Matrix Biology. May/31/2009; 28(5-3): 263-272

PMID: 19375502

PMC: 2724077

Expression and function of matrix metalloproteinase (MMP)-28

Abstract

Matrix metalloproteinase-28 (MMP-28, epilysin) is highly expressed in the skin by keratinocytes, the developing and regenerating nervous system and a number of other normal human tissues. In epithelial cells, over-expression of MMP-28 mediates irreversible epithelial to mesenchymal transition concomitant with loss of E-cadherin from the cell surface and an increase in active transforming growth factor beta. We recently reported the expression of MMP-28 in both cartilage and synovium where expression is increased in patients with osteoarthritis.

In human chondrosarcoma cells MMP-28 was activated by proprotein convertases and the active form of the enzyme preferentially associated with the extracellular matrix in a C-terminal independent manner. over-expression of MMP-28 in chondrosarcoma cells led to altered cell morphology with increased organisation of actin. Adhesion to type II collagen and fibronectin was increased, and migration across the former was decreased. MMP-28 was localised to the cell surface, at least transiently, in a C-terminal dependent manner. Heparin prevented both extracellular matrix association and cell surface binding of MMP-28 suggesting that both are via heparan sulphate proteoglycans. Over-expression of activatable MMP-28, but not catalytically inactive EA mutant increased the expression and activity of MMP-2, and all forms of MMP-28 tested increased expression of MMP19 and TIMP3 mRNA.

These data demonstrate that expression of MMP28 alters cell phenotype towards a more adhesive, less migratory behaviour. Further, MMP-28 activity may reside predominantly in the extracellular matrix, and we are currently searching for substrates in this compartment.

1Introduction

The matrix metalloproteinases (MMPs) comprise a family of 23 structurally related, zinc-dependent endopeptidases and play a significant role in the degradation of proteins in the extracellular matrix (ECM) (Ra and Parks, 2007). MMPs are also capable of targeting non-ECM proteins, such as growth factors and cytokines that may bind either to the ECM or the cell surface, leading to their release and activation (Egeblad and Werb, 2002). The activity of MMPs influences cell function and response to the microenvironment with important implications for both normal development, tissue remodelling and repair, as well as in disease progression (Lopez-Otin and Matrisian, 2007; Martin and Matrisian, 2007).

MMP-28 (also called epilysin) was originally cloned from keratinocyte, testis and mixed tumour cDNA libraries (Lohi et al., 2001; Marchenko and Strongin, 2001), however, its expression has been documented in a number of normal tissues, e.g. testis, lung, intestine, heart brain and skin (Lohi et al., 2001; Marchenko and Strongin, 2001), as well as in a variety of tumours and tumour cell lines (Marchenko and Strongin, 2001; Nuttall et al., 2003). MMP28 is also expressed in the developing and regenerating nervous system prior to myelination (Werner et al., 2007).

MMP-28 is expressed in the basal and suprabasal epidermis in intact skin, whilst in injured skin, expression is seen in basal keratinocytes both at and some distance from the wound edge (Lohi et al., 2001; Saarialho-Kere et al., 2002). In epithelial cells, recent evidence suggests that over-expression of MMP-28 can induce an epithelial to mesenchymal transition (EMT) via activation of latent TGFβ (Illman et al., 2006, 2008). This MMP-28-induced EMT is associated with the loss of E-cadherin, a major mediator of cell–cell adhesion as well as increased expression of MMP-9 (gelatinase B) and MMP-14 (MT1-MMP).

In primary keratinocytes, expression of MMP-28 is induced by tumour necrosis factor α (TNFα), but not by bFGF, EGF, GM-CSF, HGF, IFNγ, IL-1β, KGF, PDGF, TGFβ, VEGF nor IGF1 (Saarialho-Kere et al., 2002). The proximal promoters of both human and mouse MMP28 genes have been isolated, and partially characterised in immortalised keratinocytes (HaCaT) (Illman et al., 2001).

The expression of MMP28 is increased in oral squamous cell carcinomas (OSCC) compared to premalignant lesions (Lin et al., 2006). Transfection with antisense oligonucleotides and knockdown of MMP-28 lead to inhibition of anchorage independent growth.

In a myelinating dorsal root ganglion co-culture system, MMP-28 reduces myelination whilst function-blocking antibodies enhance myelination. MMP-28 protein expression is also increased in demyelinating lesions in both human multiple sclerosis and mouse experimental autoimmune encephalitis (Werner et al., 2008, 2007).

Our own data demonstrate that MMP28 expression is increased in the joints of osteoarthritis patients in both cartilage and synovium (Davidson et al., 2006; Kevorkian et al., 2004).

MMP-28 has a typical MMP domain structure with signal sequence, propeptide, zinc-binding catalytic domain and haemopexin-like C-terminal domain. Within the propeptide is a furin activation sequence suggesting the enzyme may be intracellularly activated and this has been demonstrated for murine MMP-28 (Illman et al., 2003). The only protein substrate reported for MMP-28 thus far, is casein, a non-specific substrate for many proteinases.

In order to provide some further insight into the function of human MMP-28, particularly in the joint, we examined the expression and activation of human MMP-28 and the phenotypic consequence of stable over-expression of the gene in human chondrosarcoma cell lines.

2Results

2.1MMP-28 over-expression and furin activation in HeLa cells

HeLa cells were transiently transfected with MMP28/pcDNA4-FLAG construct (WT) or empty vector and protein expression assessed by western blot. Two bands were detected in the conditioned medium of WT transient transfects, at approximately 60 kDa and 30 kDa (Fig. 1A). These bands were independently confirmed to be the pro form and C-terminal domain respectively, using domain-specific antibodies (see Experimental procedures, data not shown). Predicted molecular weights for the unmodified polypeptides of these two forms are 56.5 kDa and 26.7 kDa respectively. An approximate 60 kDa band was also detected in cell lysates (Fig. 1B), again verified as the pro form with a domain-specific antibody (data not shown). In the ECM, two bands of approximately 60 kDa and 50 kDa were detected (Fig. 1C) and were independently confirmed to be the pro and active forms of MMP-28 (data not shown); in this study, the term ‘active’ refers to MMP-28 following removal of the pro domain. Predicted molecular weight for the active form is 44.9 kDa.

In order to ascertain the mechanism of processing, the effect of a proprotein convertase inhibitor, Decanoyl Arg-Val-Lys-Arg Chloromethylketone, was tested. As the active (processed) form of MMP-28 appeared to be largely associated with the ECM, samples from the ECM were assessed. Samples treated with the inhibitor demonstrated only the presence of the pro form of MMP-28 (Fig. 2), confirming that MMP-28 processing is mediated by a furin-like proprotein convertase. The anti-FLAG antibody also detected a high molecular weight band present in both vector only and MMP-28 transfected cells.

