Proteinase-activated Receptor-1 Regulation of Macrophage Elastase (MMP-12) Secretion by Serine Proteinases*

The serine proteinases plasmin and thrombin convert proenzyme matrix metalloproteinases (MMPs) into catalytically active forms. In addition, we demonstrate that plasmin(ogen) and thrombin induce a significant increase in secretion of activated murine macrophage elastase (MMP-12) protein. Active serine protease is responsible for induction, as demonstrated by the absence of MMP-12 induction in plasminogen(Plg)-treated urokinase-type plasminogen activator-deficient macrophages. Since increased MMP-12 protein secretion was not accompanied by an increase in MMP-12 mRNA, we examined post-translational mechanisms. Protein synthesis was not required for early release of MMP-12 but was required for later secretion of activated enzyme. Immunofluorescent microscopy demonstrated basal expression in macrophages that increased following serine proteinase exposure. Inhibition of MMP-12 secretion by hirudin and pertussis toxin demonstrated a role for the thrombin G protein-coupled receptor (protease-activated receptor 1 (PAR-1)). PAR-1-activating peptides were able to induce MMP-12 release. Investigation of signal transduction pathways involved in this response demonstrate the requirement for protein kinase C, but not tyrosine kinase, activity. These data demonstrate that plasmin and thrombin regulate MMP-12 activity through distinct mechanisms: post-translational secretion of preformed MMP-12 protein, induction of protein secretion that is protein kinase C-mediated, and extracellular enzyme activation. Most importantly, we show that serine proteinase MMP-12 regulation in macrophages occurs via the protein kinase C-activating G protein-coupled receptor PAR-1.

Serine and matrix metalloproteinases (MMPs) 1 have important roles in hemostasis and remodeling of extracellular matrices during fibrinolysis and tissue repair. Abnormal regulation of these proteinases may cause tissue destruction. MMPs are typically secreted in zymogen form and require extracellular activation. The serine proteinases plasminogen, plasmin, urokinase-type plasminogen activator (uPA), and tissue-type plasminogen activator (tPA) have classically been shown to function in fibrinolysis (tPA/uPA/plasmin), cell motility and invasion (uPA/plasmin), and MMP activation. The MMPs are a family of structurally related zinc-containing enzymes that are either secreted or membrane-associated (membrane-type MMPs; MT-1-5 MMPs). As a group, the MMPs are capable of degrading all extracellular matrix components.
In addition to fibrinolysis and extracellular matrix degradation, MMPs and serine proteinases can cleave nonmatrix proteins with significant biological ramifications. For example, MMPs and related ADAMs (a disintegrin and metalloproteinase domain) can cleave and release a variety of active molecules from cell surfaces, such as tumor necrosis factor-␣ (1), and plasmin can activate transforming growth factor-␤ in plasma (2). In addition, both MMPs and serine proteinases generate angiostatin, an inhibitor of endothelial cell proliferation, from plasminogen (3)(4)(5)(6)(7)(8)(9).
Such redundant activity and interaction of the serine and metalloproteinase families have precedence in previous work. Plasmin cleaves many pro-MMPs, including the gelatinases MMP-2 and -9 (progelatinase A and B) (10) and MMP-12 (macrophage elastase) (11) within the N-terminal domain, altering conformation and exposing the active site zinc, which in turn releases the remainder of the proenzyme domain, resulting in a fully active MMP. At the cellular level, uPA-generated plasmin controls gelatinase activity in HT1080 cells (10). Carmeliet et al. (11) have recently shown that MMP activation in macrophages in culture is uPA/plasmin-dependent, and data suggests similar activation in a murine model of atherosclerotic microaneurysm formation. Another serine proteinase, thrombin, has also been reported to activate MMPs (progelatinase A, MMP-2) in microvascular endothelial cells (12) and to actually increase collagenase (MMP-1) protein and stromelysin (MMP-3) mRNA and protein in large vessel endothelial cells (human saphenous vein and mammary artery) (13).
Protease-activated receptor (PAR-1) is a unique cell surfaceassociated receptor activated by thrombin proteolysis that is characteristically coupled to G protein heterodimers (14) and is also activated by plasmin (15). It is a member of the seventransmembrane domain family that is activated following proteolytic cleavage of its N terminus (16). This newly formed N terminus acts as a tethered ligand to activate the receptor. Peptides homologous to this tethered ligand can also activate the receptor but are not as effective as thrombin that cleaves between residues Arg 41 and Ser 42 . Many cell types, including endothelial cells (17), macrophages (18), smooth muscle cells, and fibroblasts (19) express PAR-1.
While the activation of pro-MMPs to their active form by serine proteinases has been previously described, the purpose of this study was to explore other mechanisms of MMP regulation by the serine proteinases plasmin(ogen) and thrombin.

