Diverse functions of matrix metalloproteinases during fibrosis

1Fibrosis is a disease state that typically results from dysfunctional wound healing in response to tissue injury. Normal repair results in the near re-establishment of baseline levels and organization of extracellular matrix (ECM), whereas, in fibrosis, ECM production is excessive and uncontrolled. The main constituents of fibrotic lesions are interstitial collagens, such as type I and III, and excessive deposition of these durable fibers can result in disruption of proper tissue structure and function. Fibrotic lesions occur in most major organs and result in substantial morbidity and mortality.

2Both the normal and fibrotic response to tissue injury, as caused by trauma, environmental insults, infection, surgery, etc., involves multiple cellular and molecular players, and many of the mechanistic elements of these processes have been reviewed elsewhere. One class of molecules that is thought to be important in the maintenance of the ECM and processes of tissue repair are matrix metalloproteinases (MMPs). Here, we discuss data demonstrating or suggesting varied roles for specific MMPs in fibrosis of liver, lung and kidney derived from experimental models that utilize genetic deletion or transgenic overexpression of specific MMPs.

3MMPs are a family of extracellular endopeptidases defined by conserved pro-domains and catalytic domains. As a family, mammalian MMPs comprise 25 members that are either secreted or membrane-bound enzymes. Although MMPs have long been considered to be primarily responsible for turnover and degradation of ECM substrates, they are now recognized as being responsible for mediating crucial functions in a variety of processes, particularly related to immunity and repair, such as cell migration, leukocyte activation, antimicrobial defense, chemokine processing and more. Thus, although limited to secreted, extracellular or membrane proteins, the spectrum of MMP substrates goes well beyond ECM components. Furthermore, studies with gene-targeted mice have demonstrated rather convincingly that individual MMPs perform functions that do not overlap with the role of other MMPs. Although degradation studies with candidate substrates show that isolated MMPs have much redundancy in vitro, in vivo the functions of individual MMPs are limited, specific and apparently unique. An important caveat of in vitro degradation assays is that they show what an MMP can do, not what it does do in vivo. Indeed, as has been widely discussed, findings from in vitro studies on MMP catalysis have been poor predicators of their physiologic function. The diverse roles that MMPs seem to carry out in fibrotic conditions underscore this conundrum.

4If MMPs only act to degrade ECM and if ECM accumulates in fibrosis, then fibrosis would be, in part, a condition of deficient proteolysis. However, as we discuss in this Review, MMPs have both inhibitory and stimulatory roles in fibrosis (Table 1). Whereas some MMPs do indeed function to reduce fibrosis, others promote it. And, when identified, the mechanism of an MMP function in fibrosis is usually unrelated to direct proteolysis of ECM.

5Although we focus on function, control of MMP catalysis is also relevant to understanding how these proteases impact fibrosis or, for that matter, any disease or repair process. Compared with normal repair, the activity of an effector MMP might be over- or underrepresented in fibrosis, and changes in overall activity can be shaped by two mechanisms.

6First, biosynthesis of the MMP might be altered. Expression of most MMPs, including essentially all members discussed here, seems to be controlled at the level of transcription. Hence, over- or downregulation of an MMP would indicate that the effectors controlling gene expression in a given cell type are also altered.

7The second mechanism(s) that shape MMP activity are various controls over enzyme activity. These mechanisms include zymogen activation and compartmentalization of the active proteinase processes, and these concepts have been discussed in detail elsewhere. In addition, the catalytic activity of MMPs can be silenced by the four tissue inhibitors of metalloproteinases – TIMP-1, TIMP-2, TIMP-3 and TIMP-4 – among other mechanisms, such as endocytosis, oxidative modifications and other inhibitors. The proteinase-antiproteinase paradigm states that net MMP proteolysis in a tissue is the sum of total active MMPs minus inhibition by TIMPs. TIMPs inhibit MMPs with a precise stoichiometry (typically 1:1); however, there are only four TIMP proteins but over 20 MMPs, as well as many other metalloproteinases that can be blocked by TIMPs. Therefore, demonstrating a metalloproteinase imbalance – that is, an excess of metalloproteinase activity over what can be blocked by the available inhibitors – in a tissue homogenate cannot be achieved by examining a few members. However, changes in total metalloproteinase activity can be demonstrated. In discussing MMP function in models of fibrosis, we will also include relevant data on TIMPs.

