|
|
||||||||
Published online before print August 20, 2007
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Departments of Molecular Biomedical Research, VIB, and Molecular Biology, Ghent University, Ghent, Belgium
2Correspondence: VIB, Ghent University, Technologiepark 927, Ghent (Zwijnaarde), East-Flanders 9052, Belgium. E-mail: claude.libert{at}dmbr.ugent.be
| ABSTRACT |
|---|
|
|
|---|
Key Words: extracellular matrix ELR
| INTRODUCTION |
|---|
|
|
|---|
Inflammatory conditions are almost always characterized by deregulated, often increased MMP activities [5 ]. The central question is whether these MMPs can influence the outcome of inflammation and if so, how they do so. Only by understanding the mechanism by which MMPs exert their function can they develop into effective drug targets. Indeed, ignorance of the relevant in vivo substrates is one of the main reasons for the disappointing and often unexpected results of trials using MMP inhibitors. Identifying the functions of specific MMPs in specific pathologies should therefore become a major goal. Initially, knowledge of MMP substrates was based on in vitro experiments, in which purified, activated MMPs were incubated with specific substrates. Although these data can be informative, the question remains whether processing of a purified substrate under optimal conditions implies that the same protein is a physiologically relevant MMP substrate in vivo. In this respect, the use of transgene and knockout technologies has meant a giant leap forward, as it enables the testing of MMP functions in vivo.
Apart from occasional, subtle developmental differences, such as a transient delay in myelination observed in MMP-9 and MMP-12 knockout mice [6 ], all MMP-deficient animals have an overall normal development and are viable. A notable exception is the MT1-MMP (MMP-14) knockout, which displays severe skeletal deformations and dies shortly after birth [7 , 8 ]. However, many MMP knockouts challenged by injury, inflammatory stimulus, infection, or cancer reveal interesting phenotypes. Indeed, numerous studies have reported that MMPs play roles in modulating inflammatory reactions by acting at different levels: leukocyte recruitment, alteration of the functions of cytokines and proteases, and clearance of the pathogen [5 ].
| MMPs AND LEUKOCYTE RECRUITMENT |
|---|
|
|
|---|
-chemokines), in which the cysteines are separated by a variable amino acid; CX3C chemokines (or
-chemokines) have three variable amino acids between the two cysteines; and C chemokines (or
-chemokines) have only one cysteine at the N terminus [9
]. Another important structural property of chemokines is the presence or absence of a specific tripeptide sequence Glu-Leu-Arg (ELR), which is important in the interaction with CXCR1 and CXCR2 [10
].
Chemokine processing by MMPs
As shown in Table 1
, MMP-mediated proteolysis of chemokines is one way by which MMPs can influence leukocyte trafficking. MMP proteolysis can affect the biological functions of chemokines in different ways. First, the proteolysis might inactivate the chemokine. Second, processing might generate antagonistic derivatives, which can still bind to the chemokine receptor but cannot promote chemotaxis. Third, the truncated chemokine is a more powerful chemotactic agent. Whatever the outcome, chemokine processing undoubtedly affects the progression of an inflammatory reaction and influences the type of cells, which are recruited and activated.
|
MMP-9 also inactivates CXCL chemokines. For instance, CXCL4 (platelet factor 4), a ELR-negative CXC chemokine, and CXCL1 (growth-related oncogene-
), a ELR motif containing CXC chemokine [14
], are proteolytically inactivated by MMP-9. This MMP also degraded CTAPIII, which is the NH2-terminally extended, inactive precursor of CXCL7 (neutrophil-activating peptide-2 [14
]). CXCL5 [epithelial-derived neutrophil-activating factor-78 (ENA-78)] is also cleaved by gelatinase B at different sites, resulting in transient potentiation of this chemokine, but eventually leading to its inactivation [15
]. Finally, CXCL9 (monokine induced by IFN-
) and CXCL10 (IFN-inducible protein 10) are processed C-terminally by MMP-8 and MMP-9 [16
]. At least in the case of CXCL9, this decreases chemotactic ability drastically and might intervene with association with the ECM [17
].