EXPERIMENTAL PROCEDURES
Reagents-Recombinant mouse macrophage elastase (MMP-12, MMP-12) was expressed and purified from Escherichia coli to homogeneity as described (20). The E. coli-derived enzyme is spontaneously active, since only its catalytic domain is expressed upon purification.
A polyclonal monospecific antibody to mouse macrophage elastase was generated in rabbits as described (21) and used at 1:1000 for Western analysis unless otherwise stated. For immunohistochemistry, the antibody was used at 1:150. A Cy3-labeled goat anti-rabbit IgG secondary antibody (Jackson Immunoresearch Laboratories, West Grove, PA) was used for immunofluorescence at 1:500.
Macrophage Culture Conditions-Peritoneal macrophages were harvested from mice deficient in MMP-12 by targeted mutagenesis (MMP-12 Ϫ/Ϫ) (129/Sv background) (21) and their wild-type littermates (MMP-12 ϩ/ϩ). Peritoneal macrophages were also harvested from mice deficient in plasminogen, urokinase plasminogen activator, and tissue-type plasminogen activator (kindly provided by Dr. Peter Carmeliet, University of Leuven, Leuven, Brussels). Peritoneal macrophages from MMP-12 ϩ/ϩ and MMP-12 Ϫ/Ϫ mice were obtained by peritoneal lavage. In some experiments, to increase macrophage yield, mice received a 1-ml intraperitoneal injection of 3% thioglycollate (22), and 5 days following injection, peritoneal macrophages were harvested and plated into 24-well tissue culture plates (Becton Dickinson, Franklin Lakes, NJ) at 1 ϫ 10 6 cells/well in 10% fetal calf serum-supplemented DMEM (Life Technologies, Inc.). After 24 h of culture at 37°C and 5% CO 2 , cells were placed in serum-free medium and treated with 20 g/ml of plasmin(ogen), 10 -100 units/ml of thrombin or buffer control in the presence or absence of the serine proteinase inhibitor aprotinin for 48 h. Cell-conditioned media were harvested and subjected to Western blot analysis and casein zymography.
Western Blot Analysis-Cell-conditioned media harvested from buffer control, plasmin(ogen)-treated (20 g/ml) or thrombin-treated (10 -100 units/ml) MMP-12 ϩ/ϩ and MMP-12 Ϫ/Ϫ macrophages were subjected to SDS-polyacrylamide gel electrophoresis following the addition of 25 M DTT in protein running buffer at 100°C for 5 min. Proteins were transferred to a Nylon membrane and blocked with 5% casein in Tris-buffered saline overnight at 4°C. The membrane was then incubated with rabbit anti-mouse IgG MMP-12-specific antibody (1:1000 dilution) (21,23) for 1 h at room temperature, followed by incubation with goat anti-rabbit IgG-horseradish peroxidase (Amersham Pharmacia Biotech) and autoradiography.
Casein Zymography-Nonreduced conditioned media from Plg-and plasmin-stimulated MMP-12-competent macrophages were subjected to an ␣-casein 10% SDS-polyacrylamide gel electrophoresis in ice-cold buffer. The gel was then placed in 2.5% Triton-X in PBS, washed, and incubated in fresh incubation buffer (0.05 M Tris, pH 8.2, 0.005 M CaCl 2 , 0.5 mM ZnCl 2 ) for 48 h. The gels were stained with 0.125% Coomassie Blue and then destained with a 5% acetic acid, 10% methanol solution to observe gel lysis.
Northern Hybridization-Total cellular RNA was isolated from cultured murine peritoneal macrophages by guanidinium isothiocyanatephenol extraction and ethanol precipitation (24) and quantified spectrophotometrically. Equivalent amounts of total cellular RNA (5 g) were subjected to gel electrophoresis, and the RNA was transferred to nylon membranes. Northern blots were probed with random primed, [ 32 P]dCTP-labeled cDNA fragments specific for MMP-12.