2.2Expression of MMP-28 in stably transfected SW1353 chondrosarcoma cells

SW1353 cells were stably transfected with the wild-type MMP28/pcDNA4-FLAG construct (WT), an E241A (EA) mutant form in which the glutamate residue in the catalytic domain was substituted for an alanine, or a truncated form, in which the hemopexin (C-terminal) domain is absent (pro-cat). Clonal cell lines were established for each construct as well as for an empty vector control. All established lines were screened for MMP-28 mRNA expression by qRT-PCR analysis (see Fig. 8A for example). Two of these clonal lines were selected for each construct and control, and analysed for MMP-28 protein expression. Full-length, wild-type MMP-28 (WT) was found (though in variable amounts in different experiments) in the conditioned medium, cell lysates and ECM (Fig. 3A). Both the pro and active (processed) forms were detected in each fraction, at molecular sizes of approximately 58 kDa and 50 kDa respectively. However, it appeared that the pro form was predominant in the cell lysates, whereas in the conditioned medium and especially the ECM, the active form was predominant. The C-terminal domain was also detected, at approximately 33 kDa. The expression of the EA mutant followed a slightly different pattern to that of WT (Fig. 3B). The EA mutant was barely detected in the conditioned medium, and instead appeared to be largely associated with the cell lysate and the ECM. Also, only the pro form of the EA mutant was detected in the cell lysates, in striking contrast to the ECM, which shows the presence of both the pro and (active) processed forms. This eliminates intramolecular autocatalytic activity of MMP-28 as a means of processing, as the EA mutant is catalytically inactive. MMP28 expression in the SW1353 cell line or vector-only transfects was undetectable by qRT-PCR, thus intermolecular autocatalysis is also unlikely. The pro-cat form of MMP-28 was detected as two smaller bands in the ECM (Fig. 3C), at approximately 37 kDa and 28 kDa. These are the pro and active (processed) forms respectively. The smaller form corresponds to the catalytic domain alone. Only the pro form was detected in the cell lysates. Neither the pro or active forms were detected in the conditioned medium.

In order to understand the nature of the interaction of MMP-28 with the ECM, we treated cultures of stable transfects expressing the WT form with heparin (100 µg/ml in serum-free medium). We detected increased levels of both the pro and active forms of MMP-28 in the conditioned medium compared to untreated cells (Fig. 3E), showing that heparin is able to compete with or prevent binding of MMP-28 to the ECM.

2.3Immunocytochemistry

Intracellular MMP-28 was detected in permeabilised SW1353 cells stably transfected with MMP28 expression constructs (Fig. 4A–C) but not in empty vector control cells (Fig. 4D). Cells expressing the WT form demonstrated strong perinuclear staining. The EA mutant transfects (Fig. 4C) also demonstrated punctate staining that appeared to localise the periphery of the cells with WT transfects demonstrating this to a lesser extent. In non-permeabilised cells both WT (Fig. 4E) and EA mutant (Fig. 4F) stable transfects showed similar cell surface staining, though the pro-cat form was barely detectable at the cell surface (Fig. 4G). Empty vector control cells showed no cell surface staining for MMP-28 (Fig. 4H). IgG controls were negative for intracellular and cell surface staining of MMP-28 (Fig. 4I and J respectively).

The addition of heparin to the cultured wild-type transfects (as above) led to a reduction in cell surface staining for MMP-28 in non-permeabilised cells (Fig. 5), demonstrating competition for binding sites in a similar manner to that seen for ECM binding.

2.4Over-expression of MMP-28 alters cell morphology and actin stress fiber formation

SW1353 cells stably transfected with MMP28 exhibited a noticeable change in morphology when compared to empty vector control cells and untransfected cells. Empty vector transfects (Fig. 6A) showed identical morphology to untransfected SW1353 cells (data not shown). Cells over-expressing the WT form demonstrated a longer, broader morphology when compared to empty vector controls (Fig. 6B). This was exhibited to a slightly lesser extent in EA mutant (Fig. 6C) and pro-cat stable transfects (Fig. 6D). The change in morphology is due to changes in the cytoskeleton, as cells over-expressing MMP-28 demonstrated clearly defined actin stress fiber formation that was not evident in control cells (Fig. 6E–H).

2.5Effects of MMP-28 on cell adhesion

The effects of over-expression of MMP-28 on adhesion of stably transfected SW1353 cells were measured. Adhesion was measured for each of two clonal stable lines generated for each construct. Cells over-expressing MMP-28 showed significantly increased levels of adhesion when plated on fibronectin (Fig. 7A) or collagen II-coated surfaces (Fig. 7B), when compared to empty vector control cells. On both matrices, adhesion was maximal in cells expressing the WT form, though adhesion remained raised compared to control for cells expressing both EA mutant or the pro-cat form. On collagen I, wild-type and pro-cat transfects both showed significantly increased adhesion after 60 min, however EA mutant transfects demonstrated inconsistent results, making the data difficult to interpret (data not shown).

The changes in adhesion on fibronectin and collagen II suggested the possibility of increased density of cell surface receptors due to increased expression of integrin subunits. As integrins are receptors for both fibronectin and collagen II, we assessed their role in mediating adhesion of wild-type MMP-28 expressing cells. EDTA (5 mM) completely inhibited adhesion to both matrix proteins (data not shown), suggesting that integrins were mediating adhesion of stably transfected cells. A screen of integrin subunit mRNA expression in empty vector control and WT cells was performed. This revealed altered expression of several integrin receptor subunits. The subunits that showed consistent increases in mRNA expression in WT cells were α₁, α_v, α₁₁ and β₁ (data not shown). Of the subunits tested, only the α₂ subunit showed a consistent decrease in expression. Subunits α₅, α₁₀, β₃ and β₅ showed inconsistent changes while the α₄ subunit was expressed at a very low level. Anti-β₁ subunit antibodies significantly decreased adhesion of empty vector control cells to fibronectin (Fig. 7C) whilst cells treated with the isotype control showed no significant difference to untreated cells (not shown). The anti-β₁ antibody did not inhibit adhesion of WT transfects to fibronectin, which was unexpected, given that the β₁ subunit plays a major role in fibronectin recognition. However in a single preliminary experiment (n = 3), a combination of antibodies to both the α_v and β₁ subunits resulted in significant inhibition of adhesion of wild-type MMP-28 expressing cells to fibronectin (data not shown). The anti-β₁ antibody decreased adhesion of both empty vector control and WT transfects to collagen II (Fig. 7D, binding of the empty vector control VO-1 was minimal, so these cells do not show statistical significance). Again, cells treated with the isotype control showed no significant difference to untreated cells (data not shown). A comparison of adhesion of IgG1-treated WT cells to IgG1-treated empty vector control cells on either fibronectin or collagen II matrices showed that cells expressing the WT form still demonstrated significantly higher adhesion.