As mentioned earlier, several inactivated chemokines are still capable of binding to their receptors and therefore, act as functional antagonists, preventing the activity of other noncleaved chemokines. MMP-2, for instance, was shown to process CCL7 (also known as MCP-3) into an antagonistic form [18 ]. Other MMPs are also capable of cleaving MCPs to generate chemotactic antagonists. In addition to MMP-2, CCL7 is cleaved efficiently by MMP-1, MMP-3, MMP-13, and MMP-14. The closely related chemokines CCL2 (MCP-1) and CCL13 (MCP-4) could also be cleaved by MMP-1 and MMP-3, the latter of which also cleaved CCL8 (MCP-2) [19 ]. Finally, MMP-8 also processed CCL2, although inefficiently. It is interesting that all the truncated forms generated were shown to function as inflammatory antagonists when administered in vivo. This means that MMPs can have anti-inflammatory effects by dampening the action of chemokines.
In contrast to these examples, MMPs have also been shown to increase the biological activity of chemokines. Indeed, MMP-9 was shown to process CXCL8 (IL-8), which led to a significant increase in its chemotactic activity [14 ]. MMP-8, MMP-13, and MMP-14 were later also shown to generate a truncated IL-8 species with increased activity [20 , 21 ]. Moreover, another human CXCR2 ligand, CXCL5 (ENA-78), was found recently to be activated by MMP-8 cleavage [21 ].
The rodent chemokine LIX, which is considered to be the sole murine counterpart of two closely related human chemokines, namely CXCL5 (ENA-78) and CXCL6 [granulocyte chemotactic protein 2 (GCP-2)], was found to be processed N-terminally by MMP-9, with a consequent twofold increase in its biological activity [15 ]. MMP-8 was later shown to cleave LIX N- and C-terminally, generating truncated forms with increased chemotactic activities [22 ]. Recently, MMP-1, MMP-2, and MMP-13 were also shown to enhance the biological function of LIX by processing it N-terminally [21 ]. These observations of MMP-mediated LIX activation can have a serious impact on the course of several neutrophil-driven pathologies, as mouse LIX is believed to play the same role as IL-8 in humans. Indeed, compared with humans, rodents have few chemokines. For instance, the mouse has only four ELR-containing CXC chemokines, which specifically target polymorphonuclear cells: LIX, keratinocyte-derived chemokine (KC), dendritic cell (DC) inflammatory protein-1, and MIP-2, of which only LIX was found to be processed by MMPs [21 ].
Finally, it is noteworthy that cleavage does not always alter the activities of chemokines. An example of this is the N-terminal cleavage of CXCL6 (GCP-2) by MMP-8 and MMP-9, which does not affect its biological activity [15 ].
MMPs modulating chemokine gradients
In vivo, however, instead of being presented to proteases as soluble proteins, chemokines are immobilized mostly on the ECM or cell surface by binding to glycosaminoglycans (GAGs) through positively charged domains. This binding to GAGs seems to be important for chemokines to exert their role effectively. IL-8 binding to heparan sulfate leads to structural stabilization of the dimeric form of the chemokine, resulting in an extended biological activity and enhanced neutrophil responses [23
]. Some chemokines with mutations in the GAG-binding sites lose the ability to recruit cells in vivo, but they retain receptor-signaling activity in vitro [24
]. This might mean that MMPs mediate the function of chemokines indirectly by releasing chemokines bound to the cell surface or ECM.
