approaches have suggested that PAI-1 promotes vitronectin-independent angiogenesis specifically by inhibition of plasmin proteolysis, thus preserving an appropriate matrix scaffold or providing required neovessel stability (Bajou et al., 2001). Although in vivo compensatory mechanisms may partly explain this discrepancy between animal and culture models, motility controls clearly vary and depend on the level of expression of participating elements, the nature of the provisional ‘matrix’ encountered, the context of the system studied and the growth factor environment.
TGF-β1 exerts concentration-dependent effects on cellular locomotion in 3D culture systems (Gajdusek et al., 1993) as well as in the more spatially restricted planar model of denudation injury (this study) (Gajdusek et al., 1993; Zicha et al., 1999). Wound repair analysis of the PAI-1-deficient 4HH cell line, in which PAI-1 synthesis is specifically ablated by antisense targeting (Higgins et al., 1997; Providence et al., 2000), supports the contention that PAI-1 is an important component in the motile program in this model. Since TGF-β1 stimulates PAI-1 synthesis and PAI-1 impacts directly on cell motility (this study) (Deng et al., 1999; Providence et al., 2000), it was important to assess TGF-β1-dependent controls on PAI-1 expression as well as on cellular migration. TGF-β1 initiates PAI-1 transcription in quiescent T2 cells via an immediate-early response, tyrosine kinase-dependent pathway that involves MEK, an upstream activator of ERK1/2. However, unlike the typical rapid ERK phosphorylation associated with serum-stimulation (i.e., within 15 minutes), TGF-β1-mediated ERK activation (as assessed by phosphorylation of the target substrate MBP) was delayed by 30-60 minutes. The MEK dependency for PAI-1 expression and TGF-β1-stimulated as well as basal migration in T2 cells suggests that these events are related. Although MEK blockade probably interferes with cell movement at several levels (Klemke et al., 1997; Rikitake et al., 2000), genetic targeting approaches confirmed that both basal and TGF-β1-stimulated cell migration (over a 24-36 hour period) requires PAI-1 expression (Providence et al., 2000; Kutz et al., 2001). Clearly, MEK inhibition may not affect basal locomotion in all cell types (Nguyen et al., 1998) but results may depend on the specific system studied. For example, in the monolayer denudation model (unlike random motility assays), ‘basal’ migration is most probably growth factor-mediated. Indeed, monolayer wounding in various cell types, in and of itself, is a sufficient stimulus to initiate autocrine growth factor expression (e.g., TGF-β1, basic FGF, heparin-binding EGF) (Sato and Rifkin, 1988) and activate MAP kinases (Dieckgraefe et al., 1997). Moreover, growth factor synthesis and ERK phosphorylation/nuclear translocation occurs specifically in cells adjacent to the injury site (Dieckgraefe et al., 1997; Song et al., 2000; Ellis et al., 2001) similar to the distribution of locomoting PAI-1-expressing cells (Providence et al., 2000).
Matrix attachment, perhaps as part of the motile response, also stimulates PAI-1 expression. This has particular physiologic relevance. Although the present data suggest that not all matrices have equivalent inductive capability, during the process of wound healing cells ‘switch’ their integrin complement to accommodate the composition of the provisional ECM (Yamada et al., 1996). In certain instances, TGF-β1 directly mediates changes in integrin availability and, therefore, cellular adhesive traits (Collo and Pepper, 1999; Dalton et al., 1999; Lai et al., 2000). Engagement of particular integrins (i.e. αvβ3, α3β1) by immobilized antibodies or ligands has been implicated in PAI-1 gene control (Ghosh et al., 2000; Khatib et al., 2001). The α3β1 ligands laminin-5 and collagen I, when presented immobilized on beads, were also effective inducers of uPA synthesis (Ghosh et al., 2000). Similar to data reported in this study with regard to adhesive controls on PAI-1 expression, β1 integrin aggregation-induced uPA synthesis was also MEK-dependent as PD98059 inhibited ERK activation and uPA expression. Perhaps not coincidentally, both uPA and PAI-1 are induced by pharmacologic disorganization of the actin-based microfilament system as is ERK activation (Higgins et al., 1992; Irigoyen et al., 1997). Since integrin ligation/clustering and cell adhesion result in various levels of cytoskeletal reorganization and recruitment of signaling intermediates (Zhu and Assoian, 1995; Lin et al., 1997; Miyamoto et al., 1998; Renshaw et al., 1997), the control of specific protease/protease inhibitors may be a common event in ‘outside-in’ signaling initiated by adhesive state and integrin engagement. Several matrices (i.e., fibronectin vs vitronectin) clearly differ in relative capacity to induce PAI-1 expression in T2 cells allowed to adhere under growth factor-free conditions. Most novel, however, is the observation that only certain matrix attachments synergize with TGF-β1 to achieve maximal PAI-1 expression. Whether matrix-type variations in the amplitude and duration of ERK signaling underlies this differential response in T2 epithelial cells is currently under study. Nevertheless, the present findings indicate that adhesive influences also modulate TGF-β1 signaling to target genes (i.e. PAI-1).
Acknowledgements
This work was supported by grants from the NIH (GM57242, GM07033, CA69612) and the Department of the Army (DAMD17-98-1-8015, DAMD17-00-1-0124).