2.6Effect of MMP-28 on cell proliferation, death and migration

Thymidine uptake experiments were conducted to assess the effect of over-expression of MMP-28 on cell proliferation, which was measured 48 h after the cells were plated. However the data showed no significant change in proliferation (data not shown). Additional experiments using the Live/dead assay™ to ascertain cell death, as well as annexin V staining to assess the effects of MMP-28 on apoptosis also showed no differences between empty vector control cells and WT-transfected cells (data not shown).

Migration of transfects was assessed in either 10% or 0.5% serum, in the presence or absence of matrix proteins. In the absence of matrix proteins, no consistent differences between empty vector controls and WT-expressing transfects were observed (data not shown). Migration was subsequently studied on fibronectin, collagen I and collagen II. WT-expressing cells showed a trend to a decrease in migration on all matrix proteins in 0.5% serum (data not shown), though only collagen II showed a significant decrease (p < 0.0001). Further study showed that migration on collagen II was significantly reduced for all MMP-28 expressing cells in 0.5% serum (Fig. 7E). No significant differences occurred when cells plated on collagen II were in the presence of 10% serum (data not shown).

2.7Effect of MMP-28 on expression of other MMP and TIMP family members

In order to ascertain if over-expression of MMP28 could affect the expression of other MMPs and members of the TIMP family, qRT-PCR analysis was employed. MMP28 itself was not detectable in the empty vector transfects (or the untransfected SW1353 cells) by this method (i.e. Ct < 40), with the stable cell lines all showing high expression for the respective transgene (Ct between 20.3 and 23.1) (Fig. 8A). Consistent changes in expression were observed for MMP2, MMP19 and TIMP3 (Fig. 8B–D). MMP2 expression was significantly higher in WT- and pro-cat-expressing cells but not EA mutant. MMP19 and TIMP3 expression was significantly higher in all MMP-28 expressing cells.

As MMP2 levels were upregulated in transfects expressing active forms of MMP-28, the expression of MMP-2 was further assessed by gelatin zymography (Fig. 9). MMP-2 pro and active forms were visually observed to be more intense in WT- and pro-cat-expressing samples, supporting the data obtained from the analysis of steady-state mRNA.

3Discussion

MMP-28, a catalytically inactive mutant and a C-terminal domain deletion mutant were cloned into an expression construct with a C-terminal FLAG tag. Expression of MMP-28 was assessed in transiently transfected HeLa cells and in stably transfected SW1353 chondrosarcoma cells. Both latent and active (processed) MMP-28 was detected in cells expressing the wild-type construct similar to those identified in studies of mouse MMP-28 (Illman et al., 2003). The proenzyme form was detected in conditioned medium, cell lysates and the ECM, however the active (processed) form was detected predominantly in the ECM, suggesting that processing of the MMP-28 zymogen may be largely extracellular. This has previously been reported for ADAMTS-4 and ADAMTS-5, with PACE4 identified as the proprotein convertase responsible in this case (Malfait et al., 2008). Mouse MMP-28 has also been shown to have a functional proprotein convertase recognition site (Illman et al., 2003) and we have now proven this for the human MMP-28. In addition to the latent and active forms of MMP-28, we also detected the isolated C-terminal domain in the conditioned medium, though we have not detected the isolated catalytic domain. Potentially, the catalytic domain is less stable than the C-terminal domain and is broken down by further proteolysis. The C-terminal domain is known to confer substrate specificity on collagenolytic MMPs, such as MMP-1 or MMP-13 (Clark and Cawston, 1989; Knauper et al., 1997) and has also been shown to regulate cell migration (Dufour et al., 2008). Our data show that this domain mediates binding to the cell surface for MMP-28 potentially mediating proenzyme activation in a manner similar to that of ADAMTS9 by furin (Koo et al., 2006). However the pro-cat mutant protein was not detected at the cell surface, yet was still activated. Our data also show that both pro and active (processed) MMP-28 are found in conditioned medium and/or ECM perhaps suggesting that cell surface localisation may be transient. Given its localisation, MMP-28 could cleave cell surface proteins and thus alter cell signalling. Cell surface binding of MMP-28 was also competed by the addition of heparin suggesting association with cell surface heparan sulphate proteoglycans. MMP-28 and MMP-19 have both previously been shown to associate with the surface of lung adenocarcinoma cells and myeloid cells respectively, in a C-terminal domain-dependent manner (Illman et al., 2006; Mauch et al., 2002). MMP-28 association with the surface of adenocarcinoma cells was lost upon EMT which occurred as a consequence of MMP28 over-expression and increase in TGFβ activity (Illman et al., 2006). We did observe a change in cell morphology upon expression of our MMP28 constructs, though SW1353 cells are not epithelial cells. SW1353 cells have a fibroblast-like morphology, which appears to be altered when cells over-express MMP28, particularly the wild-type enzyme. The change in morphology is associated with actin stress fiber formation, in these cells, usually associated with cell spreading and increased adhesion due to integrin engagement, clustering and focal contact formation (Arnaout et al., 2007; Pellegrin and Mellor, 2007).

In the stably transfected cells, neither the EA nor pro-cat mutants were detected in the conditioned medium, though these forms were detected in the ECM, similar to WT. It is possible that saturation of MMP-28 binding sites in the matrix could result in the release of MMP-28 into the conditioned medium, and therefore detection in this latter depends on expression level. The binding of all three MMP-28 protein isoforms to the ECM, including the pro-cat form, shows that MMP-28 binds via its catalytic domain. The predominance of the active form in the ECM fraction is an intriguing novel finding. Other metalloproteinases, such as ADAMTS-1 and ADAMTS-4, have also been localized to the ECM (Gao et al., 2002; Kuno and Matsushima, 1998). When expressed in a chondrosarcoma cell line, the mature form of ADAMTS-4 bound most abundantly to the ECM (Gao et al., 2002), however the smaller molecular weight active forms were found predominantly in the conditioned medium. The active form of ADAMTS-1 was found to associate predominantly with the ECM where, along with ADAMTS-4, it cleaves versican (Russell et al., 2003). Interestingly, heparin competes with binding of ADAMTS-1 and ADAMTS-4 to the ECM, resulting in detection of the protein in the conditioned medium (Kashiwagi et al., 2004; Kuno and Matsushima, 1998). When we cultured MMP-28-expressing cells in the presence of heparin, we detected increased levels of both the pro and active forms of MMP-28 in the conditioned medium compared to untreated cells, suggesting that binding of MMP-28 to the ECM is mediated through sulphated glycosaminoglycans such as heparan sulphate in a similar manner to cell surface binding. The abundance of MMP-28 in the ECM may also suggest that it cleaves ECM substrates or regulates cell–ECM interactions.