The best example of this comes from a study describing the role of MMP-7 in a bleomycin-induced model of lung inflammation [25 ]. In response to the mucosal damage, epithelial cells release KC, which upon secretion, binds to syndecan-1, a cell-bound heparan sulfate proteoglycan (HSPG). It was already known that a tissue inhibitor of metalloproteinase 3 (TIMP-3)-sensitive protease could mediate the cleavage of the intact syndecan-1 and syndecan-4 ectodomains (shedding), thereby converting the cell-surface molecules into soluble effectors [26 ]. Indeed, together with CXCL1, MMP-7 is produced by these epithelial cells and cleaves the ectodomain of syndecan-1, thereby releasing the syndecan-1/KC complex. This creates a chemokine gradient, which triggers neutrophil infiltration into the alveolar space. As seen after bleomycin injury, neutrophils of the MMP-7-deficient animals do extravasate, but in contrast to their wild-type counterparts, they do not enter the lumen of the lung. Instead, they are confined to an expanded perivascular space. As a consequence, survival of MMP-7 knockouts was much better than that of the wild-types. It is likely that MMP-7 itself first binds to syndecan-1, as MMP-7 has been known to interact with GAGs, which markedly enhanced the activity of MMP-7 [27 ]. Therefore, interaction with the GAG side-chains of syndecan-1 would make it easier for MMP-7 to cleave the core protein of this HSPG.
Other experiments have confirmed that MMP-7 is indeed needed to generate a chemokine gradient, rather than being indispensable for neutrophil migration as such. Instillation of a bacterial chemotactic protein in the lungs of MMP-7-deficient mice did trigger neutrophils to infiltrate the luminal tissue, and the outcome was worse in these animals than in their wild-type counterparts [25 ].
Syndecan shedding is not restricted to MMP-7 and can be performed by various proteases, including other MMPs. MMP-9 can shed syndecan-1 and syndecan-4 [28 ], and MT1-MMP and MT3-MMP can release the ectodomain of syndecan-1 [29 ]. As chemokines, as well as many cytokines and antimicrobial substances, have the ability to bind to GAGs, this protease-controlled release of syndecans can affect the progression of different pathologies.
Besides the study describing the involvement of MMP-7 in generating a CXCL1 chemokine gradient, other studies report involvement of MMPs in establishing chemokine gradients. Our own research has shown that in the liver, following TNF-induced hepatitis, MMP-8 appears to be indispensable for the release of yet another neutrophil chemoattractant, LIX [30
]. As a result, neutrophil influx is seriously impaired in mice lacking a functional Mmp8 gene, which improves survival dramatically. Moreover, in a model of allergen-induced asthma, leukocytes accumulated massively in the lung parenchyma but in contrast to wild-type animals, failed to reach the airway lumen in MMP-2-deficient animals [31
]. This was accompanied by a large reduction in the levels of CCL11 (eotaxin), a potent eosinophil chemoattractant, in bronchoalveolar lavage (BAL) fluids of MMP-2 knockouts. The same group extended the study by applying the model to MMP-9 knockouts and MMP-2/9 double-knockouts [32
]. In contrast to the results obtained with the MMP-2 knockouts, in which the reduction in luminal leukocytes could be attributed entirely to reduced eosinophil numbers, MMP-9- and MMP-2/9-deficient animals had fewer eosinophils and neutrophils in their BAL. Furthermore, although absence of MMP-2 affected only CCL11 levels, lack of MMP-9 led to significantly lower levels of CCL11, CCL7 (MCP-3), and CCL17 (thymus and activation-regulated chemokine). Unfortunately, the mechanism by which both gelatinases modulate levels of these chemokines in BAL is still unclear. The fact that MMP-9 seems to promote neutrophil migration in this model is interesting in view of another study by Stefanidakis et al. [33
]. They found that MMP-9 forms a complex with the
Mβ2 integrin in the intracellular granules of neutrophils and that this complex becomes localized to the cell surface upon cellular activation. Furthermore, blocking the interaction between MMP-9 and
Mβ2 inhibits leukocyte migration in vitro as in vivo, suggesting that integrin association might help MMPs in promoting cell migration. Another study, using a similar asthma model, also showed a marked reduction in CCL17 BAL levels in MMP-9 knockout mice [34
]. However, this study did not report a difference in the number of recruited eosinophils and neutrophils [35
] but did show fewer DCs migrating into the airway lumen in the absence of MMP-9 [34
].
