PAF-Induced Furin and MT1-MMP Expression Is Independent of MMP-2 Activation in Corneal Myofibroblasts

Abstract

purpose. Corneal stromal myofibroblasts express the platelet-activating factor (PAF) receptor, but its role is unclear. In the present study, the effect of PAF on induction of metalloproteinases (MMPs) was investigated.

methods. Rabbit corneal myofibroblasts were identified by immunodetection of α-smooth muscle (α-SM)-actin. MT1-MMP, MMP-2, MMP-9, and tissue inhibitor of matrix metalloproteinase (TIMP)-2 were detected by immunofluorescence. Cells were treated with 100 nM cPAF, with or without the PAF antagonist BN 50730 or the furin inhibitor nona-d-arg-NH₂. Gene-expression levels for furin, urokinase plasminogen activator, MMP-2, MMP-9, MT1-MMP, and TIMP-2 were determined by real-time PCR. Protein expression was assessed by Western blot. MMP-2 and -9 activity was determined by gelatin zymography. Active MT1-MMP levels were measured by ELISA.

results. cPAF triggered significantly increased MT1-MMP, MMP-2, MMP-9, and TIMP-2 mRNA expression, followed by increased active MT1-MMP protein expression at 12 hours, whereas TIMP-2 protein increased at 24 hours. PAF also induced furin gene expression, followed by increased protein expression. Nona-d-arg-NH₂ blocked cPAF induction of MT1-MMP activity. PAF-treated myofibroblasts showed increased active MMP-9 protein, but unchanged MMP-2 activity. Pretreatment with BN 50730 blocked PAF-induced transcription and translation of these proteins.

conclusions. PAF, through a receptor-mediated mechanism, induces a specific pattern of furin, MMP, and TIMP-2 expression in corneal myofibroblasts. MMP-2 activity was unchanged by PAF treatment. These results suggest that in response to the inflammatory mediator PAF, induction of MT1-MMP is independent of MMP-2 activity in corneal myofibroblasts. Thus, PAF-mediated changes in extracellular matrix composition surrounding the myofibroblasts could be important in regulating the corneal scarring process. Moreover, PAF antagonists could be useful in maintaining corneal transparency.

Corneal wound healing is a complex process relying on the coordinated actions of distinct cells and mediators; however, the mechanism(s) that regulate it are unclear. The corneal stroma is composed of a dynamic network of collagens (20%–25%)—mainly collagen types I, V, and VI—and glycoproteins and proteoglycans, organized in a defined spatial arrangement. ¹ The architecture and uniquely transparent nature of the corneal matrix are maintained by a sparse population of keratocytes. Under normal conditions, the extracellular matrix (ECM) turnover is low; however, this increases significantly after an injury. Evidence suggests that during corneal wound healing, quiescent keratocytes adjacent to the wound site are stimulated to proliferate and undergo phenotypic transition to active fibroblasts and myofibroblasts. ^{2 3 4} This process is characterized by dramatic changes in cellular morphology and induction of a number of new genes. One of these cell types, the corneal myofibroblast, is essential to connective tissue remodeling where its function is to establish tension during the latter stages of wound healing and contracture. ^{4 5 6} Movement of these cells through the ECM and contraction of collagen-containing tissue are crucial components of the repair process; however, excessive remodeling and contraction of corneal stroma can lead to loss of corneal transparency. ^{4 7}  

The remodeling of the ECM during corneal wound healing is functionally linked to a family of multidomain, zinc-containing neutral enzymes known as matrix metalloproteinases (MMPs) and serine proteases, such as urokinase type (uPA), and tissue type (tPA) plasminogen activators. ⁸ These enzymes, with different substrate specificities and various optimal conditions, are tightly regulated by environmental stimuli coupled with the intracellular signaling mechanisms in the cells that produce them. Moreover, the activities of MMPs are precisely regulated at various levels, such as transcription, interaction with specific ECM components, and activation of precursor zymogens. ⁸  

Findings in studies have suggested that myofibroblasts are involved in degrading some of the key matrix proteins (such as type I collagen) and may play an important role in tissue contraction and corneal injury through production of MMPs. ^{9 10 11} One of these enzymes, MMP-2, digests types I, IV, and V collagens, in addition to gelatin, fibronectin, and laminin, ^{12 13} and is expressed in the anterior stroma of patients who undergo complicated laser-induced in situ keratomileusis (LASIK). ¹⁰ More recently, another MMP, membrane type 1 (MT1)-MMP, was reported to regulate matrix turnover either directly through collagenolytic activity against collagen types 1, II, and III, or by forming a complex with tissue inhibitor (TIMP)-2 and activating downstream MMPs such as MMP-2. ^{14 15 16} Other studies have shown the existence of a furin convertase/T1-MMP/MMP axis in the regulation of MMP-2 activity. ¹⁷ Furin convertase is a ubiquitously expressed calcium-dependent endoprotease thought to activate pro-MT1-MMP through cleavage of its prodomains. ^{18 19}  

Recent studies in our laboratory have shown that transformation of fibroblasts to the myofibroblast phenotype results in the expression of platelet-activating factor (PAF) receptor (He J, et al. IOVS 2003;44:ARVO E-Abstract 877). PAF is a potent bioactive lipid mediator that accumulates rapidly in the cornea after an injury. ^{20 21} In corneal epithelial cells, PAF stimulates the expression and activity of selective MMPs ^{22 23 24 25} and their TIMPs. ²⁴ We have recently reported that in vascular endothelial cells, PAF induces the activation of MMP-2 via the stimulation of MT1-MMP and TIMP-2 complex formation. ²⁶ Furthermore, we have demonstrated that in a rabbit model of diffuse lamellar keratitis, treatment with a PAF-receptor antagonist decreased the number of myofibroblasts in the stromal interface. ²⁷  

The goal of the present study was to investigate whether PAF, through a receptor-mediated mechanism, induces a specific pattern of MMP and TIMP gene expression in isolated corneal myofibroblasts. The results showed that PAF activated the expression of MMP-2, MMP-9, MT1-MMP, and TIMP-2 in these cells, and that PAF, through the stimulation of furin activity, led to increased MT1-MMP, but not MMP-2, activity. 