Nuclear Run-on Assay-Nuclear run-on transcription assays were performed on murine macrophages, following a protocol modified from a procedure previously detailed (25). Briefly, murine macrophages (3 ϫ 10 8 cells) were plated on four T75 tissue culture flasks in serum-free DMEM in the presence or absence of plasmin (20 g/ml) for 12 h. At 12 h, the cells were washed twice with cold PBS and harvested by trypsinization. Cells were resuspended in ice-cold hypotonic lysis buffer (10 mM Tris, pH 7.4, 3 mM MgCl 2 , 10 mM NaCl, 0.5% Nonidet P-40) and incubated for 15 min on ice. The cell suspension was then sheared three times using a 22-gauge needle. Trypan blue incorporation was used to assure nuclear isolation and the absence of nuclear lysis. The suspension was centrifuged at 2000 ϫ g for 10 min at 4°C, and the supernatant containing cytoplasm and membranes was removed. The nuclei were then resuspended in ice-cold transcription buffer (25 mM Tris, pH 8.0, 5 mM MgCl 2 , 120 mM KCl, 10 mM dithiothreitol, 10% glycerol) and centrifuged at 2000 ϫ g for 3 min at 4°C. The supernatant was removed, and nuclei were resuspended in ice-cold transcription buffer, rNTP mix, 200 Ci of [␣-32 P]GTP. This mixture was incubated 20 min at 30°C. Nuclei were then spun, supernatant was removed, and the pellet was resuspended in 4 M guanidine thiocyanate/sarkosyl, ␤-mercaptoethanol, 25 mM sodium citrate, 3 M sodium acetate, pH 4.0. Phenol extraction and ethanol precipitation was performed, and the RNA was incubated overnight at Ϫ20°C. The pellet was dried and resuspended in hybridization mixture (50% deionized formamide, 4ϫ SSC, 2ϫ Denhardt's solution, E. coli tRNA, 50 mM Na 2 PO 4 , 0.1% SDS, pH 7.5), placed on ice, and pulse-spun, and radioactivity was counted. 1 ϫ 10 7 cpm of labeled nascent transcripts from control and treated cells were placed in a fresh tube, heated at 80°C for 10 min, and then hybridized for 24 h to a previously prepared nitrocellulose membrane slot blotted with 3 g of MMP-12 cDNA, glyceraldehyde-3-phosphate dehydrogenase cDNA, and Bluescript (pBS) plasmid cDNA.
Immunofluorescence and Confocal Microscopy-Peritoneal murine macrophages (control and thioglycollate-stimulated) from MMP-12-deficient mice (MMP-12 Ϫ/Ϫ) and their wild type littermates (MMP-12 ϩ/ϩ), harvested as described, were plated onto two-chambered Lab Tech slides (Nalgen Nunc, Naperville, IL) at 5 ϫ 10 5 cells/well in 10% fetal calf serum DMEM for 2 h. The macrophages were stimulated with plasmin (20 g/ml) or thrombin (10 units/ml) in serum-free DMEM overnight and fixed in 4% paraformaldehyde for 10 min at room temperature. The cells were then washed in PBS four times and incubated with MMP-12 polyclonal antibody (1:150) in 2% fish gelatin (Sigma) in PBS or irrelevant rabbit serum (Vecta Labs, Burlingame, CA) overnight at 4°C. Cells were then washed in PBS four times, incubated with a Cy3-labeled secondary goat anti-rabbit IgG antibody (1:500 in 2% fish gelatin in PBS) for 45 min at room temperature, washed in PBS four times, and coverslipped. The slides were examined under a Nikon confocal microscope equipped with fluorescent filters under a ϫ 60 oil immersion lens and analyzed with Laser Sharp 2000 (Bio-Rad).

Plasmin(ogen) Induces MMP-12 Secretion and Activation-
Murine peritoneal macrophages were harvested from MMP-12competent mice and incubated with Plg (20 g/ml) (Sigma) in the presence or absence of aprotinin (Sigma) for 48 h. Western analysis of the macrophage-conditioned media for MMP-12 demonstrated small amounts of pro-MMP-12 at 54 kDa in control macrophages and the generation of active MMP-12 (29 kDa) in the Plg-stimulated MMP-12 ϩ/ϩ macrophages (Fig.  1A). Active MMP-12 was not, however, secreted by the Plgstimulated MMP-12 ϩ/ϩ macrophages in the presence of aprotinin, demonstrating dependence upon serine proteinase activity.
Peritoneal murine macrophages were stimulated with plasmin (Sigma) (20 g/ml), 48-h conditioned medium was collected, and Western analysis for MMP-12 was performed. As shown in Fig. 1B, treatment with either Plg or plasmin resulted in increased MMP-12 release and activation to a 29-kDa form. Densitometry of autoradiographs (n ϭ 4) demonstrated a 2.5-3.6-fold increase in secreted protein over control levels (p Ͻ 0.05 for plasmin) (Fig. 1C). Casein zymography confirmed the 29-kDa band seen on Western in addition to the typical 22-kDa fully processed form (Fig. 1D). We presume that the 29-kDa form represents an active enzyme with an N-terminal AA sequence at the beginning of the catalytic domain extending through part of the C-terminal domain as described previously (23). Conversion on zymography probably is a function of continual processing of the C-terminal domain while running through the gel. Control macrophage-conditioned media produced a minimal lysis at 22 kDa, and Plg-treated MMP-12 Ϫ/Ϫ macrophages did not produce lytic bands at any molecular weight (data not shown).