Furthermore, chondrocyte-derived MMP-3 generated an unidentified macrophage chemotactic factor, which was required for disc degradation in a model of herniated disc resorption [36 ]. MMP-3-deficient mice also showed reduced neutrophil recruitment to the airway lumen in a model of IgG-induced acute lung injury, as well as decreased lung injury [37 ]. The mechanism by which MMP-3 influences neutrophil migration is unknown. However, as neutrophils are not known to produce MMP-3, it is likely that this protease facilitates neutrophil migration indirectly, for instance, by generating a chemokine gradient.
MMP-12, which is predominantly a macrophage protease, is required for the influx of these cells in a model of smoke-induced emphysema [38 ]. As additional instillation of CCL2 resulted in macrophage migration comparable with that induced in wild-type animals by exposure to cigarette smoke alone, MMP-12 seems important in establishing a chemotactic gradient, rather than in macrophage migration as such. Later, the chemotactic proteins responsible for this phenotype were identified as being elastin fragments (EFs) [39 ]. Although it has been long known that EFs are chemotactic for monocytes in vitro [40 , 41 ], the study by Houghton et al. [39] is the first report that shows that MMP-generated elastin fragments can drive the progression of a disease. Furthermore, neutrophil migration too seems to be driven partially by the release of cryptic ECM fragments. Recently, it was shown that pulmonary ECM proteolysis, following exposure to a bacterial component, gives rise to a tripeptide chemoattractant, which promotes neutrophil recruitment [42 ].
These studies are important, as they show that ECM breakdown can promote chemotaxis, not only by releasing ECM-bound chemotactic factors but also by exposing cryptic ECM sites possessing chemotactic properties. Indeed, after so many chemotactic chemokine family members had been identified, the chemotactic properties of several ECM-derived fragments were believed to be, at best, of minimal importance in vivo. Therefore, these data might ask for a re-evaluation of the in vivo relevance of cryptic sites of several ECM components, such as fibronectin, collagen, and laminin [43 44 45 46 ].
Finally, nonmatrix proteins too, following proteolysis, can give rise to unexpected chemotaxis-promoting fragments. An example of this is the MMP-12-mediated cleavage of the
1-proteinase inhibitor, a serine proteinase inhibitor processed by several MMPs, which generates a fragment that promotes chemotaxis of neutrophils [47
].
MMPs and mobilization of hematopoietic progenitor cells (HPCs)
The influence of MMPs on leukocyte migration is not limited to the fate of mature circulating leukocytes but also affects the trafficking of HPCs from the bone marrow into circulation. Administrating anti-MMP-9 antibodies to rhesus monkeys totally inhibited the IL-8-induced mobilization of HPCs [48
]. Although a similar role for MMP-9 in IL-8-mediated HPC release could not be shown in mice [49
], MMP-9 was shown to be involved in hematopoietic recovery after depletion of hematopoietic cells [50
]. This could be explained by the role MMP-9 plays in shedding of the membrane-bound kit ligand, also known as the stem cell factor. Another study indicated a possible role for MMP-2 in releasing proteoglycan-bound CXCL12 from the surface of bone marrow stromal cells, thereby promoting pro-B cell migration [51
].
It should be noted, however, that the influence of MMPs on leukocyte trafficking is not merely restricted to chemokine processing and release. Indeed, by degrading a variety of proteins, which constitute interstitial ECMs, MMPs help leukocytes cross otherwise impassable basement membranes, such as the blood-brain barrier. For instance, MMP-2 and MMP-9 were shown to degrade dystroglycan, a critical component of the blood-brain barrier, thereby compromising its integrity and allowing leukocyte trafficking into the CNS during experimental autoimmune encephalomyelitis [52 ].