Adhesion experiments revealed that cells over-expressing MMP28, particularly those expressing the WT form, were able to adhere with significantly increased efficiency to fibronectin and collagen type II when compared to empty vector transfects. Antibodies against the β₁ integrin subunit (Kurtis et al., 2003) mediated greater inhibition of adhesion to collagen II than to fibronectin, with less impact on adhesion of the WT-expressing cells compared to control. It is likely that multiple integrins or other receptors are involved in mediating adhesion, particularly of the WT transfects. Alternatively, an increase in specific integrin subunits at the cell surface, either through increased expression and/or clustering could explain a reduced or lack of effect of anti-β₁ antibody in these cells. A screen of integrin subunit expression by qRT-PCR revealed significantly increased expression of the β₁ subunit in the MMP-28 expressing cells. A significant increase in the expression of the α₁, α_v and α₁₁ subunits was also observed. Chondrocytes are able to adhere to collagen II via integrins α₁β₁ and α₂β₁ (Loeser et al., 2000), and to cartilage tissue sections via the β₁ subunit and α_vβ₅ (Kurtis et al., 2003). Our preliminary data (not shown) demonstrate that a combination of anti-α_v antibodies with anti-β₁ antibodies resulted in almost complete inhibition of empty vector transfects to fibronectin with wild-type MMP-28 expressing cells inhibited by approximately 50% compared to isotype controls.

We assessed the effect of over-expressing MMP28 on other aspects of cell function. We measured no alteration in cell proliferation or death. When we measured cell migration on collagen II, we saw a significant decrease for all MMP28 transfects when cells were assayed in low serum. This may be related to the increased adhesion of the transfects to collagen II, though we saw no change in migration on fibronectin.

Over-expression of MMP-28 also led to an upregulation of MMP2, MMP19 and TIMP3 steady-state mRNA. We were able to demonstrate that active MMP-2 is increased in the conditioned medium produced by cells expressing MMP-28. Interestingly, only the cells expressing active forms of MMP-28 (WT and pro-cat) showed an increase in MMP-2 expression and activity. In contrast, MMP19 and TIMP3 expression was altered in all transfects, suggesting that MMP-28 activity is not a requirement for altering the expression of these genes. In our previous studies, we have seen a correlation between the expression of MMP28 and MMP2 (though not MMP19 or TIMP3) in human articular cartilage (Kevorkian et al., 2004). Contrary to previous findings in adenocarcinoma cells (Illman et al., 2006), we did not see changes in MMP9 or MMP14.

In summary, we have shown that human MMP-28 is activated by proprotein convertases and the active form of the enzyme preferentially associates with the ECM likely via heparan sulphate proteoglycans. A similar interaction localizes MMP-28 to the cell surface. Over-expression of MMP-28 in chondrosarcoma cells leads to changes in cell morphology, increased adhesion and decreased migration to and across ECM. This is, to some extent, the opposite of the change in phenotype driven by MMP-28 over-expression in epithelial cells (Illman et al., 2006).

4Experimental procedures

4.1Cloning of full-length transcript

Human MMP28 consists of 8 exons. Two human transcripts of MMP28 are described, variant 1 (NM_024302) and variant 2 (NP_001027449) with alternate splicing in exon 8 leading to a premature stop codon in variant 2. A MMP28 variant 2 cDNA clone was purchased from the Medical Research Council (MRC) (London, UK) and used as the template for creating the MMP28 variant 1 construct. Extension overlap PCR was used to remove the section of exon 8 containing the premature stop codon. PCR was performed using 20 ng cDNA template, 0.4 μM each primer, 200 μM dNTPs and 0.5 units of AccuTaq™ LA polymerase (Sigma-Aldrich) in the manufacturer's buffer. Two separate PCR reactions were performed. The first contained a forward primer for MMP28 variant 1 containing a BamHI restriction site (underlined; 5′GCGGGATCCCAGCCCGCGGAGCGCGGAG 3′) and an internal reverse primer (5′CTCCAGCATCGACCCCCTTTGAAGAAGTAGAAATC 3′). The second reaction contained a reverse primer for MMP28 variant 1 which contained an EcoRI restriction site (underlined; 5′CCGGAATTCAGAACAGGGCGCTCCCCGAGTTG 3′) and an internal forward primer (5′CTCCAGCATCGACCCCCTTTGAAGAAGTAGAAATC 3′). The two products generated from these reactions were used as the template to obtain the final desired product. The primers containing the BamHI and EcoRI restriction sites were used. The final product was subcloned into a pRSETA vector (Invitrogen) and used as the template to create a construct suitable for mammalian expression.

PCR primers were designed for cloning into the mammalian expression vector pcDNA 4 FLAG which was a kind gift from Dr. V. Knäuper (University of Cardiff, Wales, UK). This vector was constructed using the pcDNA 4-V5-His vector (Invitrogen) and substituting the V5 epitope for an in-frame FLAG sequence. Each primer was tagged with three random nucleotides (italics) and a BamHI (5′) and XhoI (3′) restriction enzyme recognition site (underlined). The 5′ forward primer also contained a consensus Kozak sequence (bold; (Kozak, 1987)). Primer sequences were as follows: 5′-CGCGGATCCGCCGCCGCCATGGTCGCGCGCGTCGGCCTC-3′ and 5′-ACGTCTCGAGCCGAACAGGGCGCTCCCCGAGTTG-3′ for forward and reverse primers respectively. Sequencing of the cloned gene was performed in both directions.