Materials and Methods

The MT1-MMP (MMP-14) activity assay kit was obtained from Amersham (Buckinghamshire, UK), 10× RIPA lysis buffer from Upstate Biotechnology (Lake Placid, NY); protease inhibitor cocktail, 1,10-phenanthroline, anti-α smooth muscle (α-SM)-actin, and phenylmethylsulfonyl-fluoride (PMSF) were from Sigma-Aldrich (St. Louis, MO); cPAF (1-alkyl-2n-methylcarbamyl-sn-glycerol-phosphorylcholine, a nonhydrolyzable PAF analogue) was from Cayman Chemical Co. (Ann Arbor, MI); human pro-MMP-9 (92 kDa), active MMP-9 (84 kDa), and recombinant human TIMP-2 standard were from Calbiochem (La Jolla, CA); human pro-MMP-2 standard (72 kDa), MT1-MMP pro-enzyme, TIMP-2, and MT1-MMP monoclonal antibodies were from Chemicon (Temecula, CA); monoclonal antibody to furin convertase (MON-150) was from Alexis (San Diego, CA); PCR master mix (TaqMan SYBR Green), reverse transcriptase (TaqMan), ribosomal RNA control reagents (18s RNA; TaqMan), deoxynucleotides (dNTPs), and DNA polymerase (Ampli-Taq Gold) were from Applied Biosystems, Inc. (ABI; Foster City, CA); agarose, ethidium bromide, and DNA mass ladder (100 bp) were from Invitrogen-Gibco-Life Technologies (Grand Island, NY); a SV total RNA isolation system was from Promega (Madison, WI); optical-quality sealing tape (iCycler iQ), PCR plates, 10% zymogram ready gel containing gelatin, gradient (4%–12%) polyacrylamide gels (Criterion XT), 1× Tris/glycine/SDS running buffer, 4× XT sample loading buffer, 20× XT reducing reagent, zymogram renaturation buffer, zymogram development buffer, Tween-20, prestained molecular weight (MWT) markers (Kaleidoscope), blotting-grade blocker nonfat dry milk, and bicinchoninic acid (BCA) protein assay reagent were from Bio-Rad (Hercules, CA); a Western blot detection kit (ECL Super Signal) and FITC-conjugated goat anti-mouse IgG antibody were from Amersham; centrifugation filters (model YM10, Centricon) were from Millipore (Bedford, MA); biotinylated protein markers (anti-biotin-horse radish peroxidase [HRP]-linked) were from Cell Signaling (Beverly, MA); and furin-specific inhibitor nona-d-arg-NH₂ was a gift from Iris Lindberg (Department of Biochemistry and Molecular Biology, LSU Health Sciences Center, New Orleans, LA). 

Myofibroblast Culture

Rabbit eyes (Pel-Freeze Biologicals, AR) were shipped to the laboratory on ice in Hanks’ solution containing antibiotics and antimycotic and used within 24 hours of enucleation. After dissection of corneas, primary myofibroblast cultures were prepared as described by Masur et al. ²⁸ Fibroblasts from passage 6 or greater were used. These fibroblasts do not express PAF receptor. ²⁹ Briefly, corneal fibroblasts in DMEM-F12 medium containing 5% fetal bovine serum (FBS) were seeded at low density (five cells per square millimeter) into 100-mm dishes and allowed to grow for 3 to 5 days. Myofibroblast phenotype was identified by the expression of α-SM-actin. The cells were starved overnight in DMEM-F12 medium containing 0.1% horse serum (HS; heat inactivated) and then stimulated with 100 nM cPAF for different times according to the experiment. In some experiments, cells were preincubated for 30 minutes with the PAF antagonists BN 50730 (10 μM, dissolved in dimethyl sulfoxide [DMSO]), CV 3988 (10 μM, dissolved in ethanol), or CV 6209 (1 μM, dissolved in water), followed by aspiration of medium and supplementation with PAF in DMEM-F12 medium containing 0.1% HS with or without antagonist. Controls contained 0.01% DMSO. For furin inhibitor studies cells were preincubated with 0.5 μM furin-specific inhibitor nona-d-arg-NH₂ (dissolved in water) for 30 minutes, followed by treatment with DMEM-F12 medium containing PAF and 0.1% HS, with or without furin inhibitor. 

RNA Extraction

Corneal myofibroblasts (two 100-mm dishes per condition) were carefully scraped and transferred to RNase-free tubes containing 175 μL lysis buffer. Samples were gently vortexed and mixed with 350 μL SV total RNA dilution buffer. After heating for 3 minutes at 70°C, extracts were centrifuged at 14,000g for 10 minutes. Supernatant was removed and mixed with 200 μL ethanol (95%), followed by transfer to spin baskets and centrifugation at 14,000g for 1 minute. Final RNA extracts were eluted from the spin columns in a total volume of 30 μL nuclease-free water. The concentration and purity of RNA were determined by spectrophotometry. Typical yields of RNA varied from 10 to 20 μg per sample. All RNA preparations had an OD260-to-OD280 ratio in the range of 1.8 to 2.0. 

