Plasminogen Activation–Induced Pericellular Fibronectin Proteolysis Promotes Fibroblast Apoptosis

Cells and Cell Culture

Normal primary human embryonic lung fibroblasts (IMR-90 [26]; Institute for Medical Research, Camden, NJ) were cultured as previously described (9) and studies performed at passages 5 to 9. Cells were plated on cell culture dishes or on 96-well plates with Dulbecco's modified Eagle's medium (DMEM) containing 5% fetal bovine serum (FBS) in 5% CO₂–95% air and then growth-arrested for 24 hours in DMEM containing 0.01% FBS before treatment in serum-free DMEM without phenol red. Primary murine fibroblasts were grown from lung explants of wild-type (WT) and PAI-1 knockout (PAI-1^−/−) mice (C57BL/6 background) and cultured as previously described (19, 27). Experiments with these primary lung fibroblasts were performed between passages 2 and 4.

Reagents

TGF-β1 derived from porcine platelets was obtained from R&D Systems (Minneapolis, MN) and acid-activated before use. Antibodies to cleaved caspase-3 were from R&D Systems. Plasmin, ɛ-aminocaproic acid and mouse monoclonal antibody to fibronectin (IST-4) were from Sigma (St. Louis, MO). Mouse monoclonal antibody to single-stranded DNA (ssDNA) was from Chemicon International (Temecula, CA). Antibody to GAPDH was from Abcam (Cambridge, MA). Secondary HRP-conjugated anti-mouse and anti-rabbit antibodies were obtained from Pierce (Rockford, IL). α2-antiplasmin and plasminogen were obtained from American Diagnostica (Stamford, CT). BB2516 (Marimastat) was from Calbiochem (San Diego, CA). A genetically engineered variant of PAI-1, 14–1, which retains all of the antiproteolytic and vitronectin binding properties of native PAI-1 but is resistant to spontaneous conversion to the latent form, was obtained from Molecular Innovations, Inc. (Detroit, MI) (28). SB431542 was from Tocris Bioscience (Ellisville, MO).

Western Immunoblotting

Cell culture supernatants were centrifuged for 3 minutes at 10,000 rpm to pellet floating cells. The soluble fraction was separated and frozen at −20°C. Adherent cells on cell culture dishes were washed with PBS and lysed with RIPA buffer. This cell lysate was combined with the insoluble fraction of the supernatant and mixed vigorously. Protein concentrations of both the supernatant and cellular fractions were determined. Equal amounts of protein were then used for SDS-PAGE immunoelectrophoresis and Western immunoblotting under reducing conditions as previously described (9).

Plasmin Activity Assay

Plasmin activity in cell culture supernatants was determined using the SPECTROZYME PL chromogenic substrate obtained from American Diagnostica according to manufacturer's instructions. Cells were treated in phenol red–free media, and 50 μl of the cell culture supernatant was transferred to a single well of 96-well plate containing 50 μl of SPECTROZYME PL chromogenic substrate and 50 μl of buffer solution. Absorbance was measured at 405 nm using a VERSAmax microplate reader from Molecular Devices (Sunnyvale, CA). In all cases, two replicates of each sample were measured.

ssDNA Enzyme-Linked Immunosorbent Assay for Apoptosis

This assay is based on the susceptibility of double-stranded DNA in apoptotic cells to form single strands when exposed to formamide (23) and was performed as previously described (9).

Affymetrix Microarray

IMR-90 fibroblasts were grown to 80% confluence, serum starved for 48 hours, and treated with TGF-β1 (2.0 ng/ml), or were not treated, for 18 hours. RNA was isolated and Affymetrix microarray analysis was performed. Values represent mean ± SEM (n = 3).

Statistical and Densitometric Analyses

Statistical analysis was performed using Student's t test when comparing two groups and one-way ANOVA with Bonferroni correction when comparing three or more experimental conditions. This analysis was done using GraphPad Prism version 3.0 for Windows (GraphPad Software, San Diego, CA).

Exogenous Plasmin Induces Fibroblast Apoptosis in Association with Fibronectin Proteolysis

Our previous studies suggest that the interactions of fibroblasts with ECM components regulate their differentiation and susceptibility to apoptosis (8, 10). To investigate the potential induction of fibroblast apoptosis by plasmin-mediated matrix proteolysis, IMR-90 fibroblasts were treated with or without exogenous plasmin in the presence or absence of TGF-β1. Cell culture supernatants collected from plasmin-treated fibroblasts demonstrated dose-dependent fibronectin proteolysis (Figure 1A). TGF-β1 treatment increased the total amount of fibronectin in the cell culture supernatants but failed to inhibit plasmin-mediated fibronectin proteolysis (Figure 1B). In addition, plasmin treatment induced fibroblast apoptosis in a dose- and time-dependent manner (Figures 1C, 1D, and 1E). As with fibronectin proteolysis, exogenous TGF-β1 failed to inhibit plasmin-induced fibroblast apoptosis (Figure 1E).