To further investigate the role of plasmin(ogen) in MMP-12 production, thioglycollate-elicited peritoneal macrophages were harvested from serine proteinase uPA-deficient mice (uPA Ϫ/Ϫ) and mice deficient in Plg (Plg Ϫ/Ϫ) and tissue-type plasminogen activator (tPA Ϫ/Ϫ). The macrophages were plated on tissue culture plastic and incubated in the presence of Plg (20 g/ml) or plasmin (20 g/ml), and Western blot analysis for MMP-12 was performed (Fig. 2). Increased amounts of activated MMP-12 (29 kDa, arrow) were produced following both Plg and plasmin stimulation in wild-type, Plg Ϫ/Ϫ, and tPA Ϫ/Ϫ macrophages. Interestingly, increased MMP-12 secretion and activation was observed in the uPA Ϫ/Ϫ macrophages in response to plasmin but not Plg. This suggests that the active serine proteinase plasmin, which is generated from plas-minogen via uPA, is required for the secretion and activation of MMP-12 in murine macrophages. Although equal numbers of macrophages were plated from wild-type and each serine proteinase-deficient animal, Plg Ϫ/Ϫ macrophages produced lesser amounts of constitutive pro-MMP-12 and less activated MMP-12 following Plg and plasmin stimulation. This was seen FIG. 2. Serine proteinase induction and activation of MMP-12 is due to plasmin. Peritoneal macrophages from wild-type (MMP-12ϩ/ϩ), Plg-deficient (Plg Ϫ/Ϫ) mice, and mice deficient in the plasminogen activators urokinase-type plasminogen activator (uPA Ϫ/Ϫ) and tissue-type plasminogen activator (tPA Ϫ/Ϫ) were plated on tissue culture plastic and stimulated with Plg (20 g/ml) and plasmin (20 g/ml). Conditioned medium was collected, and Western blot was performed with antibody specific for MMP-12. Activated MMP-12 (29 kDa, arrow) was produced following both Plg and plasmin stimulation in wild-type, Plg Ϫ/Ϫ and tPA Ϫ/Ϫ macrophages. In contrast, plasmin but not Plg stimulation induced activated MMP-12 in the uPA Ϫ/Ϫ macrophages.

FIG. 1. Plasmin(ogen) induction of MMP-12.
A, plasmin(ogen) induction of MMP-12 requires serine proteinase activity. Peritoneal macrophages from MMP-12 ϩ/ϩ mice cultured with Plg (20 g/ml) in the presence and absence of aprotinin (Ap, 50 kallikrein-inactivating units/ml, 100 kallikrein-inactivating units/ml). Western blot analysis was performed on the conditioned media for MMP-12 (activated form, 29 kDa, arrow). Aprotinin blocked the generation of activated MMP-12. B, plasmin induces MMP-12 secretion and activation. Peritoneal macrophages from MMP-12 ϩ/ϩ mice were cultured alone or in the presence of Plg (20 g/ml) or plasmin (20 g/ml), and Western blot analysis for MMP-12 was performed. Control macrophages produce small amounts of pro-MMP-12 (54 kDa, arrowhead). Both plasmin-and Plg-stimulated macrophages produced increased amounts of active MMP-12 (29 kDa, arrowhead). C, quantification of MMP-12 secretion. Densitometry was performed on autoradiographs of Western blot analysis for MMP-12 on plasmin(ogen)-stimulated macrophage-conditioned media. Western blot analysis was performed as described for A in four separate experiments. Densitometry and statistical analysis were performed. Error bars, S.E. *, p Ͻ 0.05. D, casein zymography. Conditioned media from control macrophages and macrophages stimulated with Plg (20 g/ml) were subjected to casein zymography. Due to sensitivity of zymography, control macrophages demonstrate small amounts of MMP-12 at 22 kDa, as seen by faint gel lysis. Plg-stimulated macrophages produce increased gel lysis at both 22 and 29 kDa.
in repeat experiments.
We then explored plasmin-mediated secretion of MMP-12 at varying time points (6,12,24,48, and 72 h). As demonstrated in Fig. 3, MMP-12 protein secretion increases over time, without a coincident increase in pro-MMP-12. This further demonstrates a true increase in secretion and activation of MMP-12, as opposed to sole activation of proenzyme.
