| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Bristol Heart Institute, University of Bristol, Bristol, United Kingdom
ABSTRACT I. INTRODUCTION II. INTIMAL THICKENING: FROM PHYSIOLOGICAL ADAPTATION TO PATHOLOGICAL FAILURE III. WHY CONSIDER A ROLE FOR MATRIX METALLOPROTEINASES IN INTIMAL EXPANSION AND PLAQUE RUPTURE? IV. MATRIX METALLOPROTEINASE EXPRESSED IN THE MAMMALIAN VASCULATURE V. CRITERIA FOR MATRIX METALLOPROTEINASE INVOLVEMENT IN INTIMAL THICKENING AND PLAQUE RUPTURE VI. MATRIX METALLOPROTEINASE AND TISSUE INHIBITOR OF METALLOPROTEINASE GENE TRANSCRIPTION IS DIFFERENTIALLY REGULATED IN ISOLATED VASCULAR CELLS IN CULTURE: MATRIX METALLOPROTEINASE UPREGULATION MAY OCCUR IN SEVERAL DISTINCT STAGES A. VSMC B. EC C. Macrophages D. T Lymphocytes E. Mast Cells F. Adventitial Fibroblasts G. Transcriptional Control Mechanisms VII. PARTICIPATION OF MATRIX METALLOPROTEINASES AND TISSUE INHIBITORS OF METALLOPROTEINASE IN INTIMAL THICKENING BY REGULATION OF CELL MIGRATION AND PROLIFERATION A. Methodologies B. Patterns of MMP Expression and Activity in Humans and in Animal Models of Neointima Formation C. Effects of MMP Overexpression and TIMP Knockout D. Effects of MMP Inhibition and Knockout E. Putative Mechanisms F. Conclusions VIII. ROLE OF MATRIX METALLOPROTEINASES IN ATHEROSCLEROTIC PLAQUE DESTABILIZATION A. Methodologies B. Patterns of MMP Expression and Activity in Human Pathological Tissues and Experimental Models of Atherosclerosis C. Effects of MMP Overexpression and TIMP Knockout D. Effects of MMP Inhibition and Knockout E. Putative Mechanisms F. Conclusions IX. PHARMACOTHERAPY A. Direct Inhibition by Synthetic Inhibitors and Gene Therapy with TIMPs B. Inhibition of MMP Production X. SUMMARY: WHAT MEDIATES THE TRANSITION FROM PHYSIOLOGICAL REMODELING TO MATRIX DESTRUCTION? ACKNOWLEDGMENTS REFERENCES
 |
ABSTRACT |
|---|
 Top NextReferences |
|---|
|
I. INTRODUCTION |
|---|
|
TopPreviousNextReferences |
|---|
Based on Nobel Prize winning work by Goldstein and Brown into the mechanisms of cholesterol uptake into cells, Steinberg et al. (263) formulated the "modified low-density lipoprotein" (LDL) hypothesis. According to this, modifications, including importantly oxidation, increase the uptake of plasma LDL into macrophages and give rise to foam cells. In 1972 Ross and Glomset (229) initially formulated the "response to injury" hypothesis, which focused on the behavior of vascular smooth muscle (VSMC) and endothelial cells (EC) and took into account the recent discovery of peptide growth factors. According to this proposal and subsequent modifications, the vessel wall responded by endothelial dysfunction and VSMC proliferation when injured mechanically, by abnormal flow, or by other noxious agents, such as cigarette smoke or glycated proteins produced in diabetic patients. Response to injury provided a satisfactory explanation for fibrous cap formation, for the localization of atherosclerosis in areas of disturbed flow and many risk factors in addition to hypercholesterolemia. The hypothesis also offered an explanation for the iatrogenic intimal thickening that had begun to be recognized as a clinical problem following the introduction of coronary bypass grafting, coronary angioplasty, and heart transplantation. Another consequence of all these perturbations is inflammation, and Libby (157) championed the possible pathogenetic role of inflammatory mediators. Hansson and Wick (reviewed in Ref. 107) formulated convincing cases for an immune component of atherosclerosis. In his later reviews, Ross (228) largely succeeded in unifying these various hypotheses by proposing oxidized lipids as one important protagonist of chronic vascular inflammation and hence response to injury.
By the beginning of the 1990s, a long list of inflammatory cytokines and growth factors that might mediate the response to injury had been identified (198). When examined in isolated tissue culture, many of these cytokines promoted the proliferation of VSMC, and a smaller subset affected migration. However, when these observations were translated into animal models of neointima formation by Reidy and Clowes' groups (58) or human organ culture models (our work) (259), it was apparent that growth factors by themselves were insufficient and that extracellular proteases might also be needed (58). Parallels between invasion of VSMC in the neointima and invasion of cancer cells and other a priori reasons itemized in section III of this review led us to consider a role for matrix metalloproteinases (MMPs) (256). Section IV provides a necessary introduction to the MMP family members that have subsequently been discovered in the vessel wall. Section V attempts to structure the wealth of data regarding the physiological and pathological regulation of MMP activity in the vasculature. A comprehensive and critical examination of the evidence for involvement of MMPs in intimal thickening is one of the major tasks of this review. Section VI lists the criteria, and section VII details the evidence.