8Development of fibrosis, as well as nonpathogenic normal scarring in response to wounding, can be broken down into four phases (irrespective of affected organ): (1) the precipitating injury or insult, (2) an inflammatory response, (3) activation and differentiation of resident fibroblasts and pericytes or other structural resident cells into α-smooth muscle actin (αSMA)-positive, interstitial-collagen-producing myofibroblasts (Box 1), and (4) tissue remodeling and resolution, i.e. the clearance of fibrotic matrix by extracellular proteolysis and/or endocytosis. In chronic fibrotic conditions, such as liver cirrhosis, idiopathic pulmonary fibrosis and chronic kidney disease, the resolution phase is defective. αSMA, an intracellular contractile protein, is an oft-used marker of myofibroblasts and as a surrogate marker of active collagen deposition and collagen-producing cells in fibrosis. However, it is not clear whether all αSMA-positive cells deposit collagen or if all collagen-producing cells are αSMA-positive.

9Box 1. EMT and αSMA: conflicting models

10Although studies in various organs of animal models have suggested that fibrogenic myofibroblasts can derive from epithelial cells that undergo epithelial-to-mesenchymal transition (EMT) and migrate into the interstitial compartment, these findings have been questioned by compelling lineage-tracing studies and other means. It is our opinion that the prevailing data indicate that fibrogenic myofibroblasts in lung, kidney and liver arise from resident interstitial cells (i.e. fibroblasts and pericytes). Myofibroblasts are typically identified by expression of α-smooth muscle actin (αSMA), an intracellular contractile protein. αSMA is an oft-used surrogate marker of active collagen deposition and collagen-producing cells in fibrosis. However, it is not clear whether all αSMA-positive cells deposit collagen or if all collagen-producing cells are αSMA-positive.

11Where and how do MMPs function in these phases? Although MMPs are thought to have primary roles in the maintenance of ECM, they also are involved in regulating the activity of a range of immunoregulatory molecules. MMPs can contribute to the degree of initial injury and repair, to the onset and resolution of inflammation, to the activation and de-activation of myofibroblasts, and to the deposition and breakdown of ECM. In other words, MMPs are involved in both augmenting and attenuating many processes that impact fibrosis. Related to the relevance of MMPs in fibrotic conditions, some groups have proposed that circulating levels of specific MMPs or possible MMP degradation products are reliable biomarkers of active fibrosis.

12Below, we discuss experimental data demonstrating or suggesting functional roles, both protective and detrimental, for specific MMPs in liver, lung and kidney fibrosis. Although there are many mechanisms for generating fibroses common among these organs, notably including the stable activation of myofibroblasts from resident interstitial cells and common immune features, it is important to note that the roles for specific MMPs are not necessarily the same among different organ systems (Table 1; Fig. 1).

13Chronic damage to hepatocytes by viral infection (hepatitis B or C) or excessive alcohol consumption can lead to liver cirrhosis, which consists of extensive fibrosis, loss of metabolic function and, eventually, death. Liver cirrhosis is a prevalent disease, responsible for over 1 million deaths worldwide in 2010. The common experimental models for studying liver fibrosis involve the administration of a chemical toxin, such as carbon tetrachloride (CCl4) or thioacetamide (TAA), to cause acute hepatocellular injury or by bile-duct ligation (BDL) to cause cholestasis-induced liver fibrosis. To cause chronic injury, CCl4 is given over a period of weeks to months, resulting in significant perivascular fibrosis. A major collagen-producing cell in the liver is the hepatic stellate cell (HSC). When injured, HSCs differentiate into myofibroblast-like cells expressing αSMA and interstitial collagens. By contrast, clearance of fibrotic scar tissue is associated with HSC apoptosis or de-differentiation of HSCs.

14A potential therapeutic strategy for fibrosis would be to induce expression of a collagenase that would degrade the fibrotic scar tissue, and this approach was demonstrated in a rat model of TAA-induced liver fibrosis, as well as in models of heart and muscle fibrosis. After the establishment of fibrosis, animals were infected with an adenovirus expressing human MMP-1 (also known as collagenase-1), which is not present in the rodent genome. In addition to reducing the fibrosis, overexpression of human MMP-1 increased hepatocyte proliferation yet led to a moderate amount of diverse tissue damage. These data are potentially important because there are no treatments on the market that reverse fibrosis.

15CCl4 and BDL studies with Mmp2−/− mice suggest that MMP-2 (also known as gelatinase-A) is anti-fibrotic; that is, fibrosis was greater in knockout mice compared with wild-type controls. However, MMP-2 does not seem to be anti-fibrotic by directly remodeling ECM and, indeed, in vitro studies have established that MMP-2 is not a collagenase. Notably, Radbill and colleagues reported that livers from wild-type and Mmp2−/− CCl4-treated mice have similar levels of collagenase activity and αSMA. These data suggest that breakdown of collagen and myofibroblast activation do not seem to be affected by MMP-2. Furthermore, proteins that regulate MMP-2 activity, such as MMP-14 (also known as MT1-MMP) and TIMP-2, were also produced at similar levels in livers from wild-type and Mmp2−/− CCl4-treated mice. What was dysregulated by loss of MMP-2 was expression of type I collagen. Compared with wild-type mice, expression of collagen α1 (ColIα1), the principal chain of type I collagen, was elevated in the fibrotic livers of Mmp2−/− mice. Furthermore, cell culture data demonstrate that exogenous MMP-2 suppresses collagen expression in HSC-like cells. Collectively, these data suggest that anti-fibrotic effects of MMP-2 predominately influence the degree of collagen deposition during activation of liver HSCs. It is unclear how MMP-2 mediates this effect, but a likely mechanism would be proteolysis of a surface protein that results in altered outside-in signaling.