Reverse Transcription and PCR

One microgram of isolated RNA was reverse-transcribed in a total reaction volume of 25 μL, containing 1× first-strand buffer, 1 U RNase, 4 mM dNTPs, 0.01 mM dithiothreitol (DTT), and 400 U Moloney murine leukemia virus (M-MLV) reverse transcriptase. Each reaction mixture was incubated at 25°C for 10 minutes, followed by 30 minutes at 42°C and a final hold for 5 minutes at 95°C. Aliquots of cDNA (5 μL) were amplified (GeneAmp 9600 series PCR system; ABI) in a 25-μL reaction mixture containing 1× PCR buffer (10 mM Tris-HCl [pH 8.3]; 50 mM KCl; 1.5 mM MgCl₂ and 0.001% [wt/vol] gelatin), 0.4 μM each dNTP, and 0.4 μM of each 5′ and 3′ primer. Table 1lists the primer sequences and PCR conditions for each gene studied. A control without reverse transcriptase and a negative control without RNA were set up. Amplification products were resolved on a 2% agarose gel containing 1 μg/mL ethidium bromide, and the product size was determined by comparison to a 100-bp ladder run on the gel. 

Real-Time PCR

For each treatment, a PCR reaction mixture containing 5 μL cDNA, 12.5 mL 2× PCR master mix (SYBR Green; ABI), and 0.5 μL of each primer (10 μM) in a total volume of 25 μL was prepared. Real-time PCR was performed (iCycler IQ Multicolor Real-Time PCR Detection System; Bio-Rad), using the thermocycler programs listed in Table 1 . Samples were run in duplicate for the internal control 18s rRNA and in triplicate for the gene of interest on the same plate. To correct for variations in the number of myofibroblast cells among flasks and treatments, all values were normalized to α-SM-actin mRNA, the phenotypic marker for myofibroblasts. The expression of α-SM-actin was not affected by PAF treatment. All quantitations were normalized to the 18s rRNA endogenous control and changes in gene expression reported as the x-fold increase relative to untreated controls, as previously described. ²⁴ The no-RNA template and the no-reverse-transcriptase enzyme controls were run to determine whether fluorescent contaminants were present in the sample. To confirm amplification specificity, the PCR products were subjected to melt-curve analysis and subsequent agarose gel electrophoresis. ²⁴  

Western-Blot Analysis

MT1-MMP.

Myofibroblasts were trypsinized for 2 minutes and then neutralized with serum-supplemented medium, washed twice with ice-cold phosphate-buffered saline, and pelleted at 10,000g for 3 minutes. Cell pellets were resuspended in 100 μL lysis buffer containing 10 mM Tris-HCl (pH 7.6), 10 mM NaCl, 3 mM MgCl₂, 1 mM PMSF, and 10 μL protease inhibitor cocktail (Sigma-Aldrich) and disrupted by mechanical shearing through a 23–1/3-gauge needle. After centrifugation at 45,000 rpm for 35 minutes at 4°C, the supernatant was discarded and the pellet resuspended in 100 μL RIPA buffer (1× PBS, 0.1% SDS, 0.5% Na deoxycholate, 1% Nonidet P-40, 1 mM PMSF, 1 mM leupeptin, and 10 μL protease cocktail inhibitor), stirred on ice for 1 hour and centrifuged at 45,000 rpm for 35 minutes at 4°C. Supernatant containing the membrane proteins was collected and proteins quantified using the BCA protein assay (Bio-Rad). 

Western-Blot analysis was performed with a monoclonal antibody to MT1-MMP, which recognizes both the latent (64 kDa) and the active (54 kDa) forms of the enzyme. Protein samples (40 μg) were mixed with XT sample loading buffer containing 1× XT reducing reagent and boiled for 5 minutes. Protein samples, MT1-MMP standard, biotinylated protein markers, and prestained markers (Kaleidoscope; Bio-Rad) were separated by electrophoresis on a gradient (4%–12%) polyacrylamide gel (Criterion XT; Bio-Rad) and transferred to nitrocellulose membrane. This procedure was followed by a 2-hour incubation with monoclonal antibody to MT1-MMP, six washes with 0.01% Tween-TBS buffer, and a 1-hour incubation with secondary antibody to MT1-MMP (anti-mouse HRP–linked) and biotinylated markers (anti-biotin-HRP-linked). Immunodetection of the antigen was performed with a chemiluminescence detection kit (ECL Super Signal Western Blot; Amersham). The molecular masses of the immunostained bands were estimated by comparing their migratory position with that of standard MT1-MMP and molecular-weight standards. 

Furin and TIMP-2.

For furin and TIMP-2 analysis, cells were homogenized as described for MT1-MMP and centrifuged at 10,000g for 30 minutes to remove cell debris. Immunoblot analysis was performed as described, either with mouse monoclonal antibodies raised against a cysteine-rich region of furin or with a mouse anti-human-TIMP-2 monoclonal antibody. 