Thanks mainly to careful autopsy observations, particularly by Davies and Thomas (64, 65) and Falk et al. (77), it has become accepted that myocardial infarction is a distinct acute pathological event superimposed on the chronic process of atherosclerosis. Two main mechanisms were identified, most often rupture of the fibrous cap ("plaque rupture") or alternatively disruption of the surface endothelium ("plaque erosion") (reviewed in Ref. 63). Plaque rupture was associated inter alia with a greater degree of inflammation and destruction of the extracellular matrix (152). This led Henney et al. (113) and Libby and co-workers (90) independently to propose a pathogenetic role for MMPs in plaque rupture. A detailed critical appraisal of this concept is presented in section IX. Section X summarizes the dual role of MMPs in intimal expansion and plaque rupture and asks how these two roles, which appear to have opposite effects on plaque stability, can be reconciled.
|
II. INTIMAL THICKENING: FROM PHYSIOLOGICAL ADAPTATION TO PATHOLOGICAL FAILURE |
|---|
|
TopPreviousNextReferences |
|---|
Intimal growth is mediated by an increase in the number and diversity of cells and is accompanied by accumulation of new extracellular matrix (57). It is an adaptive or repair response to a variety of adverse physical and biochemical stimuli acting on the vessel wall. For example, when pulmonary hypertension occurs secondary to congenital or acquired abnormalities of the heart, VSMC hyperplasia is a direct response to increased tangential wall stress (289). A similar response occurs also in arteriovenous fistulas and surgical vein-grafts into the systemic circulation (10, 70). Mechanical injury, particularly to the luminal surface of arteries during percutaneous transluminal angioplasty, also triggers VSMC hyperplasia as a repair response, at least in healthy animals' arteries (27, 28). Intimal remodeling is almost always accompanied by transient inflammation. For example, immunologically driven processes, such as transplant arteriosclerosis (232) and foreign body reaction to stent implantation after angioplasty (79, 138, 281), provoke intimal thickening.
The most common and arguably most complex chemical insult that provokes intimal expansion is atherosclerosis. Deposition of LDL, which subsequently becomes oxidized, is the most likely initial event. This is followed by infiltration of circulating monocytes, which convert to macrophages and take up oxidized LDL to become foam cells (228). Activated lymphocytes also occur in atherosclerotic plaques at all stages of its progression, and other inflammatory cells including mast cells are also present (107). Formation of a distinct fibrous cap over a large lipid core occurs only as a late consequence of atherosclerosis (261). The lipid core arises because macrophages die by apoptosis and spill their lipid contents. The plaque cap could form by migration of typical, contractile VSMC from the media, which would account for the frequent medial wasting observed at the base of atherosclerotic plaques (66). However, this paradigm has been challenged recently. First, many plaque caps appear to have arisen by clonal expansion of one or a few cells (26, 194). This could arise by selection of a subpopulation of medial VSMC that express a proliferative phenotype (108) or expansion from circulating VSMC precursors (stem cells) that have invaded the plaque (44, 118, 235). Alternatively, it has been suggested that some neointimal VSMC may be transdifferentiated adventitial fibroblasts (308). The atherosclerotic intima also contains capillaries, formation of which contributes to plaque growth in animal models (193).
If intimal growth is such that the initial lumen is narrowed by >70%, distal blood flow is restricted and chronic tissue ischemia results. This occurs in native coronary arteries and during restenosis after coronary angioplasty or failure of some coronary vein grafts: it then results in stable angina pectoris. Constrictive vessel wall remodeling (shrinking of the whole vessel wall), which involves the fibroblast-rich adventitia as well as the media, may exacerbate the unwanted flow restriction caused by intimal thickening (71, 188). Conversely, expansive remodeling, i.e., enlargement of the total vessel wall cross-sectional area, ameliorates these adverse hemodynamic consequences (97). On the other hand, pathological outward expansion, for example, in aortic aneurysms, can lead to catastrophic medial dissection and rupture. In humans, aortic aneurysms are considered the consequence of invasion of atherosclerotic inflammation into the media (45, 224). Recent histological and in vivo intravascular ultrasound studies showed that ruptured plaques are more likely to have undergone outward remodeling (209, 240). Moreover, apolipoproteinE (ApoE) knockout mice show frequent aortic microaneurysms at the base of atherosclerotic plaques (47). It is tempting, therefore, to conclude, as Pasterkamp et al. did (209), that aneurysmal dilatation, expansive remodeling, and plaque rupture actually have a common inflammatory origin and can be regarded as aspects of the same phenomenon.
Catastrophic rupture or surface erosion of the fibrous cap underlies fatal coronary thrombosis and myocardial infarction (MI), which is the leading cause of premature death in developed countries (see http://www.who.int/whosis/). Plaque rupture, which accounts for more than 80% of fatal MI in men, occurs in regions of high tangential stress and where collagen is depleted (63). The implication is that matrix destruction weakens the plaque to the point where it can no longer resist the cyclical strain caused by the cardiac cycle (158). Plaque erosion accounts for as much as 50% of MI in young women (63). It is apparently a distinct pathology triggered by large-scale loss of the endothelium and exposure of the thrombogenic underlying basement membrane. Because much less is known about its underlying basis, it will not be considered further in this review.
|
III. WHY CONSIDER A ROLE FOR MATRIX METALLOPROTEINASES IN INTIMAL EXPANSION AND PLAQUE RUPTURE? |
|---|
|
TopPreviousNextReferences |
|---|
Intimal thickening constitutes the generation of new tissue at least in part using VSMC derived from the media. As illustrated in Figure 1, MMPs could catalyze removal of the basement membrane around VSMC and facilitate contacts with the interstitial matrix. This could promote a change from quiescent, contractile VSMC to cells capable of migrating and proliferating to mediate repair. Freeing of sequestered growth factors and production of new, permissive matrix components could further promote VSMC migration and proliferation.
|
Remodeling of the vascular ECM occurs at neutral pH, where MMPs are active.
The MMPs (also known as matrixins) are a large family of proteases that each has at least one ECM component as a putative substrate.