16Similarly, Onozuka and colleagues reported increased expression of ColIα1 in CCl4-injured Mmp2−/− mice, as well as elevated levels of αSMA, TGFβ1, MMP-14, PDGF and TIMP-1. Although these data are not inconsistent with a role for MMP-2 in influencing collagen expression, they do suggest that other MMP-2-dependent pathways contribute to liver fibrosis. It is important to note that Onozuka and colleagues used a longer treatment regime (12 weeks) compared with the 6-week regime used by Radbill and colleagues. Therefore, the overexpression of collagen by HSCs in the absence of MMP-2 might drive increased HSC activation and subsequent dysregulation of other MMPs and TIMPs. Furthermore, increased ColIα1 expression was also observed in the BDL model, suggesting that MMP-2 plays a similar role in different forms of liver fibrosis.

17In vitro assays demonstrate that MMP-8 (also known as collagenase-2 or neutrophil collagenase) can cleave interstitial collagens types I and III, and, thus, expression of this proteinase would be expected to be anti-fibrotic by means of promoting collagen breakdown and deactivation of HSCs during the resolution phase. Similar to the MMP-1 transduction studies described above, in both rat CCl4 and BDL models of fibrosis, adenoviral transduction of MMP-8 into liver after the establishment of fibrosis resulted in reduction of fibrotic tissue. Mechanistically, it appears that MMP-8 is acting primarily by breaking down collagen, because expression of mRNAs for collagens I and III and transforming growth factor β1 (TGFβ1), an HSC activator and well-established profibrotic cytokine, was unchanged by addition of MMP-8. In addition to degradation of the fibrotic scar tissue, transduction of MMP-8 resulted in enhanced hepatocyte growth factor (HGF) expression and hepatocyte proliferation. These data are consistent with the tissue repair paradigm by which breakdown of interstitial collagens results in apoptosis of HSCs and promotes entry into the hepatocyte cell cycle. Siller-López and colleagues reported increased MMP-2 expression in CCl4-treated rats expressing transduced MMP-8, whereas there was no change in MMP-2 expression in their BDL model. Based on the Mmp2−/− mouse studies, it is possible that MMP-2 contributes to reducing fibrosis by downregulating collagen expression. How MMP-8 expression translates to increased MMP-2 expression in one injury model (CCl4) but not another (BDL) remains unclear. Because MMP-8 can also cleave the chemokines CXCL8 and LIX, resulting in enhanced chemoattractant activities, inflammatory roles for this MMP in fibrosis need to be considered.

18MMP-12 (macrophage metalloelastase) and MMP-13 (collagenase-3) are macrophage proteinases (although both are expressed by other cell types, too) that are upregulated in response to the T helper 2 (Th2) cytokines IL-4 and IL-13, but not Th1 cytokines, specifically IFNγ and TNFα. Although MMP-12 deficiency has yet to be tested in a CCl4 model of liver injury, one study has evaluated fibrosis induction by the parasite Schistosoma mansoni in Mmp12−/− mice. Infection by S. mansoni eggs results in a granuloma-associated fibrosis that, in contrast to injury-induced fibrosis, appears to be driven by the Th2 cytokine IL-13, with little-to-no role for the traditional fibrotic stimulant TGFβ1. Loss of MMP-12 production resulted in a modest reduction in granuloma-associated fibrosis at the peak of the response, but similar pathology was also observed at later time points, indicating that MMP-12 plays a modest role in generating fibrosis.

19Mechanistically, Madala and colleagues demonstrated that the loss of MMP-12 results in an overall increase in metalloproteinase activity in the liver associated with elevated levels of MMP-2, MMP-9 (gelatinase-B) and MMP-13. Although MMP-2 can suppress type I collagen expression, loss of MMP-12 does not seem to affect ColIα1 transcription. In addition, it is likely that elevated MMP-9 does not impact the altered fibrotic response in Mmp12−/− mice. A report from the same group using Mmp9−/− mice infected with S. mansoni demonstrated no change in fibrosis. Therefore, Madala and colleagues focused on MMP-13 and proposed a model in which IL-13-driven MMP-12 production promotes the expression of the decoy IL-13Rα2, which acts as a negative regulator of MMP-13. The loss of MMP-12 thus results in reduced production of the IL-13 decoy receptor with subsequent increase in IL-13 signaling and elevated MMP-13-mediated collagenase activity. Although data from Mmp13−/− mice are lacking, the authors demonstrated first that MMP-13 collagenolytic activity is increased and IL-13Rα2 expression reduced in Mmp12−/− livers, second that MMP-13 colocalizes with macrophages and myofibroblasts within granulomas, and third that the loss of IL-13Rα2 results in increased MMP-13 expression. Together, these data suggest a model whereby MMP-12 indirectly promotes fibrosis via suppression of MMP-13, an anti-fibrotic MMP.