| CYTOKINE PROCESSING BY MMPs |
|---|
|
|
|---|
-converting enzyme (TACE) suggests, the main TNF sheddase. The release of active TNF is reduced by 90% in cells from ADAM-17 knockouts, indicating that ADAM-17 is the main TACE in vivo [55
]. However, it seems that in specific cellular settings, ADAM-17-independent release of TNF can become important. In a model of macrophage-mediated herniated disc resorption, macrophage MMP-7 was found to be indispensable for TNF shedding [56
]. As a result, MMP-7-deficient macrophages were unable to infiltrate the disc. Apart from MMP-7, also MMP-1, MMP-2, MMP-3, MMP-9, MMP-12, MMP-14, and MMP-17 were shown to release active TNF from the cell surface by a mechanism similar to that of ADAM-17-mediated TNF release [20
, 57
, 58
]. Another cytokine, IL-1β, is produced as an inactive precursor protein, which is activated by proteolytic removal of the N-terminal part. Caspase-1, also known as IL-1-converting enzyme (ICE), is an intracellular cysteine protease, which has been identified as the primary IL-1β activator. However, there is evidence that IL-1β can be activated in a caspase-1-independent manner in vitro and in vivo. For instance, in response to turpentine injection, the levels of mature IL-1β were not diminished in ICE–/– mice [59 ]. Furthermore, human keratinocytes, which express IL-1β, but not active ICE, were capable of producing mature IL-1β [60 ]. Some MMPs, namely MMP-2, MMP-3, and MMP-9, were found to activate pro-IL-1β, but MMP-1 could not [61 ]. This study also showed that further proteolytic degradation during prolonged incubation with MMP-3 could eventually inactivate mature IL-1β. Indeed, MMP-mediated IL-1β degradation had been reported already a few years earlier by Ito et al. [62 ], who found that 4-aminophenyl mercuric acetate-activated, conditioned medium from uterine cervical fibroblasts could degrade mature IL-1β and that this activity could be inhibited by adding TIMP-1 to the reaction. Using purified MMP extracts, MMP-1, MMP-2, MMP-3, and MMP-9 were shown to diminish the biological activity of IL-1. As MMPs and IL-1 often colocalize during inflammatory conditions, MMPs might influence positively and negatively the inflammatory process, by activating pro-IL-1β or inactivating the mature form of this cytokine, respectively.
TGF-β is an anti-inflammatory cytokine, known to restrain the mononuclear inflammation, whose activity is tightly regulated. TGF-β is produced initially as a precursor protein, which is cleaved in the endoplasmic reticulum by furins into an amino-terminal fragment, called latency-associated protein, and a shorter, carboxy-terminal fragment, which is the mature cytokine. These fragments are assembled as a double-homodimer, called the small latent complex, which is modified further by disulfide linkage to so-called latent TGF-β-binding proteins (LTBPs), thereby forming the large latent complex. After secretion, this latent TGF-β complex is cross-linked to the ECM, forming a reservoir of latent TGF-β in the extracellular environment [63 ]. Several mechanisms, including MMP proteolysis, have been implicated in the release of mature TGF-β from the latent complex. MMP-2 and MMP-9 [64 ] as well as MMP-3 [65 ] and MMP-14 [66 ] have been identified as TGF-β activators. Moreover, MMPs have also been implicated in the cleavage of LTBPs, thereby releasing TGF-β from the ECM [67 ]. MMPs can also cause release of TGF-β by degrading decorin, a small, collagen-associated proteoglycan, known to act as a depot for TGF-β in the ECM [68 ]. If shown that these pathways of MMP-mediated TGF-β activation would be relevant in vivo, this might be another mechanism by which MMP activity restrains rather than augments inflammation. In contrast, MMP activity might also down-regulate TGF-β signaling. MT1-MMP was shown to shed β-glycan [69 ], which is a membrane-bound protein, functioning as a coreceptor for TGF-β and regulating the access of TGF-β to its signaling receptors. However, if released from the cell surface, at least in vitro, β-glycan functions as a TGF-β inhibitor by blocking the interaction between TGF-β and its cell surface receptors [70 ]. Finally, processing and subsequent inactivation of IFN-β by MMP-9 offer another example of how MMP-mediated cytokine proteolysis might influence the progression of an inflammatory reaction [71 ].
