Regulation of Matrix Metalloproteinases and Plasminogen Activator Inhibitor-1 Synthesis by Plasminogen in Cultured Human Vascular Smooth Muscle Cells
Abstract Plasmin and matrix metalloproteinases (MMPs) both participate in extracellular matrix remodeling. This study examined the effects of tumor necrosis factor-α (TNF-α) and plasminogen on collagenase, stromelysin, and plasminogen activator inhibitor-1 (PAI-1) synthesis by cultured human vascular smooth muscle cells (SMCs). TNF-α induced the concentration-dependent synthesis of collagenase and stromelysin, which remained predominantly in proenzyme forms, as determined by Western analysis of culture media. In contrast, plasminogen and plasmin not only increased secretion of MMPs but also induced cleavage to their active forms. The serine protease inhibitor aprotinin inhibited this activation of MMPs by plasminogen and plasmin. TNF-α reduced plasminogen-induced activation of MMPs, suggesting induction of an inhibitor of plasmin generation, such as PAI-1. Enzyme-linked immunosorbent assay of culture media showed that TNF-α (10 ng/mL) increased PAI-1 secretion by 4.2-fold compared with control (105.5±9.6 versus 24.9±1.7 ng/mL, n=3). Surprisingly, plasminogen also increased PAI-1 secretion by vascular SMCs (3.6-fold over control). These results demonstrate coordination of cytokines and serine proteases in regulating MMP secretion and activation. In addition, the induction of PAI-1 by TNF-α and plasminogen suggests a negative-feedback mechanism to limit both plasmin-mediated and MMP-mediated matrix degradation.
Vascular remodeling occurs in response to hemodynamic stimuli, biochemical mediators, or injury.1 Remodeling may involve changes in the number and distribution of cells as well as changes in the quantity and composition of extracellular matrix. The modification of extracellular matrix depends on both synthesis and degradation of matrix molecules. In vascular tissue, SMCs synthesize collagen and proteoglycans, the principal components of vascular matrix.2 3 Interstitial collagenase and stromelysin are MMPs, which degrade matrix components, and SMCs can produce these proteases.4 5 Macrophages, which can accumulate in diseased vascular tissue, may also be an important source of MMPs.6
The MMPs are secreted in zymogen form and require extracellular activation. The serine protease plasmin can activate latent collagenase and stromelysin in vitro and may perform this role in vivo in some tissues.7 8 Plasmin, in turn, must be generated from plasminogen by plasminogen activators, and plasmin may directly degrade several components of extracellular matrix. Vascular SMCs can synthesize plasminogen activators as well as plasminogen activator inhibitors.9 10 Plasmin may also regulate remodeling beyond its role in activating latent MMPs; Werb and Aggeler11 reported that cultured fibroblasts treated with plasmin increase collagenase secretion. Therefore, regulation of vascular extracellular matrix degradation can occur at multiple levels involving the secretion and activation of MMPs. Furthermore, TIMPs may inhibit the activity of the activated enzymes.
Several cytokines potently modulate extracellular matrix synthesis and degradation. For example, interleukin-1, platelet-derived growth factor, and transforming growth factor-β increase the synthesis of collagen and proteoglycans by cultured vascular SMCs.12 13 14 15 16 Interleukin-1 and TNF-α also induce interstitial collagenase and stromelysin synthesis by vascular SMCs.17 Since these cytokines are present in the blood vessel wall in certain pathological conditions, these observations may have in vivo relevance.
The present study investigated the role of plasminogen, with and without TNF-α, on the synthesis and activation of MMPs by cultured human vascular SMCs. We report that both plasminogen and TNF-α induce the synthesis of collagenase and stromelysin and that plasminogen also induces cleavage of the latent MMPs to the active forms. In addition, both plasminogen and TNF-α induce PAI-1, providing a negative-feedback mechanism for the matrix degradation response. These results demonstrate coordination of cytokines and serine proteases in regulating matrix remodeling by MMPs.