The remainder of this review discusses the structure of MMPs and the ways in which their activity is regulated and critically appraises their physiological roles in blood vessels.
|
IV. MATRIX METALLOPROTEINASE EXPRESSED IN THE MAMMALIAN VASCULATURE |
|---|
|
TopPreviousNextReferences |
|---|
|
Most pro-MMPs are likely to be activated biologically by tissue or plasma proteinases, including other MMPs (Fig. 3). One well-established case is activation of MMP-2 at the cell surface by MT-MMPs, which is best understood for MT1-MMP (264, 267). TIMP-2 bound to the catalytic site of one MT1-MMP molecule acts as a receptor for MMP-2 by interacting with the noncatalytic COOH-terminal domain. A second molecule of MT1-MMP then partially cleaves the MMP-2 prodomain, which is then fully cleaved by autoproteolysis. None of the other TIMPs can substitute for TIMP-2 during the activation of MMP-2 (299). One consequence is that MT-MMP-mediated MMP-2 activation is functionally impaired in TIMP-2 knockout mice (264). Secreted MMPs, particularly, MMPs-3, -7, and -10, can activate other secreted MMPs (283). Serine proteases, such as trypsin and mast cell-derived chymase and tryptase, activate several MMPs in the test tube (149). Hence, the mast cell enzymes could be significant activators of MMPs in atherosclerotic plaques in vivo (131). The serine proteases, thrombin (87, 145), and Factor Xa (223) promote activation of MMP-2, which suggests a link between thrombosis and activation of ECM turnover. Tissue but not plasma kallikrein activates purified MMP-9 (69). The abundant serine protease, plasmin, activates pro-MMPs-1, -3, -7, -9, -10, and -13 in vitro (159). Moreover, plasminogen knockout in mice impairs MMP-9-mediated elastin degradation and aneurysm formation, showing that plasmin is an essential MMP activator in vivo (47). Plasmin is, in turn, activated by plasminogen activators that can be secreted or located at the cell surface by binding to specific receptors (74). From knockout studies, the urokinase-type plasminogen activator seems to be particularly important for vascular remodeling (46). This must be activated by yet other proteases, perhaps plasma kallikrein (74). Hence, MMP activation is likely to require a complex cascade of catalytic activation during which the final proteolytic effect could be greatly amplified (Fig. 3). The question arises as to which processes initiate the cascade during vascular injury. Possibilities worthy of consideration are release of lysosomal proteases during cell death or production of ROS during oxidative stress (220).
|
A variety of protein inhibitors of MMP have been described, among which the TIMPs are by far the most potent and selective (15). There are at least four members of the TIMP family that form tight inhibitory 1:1 complexes with MMPs. These interactions are generally rather nonselective, although, exceptionally, TIMP-1 is a much weaker inhibitor of MT-MMPs than other TIMPs (248, 295). TIMPs-1 to -4 are each made up of two domains containing three disulfide bonds (195). The NH2-terminal cysteine is particularly important in inhibition, since its free 
-amino group and carbonyl function displace the catalytic water molecule from the essential zinc ion in the MMP active site (99). The three COOH-terminal loops of TIMPs mediate complex formation with the progelatinase hemopexin domain and are largely responsible for the unique matrix binding capacity of TIMP-3 (146). The importance of the COOH terminus of TIMP-3 is further underlined by the fact that any of six mutations that generate a free SH-group in this portion of the molecule leads to the autosomal-dominant blindness, Sorsby's fundus dystrophy (288). In plasma, the general protease inhibitor 2-macroglobulin may also be important. The physiological relevance of other inhibitors including the RECK protein (204) is unknown.
Disintegrin metalloproteinases (ADAMs) also contain an MMP-like catalytic domain, which in some cases remains catalytically active. In general, TIMPs do not bind to or inhibit the catalytic site of ADAMs, although TIMP-3 inhibits a number of ADAMs and TIMP-1 inhibits ADAM-10 (8). Of the more than 30 members of this family, only ADAM-10 (36), ADAM-15 (115), and ADAM-17 (31) have been characterized in vascular cells. ADAMs will not be discussed further, since there is insufficient evidence to decide whether or not they have a specific role in intimal thickening or plaque rupture.
Based on substrate specificity and structural homology, MMPs can be assigned to five subgroups: interstitial collagenases, gelatinases, stromelysins/matrilysins, membrane-type MMPs (MT-MMPs), and others (Table 1). MMPs appear to have distinct but overlapping specificities against purified matrix proteins in the test tube (see Refs. 264, 283). Interstitial collagenases (MMP-1, MMP-8, and MMP-13) cleave intact fibrillar interstitial collagens I, II, and III, and MMP-14, an MT-MMP, also has this ability (205). The gelatinases (MMP-2 and -9) have prominent activity against basement membrane components including type IV collagen and laminin and also elastin (5, 134). Stromelysins-1 and -2 (MMPs-3 and -10) and matrilysin (MMP-7) have a wide substrate repertoire, which includes most of the ECM proteins and proteoglycans except intact fibrillar collagens. Stromelysin-3 (MMP-11) is related in sequence to the other stromelysins, but its substrate specificity is unusual. Its main substrates may be serum protease inhibitors, not matrix proteins (241, 264). Metalloelastase (MMP-12) is usually set apart from the other groups of MMPs on sequence criteria and because of its limited activity as a collagenase (264). The MT-MMPs were first identified by their ability to activate progelatinase A (236), although MMP-17 does not, and all of the MT-MMPs also cleave at least one matrix protein (283). MMPs also cleave a variety of nonmatrix substrates (for comprehensive listings, see Refs. 264, 283).
|
|
V. CRITERIA FOR MATRIX METALLOPROTEINASE INVOLVEMENT IN INTIMAL THICKENING AND PLAQUE RUPTURE |
|---|
|
TopPreviousNextReferences |
|---|
1) MMPs are expressed and regulated in vascular cells.
2) MMP gene expression and activity are temporally and spatially related to intimal expansion and plaque rupture.
3) Introduction of MMP genes mimics the events of matrix remodeling.
4) Inhibition or knockout of MMP genes suppresses remodeling.
5) Plausible mechanisms explain the actions of MMPs.