Immunoprecipitation

Corneal myofibroblasts were stimulated with PAF for 12 hours and lysed as described for Western blot. An 80-μg aliquot of total protein was immunoprecipitated with 2 μg anti-MT1-MMP overnight at 4°C, using an immunoprecipitation kit (Protein-G; Sigma-Aldrich), according to the manufacturer’s recommendations. Precipitates were diluted with XT sample loading buffer and analyzed by Western blot with anti-TIMP-2 and anti-MT1-MMP, as described in the prior section. 

Analysis of MMP-2 and -9 by Zymography

Zymography was conducted with use of gelatin-containing SDS-PAGE. The gels were prefocused for 30 minutes at 100 V. Conditioned medium (2 mL) from 24-hour PAF-treated and untreated corneal myofibroblasts was collected and concentrated using micrometer-pore filters (10-kDa cutoff; Centricon; Millipore). Aliquots (10 μg) of protein were diluted 1:1 with zymography sample loading buffer and electrophoresed with a 1× Tris/glycine/SDS running buffer for 90 minutes at 100 V. The procedure was followed by incubation in zymogram renaturation buffer for 1 hour at room temperature. The buffer was then replaced with 1× zymogram development buffer, and incubation continued overnight. The following day, gels were stained with Coomassie blue, and the positions of the active MMP-2 and -9 enzymes were visualized as clear bands against a uniformly dark-stained background. The identities of the bands were confirmed by comparison with pro-MMP-2 (human, 72 kDa) and pro-MMP-9 (human neutrophil granulocyte, 92 kDa) standards run on the gel. Images were recorded on a gel reader (Gel Doc-1000; Bio-Rad) fitted with a white-light conversion screen. To verify that proteolytic activities detected were due to MMPs, duplicate zymographs were developed in zymogram renaturation buffer containing 10 mM of the MMP inhibitor 1,10-phenanthroline. 

MT1-MMP ELISA Assay

In this assay, solubilized membrane-bound MT1-MMP was bound to a specific antibody coating a 96-well microplate, and activity was assayed with a chromogenic peptide substrate. Duplicate 100-mm dishes of myofibroblasts, approximately 80% confluent, were scraped and cells pelleted at 10,000g for 3 minutes. The pellets were resuspended in extraction buffer, as described in the assay kit. Colorimetric analysis was performed on a microtiter plate reader with a computer (Soft Pro; Bucher Biotec, Basel, Switzerland). The optical densities of the samples were measured at 405 nm. A standard curve for active MT1-MMP was generated and used to determine the levels of MT1-MMP activity. 

Immunofluorescence

Rabbit corneal myofibroblasts plated at a concentration of 5 cells/mm² were allowed to grow for 72 hours in DMEM containing 5% FBS. The cells were washed once in ice-cold PBS, followed by fixation and permeabilization with precooled methanol for 5 minutes. Primary antibodies for MT1-MMP, MMP-2, TIMP-2, and α-SM-actin diluted in PBS containing 1.5% goat serum were added to myofibroblasts and incubated for 1 hour at room temperature, followed by further incubation with FITC-conjugated goat anti-mouse IgG antibody (1:50 dilution; Amersham). Cells were then observed under a fluorescence microscope (Eclipse TE 200; Nikon, Tokyo, Japan). 

Results

PAF-Induced MT1-MMP, MMP-2, MMP-9, and TIMP-2 Gene Expression in Corneal Myofibroblasts

The presence of MMPs and their localization in rabbit corneal myofibroblasts were determined by immunostaining for MT1-MMP, TIMP-2, MMP-2, and MMP-9 (Fig. 1) . Immunostaining for α-SM-actin, a marker for the myofibroblast phenotype, confirmed the presence of myofibroblasts in culture. Staining for MT1-MMP and TIMP-2 was localized mainly in the plasma membrane. MMP-2 staining was observed in the cytosol surrounding the nucleus and in the membranes of cells. MMP-9 immunostaining was localized in the cytoskeleton. 

Because myofibroblasts express the PAF receptor (He J, et al. IOVS 2003;44:ARVO E-Abstract 877) and PAF is an activator of selective MMPs in corneal epithelium, ^{22 23 24} we investigated whether PAF is able to stimulate a specific pattern of MMPs and their inhibitors in corneal myofibroblasts. For these experiments, we used traditional PCR and real-time PCR. RT-PCR analysis confirmed the immunofluorescence, in that myofibroblasts expressed MT1-MMP, MMP-2, TIMP-2, and MMP-9 mRNA (Fig. 2) . They did not express uPA. Stimulation with cPAF induced MT1-MMP, MMP-2, and TIMP-2 mRNA expression at 4 hours, followed by a decrease to basal levels by 12 hours. PAF also increased MMP-9 mRNA levels as early as 2 hours, with a peak at 4 hours and decrease to control levels by 12 hours. PCR amplification using total RNA samples before the RT step (RT–) revealed no genomic DNA contamination. 

To quantify the x-fold increases in the gene expression of these MMPs and their inhibitors, we used real-time PCR (Fig. 3) . cPAF induced a twofold increase in MT1-MMP (P ≤ 0.05) and TIMP-2 (P ≤ 0.05) gene expression at 4 and 8 hours. cPAF treatment also produced a twofold increase in MMP-2 mRNA expression at 4 hours (P ≤ 0.01), whereas at 8 hours there was a modest but significant 1.6-fold (P ≤ 0.02) increase in expression. MMP-9 mRNA expression was induced by PAF as early as 2 hours with a twofold increase (P ≤ 0.02), followed by a 3.6-fold (P ≤ 0.01) increase at 4 hours. A nonsignificant increase was also detected at 8 hours. 