Materials and Methods
DMEM, Ham’s F-12 medium, and Limulus amebocyte lysate test kit were obtained from BioWhittaker; fetal calf serum, from Hyclone Laboratories; tissue culture dishes, from Costar; ascorbate-2-phosphate, from Wako Pure Chemical Industries; nonessential amino acids, human Lys-plasminogen, kringle 1-3 and kringle 4 from human plasminogen, aprotinin, and endotoxin from Escherichia coli 0111:B4, from Sigma Chemical Co; insulin and transferrin, from Collaborative Research; TNF-α, from Genzyme; plasmin and goat anti-human plasminogen IgG, from Enzyme Research Laboratories; Centricon-10 microconcentrators, from Amicon; enhanced chemiluminescence detection system, from Amersham; goat anti-rabbit IgG antibody with horseradish peroxidase, from Bio-Rad Laboratories; and ELISA kits, from Biopool AB. Rabbit anti-human antibodies to human collagenase and stromelysin were prepared as previously described.18
Medial-layer explants were cultured from unused portions of human saphenous veins from coronary bypass surgery. The culture medium was DMEM with 10% fetal calf serum, 25 mmol/L HEPES, 100 U/mL penicillin, 100 μg/mL streptomycin, 1.25 μg/mL amphotericin B, and 2 mmol/L l-glutamine. This medium is selective for SMCs over endothelial cells.19 Cells in some experiments were stained for α-actin. Unlike cultured rat vascular SMCs, cultured human vascular SMCs do not uniformly stain for α-actin. Approximately 50% of cells in the present study stained positively; this level is similar to other human vascular SMC preparations in our laboratories. Human dermal fibroblasts used as a control did not stain for α-actin (<5%).
SMCs from the explants were grown in tissue culture flasks. At confluence, the cells were split 1:3 by using 0.25% trypsin and EDTA. The cells were cultured in monolayers for two to six passages before use in experiments.
Treatment With Plasminogen and TNF-α
Cells were plated in 22- or 35-mm culture wells at 1 to 2.5×104 cells per square centimeter and 0.26 mL/cm2 of medium. The medium was the same as that described above and was further supplemented with 0.07 mmol/L ascorbate-2-phosphate, 0.1 mmol/L nonessential amino acids, and 0.75 mmol/L sodium sulfate. After 2 days, the medium was removed and replaced with 0.13 mL/cm2 of fresh medium. After 2 more days, serum was removed from the wells by washing four times with 0.26 mL/cm2 of IT medium (equal volumes of DMEM and Ham’s F-12, with 12.5 mmol/L HEPES, 100 U/mL penicillin, 100 μg/mL streptomycin, 1.25 μg/mL amphotericin B, 1.5 mmol/L l-glutamine, 0.07 mmol/L ascorbate-2-phosphate, 0.1 mmol/L nonessential amino acids, 0.75 mmol/L sodium sulfate, 1 μmol/L insulin, and 5 μg/mL transferrin). After the last wash, 0.13 mL/cm2 of IT medium was added to each well. The IT medium was changed at 24 hours. At 48 hours, the IT medium was changed again, and human plasminogen and/or recombinant human TNF-α was added to the wells. Cells were also treated with human plasminogen kringles, human plasmin, or aprotinin in some experiments. The media were collected after 24 hours of treatment. Some wells were counted for total and viable cells by using trypan blue exclusion after collagenase and trypsin digestion of the cell layer; none of the treatments significantly changed cell numbers or cell viability. Each treatment was performed at least twice in independent experiments.
Because bacterial endotoxin can induce a variety of products, including MMPs and PAI-1,20 21 22 unused media containing 100 μg/mL of plasminogen or plasmin were assayed for endotoxin by using the chromogenic Limulus amebocyte lysate test.23 The plasminogen sample contained 8 pg/mL, and the plasmin sample contained 56 pg/mL endotoxin. Unused medium containing 2 trypsin inhibitor units per milliliter of aprotinin was similarly assayed and contained 150 pg/mL endotoxin. TNF-α is tested by the manufacturer for endotoxin level by gel clot test, and a concentration of 10 ng/mL TNF-α contains 0.02 pg/mL endotoxin. To test for possible effects due to the contaminating endotoxin, some cells were treated for 24 hours with up to 10 ng/mL of endotoxin from E coli; no significant effect on PAI-1 or MMP synthesis was seen at the concentrations of endotoxin in these studies.
Electrophoresis and Immunoblotting (Western Analysis)
Because the levels of response varied with the cell source in these primary cultures of SMCs, media samples from a TNF-α concentration-response experiment were concentrated up to 20-fold in Centricon-10 centrifugal microconcentrators. All other samples were analyzed without concentration. Samples underwent electrophoresis on SDS–10% polyacrylamide gels under reducing conditions (SDS-PAGE). The proteins were transferred to nitrocellulose membranes, and an enhanced chemiluminescence detection system was used to detect collagenase or stromelysin. A 5% solution of nonfat dried milk in PBS containing 0.1% Tween 20 was used to block nonspecific binding and to dilute the primary and secondary antibodies. The primary antibodies were rabbit anti-human collagenase IgG (0.97 μg/μL, diluted 1:800 or 1:600) and rabbit anti-human stromelysin antiserum (diluted 1:400). The secondary antibody was goat anti-rabbit IgG conjugated with horseradish peroxidase.