4.2PCR and cloning of EA mutant and pro-cat transcripts

Extension overlap PCR was employed to create the catalytically inactive E241A mutant (EA mutant) form of MMP28. The full-length wild-type sequence was used as the template for the PCR. For the first stage of creating the mutant, two separate PCR reactions were performed, the first containing the forward primer containing the BamHI restriction site as described previously, and an internal reverse primer containing the desired mutation (5′-CAAGCGTGTGACCGATCGCGTGCGCCAGCAC-3′). The second reaction contained the reverse primer containing the XhoI restriction site as described previously, and an internal forward primer containing the desired mutation (5′-GTGCTGGCGCACGCGATCGGTCACACGCTTG-3′). For the second stage, the two products were used as templates for the extension overlap PCR. Primers containing the BamHI and XhoI restriction sites (described previously) were used to create the final EA mutant transcript. PCR products were digested with BamHI and XhoI and ligated into the modified pcDNA-4-FLAG expression vector as described previously. Sequencing of the cloned cDNA was performed.

A truncated form of MMP28, containing only the pro and catalytic domains (pro-cat) was also created by PCR from the full-length template. For cloning, PCR primers at each end of the gene were tagged with three or four random nucleotides (italics) and a BamHI (5′) and XhoI (3′) restriction enzyme recognition site (underlined). The 5′ forward primer also contained a consensus Kozak sequence (bold; (Kozak, 1987)). Primer sequences were as follows: 5′-CGCGGATCCGCCGCCGCCATGGTCGCGCGCGTCGGCCTC-3′ and 5′-ACGTCTCGAGCCATACAGGCTCTGCACGGCCAGCAC-3′ for forward and reverse primers respectively. PCR products were digested with BamHI and XhoI and ligated into the pcDNA-4-FLAG expression vector as described.

4.3Transient transfection, western blot and furin analysis

HeLa cells were cultured in DMEM plus GlutaMAX™ (Invitrogen) with 10% fetal calf serum (FCS; Invitrogen). Cells were plated at 100 000/well of a 6-well plate. After 24 h 1 μg of wild-type MMP28/pcDNA4-FLAG construct or control plasmid was transfected into cells using FuGENE®6 transfection reagent (Roche) according to the manufacturer's recommendations. After another 24 h, cells were switched to serum-free medium. For studies where cells were treated with heparin, 0.1 mg/ml (w/v) heparin (sodium salt, Sigma-Aldrich) was added to serum-free medium. After a further 48 h, the conditioned media was collected and proteins precipitated with 0.5 volumes of 10% (v/v) trichloroacetic acid (TCA) for 1 h on ice. Samples were centrifuged for 15 min at 13 000 rpm, and pellets washed with cold acetone and pelleted for a further 15 min at 13 000 rpm. Protein pellets were air-dried and resuspended in 1× SDS final sample buffer (0.058 M Tris–HCl, pH 6.8; 5% v/v glycerol; 1.7% w/v SDS and 0.002% w/v bromophenol blue) containing 50 mM DTT and 6 M urea (FSB/DTT/urea). Cells were detached with 0.5 mM EDTA in PBS. Cell pellets were collected by centrifugation at 13 000 rpm for 5 min, resuspended in FSB/DTT/urea and sonicated prior to analysis. ECM proteins were harvested by scraping each well with FSB/DTT/urea for approximately 30 s/well. All samples were boiled for 5 min and subjected to western blotting. Proteins were separated by SDS-PAGE, followed by transfer to PVDF membrane. MMP-28 was probed using a 1:5000 dilution of mouse ANTI-FLAG® M2 IgG monoclonal antibody (Sigma-Aldrich) followed by a 1:1000 dilution of HRP-labelled rabbit anti-mouse IgG secondary antibody (Dako, Denmark). Antibodies raised against the pro, furin and C-terminal domains were used to verify specific MMP-28 protein expression (RP1MMP28, RP2MMP28 and RP4MMP28 respectively, Triple Point Biologics, Portland, OR). An antibody against the catalytic domain (Lohi et al., 2001) was also used. LumiGLO reagent and peroxide (Cell Signalling Technology, Beverley, MA) were used for detection on a Kodak Biomax MS film (Sigma-Aldrich) by chemiluminescence.

To assess the role of furin in processing MMP-28 into an active form, a furin inhibitor (Decanoyl Arg-Val-Lys-Arg Chloromethylketone) was employed (Bachem). The furin inhibitor (100 μM) was added to HeLa cells prior to transient transfection with wild-type MMP-28. The same concentration of inhibitor was added 24 h after the first dose. ECM samples were harvested and assessed by western blotting as described.

4.4Creation of stably expressing cell lines

SW1353 human chondrosarcoma cells (ATCC) were cultured in DMEM plus GlutaMAX™ (Invitrogen) with 10% fetal calf serum (FCS; Invitrogen). 100 000 cells/well were plated into 6-well plates and incubated overnight. The wild-type, EA mutant or pro-cat MMP28/pcDNA4-FLAG constructs or pcDNA4-FLAG vector only, were linearised by digesting with BglII and 1 μg was transfected as described above. After 48 h, cells were switched to culture media containing 200 μg/ml zeocin (Melford Laboratories). After seven days, surviving cells were plated into 96-well plates at dilutions of three, one and 0.3 cells/well. Wells containing confluent cell layers were expanded and plated again into 96-well plates containing 0.3 cells/well to select for clonal colonies. Cells were maintained in zeocin-containing media. Protein expression in conditioned medium, cell lysates and ECM was determined by western blotting as above.

4.5Quantitative real-time PCR

Cells were plated at 200 000 cells/well into 6-well plates. After 24 h, RNA was harvested using TRIzol® reagent (Invitrogen). Following the addition of 0.5 volumes of chloroform, the aqueous phase was recovered after centrifuging at 12 000 rpm for 15 min at 4 °C and added to an equal volume of isopropanol. RNA was pelleted by centrifuging at 12 000 rpm for 30 min at 4 °C and washed with 70% (v/v) ethanol. RNA was dissolved in distilled water and quantified using a NanoDrop® spectrophotometer (Nanodrop Tecnologies). cDNA was synthesized from 1 μg of RNA using Superscript™ II reverse transcriptase (Invitrogen) and random hexamers, according to the manufacturer's recommendations. Primers for MMP28 were designed using the Universal Probe Library (Roche) using the appropriate probe. The primers were designed specifically for detection of pro-cat transcripts, but were also used for detection of full-length wild-type and EA mutant forms. All of the primers and probes for the other MMPs and TIMPs were designed using Primer Express 1.0 software (Applied Biosystems) and are described elsewhere (Nuttall et al., 2003; Porter et al., 2004). Primers for the integrin subunits α₁, α_2, α₄, α₅, α₅, α₁₀, α₁₁, α_v, β₁, β₃ and β₅ were designed using the Roche Universal Probe Library and were a gift from Dr Rosemary Bass (University of East Anglia, U.K.). The 18S rRNA gene was used as an endogenous control to normalise for differences in the amount of RNA between samples; primers and probes were from Applied Biosystems. Quantitative PCRs were performed and analysed as described previously (Davidson et al., 2006).