Plasminogen Activation by Fibroblasts Is Associated with Fibronectin Proteolysis and Induction of Anoikis

We next sought to determine if endogenous mechanisms of plasminogen activation in IMR-90 fibroblasts would recapitulate the effects of exogenous plasmin on fibronectin degradation and fibroblast anoikis. Fibroblasts were treated with increasing plasminogen concentrations for various periods of time, and cell culture supernatants were assessed for plasmin activity. The addition of plasminogen with concentrations ranging from 5.0 to 50 μg/ml led to detectable plasmin activity at 6 hours, with maximal activity between 16 and 24 hours (Figures 2A and 2D, open bars). The appearance of fibronectin degradation products in the cell culture supernatant closely correlated with plasminogen activation (Figures 2B, 2C, and 2E). At all doses and time points indicated, TGF-β1 co-treatment significantly attenuated plasminogen activation (Figures 2A and 2D, solid bars) and fibronectin proteolysis (Figure 2C).

TGF-β1 Inhibits Plasminogen-Induced Fibroblast Apoptosis

Having shown that plasmin induces fibroblast apoptosis, that fibroblasts activate plasminogen, and that plasminogen activation leads to fibronectin proteolysis, we next investigated the effects of exogenous plasminogen on fibroblast apoptosis. Treatment of IMR-90 fibroblasts with exogenous plasminogen induced apoptosis with a dose response (Figures 3A and 3B) and time course (Figures 3C and 3D) similar to that of plasminogen activation and plasminogen-mediated fibronectin proteolysis. To determine if TGF-β1 would also attenuate plasminogen-induced apoptosis, fibroblasts were exposed to plasminogen with or without TGF-β1 (2 ng/ml) for 18 hours and apoptosis was assessed (Figures 4A, 4B, and 4C). In contrast to its inability to limit plasmin-induced apoptosis, TGF-β1 conferred significant protection from the pro-apoptotic effects of plasminogen (Figures 4A, 4B, and 4C). TGF-β1–mediated protection from plasminogen-induced apoptosis was maintained when its administration was delayed up to 4 hours after plasminogen treatment, but the protection was lost if TGF-β1 was administered 8 or more hours after plasminogen (Figure 4D). These findings suggest that to inhibit apoptosis, the TGF-β1–induced “protective factor(s)” must be present before significant plasminogen activation has occurred.

Next, we compared the effects of TGF-β1 and α2 antiplasmin or ɛ-aminocaproic acid on plasminogen-induced apoptosis (Figure E1 in the online supplement). α2-antiplasmin, a plasmin inhibitor, completely blocked plasminogen-induced fibroblast apoptosis and TGF-β1 conferred nearly equivalent protection (Figure E1-A and E1-B). The lysine analog, ɛ-aminocaproic acid also blocked plasminogen-induced apoptosis (Figure E1-C) (19, 29). These experiments confirm that fibroblast apoptosis induced by plasminogen requires plasminogen activation.

Plasmin has been shown to activate latent TGF-β1 from the ECM (30), providing a potential feedback mechanism to negatively regulate plasminogen-mediated apoptosis. To investigate whether plasmin-mediated TGF-β1 activation may, in the absence of exogenous TGF-β1, limit plasmin-mediated apoptosis, we treated IMR-90 fibroblasts with plasmin or plasminogen in the presence/absence of a pharmacologic inhibitor of the type 1 TGF-β receptor (ALK-5), SB431542. We have previously shown that this inhibitor blocks TGF-β1 SMAD-dependent signaling in IMR-90 fibroblasts (10). Neither plasmin nor plasminogen-mediated apoptosis increased with inhibition of TGF-β1 signaling (Figure 5). This finding suggests that any activation of latent TGF-β1 by plasmin is insufficient to overcome apoptosis induced by pericellular matrix proteolysis.