Plasminogen and Plasmin Do Not Induce Gene Transcription of MMP-12 mRNA in Murine Macrophages-Although the activation of MMP-12 by plasmin(ogen) has previously been described (11), we demonstrated an increase in total MMP-12 protein by Western analysis in the Plg-and plasmin-stimulated macrophages compared with control. Therefore, we investigated MMP-12 mRNA levels by Northern analysis following stimulation with Plg and plasmin. There was no change in MMP-12 mRNA levels at 12 h (data not shown) or at 24 h in the presence of Plg or plasmin (Fig. 4A). Similarly, nuclear run-on assays did not demonstrate Plg or plasmin induction of MMP-12 transcription (Fig. 4B).
Active Protein Synthesis Is Not Required for Early Plasminogen-induced Secretion of MMP-12 Protein-To determine if plasmin promotes an active secretion of basally expressed MMP-12 protein, macrophages were pretreated with either monensin (1 M) or cycloheximide (10 g/ml) 1 h prior to plasmin or plasminogen treatment. As demonstrated by Western analysis of conditioned media, monensin pretreatment blocked secretion of pro-MMP-12 (54 kDa) in control macrophages and both pro-and activated (29 kDa) MMP-12 in Plg-and plasmintreated cells (Fig. 5A), demonstrating active secretion of protein. Pretreatment with cycloheximide limited the amount of activated MMP-12 secreted in Plg-treated macrophages, and this amount remained constant (Fig. 5B), indicating that active protein synthesis is required for the robust induction of secretion seen at later time points (24 and 48 h; Figs. 1 and 3). The slower initial induction of secretion in plasminogen-treated (12-18 h; Fig. 5B) compared with plasmin-treated (6 -12 h; Fig.  3) macrophages may be reflective of the time required for the macrophage-mediated generation of effective amounts of plasmin from the parent protein plasminogen. Compared with control, monensin-and cycloheximide-treated cells showed no evidence of cell damage as examined by inverted microscopy (data not shown).
Macrophages Basally Express Cytoplasmic MMP-12 Protein That Increases following Plasmin and Thrombin Treatment-Control and plasmin-stimulated thioglycollate-elicited peritoneal macrophages and their cell-conditioned media that were examined at 6, 12, 24, 48, and 72 h for MMP-12 secretion by Western analysis (Fig. 3) were also examined by confocal immunofluorescent microscopy for MMP-12 protein expression. Non-thioglycollate-elicited peritoneal macrophages were exam-ined at 24 h following plasmin stimulation. Western analysis of conditioned media from plasmin-treated thioglycollate-elicited peritoneal macrophages demonstrated a progressive increase in the active form of MMP-12 (29 kDa) as demonstrated in Fig.  3. Corresponding immunofluorescent confocal microscopy of control macrophages at 24 h (Fig. 6, upper left panel) demonstrated a positive signal for MMP-12 in a cytoplasmic pattern (red signal demonstrates positive staining for MMP-12 protein; Peritoneal macrophages from MMP-12 ϩ/ϩ and MMP-12 Ϫ/Ϫ mice were cultured in the absence or presence of plasmin (20 g/ml). Conditioned medium was collected at varying times, and Western analysis was performed using an antibody specific for MMP-12.
FIG. 4. Plasminogen and plasmin do not induce gene transcription of MMP-12 mRNA in murine macrophages. A, peritoneal macrophages from MMP-12 ϩ/ϩ mice were cultured in the absence or presence of Plg (20 g/ml) and plasmin (20 g/ml); total RNA was extracted at 24 h and subjected to Northern blot analysis. Blots were probed with 32 P-labeled MMP-12 and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) cDNA and subjected to autoradiography. No induction of steady state MMP-12 mRNA was seen at 24 h. The lower panel demonstrates ethidium bromide-stained 28 and 18 S RNA bands of Northern gel. B, nuclear run-on transcription assays were performed on control and plasmin-treated macrophage nuclei as described under "Experimental Procedures." ␣-32 P-GTP-labeled nascent transcripts from control and plasmin-treated cells were hybridized to a nitrocellulose membrane slot-blotted with 3 g of MMP-12 cDNA, glyceraldehyde-3-phosphate dehydrogenase cDNA, and Bluescript (pBS) plasmid cDNA.

FIG. 5. Plasmin(ogen) induces secretion of MMP-12, and active protein synthesis is not required for the early release of protein.