These criteria are considered in the following sections.
|
VI. MATRIX METALLOPROTEINASE AND TISSUE INHIBITOR OF METALLOPROTEINASE GENE TRANSCRIPTION IS DIFFERENTIALLY REGULATED IN ISOLATED VASCULAR CELLS IN CULTURE: MATRIX METALLOPROTEINASE UPREGULATION MAY OCCUR IN SEVERAL DISTINCT STAGES |
|---|
|
TopPreviousNextReferences |
|---|
Upregulation of MMP production from VSMC fits the pattern of a series of stages (Fig. 4). VSMC constitutively express the gelatinase MMP-2 (88, 211, 256, 304). Stretch further upregulates MMP-2 via production of ROS in an NAD(P)H oxidase-dependent manner (101). Upregulation of MMP-2 production and dramatic induction of the other gelatinase, MMP-9, occur rapidly after mechanical injury (25, 95, 125, 257). These findings imply that stretch and injury can promote MMP-mediated breakdown of basement membranes, for which there is direct evidence (1). MMP-2 and -9 also have significant elastolytic activity that could be important in remodeling of arteries in response to altered hemodynamics (55). Inflammatory cytokines, such as interleukin (IL)-1, IL-4 and tumor necrosis factor- (TNF-), coordinately induce a broad range of MMPs, including MMPs-1, -3, and -9 (88, 147, 234). Cytokines act synergistically with growth factors, such as platelet-derived growth factor (PDGF) and fibroblast growth factor-2 (FGF-2) (34, 75, 137, 213, 304). The presence of both inflammatory mediators and growth factors in injured and atherosclerotic blood vessels could therefore drive the production of MMPs with the ability to efficiently remodel both basement membranes and many components of the interstitial matrix. Cell contact with T-lymphocyte membranes and addition of recombinant CD40 ligand induce MMPs-1, -3, -8, and -9 in VSMC (114, 241, 243), which implies a relationship between immune activation and wide-ranging matrix remodeling.
|
(TGF-) (76). TIMP-2 secretion is constitutive (75, 88), while TIMP-3 expression is upregulated by a combination PDGF and TGF- (75). Together these findings suggest that stretch, injury, inflammation, and immune activation progressively move the MMP/TIMP balance in VSMC towards proteolysis, while fibrogenic stimuli tend to reverse this. However, MMP activity cannot be directly measured in VSMC conditioned media (50) because of an excess of TIMPs (142). This may imply that MMPs produced from VSMC alone are unable to remodel the ECM. On the other hand, MMP activity can be detected in VSMC in vascular tissues by in situ zymography, a powerful methodology whereby frozen histological sections are coated with and allowed to degrade MMP substrates that are subsequently visualized (90). This suggests that MMP activity may not be completely abolished by TIMP binding but could be confined temporally, or spatially, to the pericellular region surrounding VSMC (309).
As in VSMC, MMP-2 expression is constitutive in EC (106) and further upregulated by ROS (120). However, induction of the MMPs is less obviously staged in EC. MMP-1 and -2 are upregulated by high glucose concentrations (67), implying links to both basement membrane and interstitial matrix degradation in diabetic vascular disease. Similarly, concomitant upregulation of MMP-1, -3, -8, and -9 occurs in response to TNF- and IL-1 (106, 114) and on coculture with macrophages (117, 314). CD40 ligation also upregulates a broad spectrum of MMPs in EC, including MMP-1, -3, -8, -9, and -11 (171, 241). Inflammatory and immune mechanisms therefore seem to be the main drivers of matrix remodeling by EC, with no obvious dissociation of basement membrane and interstitial matrix degradation. Particularly relevant to atherosclerosis, oxidized LDL increases MMP-1 and -3 expression (119, 155) via the Lox-1 receptor and activation of protein kinase C-. Oscillatory but not laminar flow induced a large increase in MMP-9 transcription and activation in a murine endothelial cell line (174), and this could have some bearing on the localization of atherosclerotic plaques. Growth in three-dimensional collagen (103) and cyclic strain (303) upregulate MT1-MMP (MMP-14), while increased shear stress downregulates MMP-14 (307). These phenomena are relevant to microvascular endothelium and angiogenesis (212).
EC express TIMP-1 and TIMP-2 constitutively (106), and this is not increased by oxidized LDL (155). TIMP-1 is further upregulated in response to CD40 ligation (171). MMP activity is nevertheless directly detectable in conditioned media from EC (120), which implies that EC are potentially more active in matrix remodeling than VSMC.
A similar multistage upregulation of MMP genes as that found in VSMC (Fig. 4) is clearly evident in macrophages. At least in human monocyte-macrophages, MMP-9 appears to be the most abundant gelatinase. Contact of peripheral blood monocytes with EC rapidly upregulates MMP-9, via a 30-kDa soluble mediator (7). Adhesion to matrix components including collagen, fibronectin, and tenascin-C also upregulates macrophage MMP-9 secretion (91, 285, 290). Integrin-dependent adhesion to collagen acts synergistically with P-selectin-dependent adhesion to platelets to increase MMP-9 secretion by both transcriptional and translational mechanisms (91). MMP-9 expression is further stimulated to a small extent by soluble mediators IL-18 and TNF- (202, 233) and also by oxidized LDL (301). Proatherogenic cytokines and growth factors, including IL-1, TNF-, macrophage colony stimulating factor (M-CSF) and PDGF, upregulate macrophage MT1-MMP (MMP-14) (221) and MT3-MMP (MMP-16) expression (280) and could therefore activate MMP-2 secreted constitutively from VSMC or EC. Because MMP-1 and -3 are not upregulated by cytokines in macrophages (233), the increased production of MMP-9 and activation of MMP-2 could selectively mediate basement membrane turnover. Cytokines and growth factors also increase macrophage MMP-12 (80), which aids monocyte migration into tissues (250). Immune activation via CD40 ligand upregulates MMPs-1, -3, -8, -9, and -11 (114, 170, 172, 176, 241, 300), which have the ability to remodel interstitial matrix also. MMP-1 secretion is also upregulated by contact with EC (117).