4.6Immunocytochemistry

Stably transfected SW1353 cells were plated into chamber slides at a density of 15 000 cells/well. After incubating for 24 h, cells were switched to serum-free media for a further 24 h. Cells were fixed in 4% (w/v) paraformaldehyde for 10 min, permeabilised with 0.1% (v/v) Triton X-100 for 10 min, blocked with 3% (w/v) BSA for 30 min and incubated with a 1:1000 dilution of mouse ANTI-FLAG® M2 IgG monoclonal antibody for 30 min. This was followed by a 1:1000 dilution of Alexa Fluor 488-conjugated goat anti-mouse IgG (Invitrogen) secondary antibody for 30 min. Cells were also incubated with normal mouse IgG of equivalent isotype (IgG1, Sigma-Aldrich) and secondary antibody as controls. For cell surface detection of MMP-28, cells were treated as described but without permeabilisation. For staining the actin cytoskeleton, cells were stained with 25 μg/ml TRITC-phalloidin (Sigma-Aldrich) for 45 min following fixing and blocking. Cells were visualised using an AxioPlan 2ie fluorescence microscope with Axiovision software (Carl Zeiss).

4.7Adhesion, proliferation, migration and morphology studies

96-well culture plates were coated with 100 μl of 50 μg/ml vitronectin or fibronectin (Sigma-Aldrich) or 100 μg/ml rat tail collagen type I (First Link (U.K.) Ltd., West Midlands, U.K.) or bovine collagen type II (Williams, 2007) overnight at 4 °C. Plates were blocked with 1% (w/v) BSA in PBS for 1 h at room temperature before adding 30 000 cells/well. Cells were plated in uncoated wells blocked with 1% (w/v) BSA as a control. After 15 min or 1 h, plates were washed six times with PBS to remove non-adherent cells. Cell adhesion was measured using methylene blue staining. Cells were fixed for 1 h in 70% (v/v) ethanol and stained in 1% (w/v) methylene blue for 30 min. Cells were lysed with 50% (v/v) ethanol in 0.1 M HCl for 15 min, and absorbance measured at 650 nm. Experiments were performed at least three times in triplicate. For proliferation assays, cells were plated in medium containing 10% (v/v) FCS at 25 000 cells/well in 24 well plates in replicates of six and allowed to adhere overnight. Each well was labelled with 0.5 μCi per well [6-³H]-thymidine for 6 h in medium containing either 10% (v/v) or 0.5% (v/v) FCS, washed twice with HBSS and fresh medium added containing 3 μM cold thymidine. After overnight incubation, medium was aspirated and 300 µl per well 0.25 M ammonia added. Plates were rocked for 2 h at room temperature then lysate was transferred into scintillant for counting tritium. A separate assessment of proliferation was made by plating cells at 10 000/well of a 24-well plate. Cells were harvested at 24 and 48 h after plating. Cell counts were determined by diluting samples 1:2 in Trypan Blue (Sigma-Aldrich) and viable cell numbers determined by counting cells which excluded the dye, using a hemocytometer. Non-viable cell numbers were also determined. Cell death was further explored using the Live/Dead Assay™ (Invitrogen), and staining for annexin V. For migration assays, cells resuspended in DMEM with 10%(v/v) serum were plated at 1000 cells/well in 24-well plates. After 24 h medium was switched to 10%(v/v) or 0.5%(v/v) serum in DMEM. Cells were subjected to time-lapse microscopy on an AxioVert 200M microscope that was enclosed in an Incubator XL (Carl Zeiss) for temperature and CO₂ control. Cells were photographed at 15 minute intervals for 13 h. Total distance moved (in μm) for 10 cells/well was measured using Axiovision Software (Carl Zeiss). In order to study the effect of specific matrix proteins on cell migration, wells were coated with 50 μg/ml fibronectin or 100 μg/ml collagens I and II overnight at 4 °C, prior to plating cells. In order to collect images that showed the effect of MMP-28 expression on cell morphology, cells were plated as described for migration assays. After 24 h, images of live cells were captured under phase contrast on an AxioVert 200M microscope as described above.

4.8Zymography

Cells expressing wild-type MMP-28 and vector-only controls were plated at 200 000 cells/well in 6-well plates. After 24 h media was changed to serum-free. After a further 48 h, the conditioned media was collected and proteins concentrated using Centricon ultrafiltration cells (3 kDa molecular weight cut-off, Millipore). Samples were loaded with equal protein concentration on a 10% (v/v) SDS-PAGE resolving gel containing either 1 mg/ml gelatin or 1 mg/ml casein. Once the gels were run, they were placed in rinse buffer (50 mM Tris–HCl pH 8.0, 5 mM CaCl₂, 2.5% (v/v) Triton X-100) overnight with shaking. Gels were then rinsed well in water then incubated in incubation buffer (50 mM Tris–HCl pH 8.0, 5 mM CaCl₂) for 24 h at 37 °C. Gels were stained with Coomassie Brilliant Blue and bands were visualised following de-staining.

Fig. 1

Expression of MMP-28 in transiently transfected HeLa cells. HeLa cells were transiently transfected with the wild-type (WT) MMP28 cDNA, previously cloned into a modified pcDNA4FLAG vector. Vector-only (VO) transfects were included as a negative control. Protein expression in conditioned medium (A), cell lysate (B) and extracellular matrix (C) was detected by western blotting, using an anti-FLAG antibody. The pro and active forms of MMP-28 are indicated. CTD = C-terminal domain.

Fig. 2

Furin activity is required for processing of proMMP-28. HeLa cells were transiently transfected with the wild-type MMP28 gene or the pcDNA4FLAG vector as a negative control. Cells were treated with 100 μM of the furin inhibitor Decanoyl-Arg-Val-Lys-Arg-chloromethyl ketone prior to transfection. The same dose was repeated 24 h after transfection. Untreated cells were included as controls. Protein expression in the extracellular matrix was detected by western blotting, using an anti-FLAG antibody.

Fig. 3

Expression of MMP-28 in stably transfected SW1353 cells. SW1353 cells were stably transfected with wild-type (A), EA mutant (B) or pro-cat mutant (C) MMP28 gene constructs. Cells stably transfected with pcDNA4FLAG vector only were included as a negative control (D). In a separate experiment, WT cells were treated with 0.1 mg/ml heparin, or left untreated as a negative control (E). Protein expression in conditioned medium (CM), cell lysate (CL) and extracellular matrix (ECM) was detected by western blotting, using an anti-FLAG antibody. The pro and active species are indicated for the WT and pro-cat forms of MMP-28. The pro and processed species are indicated for the catalytically inactive EA mutant. CTD = C-terminal domain.