A, murine peritoneal macrophages were pretreated with monensin (1 M) 1 h before treatment with Plg or plasmin (20 g/ml) for 48 h, and conditioned media were collected for Western analysis using antibodies specific for MMP-12. An increase in activated MMP-12 protein (29 kDa) was seen in the both the Plg-and plasmin-treated macrophage-conditioned medium that was inhibited by monensin. B, murine peritoneal macrophages were pretreated with cycloheximide (CHX) (10 g/ml) 1 h before treatment with plasminogen (20 g/ml), and Western analysis was performed for MMP-12. An increase in activated MMP-12 protein (29 kDa) was seen in the plasminogen-treated macrophage-conditioned media in the presence of cycloheximide at 18 h that remained constant at 24 and 48 h. laser reflectance of cell morphology depicted in green). Following plasmin treatment, the MMP-12 signal increased in intensity as early as 12 h following treatment and was intense by 24 h (Fig. 6, lower left panel). Additionally, in the plasmintreated cells, positive staining for MMP-12 accumulated within the extracellular matrix at 24 h (Fig. 6, asterisk, lower left panel). This was not seen in untreated cells or in cell-free plasmin-or plasminogen-containing tissue culture wells that were similarly examined (data not shown). This demonstration of extracellular MMP-12 is suggestive of macrophage "footprinting," where membrane and cytosol fragments, possibly containing enzyme, are left behind following cell migration. In the continued presence of plasmin, increased MMP-12 protein expression was demonstrated at both 48 h (Fig. 6, lower right panel) and 72 h when compared with control, untreated cells (Fig. 6, lower right panel, inset). To determine that MMP-12 protein was truly constitutively expressed, we examined peritoneal macrophages from non-thioglycollate-treated mice. In fact, staining for MMP-12 protein was demonstrated in a cytoplasmic pattern in control cells (data not shown).
Similarly examined serine proteinase thrombin-stimulated macrophages at 24 h (Fig. 7, shown without green signal laser reflectance) demonstrated a significantly increased signal for MMP-12 in the cytoplasm (arrow, right panel) compared with control, constitutive MMP-12 expression (arrow, left panel). As in plasmin-treated cells at 24 h, extracellular expression of MMP-12 was again seen following thrombin treatment (asterisk, right panel).
The Serine Proteinase Thrombin Activates and Induces Secretion of MMP-12 Protein and Involves the Thrombin Receptor, PAR-1-We therefore further explored the serine proteinase thrombin, due to its well characterized signal transduction mechanism and its effect on of MMP-12 secretion. Western analysis of conditioned media from thrombin-stimulated murine peritoneal macrophages shows that thrombin also induces MMP-12 secretion and activation (Fig. 8A), consistent with our findings by confocal microscopy. Thrombin and plasmin are capable of activating the thrombin G protein-coupled receptor, proteinase activator receptor-1 (PAR-1). To determine if induction of MMP-12 was occurring through thrombin receptor activation, we attempted to block thrombin activity by preincubating thrombin with hirudin (Sigma) prior to exposure to macrophages. Hirudin, a leech-derived Thr-specific inhibitor, neutralizes the activity of thrombin by binding to its anion site, blocking receptor activation. On Western blot analysis, hirudin inhibited the thrombin-induced secretion of activated MMP-12 protein (Fig. 8A). In addition, the PAR-1-binding but proteolytically inactive Thr-D-phenylalanyl-L-prolyl-L-arginine chloromethylketone 2 HCl complex, totally inhibited Thr induction (data not shown) and suggests that the serine proteinase proteolytic activity of Thr is important for inducing MMP-12.
To confirm that Thr was acting through activation of PAR-1 to induce MMP-12 protein, PAR-1 (thrombin receptor)-specific peptides were generated (TRAPs; TFLLR-NH 2 and TFRIFD-NH 2 ) (26, 27) as was an irrelevant peptide (FLTRL-NH 2 ). PAR-1-activating peptide sequences are identical to the protein sequence C-terminal to the receptor cleavage site and duplicate the action of thrombin. LDPR 41 S 42 FLLRN, the NH 2 terminus of PAR-1, includes the thrombin cleavage site (14). The human TRAP SFLLRNP-NH 2 is a more potent activator of PAR-1, but has also been found to activate PAR-2 (27) and therefore was not used in these experiments. Lower potency TRAPs having specificity for PAR-1 in platelets and in human embryonic kidney cells include TFLLR-NH 2 and TFRIFD-NH 2 ; TFLLR-NH 2 is the more potent activator of PAR-1. Although these TRAPs are also less potent activators of PAR-1 than Thr itself and differences in concentration have been reported to be required to elicit similar responses (28), the activating peptides were capable of inducing active and intermediate forms of MMP-12 protein (Fig. 8A). The irrelevant peptide had no effect.
PAR-1-mediated Induction of MMP-12 Protein Is G Proteinmediated-To confirm that thrombin induction of MMP-12 secretion via PAR-1 was G protein-coupled, thrombin effect on MMP-12 secretion was determined following pretreatment with pertussis toxin (PTX). PTX prevents signaling through the G i -derived G ␤␥ subunit of GPCRs. PTX entirely inhibited thrombin-induced MMP-12 secretion (Fig. 8B), suggesting signal transduction through a GPCR.