While these mechanisms are no doubt relevant to atherosclerosis, they probably do not completely account for the increase in MMP secretion observed from foam cell macrophages. Isolated, lipid-laden, macrophages from aortic atheroma or from subcutaneous granulomas in cholesterol-fed rabbits continue to express MMPs-1, -3, -9, and -13 for long periods in culture (50, 89). The upregulation of MMPs-1 and -3 depends on engorgement with lipid because it does not occur in nonfoamy alveolar macrophages from lipid-fed rabbits (89) or granuloma macrophages from chow-fed rabbits (50). The mechanism is apparently complex because simple addition of ox-LDL to human macrophages does not upgregulate MMPs-1 and -3 (191). This implies that foam cell formation somehow activates the intrinsic transduction pathways leading to MMP secretion or constitutive production of autocrine factors. Consistent with the first possibility, treatment with the scavenger N-acetyl-L-cysteine decreases MMP-9 expression, which implies that ROS are involved (85), and foam cell macrophages show constitutive activation of the redox-sensitive transcription factor NF-
B (50). Consistent with the second possibility, preliminary work from our laboratory suggests that upregulation of MMP-1 but not MMP-3 from foam cells is in part mediated by a soluble autocrine factor unrelated to CD40 ligand (22).
Human macrophages and rabbit foam cells secrete TIMP-1, TIMP-2, and TIMP-3 constitutively (50, 76, 168), while ox-LDL treatment decreases TIMP-1 secretion, partly at least through autocrine action of IL-8 (191, 301). Despite TIMP secretion, MMP activity is measurable in macrophage and foam cell conditioned media (50, 168), and macrophage-conditioned media degrade collagen from isolated plaque caps in vitro (245). MMPs secreted by macrophages could also function as activators of MMPs derived from other cell types including VSMC.
T lymphocytes in atherosclerotic plaques stain positive for MMPs-1, -2, -3, and -9 by immunocytochemistry (90), but relatively few studies of their MMP secretion have appeared. Interaction with fibronectin through 41-integrins upregulates MMP-2 and -9, while ligation of vascular cell adhesion molecule (VCAM)-1 selectively increases MMP-2 (302). The resulting conditioned media contain MMP activity (302), and these proteases may therefore have a role in transmigration of T cells into sites of inflammation. MMP-2 is also increased in response to the cytokine IL-2 (153).
Mast cells constitutively secrete MMP-2 and upregulate MMP-9 secretion in response to TNF- kit ligand (KL, stem cell factor) and contact with activated T cells (17, 78). Induction of the MMP-9 gene and secretion of MMP-9 protein appear to be separately regulated by T-cell contact (17). T-cell contact also increases TIMP-1 secretion (17); the net effect on MMP activity was not studied.
Overall the pattern of constitutive MMP-2 secretion and synergistic upregulation by cytokines and growth factors of MMP-1, -3, and -9 seems very similar in fibroblasts and VSMC (33, 35). However, the levels of MMP-2 and -9 secretion from pig coronary adventitial fibroblasts are significantly greater than from VSMC (247). Even more strikingly, the levels of TIMP-1 and -2 expression are more than 10-fold less in the fibroblasts (247). Fibroblasts may therefore have a greater propensity to generate MMP activity, which could give them an advantage over VSMC when migrating into areas of vascular injury (247).
G. Transcriptional Control Mechanisms
The basis of the staged induction of different MMP genes by stretch, injury, inflammation, and immune activation can be partially understood from the known structures of their proximal promoter regions (51). Among the secreted MMPs, MMPs-2 and -11 are unusual in not having a TATA or CAAT box, which implies that they are largely constitutively expressed, housekeeping genes. Nevertheless, the transcription of MMP-2 can be upregulated by two far upstream enhancer regions that respond to Y-box protein-1, p53, activator protein-2, and GATA-2 transcription factors (105). The promoters of many secreted MMPs have a TATA box and proximal activator protein-1 (AP-1) binding site, which is essential for upregulation (reviewed in Refs. 21, 80). There are also multiple AP-1 and polyomavirus enhancer-A-binding protein-3 (PEA-3)/ets (PEA-3/ets) sites further upstream. Individual MMP gene promoters contain other unique transcription factor binding sites. For example, a STAT-1 binding element (SBE) in MMP-1, stromelysin PDGF response element (SPRE) in MMP-3, Tcf/LEF in MMP-7, and nuclear factor -B (NF-B) sites in MMP-9 promoters, respectively, probably contribute to their staged induction (reviewed in Ref. 51). These transcription factors apparently act cooperatively with each other and the core AP-1 site, which in part accounts for the synergistic effects of agonists (34, 75). Oddly, activation of the NF-B transcription factor is required for induction of MMP-1 and -3 in VSMC (34) and macrophages (50), even though their proximal promoters lack functional sites for binding NF-B. One explanation for this could be induction through NF-B-dependent pathways of other unknown transcription factors, although this has not been demonstrated. The MMP-9 promoter also has a c-Myc binding site, which participates in responses to oscillatory flow (174). The promoter region of MT1-MMP (MMP-14) is distinctly different from those of the secreted MMPs, since it lacks a TATA box and contains a proximal Sp-1 rather than AP-1 site (165). Activity of the MMP-14 promoter appears to be upregulated either by Sp-1 phosphorylation or by its displacement with the Egr-1 transcription factor (303, 307).
|
VII. PARTICIPATION OF MATRIX METALLOPROTEINASES AND TISSUE INHIBITORS OF METALLOPROTEINASE IN INTIMAL THICKENING BY REGULATION OF CELL MIGRATION AND PROLIFERATION |
|---|
|
TopPreviousNextReferences |
|---|
Well-established methodologies have been developed to measure intimal expansion and its component processes. Endothelium-denuded medial tissue can be stripped from the adventitia and VSMC isolated by outgrowth from explants (187) or by digestion with collagenase (43). Such cells progressively lose their normal complement of contractile and cytoskeletal proteins and increase their content of synthetic organelles, a process known as phenotypic modulation (273). In fact, a variety of different VSMC phenotypes have been isolated from animal and human arteries and veins (108). Some are fibroblast-like in shape and grow over each other in a typical hill and valley formation, while others are epithelioid and grow as monolayers. EC can be selectively removed from vessels by brief collagenase digestion; depending on the culture conditions, they grow as monolayers or as capillary-like tubes (212).
Freshly isolated VSMC can be allowed to undergo contractile to synthetic phenotype modulation under controlled tissue culture conditions, especially when grown on the interstitial matrix component fibronectin (111, 112). Phenotypic change can be delayed by culture on the basement membrane component laminin (110), while addition of heparin, TGF-, and, at least for human cells, growth at high density tends to reverse the phenotypic change (100). Growth of human SMC at high density can give models for other pathological features including calcification (216).