Pretreatment of corneal myofibroblasts with the PAF antagonist BN 50730 completely abolished cPAF-induced increases in MT1-MMP, MMP-2, TIMP-2, and MMP-9 mRNA expression (Fig. 4) . This suggests that cPAF induction of these MMPs and their inhibitors in corneal myofibroblasts is through a PAF-receptor–mediated mechanism. 

PAF Induces MT1-MMP and TIMP-2 Protein Expression

To determine whether the observed increases in MMP gene expression were translated into increased protein expression, corneal myofibroblasts were stimulated with 100 nM cPAF for 12 and 24 hours. These times were chosen to obtain enough protein to be analyzed by Western blot analysis as described in Methods. Blotting with the MT1-MMP antibody yielded two sharp bands corresponding to the pro- (62 kDa) and active (56 kDa) forms of MT1-MMP (Fig. 5A) . Addition of 100 nM cPAF increased MT1-MMP protein levels at 12 hours in membrane preparations of corneal myofibroblasts. Treatment of the cells with 10 μM BN 50730 blocked the stimulatory effect of cPAF on MT1-MMP protein production. 

To corroborate these results, and to support our hypothesis that active membrane bound MT1-MMP expression is induced by PAF treatment in corneal myofibroblasts, we measured changes in active MT1-MMP activity by ELISA after 6, 12, or 24 hours of PAF stimulation. There were no differences between the controls and cPAF at 6 and 24 hours. Stimulation with 100 nM cPAF for 12 hours produced a 40% to 50% increase in MT1-MMP activity compared with controls (Fig. 5B) . Pretreatment with the PAF antagonist BN 50730 blocked the stimulatory effect of cPAF on MT1-MMP activity. 

Exposure of corneal myofibroblasts to 100 nM cPAF for 24 hours also stimulated TIMP-2 protein expression (Fig. 5C) , followed by a return to control levels at longer incubation times. Earlier time points (12-hour incubation) did not produce changes in TIMP-2 levels. The induction of TIMP-2 was inhibited by the PAF antagonist. In an attempt to detect the TIMP-2/MT1-MMP complex, immunoprecipitation with anti-MT1-MMP antibody was performed. By Western blot analysis we were unable to detect positive staining under the conditions of our experiments (data not shown). 

PAF Stimulates MMP-9 but not MMP-2 Activity in Corneal Myofibroblasts

Recent studies suggest that membrane-anchored MT1-MMP is an activator of pro-MMP-2. ^{14 15} To examine whether the PAF-induced increase in MT1-MMP expression correlates with MMP-2 activation in corneal myofibroblasts, we studied the effect of cPAF on MMP-2 activity. Cells were stimulated with PAF for 12, 18, or 24 hours and medium collected and analyzed by gelatin zymography. At all the times analyzed, rabbit corneal myofibroblasts released only pro-MMP-2 enzyme. These results were further confirmed by Western-blot analysis using a monoclonal antibody to MMP-2 (data not shown). Figure 6represents the zymogram after 24 hours, when enough protein for MMP-9 and -2 could be detected in the medium. Moreover, treatment with other cytokines, such as IL1-α and TNF-α did not induce MMP-2 activation in corneal myofibroblasts in culture (data not shown). However, treatment with 100 nM cPAF increased active MMP-9 levels in these cells. This increase was inhibited by the PAF antagonist BN 50730. 

PAF-Induced Expression of Furin mRNA and Protein in Rabbit Corneal Myofibroblasts

The absence of change in MMP-2 activity in these cells after PAF treatment led us to examine which other mechanism(s) could account for the activation of MT1-MMP. Recent studies have suggested a role for furin in this activation. ^{17 18 19} This was further strengthened by the discovery of two potential furin cleavage sites in the propeptide domain of MT1-MMP. ¹⁸ We investigated whether PAF treatment in corneal myofibroblasts altered furin gene expression. PCR analysis (Fig. 7A)showed that rabbit corneal myofibroblasts constitutively expressed furin mRNA. When compared with untreated controls, 100 nM cPAF stimulated furin expression as early as 2 hours, with sustained increases occurring up to 8 hours. TGF-β, a positive inducer of furin in other cells ^{31 32} also stimulated increases in furin mRNA levels in corneal myofibroblasts. 

By real-time PCR analysis, we showed that PAF induced more than a twofold increase in furin mRNA expression at 2 hours, with higher increases occurring at 4 (3.2-fold) and 8 (5.4-fold) hours. At 8 hours, furin gene induction by cPAF was higher than that induced by TGF-β treatment. The increase was reversed by pretreatment with the PAF antagonist BN 50730. Other PAF antagonists, CV 3889 and CV 6209, were also tested and also inhibited furin gene induction (data not shown). Control cultures treated with PAF antagonist alone did not produce changes in gene expression when compared to vehicle controls. The experiments demonstrated that cPAF was a significant inducer of furin mRNA through activation of its receptor. 

The gene induction of furin by PAF was translated into the expression of a protein (Fig. 7C)with a molecular weight of approximately 100 kDa. Treatment with 100 nM cPAF produced a marked increase of furin protein at 6 hours; higher expression occurred at 12 hours. Pretreatment with a PAF antagonist inhibited this increase. 