ELISA for PAI-1
Media samples were assayed for PAI-1 as previously described.24 In brief, samples were incubated in microtiter plates coated with monoclonal antibodies against PAI-1, unbound antigen was washed off, and bound antigen was detected by addition of a second specific antibody conjugated to horseradish peroxidase. Standard curves were constructed from dilutions of purified PAI-1.
Effect of Plasminogen and Plasmin on MMP Secretion and Activation
Western blot analysis demonstrated that plasminogen induced secretion of both collagenase and stromelysin in a concentration-dependent manner in vascular SMCs (Fig 1⇓). In the absence of plasminogen, SMCs secreted little collagenase in its latent glycosylated and unglycosylated forms, although the lower molecular weight activated forms of the enzymes were not detectable (Fig 1A⇓). Total collagenase secretion increased with plasminogen concentration, although the absolute amounts varied among experiments using SMCs from different sources. These differences are explained by the variability in the magnitude of response to stimuli in primary cultures of SMCs. Plasminogen levels of 50 or 100 μg/mL, depending on the particular experiment, were sufficient to cause activation of the collagenase to the smaller forms. The response of stromelysin secretion to plasminogen was similar (Fig 1B⇓). Total stromelysin levels increased with plasminogen concentration. The unglycosylated and glycosylated activated forms appeared with 50 or 100 μg/mL plasminogen, but only the latent forms appeared at lower concentrations. Some cross-reactivity of the stromelysin antiserum with plasmin is evident near 60 kD.
To explore further the induction of MMPs, vascular SMCs were cultured with plasmin (Fig 2⇓). As in the experiments with plasminogen, Western blot analysis of the culture media showed that both collagenase and stromelysin were induced in a concentration-dependent manner by plasmin; furthermore, activation of the proenzymes to the active forms was also concentration dependent. Some cross-reactivity of the collagenase antiserum with plasmin is evident near 60 kD.
Treatment of the SMCs with plasminogen lysine-binding-site fragments was performed to examine the possibility of a direct effect of plasminogen on MMP secretion. Kringle 4 did not affect MMP secretion, even at 10 times the molar concentration of plasminogen that increased MMP production (data not shown). Kringle 1-3 cross-reacted with the anti-collagenase and anti-stromelysin antisera and comigrated as a smear in the same molecular weight range as the activated MMPs.
Effect of Aprotinin on Induction of MMPs by Plasminogen and Plasmin
The activation of collagenase and stromelysin by 100 μg/mL plasminogen or plasmin was completely blocked by treatment with 2 trypsin inhibitor units per milliliter of the serine protease inhibitor aprotinin (Fig 3⇓). Unexpectedly, aprotinin itself caused the induction of latent MMPs in the absence of cytokines, plasminogen, or plasmin. Plasminogen or plasmin combined with aprotinin did not increase collagenase levels above those induced by aprotinin alone. The results for the effects of aprotinin on stromelysin activation were similar (data not shown).
Effect of TNF-α on MMP Secretion and Activation
Vascular SMCs responded to 0, 0.1, 1, or 10 ng/mL TNF-α with a concentration-dependent increase of collagenase and stromelysin (data not shown). Most of the induced MMPs were in the latent forms, but a small amount appeared in the activated forms. These results are consistent with those of Galis et al,17 who reported that TNF-α induced de novo synthesis of these two enzymes by vascular SMC cultures. In the presence of 100 μg/mL plasminogen, the concentration of TNF-α had varying effects on total collagenase secretion in experiments using SMCs from different sources. However, the trend in activation state of collagenase was consistent among independent experiments. Although the collagenase was primarily in the active forms with 0, 0.1, and 1 ng/mL TNF-α, most of the collagenase appeared in the latent forms with 10 ng/mL TNF-α (Fig 4A⇓), suggesting that TNF-α induced an inhibitor of MMP activation. The addition of TNF-α to 100 μg/mL plasminogen also had varying effects on total stromelysin secretion by SMCs in independent experiments. However, as with collagenase, 10 ng/mL TNF-α consistently inhibited plasminogen-induced activation of stromelysin, whereas lower concentrations of TNF-α allowed most of the stromelysin to be activated (Fig 4B⇓).