Fig. 4

Detection of MMP-28 expression in SW1353 cells using immunofluorescence. SW1353 cells stably transfected with MMP28 constructs were probed for protein expression with an anti-FLAG primary antibody and an Alexa 488-conjugated secondary antibody. Cells were permeabilised (A to D) or non-permeabilised (E to H) to detect cell surface localisation of MMP-28. Wild-type cells treated with an IgG isotype control primary antibody were also included as a control, both permeabilised (I) and non-permeabilised (J). Cells were counterstained with DAPI and viewed under 20× magnification.

Fig. 5

Heparin competes for cell surface-associated MMP-28. SW1353 cells stably transfected with the wild-type MMP28 expression construct were probed, non-permeabilised, for protein expression with an anti-FLAG primary antibody and an Alexa 488-conjugated secondary antibody. Cells were either untreated (A, D) or treated with 0.1 mg/ml heparin (B, E). Wild-type cells treated with an IgG isotype control primary antibody were also included as a control (C). Cells were counterstained with DAPI and viewed under 20× magnification.

Fig. 6

Effect of over-expression of MMP-28 on SW1353 morphology. Phase contrast images of live SW1353 cells stably transfected with MMP28 cDNA constructs were obtained when the cells were at low confluence, 24 h after plating (A–D). Fixed cells were stained with phalloidin (E–H). Cells were viewed under 20× magnification.

Fig. 7

Effect of over-expression of MMP-28 on adhesion and migration of SW1353 cells. For adhesion experiments, two clones each of cells stably transfected with MMP28 cDNA constructs (wild-type (WT), EA mutant (EA), pro-cat, (PC)) or vector-only control (VO) were plated in wells of a 96-well plate coated with either 100 µl/well of 50 µg/ml fibronectin (A) or 100 µg/ml of collagen II (B). To study the role of the integrin β₁ subunit in mediating adhesion, cells were treated with anti-β₁ antibody for 30 min prior to plating on fibronectin (C) or collagen II (D). Cells were allowed to adhere for 1 h, followed by washing, fixing and staining with methylene blue. Absorbance of lysed cells was measured at 630 nm. Data were corrected for blank readings and control adhesion to 1% BSA, and are presented as fold-change relative to the mean of the empty vector controls. For migration experiments, stably transfected cells were plated in DMEM/10% serum in wells of a 24-well plate coated with 100 μg/ml collagen II (E). Media was switched to DMEM/0.5% serum after 24 h, and cell migration measured at 15 minute intervals over a 13 hour period. Data are expressed as μm per hour. Values for adhesion and migration experiments were compared to mean values for the empty vector control. Experiments were performed in triplicate, data are plotted as mean +/− s.e.m. Statistical significance indicated by: ⁎⁎⁎⁎, p < 0.0001; ⁎⁎⁎, p < 0.001: ⁎⁎, p < 0.01 and ⁎, p < 0.05.

Fig. 8

Effect of over-expression of MMP28 on expression of other MMP and TIMP family members. Expression of all of the members of the MMP and TIMP family, including MMP28, was measured using qRT-PCR and normalised to 18S rRNA in two clones each of cells stably transfected with MMP28 cDNA constructs (vector only (VO), wild-type (WT), EA mutant (EA) and pro-cat (PC)). Expression of each of the MMP28 constructs was verified (A) with MMP28 undetectable in the vector-only transfects. The expression of three genes altered in response to MMP28 over-expression: MMP2 (B), MMP19 (C) and TIMP3 (D). Data are plotted as mean +/− s.e.m. (n = 6). Statistical significance indicated by: ⁎⁎⁎⁎, p < 0.0001; ⁎⁎⁎, p < 0.001: ⁎⁎, p < 0.01 and ⁎, p < 0.05.

Fig. 9

Analysis of MMP-2 activity. Activity of MMP-2 was analyzed by gelatin zymography. Two clones each of cells stably transfected with empty vector or MMP28 cDNA constructs were plated in 6-well dishes. Medium was exchanged to serum-free 24 h after plating. Conditioned medium was collected for a further 48 h later and concentrated. Protein concentration was determined and equal amounts of protein for each sample were loaded (vector only (VO), wild-type (WT), EA mutant (EA) and pro-cat (PC)).

Acknowledgements

This study was funded by the Arthritis Research Campaign, UK by a grant to URR and IMC. We would also like to thank Vera Knäuper (University of Cardiff, Wales, UK) for mammalian expression vectors and Rosemary Bass (University of East Anglia, UK) for integrin primers.

URR, LK, AKS, JGW, TES and KC undertook all aspects of the experiments described; SI and JL provided essential reagents and intellectual input into the study design and write up; IMC and AEP conceived of and designed the study. URR and IMC wrote the manuscript.