Effect of Inhibition of Furin on PAF-Mediated MT1-MMP Activity

To determine the role of furin in the activity of MT1-MMP, corneal myofibroblasts were preincubated with 0.5 μM nona-d-arg-NH₂, a nontoxic inhibitor of furin ³³ for 30 minutes and then stimulated with 100 nM cPAF. Results from ELISA experiments revealed a 40% to 50% increase in MT1-MMP activity at 12 hours after cPAF stimulation (P < 0.02; Figs. 5 8 ). This increase was blocked in the presence of the furin inhibitor_, suggesting that PAF-induced MT1-MMP activity is dependent on furin activation. 

Discussion

The wound-healing process is a highly regulated event involving the release of growth factors, secretion of ECM components, and the eventual formation of granulation tissue (scar formation). Scarring is involved in the pathogenesis or failure of treatment of most visually disabling and blinding conditions. Over the past few years, studies have sought to elucidate the complex cascade that compromises the corneal stromal wound-healing response at the molecular and cellular levels. ^{2 3 4 7} Myofibroblasts are believed to play a role in these wound-healing events, through the production of MMPs, ^{9 11} which appear to be involved in ECM remodeling and the eventual contraction of the collagen-containing tissue. ^{9 34} At present, there is no clear understanding of which MMPs are synthesized by corneal myofibroblasts. Most of the published studies on MMP production have dealt with the fibroblast phenotype (for review, see Ref. ³⁵ ); however, these cells do not respond to PAF due to lack of PAF-receptor expression. ²⁹ We found that nonstimulated myofibroblasts expressed MT1-MMP, MMP-2, MMP-9, and TIMP-2, but not uPA. Colocalization studies were not performed, because the available primary antibodies are from the same host (mouse). PAF, a product of corneal injury, ^{20 21} induced the gene expression of MT1-MMP, MMP-2, MMP-9, and TIMP-2 in these cells, which was translated into increases in protein expression, with the exception of MMP-2. It is important to point out that although the gene induction of MMP-2 was significant, the increase was less than twofold at 8 hours. 

MMPs are synthesized as inactive zymogens (pro-MMPs) and their activation by proteolytic cleavage is a rate-limiting step in their catalytic function. MT1-MMP is a membrane-anchored MMP with a pivotal function in connective tissue metabolism ^{18 36} and pro-MMP-2 activation. Several studies have shown that activation of pro-MMP-2 by MT1-MMP involves the formation of a ternary complex between activated MT1-MMP, TIMP-2, and pro-MMP-2 at the cell surface. ^{14 15 16} TIMP-2 frees MT1-MMP at the cell surface, then cleaves the propeptide of pro-MMP-2, generating the active species. We have shown recently that in vascular endothelial cells, PAF stimulates MMP-2 activation through a MT1-MMP/TIMP-2 complex. ²⁶ Indeed, nonstimulated corneal myofibroblast cells stained with antibodies to MT1-MMP and TIMP-2 proteins exhibited positive immunostaining in the plasma membrane, whereas MMP-2 appeared to be mainly localized to the cytosol, although some was visible in the plasma membrane. 

Our studies revealed no transformation of pro-MMP-2 to the active enzyme in medium collected from corneal myofibroblasts treated with cPAF, although there was marked activation of MMP-9 on the same zymogram. One possible explanation for this result is that there are some as yet unidentified differences among the regulation and/or degradation of various mRNAs. It has been reported that many gene transcripts contain an adenine-uridine (AU)-rich destabilizing element in their 3′ termini. ³⁷ Notably, in human and rabbit collagenase genes, this motif is repeated three times. ³⁸ Thus, there is a possibility that mRNAs of various MMPs are more susceptible to degradation, due to the presence of these AU-rich motifs. Another possibility is that the PAF-induced increase in TIMP-2 expression in corneal myofibroblasts inhibits active MMP-2 in these cells, which could have important consequences in the wound-healing process, with TIMP-2 regulating active MMP-2 in corneal myofibroblasts and preventing excessive tissue remodeling. A third possibility is that pro-MMP-2 migrates to the nucleus, where it is activated and remains membrane bound as a trimolecular complex or bound to cell-surface integrins, such as αvβ3. ^{39 40} However, zymographic analysis of pure membrane isolates from PAF-treated myofibroblasts did not detect any conversion of pro-MMP-2 to its active form (data not shown). 

Our attempts to determine the formation of MT1-MMP/TIMP-2 by immunoprecipitation were not successful, even under nonreducing conditions. There is always the possibility that, with the available antibodies, we were unable to precipitate the complex after PAF stimulation. In addition, the time course of cPAF induction of TIMP-2 protein did not correlate with the expression of MT1-MMP protein, suggesting that MMP-2 activation was not linked to MT1-MMP activity through the formation of a ternary complex at the cell surface. 

Our results showed that PAF stimulated MT1-MMP as measured by ELISA. It has been reported that plasmin, trypsin, and urokinase may play an important role in converting MT1-MMP to its active form. ⁴¹ In rabbit corneal myofibroblasts, uPA levels were undetectable and remained unchanged by PAF treatment. One important finding, described herein for the first time in any tissue, is that PAF significantly stimulated the induction of furin gene and protein expression. However, regulation of furin gene expression by external stimuli is poorly understood, although evidence suggests a role for cytokines. Studies of the gene promoter of furin in humans have revealed the presence of AP-1 and SP-1 regions ^{42 43} and, in rabbit corneal epithelial cells, PAF increases the transcriptional activator AP-1–responsive elements c-fos and c-jun. ²² Furthermore, studies in our laboratory have shown that PAF induces SP-1 DNA-binding activity in rabbit corneal epithelial cells (Taheri F, Bazan HEP. IOVS 2003;44:ARVO E-Abstract 3833). 

The kinetics of PAF induction of furin and MT1-MMP expression in myofibroblasts suggests a relationship between these two proteins. Indeed, studies by Yana and Weiss ¹⁸ showed that MT1-MMP processing and activation are regulated by furin, which recognizes one of two potential recognition motifs in the MT1-MMP prodomain. The cDNA amino acid sequence of rabbit MT1-MMP shows 98% identity to human MT1-MMP and contains the furin-recognition sequence in the propeptide. Further evidence for furin’s role in activation of MT1-MMP was demonstrated by inhibition of its activity with a furin inhibitor. Although this inhibitor could also inhibit, with less potency, another related convertase, PACE 4, ³³ this peptidase does not play a major role in the processing of MT1-MMP. ¹⁸  

The induction of active MT1-MMP by PAF can remodel components of the surrounding ECM. The catalytic domain of this enzyme is structurally similar to that of other MMPs, suggesting that MT1-MMP directly degrades various ECM components. Indeed, studies by Ohuchi et al. ⁴⁴ have shown that deleted mutants of MT1-MMP from stable transfectants and native MT1-MMP secreted from a human breast carcinoma cell line (MDA-MB-231) degrade various fibrillar collagens (i.e., types I, II, and III), as well as other ECM components, including gelatin, proteoglycan, fibronectin, vitronectin, and laminin-1. Several of these proteins are components of the stroma and the basement membrane. 

Another MMP induced by PAF in myofibroblasts was gelatinase B (MMP-9). This enzyme is consistently induced in the stroma after deep keratectomy and photorefractive keratectomy (PRK) injury, ⁴⁵ and immunohistochemical studies in rabbit corneas showed the clear expression of MMP-9 in rabbit myofibroblasts after corneal injury. ¹¹ Although in this study PAF induced MMP-9 mRNA expression as well as activity in corneal myofibroblasts, it has been suggested that MMP-9 expression does not correlate with stromal remodeling, but rather that this enzyme plays a part in the resynthesis of the epithelial basement membrane. ⁴⁶  

In summary, in the current study, PAF, a potent bioactive lipid mediator that accumulates rapidly in the cornea after an injury, stimulated in corneal myofibroblasts the gene and protein expression of the convertase furin and MT1-MMP, MMP-9, and TIMP-2 through a receptor-mediated mechanism. The kinetics of furin and MT1-MMP expression suggest a relationship between these two enzymes. This was further evidenced when inhibition of furin activity blocked PAF stimulation of MT1-MMP activity in these cells. MMP-2 activity, however, remained unchanged. Moreover, the time course of cPAF induction of TIMP-2 protein did not correlate with the expression of MT1-MMP, and we were unable to detect MT1-MMP/TIMP-2 by immunoprecipitation, suggesting that the MT1-MMP/TIMP-2/pro-MMP-2 ternary complex was not formed in these cells. Therefore, our results suggest that, in response to the inflammatory mediator PAF, MT1-MMP induction is independent of MMP-2 activity in corneal myofibroblasts. PAF inhibition could be an important approach to attenuating or even preventing excessive remodeling of the surrounding ECM by myofibroblasts during corneal wound healing, which could lead to loss of tissue transparency. 

 Supported by National Eye Institute Grant R01EY04928.

 Submitted for publication July 21, 2004; revised October 25, 2004; accepted November 3, 2004.

 Disclosure: P. Ottino, None; J. He, None; T.W. Axelrad, None; H.E.P. Bazan, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Table 1.

 

Primer Sequences and Amplification Conditions for Various Genes Studied

Gene	Primer Sequence	Annealing Temp (°C)	Optimal Cycle No.	Product Size (bp)	Ref No.
MT1-MMP	accaatggatggacacagagaact	62	35	261	17
	ggttgttcctcaccctccagaa
Gelatinase (MMP-2)	ccccaaaacggacaaagag	60	30	331	24
	cacgagcaaaggcatcatcc
Gelatinase (MMP-9)	aaactggatgacgatgtctgcgtcccg	58	30	362	24
	acctgttccgctatggttacacccgcgta
TIMP-1	gcaactccggaccttgtcatc	60	32	326	24
	agcgtaggtcttggtgaagc
TIMP-2	gtagtgatcagggccaaag	60	27	416	24
	ttctctgtgacccagtccat
Urokinase	tggtttgcagccatctac	60	36	384	24
	tccaaagccagtgatctc
Furin	cagatcttcggggactattaccac	54	35	405	30
	cctgttgtcattcatctgtgtgta

 The sequences listed in the table correspond to primers taken from cited references.

Figure 1.

 

Immunostaining of MT1-MMP, MMP-2, MMP-9, and TIMP-2 in rabbit corneal myofibroblasts. Corneal myofibroblast cells were stained with the antibodies against MT1-MMP, TIMP-2, MMP-2, and MMP-9. Immunostaining for α-SM-actin protein confirmed the presence of myofibroblasts in culture. There was no staining when the cells were treated without the primary antibody or without IgG (negative controls).

Figure 2.