Effects of TNF-α and Plasminogen on PAI-1 Secretion
The TNF-α inhibition of MMP activation by plasminogen suggested that TNF-α increased serine protease inhibitory activity. TNF-α has been shown to induce PAI-1 secretion in endothelial cells.22 Therefore, measurements of PAI-1 in the culture media were performed. As expected, TNF-α induced PAI-1 in a concentration-dependent manner; at 10 ng/mL, PAI-1 levels were 4.2-fold over control levels (Fig 5A⇓). Surprisingly, plasminogen (Fig 5B⇓) also induced PAI-1 in a concentration-dependent manner (3.6-fold over control at 100 μg/mL). The combination of TNF-α and plasminogen led to levels of PAI-1 that were significantly higher than levels reached by either one alone (Student’s t test, P<.05) but less than would be expected if their effects were simply additive (Fig 5C⇓). In a separate experiment, plasmin (100 μg/mL) induced PAI-1 levels more than ninefold over control levels.
Activity of PAI-1 Induced by TNF-α
To test the activity of the TNF-α–induced PAI-1, the conversion of plasminogen to plasmin in the absence or presence of TNF-α in SMC cultures was determined by Western blot analysis (Fig 6⇓). In the absence of TNF-α, plasminogen was cleaved to plasmin, although the efficiency of the conversion varied among experiments using different sources of SMCs. In the presence of 10 ng/mL TNF-α, and therefore in the presence of PAI-1, plasminogen activation was almost completely blocked.
The matrix degradation process is regulated at several levels. The MMPs are secreted in latent form and require activation, possibly by plasmin. Once the MMPs are activated, their activity may be inhibited by TIMPs. Vascular SMCs can produce not only matrix molecules and MMP zymogens but also plasminogen activators, plasminogen activator inhibitors, and TIMPs.4 Furthermore, vascular cells can vary the expression of plasminogen activator receptors to facilitate or decrease the activation of plasminogen.25 SMCs may also influence the degradation of plasminogen activators.26
In the present study, we found that plasminogen not only increases MMP secretion and activation but also induces PAI-1 secretion in SMCs. This concentration-dependent induction of PAI-1 synthesis by plasminogen suggests a negative-feedback mechanism for downregulating the effects of plasmin (Fig 7⇓). If plasmin and MMPs were not tightly regulated, matrix degradation could proceed too rapidly for appropriate remodeling responses by the SMCs. The secretion of PAI-1 could have the dual negative-feedback effect of both downregulating plasmin-mediated degradation and downregulating MMP activation. This may provide the cell with a mechanism to control MMP activity even after the MMPs have been synthesized and secreted into the extracellular space. The cell specificity of this negative-feedback mechanism warrants further study, since Etingin et al27 found no effect of plasminogen on PAI-1 secretion by cultured vascular endothelial cells.
Although TNF-α increased MMP secretion in SMCs, the MMPs were largely in the latent zymogen forms. We also found that TNF-α at 1 ng/mL and higher concentrations can partially inhibit the activation of MMPs, even in the presence of plasminogen levels sufficient to activate MMPs in the absence of TNF-α. These results are consistent with the finding that TNF-α increases PAI-1, decreasing the amount of plasmin available to activate MMPs. Analysis by immunoblotting confirmed that PAI-1 induced by TNF-α almost completely inhibited the conversion of plasminogen to plasmin in SMC cultures. Consequently, at higher concentrations of TNF-α, there was much less plasmin available to induce MMP secretion, such that TNF-α played a greater role in increasing MMP levels. On the other hand, PAI-1 induced by plasminogen (or by plasminogen-derived plasmin) did not inhibit plasminogen-induced MMP activation, probably because sufficient plasmin was already generated to activate MMPs.
The level of matrix-bound PAI-1 may respond to stimuli differently from PAI-1 in the media. Functionally, matrix-bound PAI-1 is more stable and active.10 In the present study, only the free forms were assayed. A comparison of the levels of free and bound forms might provide more insight into their relative importance in matrix remodeling. Furthermore, changes in plasminogen activator levels may be occurring in parallel with changes in PAI-1 levels. Thus, although circumstantial evidence indicates that PAI-1 is inhibiting plasminogen activation, other plasminogen activator inhibitors synthesized by vascular SMCs such as PAI-2 and protease–nexin I28 may play significant roles as well.