Serine Proteinase-stimulated MMP-12 Secretion Is Nega-tively Coupled to Adenylyl Cyclase and Is PKC-and MAPK-dependent-Plasmin(ogen) stimulation was performed in the presence of pharmacologic agents that alter signal transduction mediators, specifically tyrosine kinase (genestein, herbimycin, orthovanadate) and cyclooxygenase (indomethacin). Secretion of active MMP-12 was not inhibited in the presence of genestein, herbimycin, or orthovanadate, indicating no dependence of signal transduction upon tyrosine kinase(s) (Fig. 9A). Thrombin-induced secretion of active and intermediate forms of MMP-12 was also demonstrated in the presence of genestein but was inhibited by forskolin, an activator of adenylyl cyclase, the PKC inhibitor calphostin C, and the MAPK/ extracellular signal-regulated kinase 1 inhibitor PD 98059 (Fig. 9B). These findings demonstrate that the thrombin response is negatively coupled to cAMP as previously demonstrated in adrenal medullary microvascular endothelial cells (27). This also demonstrates PKC dependence of the serine proteinase-mediated MMP-12 effect. Failure of the PKC inhibitor H7 to block this response may reflect selective effects on different PKC isoforms by these inhibitors. The effect of inhibition of these signal transduction mediators on serine proteinase-induced MMP-12 secretion is summarized in Table I. DISCUSSION Macrophage elastase is expressly produced by macrophages and is the primary metalloproteinase produced by murine macrophages (21). Macrophages also express the serine proteinases uPA and the urokinase receptor, uPAR, and are capable of generating plasmin from plasminogen. It has previously been demonstrated that plasmin is capable of extracellularly activating pro-MMP-12 (11) and generates angiostatin from plasminogen in tumor (7) and HT1080 cells (9). We have recently reported that macrophage-mediated generation of the  9. A, plasminogen-stimulated MMP-12 secretion is not blocked by inhibition of tyrosine kinase or cyclooxygenase activities. Murine peritoneal macrophages were harvested and cultured as described previously. Cells were then placed in serum-free media, pretreated with genestein (50 M), herbimycin (500 nM), orthovanadate (100 nM), or indomethacin (6 g/ml), for 1 h, and then stimulated with Plg (20 g/ml) or plasmin (20 g/ml), and supernatants were harvested for Western analysis for MMP-12. Total protein synthesis was not significantly affected. B, Thr-stimulated MMP-12 secretion is negatively coupled to adenylyl cyclase and is PKC-and MAPK-dependent. Murine peritoneal macrophages were harvested and cultured as described previously. Cells were then placed in serum-free media; pretreated with genestein (50 M), forskolin (100 M), PD 98050 (50 M), calphostin C (2 M), or H7 (100 M) for 1 h; and then stimulated with thrombin (10 units/ml), and supernatants were harvested for Western analysis for MMP -12. angiogenesis inhibitor angiostatin from plasminogen is dependent upon the metalloproteinase MMP-12 (3). In the present study, we demonstrate that the serine proteinases plasmin-(ogen) and thrombin induce release of preformed MMP-12 stores and increase MMP-12 protein secretion and conversion of pro-MMP-12 to its active form. In our system, activation of pro-MMP-12 (54 kDa) results in a 29-kDa form, and detectable amounts of lower molecular weight forms by Western analysis, as previously reported (22 kDa) (23), are not seen. This may be reflective of the serine proteinase activity on secreted enzyme at the cell surface. Our investigations into the mechanism of this secretion demonstrate that it does not require tyrosine kinase activity and further show G protein-mediated protein kinase C/MAPK regulation via the thrombin receptor, PAR-1.
Interestingly, we found decreased amount(s) of control and active MMP-12 in the plasminogen-deficient (Plg Ϫ/Ϫ) macrophages following plasminogen and plasmin stimulation (Fig.  2). Although the mechanism of this is unclear, we postulate that in plasminogen-competent animals, macrophages may be "primed" by the presence of plasmin(ogen) to respond to plasmin(ogen) stimulation. Supporting this hypothesis, we demonstrate that MMP-12 secretion involves PKC and MEK signaling mechanisms that may not be activated in the absence of plasmin(ogen), and that inhibition of these signaling pathways inhibits, and in some cases totally blocks, not only activated but also constitutive MMP-12 secretion (Fig. 9).
In fact, the effect of decreased levels of plasmin on MMP activity in vivo has significance for recent findings in certain animal disease models. We have demonstrated that macrophages from uPA-deficient animals cannot generate plasmin in vitro (Fig. 2). In vivo, however, these animals are capable of generating plasmin (although possibly in decreased amounts) via endothelial tPA. Consistent with this, previous studies have demonstrated that plasmin-dependent activation of pro-MMPs in uPA-deficient animals does occur, but in reduced amounts (11). Interestingly, it has been shown that lower amounts of MMP-12 are expressed following vascular injury in these mice compared with wild type, and this finding is associated with lower macrophage numbers in the neointima (29). In that MMP-12 is required for macrophage matrix invasion in mice (21), our data suggest that decreased amounts, together with decreased activation, of secreted MMP-12 in uPA-deficient animals contribute to the impaired macrophage neointimal infiltration.