 

Effect of cPAF on MT1-MMP, MMP-2, MMP-9, and TIMP-2 mRNA expression in rabbit corneal myofibroblasts. RT-PCR was performed to determine MT1-MMP, MMP-2, MMP-9, and TIMP-2 mRNA expression in rabbit corneal myofibroblasts treated with 100 nM cPAF for 1 to 12 hours. The primer sets used revealed a 261-bp product for MT1-MMP, 331-bp for MMP-2, 362-bp for MMP-9, and 416-bp for TIMP-2. Expression of the housekeeping gene 18s rRNA did not change under the conditions tested in these experiments. The results in these gels are representative of two separate experiments. On each gel, no-reverse-transcriptase (RT−) and no-template (RNA−) controls resulted in no bands.

Figure 3.

 

Real-time PCR analysis of MT1-MMP, MMP-2, MMP-9, and TIMP-2 gene expression stimulated by PAF. Myofibroblasts were stimulated with 100 nM cPAF for up to 12 hours. Data are the increases (x-fold) in gene expression for MT1-MMP, MMP-2, MMP-9, and TIMP-2 relative to unstimulated controls (dashed line) as measured by real-time PCR and are the mean ± SEM of results in three independent experiments (each experiment run in triplicate). *Significant increase relative to the untreated control (P < 0.01).

Figure 4.

 

Effect of PAF antagonist on cPAF-induced gene expression of MMPs and TIMP-2 analyzed by real-time PCR. Corneal myofibroblasts were preincubated with BN 50730 for 1 hour, followed by cPAF stimulation for different times. The data represent increases (x-fold) in mRNA levels relative to vehicle controls (dashed line). *Significant (P < 0.01) difference relative to vehicle control. **Significant difference (P < 0.01) relative to cPAF-treated cells. The data are the mean ± SEM of results in three independent experiments (each experiment run in triplicate).

Figure 5.

 

PAF stimulated the expression of MT1-MMP and TIMP-2. (A) Western immunoblot analysis is shown of MT1-MMP protein expression in rabbit corneal myofibroblasts treated with the cPAF and PAF antagonist BN 50730. The positions of the 62-kDa Pro-MT1-MMP and 54-kDa active MT1-MMP were determined by comparison to 58-kDa human recombinant pro-MT1-MMP enzyme. The results are representative of three separate experiments. (B) Effect of cPAF stimulation on active MT1-MMP levels in corneal myofibroblasts assayed by ELISA. The data represent the percentage increase in active MT1-MMP levels relative to untreated controls and are the mean ± SEM of results of three independent experiments (each experiment run in triplicate). *Significant difference (P < 0.001) with respect to vehicle controls; **Significant difference (P < 0.05) relative to cPAF-treated cultures. (C) Shown are Western immunoblots of TIMP-2 protein in corneal myofibroblast cultures treated with cPAF and combined cPAF/BN 50730. The positions of the 24-kDa TIMP-2 protein bands were confirmed by comparison with recombinant human TIMP-2 (24-kDa) standard and biotin-labeled and prestained molecular weight protein markers run on the gel. The immunoblot is representative of in two separate experiments.

Figure 6.

 

PAF induced MMP-9 but not MMP-2 activation in corneal myofibroblasts. Gelatin zymograph of MMP-2 and -9 activity in conditioned medium collected from corneal myofibroblasts treated with 100 nM of the cPAF and PAF antagonist BN 50730 for 24 hours. The position of the 72-kDa MMP-2 enzyme, visible as a clear band, was confirmed by comparison to 72-kDa human granulocyte MMP-2 and prestained protein molecular weight markers. The position of the active MMP-9 enzyme is visible as an 84-kDa band and confirmed by comparison to a standard mixture of human pro-MMP-9 (92 kDa) and active MMP-9 (84 kDa) run on the gel. A phorbol ester tumor promoter activator (TPA) was used as a positive control for expression of MMP-9. The results are representative of two independent experiments.

Figure 7.

 

PAF-induced furin expression in corneal myofibroblasts. (A) Time-dependent effect of cPAF on furin expression as determined by PCR. Separate controls lacking reverse transcriptase (RT−) and template (RNA−) were run on each gel (not shown). Included on the gels were positive controls for furin expression. TGFβ1 was used as a positive control. The expression of 18s rRNA did not change under these conditions. The experiments were repeated twice with similar results. (B) Analysis of furin gene expression by real-time PCR. The data represent the increase (x-fold) in furin gene expression relative to untreated controls at 2, 4, and 8 hours and are the mean ± SEM of results in three independent experiments (each experiment run in triplicate). Furin expression correlates (R ² = 0.9997) positively with length of cPAF treatment (inset). Controls and cPAF-treated samples were supplemented with the vehicle (DMSO) for the PAF antagonist, to a final concentration not exceeding 0.01%. *Significant (P < 0.01) difference relative to vehicle controls. **Significant difference (P < 0.05) relative to the cPAF-treated group. (C) Immunoblot shows the expression of a 100-kDa protein corresponding to furin.

Figure 8.

 

PAF-induced MT1-MMP activity in corneal myofibroblasts is blocked by furin inhibitor nona-d-arg-NH₂. Myofibroblasts were preincubated with nona-d-arg-NH₂ (0.5 μM) for 30 minutes before addition of 100 nM cPAF and further incubation for 12 hours. Data represent the percentage increase in active MT1-MMP levels relative to untreated controls and are the mean ± SEM of three independent experiments (each experiment run in triplicate). *Significant (P < 0.03) difference relative to untreated controls. **Significant difference (P < 0.02) relative to the cPAF-treated group.