MMPs might alter the biological activity of cytokines, not only by direct proteolytic processing but also by shedding their receptors. Some data suggest that MMPs contribute to the shedding of soluble Type II IL-1 decoy receptor (sIL-1RII) from the cell surface, as this shedding was blocked by BB-94, a broad-spectrum MMP inhibitor [72
]. The soluble receptor, by retaining its ability to interact with IL-1β, neutralizes the biological effects of this cytokine. Thus, MMP release of sIL-1RII offers another mechanism of how MMPs might dampen inflammation. In contrast, other data indicate that MMPs can degrade and therefore inactivate this soluble receptor [73
]. This would mean that MMPs support rather than inhibit IL-1β-driven inflammation, but all these data have to be verified in vivo. Other in vitro studies showed that MMP-9 cleaves IL-2R
and thereby, down-regulates the proliferative capability of activated T cells by generating antagonistic, sIL-2R
chains [74
]. The release of the receptor of a structurally related cytokine, namely sIL-15R
, could also be blocked by using a broad-spectrum MMP inhibitor; however, the specific MMPs responsible remain to be identified [75
].
| CONCLUSION |
|---|
|
|
|---|
| ACKNOWLEDGEMENTS |
|---|
| FOOTNOTES |
|---|
Received June 1, 2007; revised July 19, 2007; accepted July 20, 2007.
| REFERENCES |
|---|
|
|
|---|
and leaves RANTES and MCP-2 intact Blood 96,2673-2681
Mβ2 integrin and promatrix metalloproteinase-9 progelatinase in neutrophils J. Immunol. 172,7060-7068
5, AQARSAASKVKVSMKF, induces inflammatory cell production of matrix metalloproteinase-9 and chemotaxis J. Immunol. 171,398-406
1-Proteinase inhibitor is a neutrophil chemoattractant after proteolytic inactivation by macrophage elastase J. Biol. Chem. 263,4481-4484
Nature 385,733-736[CrossRef][Medline]
converting enzyme (TACE): a unique metalloproteinase with highly defined substrate selectivity Biochemistry 41,9462-9469[CrossRef][Medline]
in a model of herniated disc resorption J. Clin. Invest. 105,143-150[Medline]
precursor by metalloproteinases Nature 370,555-557[CrossRef][Medline]
convertase activity but does not activate pro-MMP2 J. Biol. Chem. 275,14046-14055
-chain that behaves as a specific, high affinity IL-15 antagonist J. Immunol. 173,1681-1688This article has been cited by other articles:
![]() |
R. A. Dean, J. H. Cox, C. L. Bellac, A. Doucet, A. E. Starr, and C. M. Overall Macrophage-specific metalloelastase (MMP-12) truncates and inactivates ELR+ CXC chemokines and generates CCL2, -7, -8, and -13 antagonists: potential role of the macrophage in terminating polymorphonuclear leukocyte influx Blood, October 15, 2008; 112(8): 3455 - 3464. [Abstract] [Full Text] [PDF] |
||||
![]() |
M. A. West, A. R. Prescott, K. M. Chan, Z. Zhou, S. Rose-John, J. Scheller, and C. Watts TLR ligand-induced podosome disassembly in dendritic cells is ADAM17 dependent J. Cell Biol., September 8, 2008; 182(5): 993 - 1005. [Abstract] [Full Text] [PDF] |
||||
![]() |
C. A. Reichel, M. Rehberg, P. Bihari, C. M. Moser, S. Linder, A. Khandoga, and F. Krombach Gelatinases mediate neutrophil recruitment in vivo: evidence for stimulus specificity and a critical role in collagen IV remodeling J. Leukoc. Biol., April 1, 2008; 83(4): 864 - 874. [Abstract] [Full Text] [PDF] |
||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |