Cardiac fibroblasts account for up to two-thirds of the total number of cells in the normal heart and are responsible for extracellular matrix homoeostasis. In vitro, type I collagen, the predominant myocardial collagen, stimulates proteolytic activation of constitutively secreted proMMP-2 (pro-matrix metalloproteinase-2). This occurs at the cell membrane and requires formation of a ternary complex with MT1-MMP (membrane-type-1 MMP) and TIMP-2 (tissue inhibitor of metalloproteinases-2). Following MI (myocardial infarction), normally quiescent fibroblasts initiate a wound healing response by transforming into a proliferative and invasive myofibroblast phenotype. Deprivation of oxygen to the myocardium is an inevitable consequence of MI; therefore this reparative event occurs under chronically hypoxic conditions. However, species and preparation variations can strongly influence fibroblast behaviour, which is an important consideration when selecting experimental models for provision of clinically useful information.
The adult heart consists of myocytes, connected and surrounded by an extracellular collagen matrix [ECM (extracellular matrix)] that maintains the heart's structural integrity and through mechanical coupling allows muscle fibre shortening to be transduced into volume and pressure changes during systole. Non-myocytes (fibroblasts, endothelial cells and smooth-muscle cells) reside within the ECM. Cardiac fibroblasts, which constitute two-thirds of total cells and 90% of non-myocytes, are responsible for regulating the synthesis and degradation of ECM components in the normal heart. The predominant collagen types in the myocardium are I (85%), III (11%) and V (3%).
Cardiac fibroblasts and adverse remodelling
Fibroblasts play a key role in the structural rearrangement (remodelling) of the myocardium which occurs in heart disease. After MI (myocardial infarction), normally quiescent fibroblasts transform into a proliferative and invasive myofibroblast phenotype, characterized by the co-expression of both smooth-muscle actin and vimentin . Invasion is accompanied by increased expression of specific matrix-degrading metalloproteinases [MMPs (matrix metalloproteinases)]. Activation of the gelatinases MMP-2 and MMP-9 allows degradation of the type IV collagen basement membrane surrounding necrotic myocytes prior to fibroblast invasion and collagen deposition. In a mouse model of MI, a broad-spectrum MMP inhibitor resulted in delayed infarct healing, leading to the suggestion that this was due to impaired myofibroblast migration to the infarct site .
MMPs and TIMPs (tissue inhibitor of metalloproteinases)
MMPs are a family of more than 23 zinc-containing enzymes with overlapping roles with respect to cleavage activity of matrix components, and are grouped according to their mechanism of activation. Soluble MMPs are secreted as latent proteins (zymogens) and undergo extracellular proteolytic cleavage to the active enzyme, whereas MT-MMPs (membrane-type MMPs) are activated intracellularly and incorporated in the cell membrane as active enzymes. TIMPs are the physiological inhibitors of MMPs which bind to the catalytic domain of both latent and active MMPs thus preventing enzymatic activity. There are four known types of TIMP (1–4) with varying tissue distribution and affinity for the different MMP subtypes.
Activation of MMP-2
MMP-2 (gelatinase A) is a 72 kDa type IV collagenase that, unlike other MMPs, is constitutively secreted by cardiac fibroblasts. It differs from other soluble MMPs in that it is resistant to activation by serine proteases. Although most MT-MMPs (MT1-, MT2-, MT3-, MT5- and MT6-MMPs) are able to activate proMMP-2, strong evidence supports a mechanism of cell-surface activation of MMP-2 via formation of a ternary complex with MT1-MMP and TIMP-2. MT1-MMP is synthesized as a proprotein, activated within the Golgi network by the proprotein convertase furin and then expressed on the cell membrane where it cleaves the latent 72 kDa proMMP-2 to an intermediate 68 kDa form, which autocatalytically converts into a 62 kDa active form (act.MMP-2) .
TIMP-2 is the major physiological inhibitor of MMP-2 and MT1-MMP and post-translationally regulates both proteins by binding and preventing proteolytic cleavage of the pro-enzyme, as well as function of the active enzyme . However, TIMP-2 is also essential for efficient activation of proMMP-2 at the cell membrane . MT1-MMP acts as a receptor for TIMP-2, which in turn binds to latent MMP-2, which is then presented to the active site of an adjacent MT1-MMP molecule (see Figure 1).
MMP-2 activation pathway
Involvement of MT1-MMP and MMP-2 in cellular invasion
Considerable evidence supports a role for MMP-2 and MT1-MMP in cellular invasion. MMP-2 activation and high levels of MT1-MMP are well correlated with tumour spread , and both MMP-2 and MT1-MMP have been localized to the migrating edge of invasive cells. Furthermore, transfection with MMP-2 and MT1-MMP promotes their invasive potential.
Collagen stimulation of MMP-2 activation
Collagen is the only known physiological stimulus for MMP-2 activation. Previous studies that have used non-physiological agents such as cytochalasin B, monensin and concanavalin A have been shown to activate MMP-2 via pathways distinct from that of collagen [6,7]. For example, the tyrosine kinase inhibitor, genistein, and the cyclo-oxygenase inhibitor, indomethacin, blocked monensin-induced MMP-2 activation by human dermal fibroblasts but had no effect on collagen-induced MMP-2 activation.
Fibroblasts cultured in the absence of collagen secrete only latent MMP-2. Fibrillar collagens including type I, II and III but not IV induce MMP-2 activation when fibroblasts are cultured in collagen gels . The addition of soluble type I collagen to culture medium of human skin fibroblasts has also been shown to stimulate MMP-2 activation, even at relatively low (12.5 μg/ml) concentrations . Collagen stimulation of MMP-2 activation therefore represents a more useful model for studying mechanisms of MMP-2 activation and its regulation in vivo. A number of cell types have been successfully cultured in three-dimensional collagen lattices in vitro. Fibroblasts cultured in these collagen gels demonstrate both morphological and phenotypic differences compared with cells cultured as monolayers . The ability of fibroblasts within these matrices to reorganize them into dense tissue-like structures has been used as an in vitro model to study wound healing and remodelling.
Collagen-induced MMP-2 activation in cardiac fibroblasts
Guo and Piacentini  studied collagen stimulation of MMP-2 activation in rat cardiac fibroblasts and demonstrated that cells cultured in a type I collagen lattice activated MMP-2 that was paralleled by up-regulation of MT1-MMP and TIMP-2. Cell membranes from collagen-stimulated cells contained active MT1-MMP and TIMP-2 and were able to activate exogenous proMMP-2. Inhibition of furin reduced proMT1-MMP processing and MMP-2 activation together with reducing fibroblast migration and invasion, thus indicating that MMP-2 activation was furin-dependent.
While studies of animal fibroblasts have furthered our understanding of both physiological and pathophysiological function, a number of studies have revealed phenotypic differences between human and non-human cultured cells. Furin inhibitor or furin siRNA (small interfering RNA) blocked MT1-MMP and MMP-2 activation in human uterine cervical fibroblasts but had no effect on rabbit dermal fibroblasts , suggesting that both furin-dependent and -independent pathways exist for MT1-MMP activation.
Few studies have been performed on fibroblasts derived from the human heart. A comparison of human and rabbit cardiac fibroblasts revealed significant differences in proliferation, collagen production and response to TGF-β (transforming growth factor-β) . Another study demonstrated that human cardiac fibroblasts express few angiotensin II receptors compared with rat cardiac fibroblasts . Such phenotypic differences may limit the use of non-human fibroblasts as suitable models for myocardial remodelling events in human.
Effect of hypoxia on MMP-2 activation and cellular invasion
Hypoxia is an inevitable consequence of MI; however, its effect on cardiac fibroblast function has not previously been investigated. Modulation of MMP-2 activation by hypoxia has revealed a marked species-, tissue- and cell-type variation. For example, in human endometrial stromal cells hypoxia induced a decrease in MT1-MMP and accompanying decrease in active MMP-2 although total MMP-2 remained unchanged . Conversely, in HT1080 cells, hypoxia decreased TIMP-2 expression, which subsequently led to a late increase in MMP-2 activation . Effects of hypoxia on invasion have been studied in the context of tumour dissemination where hypoxia was shown to correlate with increased metastasis and poor clinical outcome .
Given the phenotypic differences between human and animal cardiac fibroblasts, it is crucial that laboratory studies determine the regulation of MMP-2 activation in appropriate experimental models. Our own studies have explored the effect of hypoxia on primary cultures of human atrial cardiac myofibroblasts with respect to both MMP-2 activity and cellular function. Consistent with previous observations in other species, culture of these human cells with type I collagen stimulates MMP-2 activation. In contrast, hypoxic culture results in attenuation of MMP-2 activation, which is mediated by changes in membrane MT1-MMP expression. Using siRNA technology, we have demonstrated that myofibroblast invasion is dependent on both MT1-MMP and MMP-2, whereas cellular motility is an MMP-independent event. Consequently, hypoxia significantly impairs myofibroblast invasion but not migration. A thorough understanding of the mechanisms underlying the effects of hypoxia on MMP-2 activation and cellular function in these clinically relevant cells may have important consequences for early adaptive remodelling following MI.
Cardiovascular Bioscience: A Focus Topic at Life Sciences 2007, held at SECC Glasgow, U.K., 9–12 July 2007. Edited by S. Kennedy (Strathclyde, Glasgow, U.K.), M. Lloyd (Bath, U.K.) and C. Wainwright (Robert Gordon University, Aberdeen, U.K.).