Increases in MMP-12 gene transcription or post-transcriptional mRNA stability did not account for the increase in MMP-12 protein. Examination of control and plasmin-treated murine macrophages by confocal immunofluorescent microscopy demonstrates a modest level of basal cytoplasmic expression in control macrophages that is not dependent upon thioglycollate elicitation. Within 24 h of treatment, plasmintreated macrophages increase their expression of cytoplasmic MMP-12 that is then secreted, with some evidence of extracellular accumulation. Consistent with secretion of basal protein, initial increases in secreted MMP-12 remained stable in the presence of cycloheximide, as determined by Western blot analysis. Later (Ͼ24 h) increases in MMP-12, however, are dependent upon active protein synthesis, as demonstrated by the absence of increased MMP-12 over time in the presence of cycloheximide. These data also demonstrate that plasmin(ogen), at concentrations that would be consistent with those at the macrophage cell surface (approximately 10-fold increase over plasma concentrations), is capable of regulating MMP-12 activity predominantly via post-translational secretion of preformed cytosolic MMP-12 protein and increase in protein secretion and extracellular enzyme activation.
To investigate the mechanism underlying this serine proteinase regulation of MMP-12, we then examined the effect of thrombin on MMP-12 secretion, due to its well characterized G protein-mediated receptor activation through PAR-1. In fact, thrombin increased the protein expression and secretion of MMP-12 from macrophages. Additionally, we recognized that plasmin has been reported to similarly cleave and consequently activate this receptor (15). Through the use of specific inhibitors of signal transduction molecules, we show that plasmin-(ogen)-and thrombin-induced MMP-12 secretion of active enzyme in macrophages is not tyrosine kinase-dependent but is dependent upon other protein kinases (protein kinase C family members, MAPK). It is indeed interesting that LPS immunomodulation of macrophage function involves similar signal transduction pathway(s) through its GPCR receptor CD14 (30). More specifically, LPS-induced tumor necrosis factor-␣ and IL-1 secretion in macrophages has been similarly reported to be PKC-dependent (31).
The control of MMP expression through several distinct regulatory mechanisms and signaling pathways has been demonstrated previously. In fact, LPS regulation of gene expression of another MMP, MMP-9, in a human monocyte-like cell line (U937) and in alveolar macrophages has been found to involve post-transcriptional mechanisms (32). Similarly, signal transduction pathways have been investigated in collagen-induced MMP-1 (collagenase) production in keratinocytes and have shown a PKC-and tyrosine kinase-dependent mechanism distinct from that of PKC-induced PMA up-regulation (33). Finally, during macrophage differentiation, PKC-␤ has been found to have a signaling role in the fibronectin matrix-mediated cell adhesion and gene expression of MMP-9 (34). Although these investigations primarily involve regulation of MMP expression via gene induction, our findings provide evidence that similar signaling mechanisms may ultimately be involved in serine proteinase-mediated mechanisms regulating MMP expression.
We propose that serine proteinase-dependent regulation of MMP-12 may have a role in biologic processes associated with inflammation, such as vascular injury, angiogenesis, tumor growth, and metastasis. Consistent with this hypothesis is the lowered concentration of MMP-12 seen in vascular injury in uPA Ϫ/Ϫ mice (29). In tumor biology, tumor cell proteinase expression has been characteristically associated with increased tumor invasion and metastasis (35). Tumor stromal and inflammatory cell proteinase expression, however, does not always contribute to tumor progression but may function in host repair and defense. Macrophages are often recruited to tumor sites (tumor-associated macrophages) where plasmin-(ogen) and thrombin are replete, due to both local hemostasis and tumor-and macrophage-derived factors (vascular endothe- Protein kinase A, C, G Ϫ lial growth factor and tumor necrosis factor-␣) that contribute to vessel wall "leakiness" (36). In this context, the serine proteinase regulation of MMP-12 in macrophages may affect tumor growth and metastasis through the MMP-12-mediated generation of the angiogenesis inhibitor, angiostatin, from its parent serine proteinase protein plasminogen. In sum, we show that the serine proteinases plasmin and thrombin direct the release of MMP-12 from the macrophage and activate the enzyme extracellularly. We hypothesize that control of MMP-12 expression through the G protein-coupled receptor PAR-1 provides a distinct and focused regulation of this MMP.