Plasminogen-Mediated Fibroblast Apoptosis Is Independent of MMP Activity

In addition to fibrinolysis and ECM proteolysis, plasmin has been implicated in the activation of matrix metalloproteinases (MMPs), a family of proteinases capable of degrading all components of the ECM. We sought to determine if fibroblast apoptosis induced by plasminogen activation was a direct effect of plasmin-mediated ECM proteolysis or secondary to MMP activation. We first assessed if a broad spectrum MMP inhibitor (marimastat, BB2516), at doses known to inhibit MMPs, would alter plasminogen activation and/or fibronectin proteolysis (31, 32). IMR-90 fibroblasts were treated with plasminogen in the presence/absence of BB2516 (5 μM) and/or TGF-β1 for 18 hours. BB2516 had no significant effect on plasminogen activation, fibronectin proteolysis, or the ability of TGF-β1 to inhibit these cellular and pericellular processes (Figures 6A and 6B). Similarly, BB2516 had no effect on plasminogen-induced fibroblast apoptosis or TGF-β1–mediated protection from apoptosis (Figures 6C and 6D).

Plasminogen-Induced Fibroblast Apoptosis Is Not Mediated by Fibronectin Degradation Peptides

Soluble fibronectin peptides have been shown to induce fibroblast anoikis through competitive blockade of integrin-dependent adhesion signaling pathways (33). We questioned if plasmin-mediated fibroblast apoptosis might be induced by the generation of soluble fibronectin degradation peptides released into the cellular microenvironment. To test this possibility, conditioned media was collected from fibroblasts treated with/without TGF-β1 or plasminogen. We confirmed that conditioned media from plasminogen-treated fibroblasts, but not the TGF-β1 or control fibroblasts, contained fibronectin proteolytic peptides (Figure 7A). α2-antiplasmin was added to the conditioned media to neutralize residual plasmin activity, and the conditioned media was then transferred to growth-arrested fibroblasts that were pre-treated with suramin (300 μM) to block activation of growth factor receptors (66). Apoptosis was assessed after 18 hours (Figures 7B and 7C). In these experiments, we found no differences in apoptosis of cells treated with conditioned media containing fibronectin peptide fragments compared with fibroblasts treated with conditioned media lacking these cleaved peptides. These studies indicate that, while pericellular fibronectin degradation is associated with fibroblast apoptosis, the soluble fibronectin peptides themselves are incapable of mediating apoptosis in this context.

TGF-β1 Protection from Plasminogen-Induced Apoptosis Is Mediated by PAI-1

Fibroblast apoptosis induced by plasminogen correlates with plasmin activity. Moreover, we found that plasminogen-mediated apoptosis, but not plasmin-mediated apoptosis, was inhibited by TGF-β1 if the TGF-β1 was given before significant plasminogen activation had occurred. These findings suggested that the apoptosis protection afforded by TGF-β1 may depend on the inhibition of plasminogen activation. Previous studies have shown that the expression of PAI-1, which inhibits plasminogen activation, is upregulated by TGF-β1 in lung fibroblasts (34). In our experimental conditions, PAI-1 expression is potently induced by TGF-β1 in IMR-90 fibroblasts (Figure 8A). To determine if PAI-1 would inhibit plasminogen-mediated apoptosis, IMR-90 fibroblasts were treated with plasminogen ± TGF-β1 or exogenous PAI-1 (Figures 8B and 8C). Analysis of the cell culture supernatants confirmed that both exogenous PAI-1 and TGF-β1 significantly blocked plasminogen activation. In addition, both TGF-β1 and exogenous PAI-1 attenuated plasminogen-induced fibroblast apoptosis (Figure 8C).

To confirm that PAI-1 is required for TGF-β1 inhibition of plasminogen-induced apoptosis, we used primary lung fibroblasts from wild-type (WT) and PAI-1 null mice (PAI-1^−/−). As with IMR-90 fibroblasts and NIH-3T3 fibroblasts, wild-type murine lung fibroblasts activate plasminogen and are susceptible to plasminogen-mediated apoptosis (Figures 8D, open bars; 8E; and 8F, open bars). The effect of TGF-β1 on inhibition of plasminogen activation was found to be greater with 5.0 ng/ml than 2.0 ng/ml. At this dose of TGF-β1, significant inhibition of plasminogen activation and blockade of plasminogen-induced fibroblast apoptosis was observed (Figures 8D, open bars; 8E; and 8F, open bars). Relative to wild-type fibroblasts, PAI-1^−/− fibroblasts generated increased plasmin activity after plasminogen administration and, importantly, TGF-β1 failed to inhibit plasminogen activation (Figure 8D, solid bars). Apoptosis in the PAI-1^−/− fibroblasts correlated with plasmin activity; PAI- 1^−/− fibroblasts displayed increased apoptosis after administration of plasminogen when compared with wild-type fibroblasts, and TGF-β1 failed to confer protection from plasminogen-induced apoptosis (Figures 8E and 8F, solid bars). Finally, the addition of exogenous PAI-1 to PAI-1^−/− fibroblasts rescued the knock-out phenotype and restored apoptosis resistance in plasminogen-treated fibroblasts (Figure 8F, solid bars).