References

1. ArnaoutM.A.GoodmanS.L.XiongJ.P.Structure and mechanics of integrin-based cell adhesionCurr. Opin. Cell Biol.192007495507[PubMed][Google Scholar]
2. ClarkI.M.CawstonT.E.Fragments of human fibroblast collagenase. Purification and characterizationBiochem. J.2631989201206[PubMed][Google Scholar]
3. DavidsonR.K.WatersJ.G.KevorkianL.DarrahC.CooperA.DonellS.T.ClarkI.M.Expression profiling of metalloproteinases and their inhibitors in synovium and cartilageArthritis Res. Ther.82006R124[PubMed][Google Scholar]
4. DufourA.SampsonN.S.ZuckerS.CaoJ.Role of the hemopexin domain of matrix metalloproteinases in cell migrationJ. Cell. Physiol.2172008643651[PubMed][Google Scholar]
5. EgebladM.WerbZ.New functions for the matrix metalloproteinases in cancer progressionNat. Rev., Cancer22002161174[PubMed][Google Scholar]
6. GaoG.WestlingJ.ThompsonV.P.HowellT.D.GottschallP.E.SandyJ.D.Activation of the proteolytic activity of ADAMTS4 (aggrecanase-1) by C-terminal truncationJ. Biol. Chem.27720021103411041[PubMed][Google Scholar]
7. IllmanS.A.Keski-OjaJ.LohiJ.Promoter characterization of the human and mouse epilysin (MMP-28) genesGene2752001185194[PubMed][Google Scholar]
8. IllmanS.A.Keski-OjaJ.ParksW.C.LohiJ.The mouse matrix metalloproteinase, epilysin (MMP-28), is alternatively spliced and processed by a furin-like proprotein convertaseBiochem. J.3752003191197[PubMed][Google Scholar]
9. IllmanS.A.LehtiK.Keski-OjaJ.LohiJ.Epilysin (MMP-28) induces TGF-beta mediated epithelial to mesenchymal transition in lung carcinoma cellsJ. Cell Sci.119200638563865[PubMed][Google Scholar]
10. IllmanS.A.LohiJ.Keski-OjaJ.Epilysin (MMP-28)–structure, expression and potential functionsExp. Dermatol.172008897907[PubMed][Google Scholar]
11. KashiwagiM.EnghildJ.J.GendronC.HughesC.CatersonB.ItohY.NagaseH.Altered proteolytic activities of ADAMTS-4 expressed by C-terminal processingJ. Biol. Chem.27920041010910119[PubMed][Google Scholar]
12. KevorkianL.YoungD.A.DarrahC.DonellS.T.ShepstoneL.PorterS.BrockbankS.M.EdwardsD.R.ParkerA.E.ClarkI.M.Expression profiling of metalloproteinases and their inhibitors in cartilageArthritis Rheum.502004131141[PubMed][Google Scholar]
13. KnauperV.CowellS.SmithB.Lopez-OtinC.O'SheaM.MorrisH.ZardiL.MurphyG.The role of the C-terminal domain of human collagenase-3 (MMP-13) in the activation of procollagenase-3, substrate specificity, and tissue inhibitor of metalloproteinase interactionJ. Biol. Chem.272199776087616[PubMed][Google Scholar]
14. KooB.H.LongpreJ.M.SomervilleR.P.AlexanderJ.P.LeducR.ApteS.S.Cell-surface processing of pro-ADAMTS9 by furinJ. Biol. Chem.28120061248512494[PubMed][Google Scholar]
15. KozakM.An analysis of 5′-noncoding sequences from 699 vertebrate messenger RNAsNucleic Acids Res.15198781258148[PubMed][Google Scholar]
16. KunoK.MatsushimaK.ADAMTS-1 protein anchors at the extracellular matrix through the thrombospondin type I motifs and its spacing regionJ. Biol. Chem.27319981391213917[PubMed][Google Scholar]
17. KurtisM.S.SchmidtT.A.BugbeeW.D.LoeserR.F.SahR.L.Integrin-mediated adhesion of human articular chondrocytes to cartilageArthritis Rheum.482003110118[PubMed][Google Scholar]
18. LinM.H.LiuS.Y.SuH.J.LiuY.C.Functional role of matrix metalloproteinase-28 in the oral squamous cell carcinomaOral Oncol.422006907913[PubMed][Google Scholar]
19. LoeserR.F.SadievS.TanL.GoldringM.B.Integrin expression by primary and immortalized human chondrocytes: evidence of a differential role for alpha1beta1 and alpha2beta1 integrins in mediating chondrocyte adhesion to types II and VI collagenOsteoarthr. Cartil.8200096105[PubMed][Google Scholar]
20. LohiJ.WilsonC.L.RobyJ.D.ParksW.C.Epilysin, a novel human matrix metalloproteinase (MMP-28) expressed in testis and keratinocytes and in response to injuryJ. Biol. Chem.27620011013410144[PubMed][Google Scholar]
21. Lopez-OtinC.MatrisianL.M.Emerging roles of proteases in tumour suppressionNat. Rev., Cancer72007800808[PubMed][Google Scholar]
22. MalfaitA.M.ArnerE.C.SongR.H.AlstonJ.T.MarkosyanS.StatenN.YangZ.GriggsD.W.TortorellaM.D.Proprotein convertase activation of aggrecanases in cartilage in situArch. Biochem. Biophys.47820084351[PubMed][Google Scholar]
23. MarchenkoG.N.StronginA.Y.MMP-28, a new human matrix metalloproteinase with an unusual cysteine-switch sequence is widely expressed in tumorsGene26520018793[PubMed][Google Scholar]
24. MartinM.D.MatrisianL.M.The other side of MMPs: protective roles in tumor progressionCancer Metastasis Rev.262007717724[PubMed][Google Scholar]
25. MauchS.KolbC.KolbB.SadowskiT.SedlacekR.Matrix metalloproteinase-19 is expressed in myeloid cells in an adhesion-dependent manner and associates with the cell surfaceJ. Immunol.168200212441251[PubMed][Google Scholar]
26. NuttallR.K.PenningtonC.J.TaplinJ.WhealA.YongV.W.ForsythP.A.EdwardsD.R.Elevated membrane-type matrix metalloproteinases in gliomas revealed by profiling proteases and inhibitors in human cancer cellsMol. Cancer Res.12003333345[PubMed][Google Scholar]
27. PellegrinS.MellorH.Actin stress fibresJ. Cell Sci.120200734913499[PubMed][Google Scholar]
28. PorterS.ScottS.D.SassoonE.M.WilliamsM.R.JonesJ.L.GirlingA.C.BallR.Y.EdwardsD.R.Dysregulated expression of adamalysin–thrombospondin genes in human breast carcinomaClin. Cancer Res.10200424292440[PubMed][Google Scholar]
29. RaH.J.ParksW.C.Control of matrix metalloproteinase catalytic activityMatrix Biol.262007587596[PubMed][Google Scholar]
30. RussellD.L.DoyleK.M.OchsnerS.A.SandyJ.D.RichardsJ.S.Processing and localization of ADAMTS-1 and proteolytic cleavage of versican during cumulus matrix expansion and ovulationJ. Biol. Chem.27820034233042339[PubMed][Google Scholar]
31. Saarialho-KereU.KerkelaE.JahkolaT.SuomelaS.Keski-OjaJ.LohiJ.Epilysin (MMP-28) expression is associated with cell proliferation during epithelial repairJ. Invest. Dermatol.11920021421[PubMed][Google Scholar]
32. WernerS.R.MescherA.L.NeffA.W.KingM.W.ChaturvediS.DuffinK.L.HartyM.W.SmithR.C.Neural MMP-28 expression precedes myelination during development and peripheral nerve repairDev. Dyn.236200728522864[PubMed][Google Scholar]
33. WernerS.R.DotzlafJ.E.SmithR.C.MMP-28 as a regulator of myelinationBMC Neurosci.9200883[PubMed][Google Scholar]
34. WilliamsR.O.Collagen-induced arthritis in miceMethods Mol. Med.1362007191199[PubMed][Google Scholar]