20Much like the MMP-1 studies mentioned above, other groups have used gene therapy technology to assess whether MMP-13 can reduce scarring in CCl4-injured liver. The idea that MMP-13 is anti-fibrotic is supported by observations of persistent fibrosis in Mmp13−/− mice. But does MMP-13 indeed ameliorate fibrosis by clearing deposited collagen? Overexpression studies by Endo and colleagues suggest that MMP-13 promotes tissue remodeling by converting latent pro-HGF to the active cytokine, thus promoting hepatocyte proliferation and repair. Interestingly, elevated levels of MMP-2 and MMP-9 were detected in fibrotic livers overexpressing MMP-13. However, it is unlikely that the protective effect of overexpressing MMP-13 is due to MMP-2 and MMP-9 acting on collagen directly. Contrary to its apparent role in CCl4-mediated fibrosis, in the BDL model of injury, loss of MMP-13 had an overall protective effect consisting of reduced cytokine production, myofibroblast activation and collagen expression during the acute phase, and reduced fibrotic areas during the chronic phase. Because pathogenesis differs between the CCl4 and BDL models, it is not surprising that certain proteins can have opposing roles in these liver injury models. However, certain common responses stand out. For instance, the loss of MMP-13 in the BDL model results in reduced MMP-2 and MMP-9 expression, which is consistent with the overexpression studies that induced these two MMPs. Although not apparent in the CCl4 studies, data from Uchinami and colleagues is supportive of published work suggesting that MMPs – particularly MMP-2 – contribute to the proliferation of stellate cells.

21MMP-19 can degrade various ECM proteins in vitro and, thus, could be predicted to act in an anti-fibrotic manner by promoting breakdown of the ECM. However, liver fibrosis is reduced in Mmp19−/− mice compared with wild-type controls, suggesting that MMP-19 is pro-fibrotic. Consistent with the diverse functionality of MMPs, liver injury (as measured by serum alanine aminotransferase levels) is reduced in Mmp19−/− mice. Supporting the idea that MMP-19 promotes hepatocellular injury, Jirouskova and colleagues demonstrated impaired TGFβ1 signaling and reduced apoptosis in Mmp19−/− mice in response to CCl4 injury. Loss of MMP-19 led to increases in the expression levels of MMP-2 and MMP-13, although not all changes were statistically significant. Despite the lack of significance, the trends observed are consistent with the aforementioned anti-fibrotic roles of these MMPs. Furthermore, Jirouskova et al. uncovered a potential dual role for MMP-19. Although Mmp19−/− animals have reduced injury and fibrosis, they appear to resolve fibrosis at a rate slower than do wild-type controls. In wild-type mice, an increase in MMP-19 expression correlates with the resolution of fibrosis. Thus, MMP-19 seems to be pro-fibrotic in the early phases post-injury, and anti-fibrotic during resolution.

22Liver-specific overexpression of TIMP-1 leads to more severe fibrosis without a significant effect on collagen synthesis. Although these data support the parsimonious explanation that TIMP-1 promotes fibrosis by inhibiting the activity of metallocollagenases, such as MMP-13 and MMP-14, increased fibrosis is also seen in Timp1−/− mice. Conversely, mice with siRNA knockdown of TIMP-2 following CCl4 injury showed reduced HSC activation and collagen deposition, suggesting a pro-fibrotic function. These roles during chemical injury do not appear to apply during Th2-mediated injury because deficiency of TIMP-1 or TIMP-2 does not seem to affect liver fibrosis in response to S. mansoni infection. Although the role of TIMP-3 has not been assessed in the liver fibrosis models discussed, perisinusoidal fibrosis is seen in Timp3−/− mice with nonalcoholic fatty liver disease. Hence, as a group, the TIMPs seem to have divergent roles during liver fibrosis and do not appear to function strictly by blocking the matrix-degrading or, more specifically, the collagenolytic activity of metalloproteinases.

23Fibrotic diseases of the lung represent diverse etiologies, resulting in similar pathophysiology. Pulmonary fibrosis can result from exposure to allergens, infections, chemicals or radiation. Most often, the source of injury is unknown, hence leading to a diagnosis of idiopathic pulmonary fibrosis (IPF). IPF affects upwards of 200,000 people in the USA, with about 50,000 new cases each year and mortality occurring usually within 3 to 5 years of diagnosis. Current therapeutics have focused predominantly on limiting pulmonary inflammation, with little success. Experimental models, potential cellular etiological mechanisms and therapeutic approaches have been extensively reviewed.

24The most frequently used model for studying pulmonary fibrosis involves the use of the chemotherapeutic agent bleomycin. Administration of a single dose of bleomycin to the lungs results in extensive epithelial injury, permeability and a profound inflammatory response, followed by development of interstitial fibrosis. The fibrotic stage lasts for about 3 to 4 weeks and resolves thereafter. Although this model has been criticized for not mimicking non-resolving IPF in humans, it has been the backbone of many potential targets that are currently in clinical trial. As an alternative, a mouse model utilizing repeated doses of bleomycin has been proposed to more closely replicate pathological features of IPF in humans. Next, we discuss experimental data examining functional roles for specific MMPs in lung fibrosis.

25MMP-3 (stromelysin-1) can process latent TGFβ1 to its active form, predicting that this MMP is pro-fibrotic. Indeed, as reported by Yamashita and colleagues, fibrosis develops in lungs following ectopic adenoviral expression of MMP-3, and bleomycin-induced fibrosis is dramatically reduced in Mmp3−/− mice. However, unexpectedly, levels of active TGFβ1 were comparable between wild-type and Mmp3−/− mice. The authors suggested that MMP-3 promotes fibrosis by inducing epithelial cells to undergo an epithelial-mesenchymal transition (EMT) to generate cells that are myofibroblast-like in function. Although it has been suggested that EMT is a source of myofibroblasts in lung and other organs, lineage-tracing studies do not support the view that any type-I-collagen-producing cells in bleomycin-induced fibrosis arise from the epithelium (Box 1). Thus, how epithelial-derived MMP-3 functions to promote fibrosis remains an open question.

26A potential role for MMP-7 (matrilysin) in pulmonary fibrosis was initially identified through gene expression analysis of human fibrotic lung samples and was found to be highly upregulated compared with its expression in healthy tissue. Although Mmp7−/− mice are protected from bleomycin-induced acute lung injury, they develop extensive fibrosis, although slightly reduced from that in wild-type mice, suggesting a pro-fibrotic role for this protease. MMP-7 cleaves syndecan-1, which carries the chemokine CXCL1 as cargo on its glycosaminoglycan chains, from injured lung epithelial cells, and that release of the syndecan-1–CXCL1 complexes is required for the transepithelial infiltration of neutrophils. Demonstrating that MMP-7-dependent neutrophil infiltration contributes to development of fibrosis, Manicone and colleagues found that recovery of transepithelial neutrophil influx in Mmp7−/− mice – through instillation of the bacterial chemotactic peptide N-formylmethionyl-leucyl-phenylalanine) – restored fibrosis to wild-type levels. Thus, it is possible that MMP-7 promotes fibrosis indirectly by facilitating neutrophil influx and activation and, in turn, enhanced epithelial damage that leads to more fibrosis. However, MMP-7 also facilitates epithelial repair, also through shedding of syndecan-1, probably balancing to some degree any added injury from allowing neutrophils to transmigrate.

27In addition to a pro-fibrotic consequence due to an acute process, Manicone and colleagues proposed that MMP-7 also functions to help resolve fibrosis. MMP-7 sheds syndecan-1 during the first 5–7 days post-bleomycin treatment, yet the expression of MMP-7 persists and increases for up to 25 days. MMP-7 begins to shed the intact 80-kDa ectodomain of E-cadherin at about 10 days post bleomycin treatment. The E-cadherin ectodomain can ligate αEβ7 integrin (CD103), which is expressed on a subset of T cells and CD11chi dendritic cells. Manicone and colleagues found that fibrosis is elevated and persists in mice lacking the αE subunit (Itgae−/−), which only pairs with the β7 integrin subunit. They proposed that shedding of E-cadherin acts to home and activate immunosuppressive dendritic cells that aid in the resolution of fibrosis. Overall, the findings with MMP-7 indicate that it serves both pro- and anti-fibrotic functions and that these opposing functions are separated in time. Early on, MMP-7 functions in the innate response by facilitating neutrophil influx and activation, leading to collateral epithelial damage and an enhanced fibrotic environment. Then later, epithelial-derived MMP-7 promotes resolution by attracting an influx of immunosuppressive leukocytes.

28Two studies have reported reduced fibrosis in Mmp8−/− mice, indicating that this proteinase has pro-fibrotic functions. These findings are opposed to observations in the liver, where overexpression of MMP-8 led to improved resolution of fibrosis. The mechanisms explaining how MMP-8 promotes pulmonary fibrosis differs between the two studies. García-Prieto and colleagues observed a reduction of active TGFβ1 that they link to the loss of MMP-8-mediated inactivation of IL-10. Hence, their model predicts that MMP-8 blocks IL-10 signals, leading to enhanced fibroblast activation and collagen production. This idea, however, is counter to data demonstrating induction of lung fibrosis by overexpression of IL-10. By contrast, other investigators did not observe any differences in levels of active TGFβ1 or IL-10 in Mmp8−/− mice with bleomycin-induced fibrosis, although they did find increased levels of CXCL10 (IP-10) and CCL3 (MIP-1α). Furthermore, when they crossed Mmp8−/− mice with Cxcl10−/− or Ccl3−/− mice, fibrosis was restored to wild-type levels. Although these data are consistent with the proposed pro-fibrotic role for MMP-8, further data are needed to uncover the underlying mechanism. However, the data presented do suggest that MMP-8 influences multiple processes by modulating cellular inflammation, cytokine inactivation and/or fibroblast activation.

29Like MMP-3, MMP-9 can activate latent TGFβ1. This activity, along with expression levels that correlate positively with experimental fibrosis, suggests a pro-fibrotic role for MMP-9. However, similar to observations in a model of liver fibrosis, bleomycin-induced pulmonary fibrosis does not differ between wild-type and Mmp9−/− mice. By contrast, bleomycin-induced fibrosis is reduced in transgenic mice constitutively expressing human MMP-9 in macrophages under control of the promoter for scavenger receptor class A. Cabrera and colleagues linked MMP-9-mediated attenuation of fibrosis to reduced insulin growth factor binding protein-3 (IGFBP-3), which normally sequesters IGFs and prevents cell growth. They hypothesized that with reduced IGFBP-3 there is increased recovery from injury and less cell death that would potentially drive fibrosis.

30Overexpression of an MMP might lead to unexpected or unnatural functions (see Box 2). It could be that excess MMP-9 acts directly on lung ECM, as was proposed for the destructive consequence of overexpression of MMP-9 in a model of atherosclerosis. In a model of IL-13-driven fibrosis, lack of MMP-9 results in reduced fibrosis, which was linked to activation of TGFβ1. It is interesting to note that the IL-13 lung-fibrosis model is dependent on TGFβ1, whereas other IL-13-dependent models of fibrosis (e.g. S. mansoni) are not.

31Box 2. Caveats of overexpressing MMPs to uncover function

32As discussed in the text, functions for some MMPs, such as MMP-1 and MMP-9, in fibrosis have been suggested from studies that involved overexpression – typically by viral transduction – of the enzyme. There are two important caveats associated with this type of experimental strategy, caveats that are relevant for many types of proteins. First, when overexpressed, an MMP might act on non-physiological substrates, mediating misleading responses. Owing to excess levels of the enzyme, the proteinase:substrate stoichiometry becomes skewed in favor of the enzyme. Because MMPs lack inherent substrate specificity (unlike, for example, coagulation cascade proteinases), their activity in vivo is likely to be limited to specific proteins by allosteric interactions that confine the proteinase to pericellular niches and, in turn, substrates. An excess of a given MMP would allow some portion of the enzyme produced to escape these regulatory mechanisms. The concept that MMP activity is confined by interactions with other extracellular macromolecules has been discussed elsewhere. The second caveat with overexpression is that – unless cell-specific promoters are used – the enzyme can be produced ectopically. That is, the MMP might end up being made by a cell type that does not make the enzyme naturally, leading to the strong potential of acting on non-physiological substrates. In addition, the cell in which the MMP is ectopically expressed could lack the allosteric control mechanisms discussed under the first caveat, thereby adding the potential artefactual responses. Overexpression approaches demonstrate what an MMP can do, not what it does do. On the positive side, overexpression might point to potential therapeutic strategies, as suggested by the overexpression of MMP-1 in mouse models of fibrosis.

33It seems that, as appears to be the case for many MMPs, the jury is still out on the nature of the role for MMP-9 in fibrosis. Thus, MMP-9 expression resulted in less fibrosis in one study, whereas in another it was pro-fibrotic, while yet others concluded it had no role. In models of allergic lung inflammation, MMP-9 has been shown to have various functions related to chemokine levels and leukocyte influx, and other studies have demonstrated an ability to cleave and alter the activity of various chemokines. Clearly, assessing immunomodulatory roles for MMP-9 in fibrosis needs to be examined in further depth.

34The role of MMP-12 during lung fibrosis has been studied in a number of different injury models, most of which attribute its role as being pro-fibrotic. With the bleomycin model, there are two conflicting studies. Whereas one group observed reduced fibrosis in Mmp12−/− animals compared with wild-type controls, another reported no difference in fibrosis between the two genotypes. However, the discrepancy between these two reports might be explained by the time point at which fibrosis was assessed. Manoury and colleagues assessed fibrosis at 2 weeks post-bleomycin, whereas Kang and colleagues looked later, at 3 weeks. After 2 weeks of bleomycin treatment, excessive collagen deposition is still barely evident (increased collagen expression can be seen by day 7). Hence, an MMP-12-driven effect might not manifest until fibrosis is more advanced. Kang and colleagues also demonstrated a similar pro-fibrotic role for MMP-12 in transgenic mice overexpressing TGFβ1. The authors stated – but did not show – that inflammation did not differ between Mmp12−/− and wild-type mice, suggesting that MMP-12 influences downstream processes such as fibroblast activation or collagen production.

35Pro-fibrotic roles for MMP-12 were also found in models of pulmonary fibrosis generated by anti-Fas antibody (Jo2), which targets the tumor necrosis factor receptor, and infection with S. mansoni. As with bleomycin studies, MMP-12 deficiency did not affect inflammation during anti-Fas-antibody-mediated injury. However, gene expression analysis of Jo2-treated Mmp12−/− mice indicated reduced expression of genes implicated in fibroblast migration and proliferation, suggesting that MMP-12 contributes to fibroblast activation. Observations from animals infected with S. mansoni suggest that MMP-12 promotes fibrosis indirectly by suppressing expression of other anti-fibrotic MMPs.

36Contrary to previous studies suggesting a pro-fibrotic role, two contrasting studies, one using a fibrosis model with transgenic mice expressing IL-13 and the other a model involving radiation injury followed by bone-marrow transfer (so-called ‘RAD+BMT’), observed either no effect or an anti-fibrotic effect of MMP-12, respectively. Although a direct mechanistic link to fibrosis is unclear, both studies reported altered expression of other MMPs in Mmp12−/− mice, similar to that found in the S. mansoni model.

37Two conflicting studies addressed the role of MMP-13 during injury-induced lung fibrosis. One study by Flechsig and colleagues observed that, after radiation-induced injury, Mmp13−/− mice had reduced fibrosis coupled with decreased inflammation. However, Sen et al., using a hyperoxia injury model, reported that the extent of fibrosis did not differ between Mmp13−/− and wild-type mice. They did, however, find that inflammation was elevated in Mmp13−/− mice, and in vitro data demonstrated that MMP-13 can cleave and inactivate monocyte chemokine CCL2 (MCP-1). Other studies have demonstrated that MMP-13-generated cleavage of specific chemokines – CCL7 and CXCL12 – leads to reduced chemotactic activity. Many MMPs, including MMP-13, are known to process various chemokines, resulting in either enhanced, diminished or altered activity. Because cellular inflammation is believed to be a key element in the development of fibrosis, the regulation of leukocyte chemotaxis by MMP-mediated chemokine modification could be a significant contributing factor to disease severity.

38MMP-19 was identified as a potential target by a gene expression array that showed that MMP-19 was strongly upregulated in hyperplastic epithelial cells of IPF patients compared with epithelial cells of normal lung tissue. However, evaluation of fibrosis in Mmp19−/− mice showed increased hydroxyproline and αSMA staining compared with the levels in wild-type animals. Thus, the overexpression of MMP-19 in IPF patient lungs could be indicative of a reparative response. Yu and colleagues proposed that MMP-19 exerts an anti-fibrotic effect by inducing prostaglandin G/H synthase 2 (PTGS2; also known as COX2), which generates prostaglandin E2 (PGE2). PGE2 suppresses fibrosis by inhibiting fibroblast migration, proliferation, collagen synthesis and differentiation into myofibroblasts. This hypothesis was supported by the loss of PTGS2 expression in bleomycin-treated Mmp19−/− mice and the induction of PTGS2 by A549 cells overexpressing recombinant MMP-19.

39TIMPs have not been extensively studied in models of lung fibrosis. In contrast to its potential protective role in liver fibrosis, TIMP-1 does not impact the development or extent of lung fibrosis in response to bleomycin. However, it has been reported that fibrosis is elevated and persists in bleomycin-injured Timp3−/− lungs. Interestingly, enhanced fibrosis occurred in Timp3−/− lungs despite an overall increase in metalloproteinase activity, further questioning the dogma that MMPs function to degrade or clear ECM. Inflammation is increased in both bleomycin-injured Timp1−/− and Timp3−/− mice, hence supporting the concept that MMPs function predominately to control immune processes rather than directly promoting or attenuating fibrotic pathways.

40Fibrosis is a complication of chronic kidney disease (CKD), which eventually progresses towards end-stage renal disease (ESRD). Like other fibrotic diseases, treatments are limited. For those with severe disease, treatment involves dialysis and kidney transplants. In 2009, there were about 90,000 individuals in the USA who died owing to ESRD. This disease is predominately associated with diabetics and the elderly, with ~26% of the population over age 60 having stage 3 CKD (moderately reduced kidney function). Cellular and molecular pathways involved in the development of renal fibrosis have been reviewed elsewhere.

41Experimental models of renal fibrosis include ureteral obstruction nephropathy, diabetic nephropathy, remnant kidney and chronic allograph nephropathy. Most studies that have addressed MMP function have used the unilateral ureteral obstruction (UUO) model, which involves the ligation of one ureter while leaving the contralateral one patent as a control. Fibrosis develops within the first 7 days, and the kidney reaches end-stage disease after 2 weeks.

42One study evaluated renal fibrosis in Mmp2−/− mice using a diabetic nephropathy model initiated by streptozotocin (STZ), a pancreatic β-cell toxin. Development of fibrosis in this model occurs after 3 months, even though increased activation of fibroblasts is detectable after a couple of weeks. Similar to increased liver fibrosis in Mmp2−/− mice, Takamiya and colleagues demonstrated increased collagen deposition and fibroblast activation – as indicated by αSMA production – in diabetic Mmp2−/− mice, supporting the view that MMP-2 has an anti-fibrotic role. Furthermore, the levels of ColIα1 mRNA did not differ between wild-type and Mmp2−/− diabetic mice, suggesting that increased fibrosis in the null mice was due, in part, to impaired degradation of collagen. This effect differs from the liver studies, where the loss of MMP-2 impacted ColIα1 transcription. As mentioned above, MMP-2 is not a collagenase and therefore is probably not directly degrading collagen. However, it could help facilitate the clearance of fibrosis by acting on other ECM components. Alternatively, because the expression of other MMPs is altered in Mmp2−/− mice, MMP-2 might regulate the severity of renal fibrosis indirectly by promoting expression of other collagen-degrading proteases.

43As discussed earlier, MMP-9 is a potential activator of latent TGFβ1 and, hence, might be a pro-fibrotic mediator. Indeed, data from a UUO model showed reduced fibrosis in Mmp9−/− mice. However, this reduction in fibrosis was not accompanied by any significant change in active TGFβ1 or TGFβ receptor production between wild-type and Mmp9−/− mice. Wang et al. observed a significant decrease in αSMA levels in Mmp9−/− mice, suggesting a role for MMP-9 in myofibroblast activation or survival. Thus, in contrast to its overall pro-fibrotic role in liver and lung, MMP-12 does not seem to impact the severity of fibrosis during UUO.

44Two studies evaluating induction of kidney fibrosis by either UUO or protein overload show that the loss of TIMP-1 does not affect disease severity, similar to the lack of an effect on lung fibrosis but in contrast to its anti-fibrotic role in liver. Similar to lung, kidney fibrosis is enhanced in Timp3−/− mice subjected to UUO for 2 weeks. The authors suggest that the protective effect of TIMP-3 is due to suppression of TNFα, which mediates renal fibrosis and regulates expression of several MMPs. These data are in contrast to an earlier study that found no difference in renal fibrosis between wild-type and Timp3−/− mice after 7 days of UUO. However, they did detect increases in metalloproteinase activity and TGFβ1 expression in Timp3−/− mice, suggesting that, as in lung tissue, TIMP-3 might serve a protective function by suppressing cytokines involved in myofibroblast activation.

45MMPs modulate a range of biological processes, especially processes related to immunity, tissue repair and tissue remodeling. Although we do not yet know precisely how MMPs function during fibrosis – that is, the identity of the protein substrate(s) an individual MMP acts on to effect a specific process – we can nevertheless discern important points that could be relevant to future studies and potential targeting of MMPs for clinical therapies. Together, the studies discussed above demonstrate that MMP-dependent functions during fibrosis are not limited to effects on ECM turnover, but rather that these proteinases influence cellular proliferation and survival, gene expression, and multiple aspects of inflammation, and can directly and indirectly influence the activation of myofibroblasts.

46As is apparent by the information presented here and summarized in Table 1 and Fig. 1, MMPs have the potential to influence many biological processes during the pathogenesis of tissue fibrosis. To understand better the mechanisms of MMP function during fibrosis, we propose three paramount issues for future experimentation to facilitate more-targeted approaches to specific processes during fibrogenesis. First, we need to identify the cell types expressing relevant MMPs and assess their roles. Although cellular sources of many MMPs are known, expanded use of conditional gene targeting strategies will provide important insights into functional roles during disease states. The establishment of floxed gene repositories, such as the knockout mouse project (KOMP; www.komp.org), facilitates the ability to perform targeted gene-deletion experiments. Second, it is difficult to surmise biological mechanisms without a clear understanding of what protein substrates individual MMPs cleave in vivo. Although the identities of many in vitro substrates are known, novel methods are needed to experimentally ascertain what targets MMPs are cleaving during injury and repair. Finally, because MMPs can affect production of other family members, future studies need to track production of each of the functionally relevant family members during gene-targeted deletion studies. By evaluating all relevant MMPs together, we will be in a much stronger position to make future well-judged pharmaceutical interventions in these debilitating disorders.