The data of the present study are consistent with the hypothesis that induction and activation of MMPs by plasminogen are consequences of plasmin activity. The ability of aprotinin to inhibit plasminogen-induced MMP activation provides further evidence for the role of plasmin. However, aprotinin inhibits a wide variety of serine proteases; therefore, it is possible that its effects on MMP activation are via a different pathway. Similarly, the unexpected induction of latent MMPs by aprotinin in control experiments may be an indirect consequence of its inhibition of endogenous active serine proteases.
A direct plasmin-independent effect of plasminogen has not been excluded by these studies. Treatment of SMC cultures with kringle 4, a lysine-binding fragment of plasminogen,29 did not affect MMP levels, but the results of treatment with kringle 1-3, another lysine-binding fragment,29 were inconclusive. Kringle 5, which has a non–lysine-binding site, is the main site of interaction of plasminogen with cultured human umbilical vein endothelial cells30 and may play a role in interactions of plasminogen with receptors on SMCs. The effects of kringle 5 were not explored in the present study.
Plasmin can directly degrade many extracellular matrix molecules, including fibronectin, laminin, and proteoglycans.31 In addition, plasmin can regulate matrix degradation by increasing the secretion and activation of MMPs. Degradation of matrix may cause changes in cell-matrix interactions, changes in the local stresses on cells, and changes in cell shape. These changes could then signal the cell to express proteins that are necessary for remodeling, including MMPs and PAI-1. For example, fibroblasts express collagenase and stromelysin in response to agents that cause loss of stress fibers.32 In addition, Werb et al33 found that fibronectin fragments but not native fibronectin induced collagenase and stromelysin gene expression in the absence of obvious changes in cell shape, indicating that fibronectin degradation stimulates remodeling activity. Thus, it is possible that both matrix-receptor mechanisms and cytoskeletal changes may explain the induction of PAI-1 by plasminogen.
Selected Abbreviations and Acronyms
|ELISA||=||enzyme-linked immunosorbent assay|
|PAI-1, PAI-2||=||plasminogen activator inhibitor-1 and -2|
|SMC||=||smooth muscle cell|
|TIMPs||=||tissue inhibitor of metalloproteinases|
|TNF-α||=||tumor necrosis factor-α|
This study was supported in part by a grant from the American Heart Association (Massachusetts affiliate) and by a Merit Award from the Department of Veterans Affairs Research Service (Dr Vaughan). Dr E. Lee was supported by a Department of Defense National Science and Engineering Graduate fellowship. We thank Dr Marysia Muszynski for her excellent assistance with the Limulus amebocyte lysate test.
- Received January 25, 1995.
- Accepted September 11, 1995.
- © 1996 American Heart Association, Inc.
Ross R, Klebanoff SJ. The smooth muscle cell, I: in vivo synthesis of connective tissue proteins. J Cell Biol. 1971;50:159-171.
Wight TN, Ross R. Proteoglycans in primate arteries, I: ultrastructural localization and distribution in the intima. J Cell Biol. 1975;67:660-674.
Henney AM, Wakeley PR, Davies MJ, Foster K, Hembry R, Murphy G, Humphries S. Localization of stromelysin gene expression in atherosclerotic plaques by in situ hybridization. Proc Natl Acad Sci U S A. 1991;88:8154-8158.
Welgus HG, Campbell EJ, Cury JD, Eisen AZ, Senior RM, Wilhelm SM, Goldberg GI. Neutral metalloproteinases produced by human mononuclear phagocytes: enzyme profile, regulation, and expression during cellular development. J Clin Invest. 1990;86:1496-1502.
He C, Wilhelm SM, Pentland AP, Marmer BL, Grant GA, Eisen AZ, Goldberg GI. Tissue cooperation in a proteolytic cascade activating human interstitial collagenase. Proc Natl Acad Sci U S A. 1989;86:2632-2636.
Sperti G, Quax P, Van Leeuwen R, Kluft C. PAI-1 and t-PA mRNAs are expressed by vascular smooth muscle cells in vivo. Circulation. 1990;82(suppl III):III-573. Abstract.
Knudsen BS, Harpel PC, Nachman RL. Plasminogen activator inhibitor is associated with the extracellular matrix of cultured bovine smooth muscle cells. J Clin Invest. 1987;80:1082-1089.