Migration of isolated VSMC and EC can be measured in wounded monolayers (161) or by seeding isolated cells onto thin or thick layers of collagen and other synthetic matrices, such as matrigel (82, 184). Measuring the rate of emigration of cells from explants (256), especially in the presence of hydroxyurea to inhibit cell proliferation (136), allows studies of the effects of native ECM barriers. Cell proliferation is commonly measured by cell counting, incorporation of DNA precursors, measurement of cell cycle specific proteins, and flow cytometry. Death is measured by cell counting, dye penetration, nuclear shape, and a variety of assays selective for apoptosis, such as caspase activation and DNA end-labeling (206, 286).
Organ culture (140, 259) gives a simple in vitro assay for mechanistic studies of neointima formation and is one of the few tools that can be directly used with human tissue. Pulsed or continuous labeling of the cultures with DNA precursors can be used to measure proliferation. However, some VSMC that migrate to the neointima do not subsequently proliferate and can be used as a measure of migration alone (92, 93, 259). In vivo, rat, rabbit, and pig models of arterial balloon injury have been widely used. These show immediate endothelial denudation, apoptosis of medial VSMC (175), and in some cases rupture of the media (246). Early proliferation of VSMC follows in the media. Migration of VSMC (57) (and fibroblasts, Ref. 246) to the neointima and their proliferation occur later. Intimal VSMC lay down excess ECM, which contributes to thickening (57). As in the organ cultures, some VSMC that migrate to the neointima do not subsequently proliferate and can be used as a measure of migration alone (59). However, more commonly, migration is quantified by measuring the number of cells that appear in the neointima at an early time point, e.g., 4 days (124). Neointimaformation in balloon injury models is self-limiting, in part by endothelial regrowth, and may actually regress by enhanced apoptosis and compaction of the ECM (57). Investigators have usually studied the response to injury of normal blood vessels, although in a minority of cases injury has been superimposed on an experimental plaque (54). Even so, the relevance of any of these models to clinical angioplasty in humans has been severely questioned. First, the ready proliferation of VSMC in the animal models does not seem to be observed in human restenotic lesions (203). Second, clinical restenosis may be more closely related to shrinking of the vessel wall (negative remodeling) rather than to neointima formation (71, 188). Recently, animal models of stent deployment in normal arteries have been developed to study the phenomenon of in-stent restenosis (81, 138). These models show foreign body reaction to the stent and localized intimal growth that mirrors the human situation. However, these models do not usually consider the important contribution of existing atherosclerosis to subsequent human in-stent restenosis (79). Rodent and large mammal models of vein grafting have also been described. We recently reviewed the comparative merits of these models, pointing out that only the large animal models reproduce the size and therefore the mechanical changes of human bypass grafting (292). However, induction of vein graft atherosclerosis is much easier in rabbit and mouse models.
A particular cautionary note is necessary when considering the available models in mice, since so much recent research has sought to harness the power of their genetic manipulation. Because of their relative size, mice are likely to have lesions less than 1,000 times smaller in terms of mass and volume compared with human lesions (62). Vessel diameters are correspondingly smaller, and the thickness may be just a few cell layers. In addition, normal mice carry most of their plasma cholesterol as high-density lipoprotein (HDL), not LDL. Species-related differences in protein expression (importantly MMP-1, see below) also make extrapolation from mouse to human a risky enterprise.
Injury to healthy mouse carotid artery has been conducted using filament loops, guide wires, dessication, adventitial cuffing, and flow cessation, although none of these is the mechanical equivalent of clinical balloon angioplasty. An ingenious mouse vena cava into carotid artery vein-grafting model was developed by Xu and colleagues (315). However, this may differ crucially from vein grafting in humans because the mouse vena cava is so thin walled that all of its cells undergo apoptosis under the strain of arterial pressure (315), and the graft becomes repopulated with cells from the graft and the host (118). The extent to which similar processes occur in the much thicker walled human vein grafts is uncertain.
B. Patterns of MMP Expression and Activity in Humans and in Animal Models of Neointima Formation
Life-threatening complications early after angioplasty are thankfully rare. Hence, there have been few histological and biochemical studies of MMPs in human restenotic lesions and none in the crucial first few days and weeks after injury. Nikkari et al. (200) observed increased collagen transcripts in restenotic lesions 6–18 mo after angioplasty. Levels of MMP-1 and TIMP-1 were reduced compared with normal tissue, which implies a balance towards fibrosis at these relatively late time points. In human saphenous veins, upregulation and activation of gelatinases (MMP-2 and -9) occurs within hours after injury caused by surgical preparation and continues during neointima formation in organ cultures (95). Active gelatinases are detected in the media and neointima by in situ zymography (95, 142), even though TIMP-1 and TIMP-2 levels are also increased (142). Upregulation of MMP-1 and MMP-3 is much less evident in human saphenous vein (132). Given the paucity of direct human pathological observations, other approaches including genetic studies are appropriate. A single genetic study identified a functional A-82G polymorphism in the MMP-12 promoter, which affected the AP-1 binding site (133). The A allele increased promoter activity and favored luminal narrowing in a small group of 71 diabetic patients, who underwent angioplasty and stenting.
There is abundant evidence of gelatinase upregulation in animal models of neointima formation. Induction of MMP-9 and activation of MMP-2 occur rapidly after balloon injury to rat (25, 311), pig (257), baboon (136), rabbit (12), and mouse (98, 160) arteries. In the pig and baboon models, induction of MMP-9 and activation of MMP-2 occurred together (136, 257), but MMP-2 activation was delayed in the rat and mouse studies (25, 98, 160, 311), perhaps pointing to a preeminent role for MMP-9 in small rodents (see below). Activation of MMP-2 is associated with upregulation of MMP-14 and -16 expression in balloon-injured rat carotid arteries, implying that activation of MMP-2 by these MT-MMPs is feasible (251). Levels of MMP-1, -3, -12, and -13 were also increased after femoral artery injury in mice (160). TIMP-1 and TIMP-2 expression are either not changed or increased after balloon injury (109, 287), while levels of TIMP-3 (unpublished observations) and TIMP-4 are clearly increased (73). Increased expression and activation of MMP-2 and -9 was also observed in other in vivo models of neointima formation, namely, pig vein grafts (258) and the rabbit iliac artery after angioplasty and stenting (81).
Hence the picture of early and possibly selective upregulation of gelatinases is highly consistent across a variety of models of neointima formation, irrespective of the provoking stimulus. Levels of TIMPs are either maintained constant or increased so that changes in TIMPs do not amplify but tend to counteract the effects of MMP upregulation. What then are the mechanisms responsible for MMP upregulation? First, this could be the direct result of injury because MMP-1 and -3 upregulation occur in injured monolayers of isolated VSMC (125). MMP-9 induction and MMP-2 activation also occur in human saphenous veins injured by hydrostatic distension ex vivo even before culturing (95). Increased intraluminal pressure produces a similar response in pig carotid arteries too (55), although this may be related to stretch per se rather than injury. Inflammation is another possible cause of MMP upregulation, although neointima formation in organ cultures of human saphenous vein proceeds in the presence of only scattered resident macrophages (60). Balloon injury to the rat carotid artery also provokes little inflammatory response (277). However, balloon injury to other arteries (270), stent implantation (138), and vein grafting (291) provoke a vigorous inflammatory response. As expected, there is initial neutrophil infiltration followed by more sustained macrophage deposition (79, 291). Lymphocytes may also be present especially after stent implantation (138). Hence, there is potential for MMP induction by soluble inflammatory mediators and CD40 ligation, as identified from in vitro studies (see above).
C. Effects of MMP Overexpression and TIMP Knockout
Clowes and colleagues (184) prepared rat VSMC stably transfected with the MMP-9 gene under a tetracycline-regulated promoter. They found that MMP-9 overexpression had no effect on VSMC proliferation but promoted migration through matrigel and three-dimensional collagen in vitro, and invasion of cells seeded onto the adventitia of cryo-injured rat carotid arteries in vivo (184). Luminal seeding of VSMC overexpressing MMP-9 after balloon injury promoted vessel wall dilatation and, as might be expected, reduced the accumulation of extracellular matrix in the neointima. However, there was no effect on the eventual neointimal size (184). Lijnen et al. (161) investigated knockout of TIMP-1 after electrical injury to the mouse femoral artery. They found significantly greater neointima formation after 1–3 wk that was associated with increased levels of the activated forms of MMP-2 and -9. Paradoxically, they found that TIMP-1 knockout promoted matrix accumulation at 1 wk. There was no effect on proliferation indices in vivo, but TIMP-1 knockout VSMC migrated faster in scratch-injured cell cultures. The authors concluded that MMPs mediated enhanced neointima formation primarily by stimulating VSMC migration (161).
D. Effects of MMP Inhibition and Knockout
Two structurally unrelated synthetic MMP inhibitors inhibited both proliferation and emigration of VSMC from rabbit aortic explant cultures (256). The effects on migration but not proliferation were also found in baboon saphenous artery explants (136). Studies with synthetic MMP inhibitors in the rat carotid balloon injury model showed a consistent effect on early SMC migration (23–25, 121, 312). Two of the five studies found an effect also on the first wave of medial (121, 312) and two on the second wave of intimal (23, 121) VSMC proliferation. However, except in one study (121), MMP inhibitors had no effect on the final intimal size (23, 24, 312), possibly because of late catch-up proliferation in the neointima. A likely explanation for this is redundancy between the MMP and other plasminogen-mediated proteolytic pathways (136). A study of balloon angioplasty to the atherosclerotic iliac arteries of Yucatan micropigs also showed no effect of MMP inhibition on intimal growth (68). However, MMP inhibition significantly antagonized constrictive remodeling and therefore led to a decrease in late lumen loss (68). In a recent, thorough preclinical study in atherosclerotic primates, a broad spectrum MMP inhibitor was infused into animals for 28 days to achieve inhibitory plasma concentrations. MMP inhibitor treatment failed to reduce neointima formation after either plain balloon angioplasty or angioplasty with stenting (54). In contrast to the study in pigs (68), no effect on constrictive remodeling was observed. In contrast to these largely negative results, it was very recently reported that a broad spectrum MMP inhibitor reduced neointima formation at the distal anastomoses of polytetrafluoroethylene arteriovenous shunts in pigs after 4 wk (230); longer follow up is clearly warranted to see whether the benefit is maintained.
Given that high concentrations of synthetic inhibitors might have effects other than on activity of MMPs, for example on mitogen-activated protein kinases (167), it was valuable to verify these results with TIMP gene transfer. Seeding of cells stably overexpressing TIMP-1 reduced neointima formation after balloon injury (82), as did adenovirus-mediated gene transfer of TIMP-1 or TIMP-2 (53, 72). TIMP-1, -2, and -3 gene transfer into human saphenous vein prevented neointima formation in organ cultures (92–94). For TIMP-1 and -2, this was achieved by inhibition of migration not proliferation (92, 93). The effects of TIMP-3 were more complex because it stimulated apoptosis as well as inhibiting migration (16). This may explain why TIMP-3 was more effective than TIMP-2 in inhibiting neointima formation in pig vein grafts in vivo (94). Interestingly, TIMP-3 also inhibits signaling through the vascular epidermal growth factor KDR receptor by a mechanism independent of MMP inhibition (218), although there is no direct evidence that this contributed to its effects on vein grafts. From these studies, catch up growth does not seem to compensate for the effects on neointima formation of TIMP gene transfer in contrast to those of synthetic MMP inhibitors, at least over the 2- to 4-wk time periods studied. The reasons for this are obscure but could involve any of the several effects of TIMPs that are unrelated to MMP inhibition (15, 218).
While synthetic inhibitors and TIMP gene transfer target a broad spectrum of MMPs, knockout studies have the potential to elucidate the role of individual MMPs. Knockout of MMP-11, the unusual MMP that degrades serum proteinase inhibitors (see Table 1), increases neointima formation after injury in mice (162). MMP-11 knockout was associated with a paradoxical increase in elastolysis and increased cellularity of the neointima rather than matrix accumulation (162). Two groups independently studied the effect of MMP-9 knockout on responses to filament loop injury (56) or arterial occlusion (86) in different background strains of mice. In wild-type mice, both groups observed early upregulation of MMP-9 and later activation of MMP-2 that was apparently unrelated to inflammation. MMP-9 knockout impaired neointima formation by inhibiting VSMC migration. One of the groups (56) also observed delayed inhibition of VSMC proliferation. Interestingly, MMP-9 knockout caused excess collagen accumulation and impaired outward remodeling in the occlusion model (86). More recently, two groups also found that knockout of the other gelatinase, MMP-2, also decreased neointima formation in the carotid ligation model and that this was associated with decreased VSMC migration in vitro (128, 144). Hence, MMP-2 and -9 appear to serve complementary rather than redundant functions in neointima formation, despite similar structure and substrate specificity and the compensatory upregulation of MMP-9 and downregulation of TIMP-1 expression evident in MMP-2 knockouts in one of the studies (144). The ability of MMP-9 but not MMP-2 to participate in CD44-mediated matrix attachment of VSMC and to organize collagen fibrils may account in part for the unique contribution of this MMP (128).
MMPs could facilitate VSMC migration by a variety of mechanisms. First, they could simply relieve the direct binding between basement membrane components and cell surface integrins that physically restrict movement. Second, MMPs could aid invasion of already freely moving VSMC by degrading the physical ECM barriers they encounter. Consistent with this, addition or gene transfer of MMPs promotes (184) and gene transfer of TIMPs inhibits (16) invasion of collagenase-isolated VSMC through synthetic matrix. Third, cleavage by MMPs could expose adhesion sites on matrix molecules over which VSMC could then gain traction. For example, Gavrilovic's group showed that cleavage of collagen I into one-fourth and three-fourth length fragments using MMP-13 produced a substrate upon which PDGF stimulated VSMC could migrate more rapidly than on intact collagen I (266). Moreover, adhesion and migration on cleaved collagen was mediated by newly induced v3-integrins, while migration on native collagen I was mediated by constitutive 21-integrins. This observation exemplifies a fourth mechanism, namely, that MMPs might modify the cell-to-matrix interactions of VSMC, which in turn changes their behavior by phenotypic modulation. Modulation of contractile VSMC towards a more synthetic phenotype accompanies neointima formation (42, 49, 274). During phenotypic modulation, basement membrane components, for example, laminin, are downregulated, while hyaluronic acid production (237, 293) and interstitial matrix components, monomeric type I collagen, elastin, and fibronectin are greatly increased (9, 273, 275) (Fig. 5). Matrix remodeling is of key importance because laminin retards or partially reverses phenotypic modulation, while fibronectin promotes it (110, 111). Hence, MMPs could promote phenotypic modulation by relieving interactions with inhibitory components or promoting interactions with modified stimulatory components. These effects of matrix proteins on migration are likely to be mediated by concomitant changes in expression of their cell surface integrin receptors (189) (Fig. 5).
|
(264). MMPs might also degrade matrix proteins with inhibitory effects on VSMC proliferation. For example, heparan sulfate proteoglycan (HSPG) components of the basement membrane, such as syndecans and perlecan, inhibit VSMC proliferation in culture and in vivo (29, 139). Hence, the ability of MMPs to degrade core proteins of HSPGs could underlie their effect on VSMC proliferation. Alternatively, basement membrane remodeling could permit new contacts with interstitial matrix components that promote cell cycle progression. Contractile VSMC in arteries exhibit very low rates of proliferation, most likely because they fail to downregulate cyclin-dependent kinase inhibitors (123, 271, 272). Components of the ECM of normal arteries, e.g., laminin and polymerized collagen, do not support downregulation of cyclin-dependent kinase inhibitors or proliferation of isolated VSMC (141). On the other hand, components of the remodeling matrix such as monomeric collagen and fibronectin support G1 cell cycle progression and VSMC proliferation (141, 244). Again, these actions of matrix proteins are believed to be mediated by binding to appropriate integrin receptors (244). Few studies have directly measured the influence of MMPs on matrix protein turnover in blood vessels. Strauss et al. (265) found that MMP inhibition decreased collagen breakdown after balloon injury to rabbit femoral arteries, as might be expected. However, collagen synthesis was also inhibited, presumably because MMP inhibition prevented VSMC transforming into the synthetic phenotype. The net effect on neointima formation was neutral (265). A reversible disappearance of basement membranes around VSMC and an increase in fibronectin synthesis were demonstrated during neointima formation in balloon-injured rat carotid arteries (275). These changes track the changes in MMP expression established in other studies (25, 311), but the influence of MMP inhibition on ECM proteins was not tested directly. To achieve this, we investigated the effect of TIMP-1 and -3 gene transfer on expression of basement membranes in human saphenous vein undergoing neointima formation in organ culture (1). Medial VSMC were normally surrounded by basement membranes, but a proportion became negative during culture. All VSMC that migrated into the neointima lacked basement membranes. TIMPs-1 and -3 reduced the numbers of medial VSMC lacking basement membranes and inhibited migration into the neointima. This implies that MMP-mediated loss of basement membrane is essential for medial to intimal migration of VSMC. Proliferating VSMC were more likely to be negative for basement membranes than nonproliferating VSMC. On the other hand, 30–40% of proliferating medial VSMC retained some basement membrane, which shows that complete basement membrane degradation is not necessary for VSMC proliferation. Furthermore, TIMPs did not affect either basement membrane degradation in proliferating VSMC or the rate of medial VSMC proliferation. Hence, MMPs are not essential for basement membrane degradation in proliferating VSMC, presumably because of redundancy with other proteases. These conclusions are summarized in Figure 6.