Werb Z, Aggeler J. Proteases induce secretion of collagenase and plasminogen activator by fibroblasts. Proc Natl Acad Sci U S A. 1978;75:1839-1843.
Amento EP, Ehsani N, Palmer H, Libby P. Cytokines and growth factors positively and negatively regulate interstitial collagen gene expression in human vascular smooth muscle cells. Arterioscler Thromb. 1991;11:1223-1230.
Schönherr E, Järveläinen HT, Sandell LJ, Wight TN. Effects of platelet-derived growth factor and transforming growth factor-β1 on the synthesis of a large versican-like chondroitin sulfate proteoglycan by arterial smooth muscle cells. J Biol Chem. 1991;266:17640-17647.
Chen J-K, Hoshi H, McKeehan WL. Transforming growth factor type α specifically simulates synthesis of proteoglycan in human adult arterial smooth muscle cells. Proc Natl Acad Sci U S A. 1987;84:5287-5291.
Schlumberger W, Thie M, Rauterberg J, Robenek H. Collagen synthesis in cultured aortic smooth muscle cells: modulation by collagen lattice culture, transforming growth factor-β1, and epidermal growth factor. Arterioscler Thromb. 1991;11:1660-1666.
Galis ZS, Muszynski M, Sukhova GK, Simon-Morissey E, Unemori EN, Lark MW, Amento E, Libby P. Cytokine-stimulated vascular smooth muscle cells synthesize a complement of enzymes required for extracellular matrix digestion. Circ Res. 1994;75:181-189.
Saarialho-Kere UK, Welgus HG, Parks WC. Distinct mechanisms regulate interstitial collagenase and 92-kDa gelatinase expression in human monocytic-like cells exposed to bacterial endotoxin. J Biol Chem. 1993;268:17354-17361.
Sawdey MS, Loskutoff DJ. Regulation of murine type 1 plasminogen activator inhibitor gene expression in vivo: tissue specificity and induction by lipopolysaccharide, tumor necrosis factor-α, and transforming growth factor-β. J Clin Invest. 1991;88:1346-1353.
Sawdey M, Podor TJ, Loskutoff DJ. Regulation of type 1 plasminogen activator inhibitor gene expression in cultured bovine aortic endothelial cells: induction by transforming growth factor-β, lipopolysaccharide, and tumor necrosis factor-α. J Biol Chem. 1989;264:10396-10401.
Ridker PM, Vaughan DE, Stampfer MJ, Manson JE, Shen C, Newcomer LM, Goldhaber SZ, Hennekens CH. Baseline fibrinolytic state and the risk of future venous thrombosis. Circulation. 1992;85:1822-1827.
Noda-Heiny H, Daugherty A, Sobel BE. Augmented urokinase receptor expression in atheroma. Arterioscler Thromb Vasc Biol.. 1995;15:37-43.
Grobmyer SR, Kuo A, Orishimo M, Okada SS, Cines DB, Barnathan ES. Determinants of binding and internalization of tissue-type plasminogen activator by human vascular smooth muscle and endothelial cells. J Biol Chem. 1993;268:13291-13300.
Etingin OR, Hajjar DP, Hajjar KA, Harpel PC, Nachman RL. Lipoprotein (a) regulates plasminogen activator inhibitor-1 expression in endothelial cells: a potential mechanism in thrombogenesis. J Biol Chem. 1991;266:2459-2465.
Sottrup-Jensen, Claeys H, Zajdel M, Petersen TE, Magnusson S. The primary structure of human plasminogen: isolation of two lysine-binding fragments and one ‘mini-’ plasminogen (MW, 38,000) by elastase-catalyzed-specific limited proteolysis. In: Davidson JF, Rowan RM, Samama MM, Desnoyers PC, eds. Progress in Chemical Fibrinolysis and Thrombolysis. New York, NY: Raven Press Publishers; 1978;3:191-209.
Mignatti P, Rifkin DB. Biology and biochemistry of proteinases in tumor invasion. Physiol Rev. 1993;73:161-195.
Werb Z, Hembry RM, Murphy G, Aggeler J. Commitment to expression of the metalloendopeptidases, collagenase and stromelysin: relationship of inducing events to changes in cytoskeletal architecture. J Cell Biol. 1986;102:697-702.
Werb Z, Tremble PM, Behrendtsen O, Crowley E, Damsky CH. Signal transduction through the fibronectin receptor induces collagenase and stromelysin gene expression. J Cell Biol. 1989;109:877-889.