Review
The regulation of matrix metalloproteinases and their inhibitors

https://doi.org/10.1016/j.biocel.2007.12.006Get rights and content

Abstract

The matrix metalloproteinases (MMP) are a family of 23 enzymes in man. These enzymes were originally described as cleaving extracellular matrix (ECM) substrates with a predominant role in ECM homeostasis, but it is now clear that they have much wider functionality. Control over MMP and/or tissue inhibitor of metalloproteinases (TIMP) activity in vivo occurs at different levels and involves factors such as regulation of gene expression, activation of zymogens and inhibition of active enzymes by specific inhibitors. Whilst these enzymes and inhibitors have clear roles in physiological tissue turnover and homeostasis, if control of their expression or activity is lost, they contribute to a number of pathologies including e.g. cancer, arthritis and cardiovascular disease. The expression of many MMPs and TIMPs is regulated at the level of transcription by a variety of growth factors, cytokines and chemokines, though post-transcriptional pathways may contribute to this regulation in specific cases. The contribution of epigenetic modifications has also been uncovered in recent years. The promoter regions of many of these genes have been, at least partly, characterised including the role of identified single nucleotide polymorphisms. This article aims to review current knowledge across these gene families and use a bioinformatic approach to fill the gaps where no functional data are available.

Introduction

The matrix metalloproteinases (MMPs) are a family of 23 enzymes in man. These enzymes were originally described as cleaving extracellular matrix (ECM) substrates with a predominant role in ECM homeostasis, but it is now clear that they have much wider functionality. Such functions include: opposing effects on angiogenesis via matrix degradation but also release of angiogenesis inhibitors (via digestion of e.g. plasminogen to generate angiostatin, and type XVIII collagen to generate endostatin); regulation of cell growth via cleavage of cell surface-bound growth factors and receptors, release of growth factors sequestered in the ECM or integrin signalling; regulation of apoptosis via release of death or survival factors; alteration of cell motility by revealing cryptic matrix signals, or cleavage of adhesion molecules; effects on the immune system and host defense; modulation of the bioactivity of chemokines (Cauwe, Van den Steen, & Opdenakker, 2007).

Traditionally, the MMPs have been subdivided into collagenases, gelatinases, stromelysins, and membrane-type MMPs, according to their substrate specificity, primary structure and cellular location. However, several more recently discovered enzymes do not fit easily into these classifications and alternatives based on domain structure, activation mechanism and cellular location have been suggested e.g. (Brinckerhoff & Matrisian, 2002). The MMPs have a common domain structure with a signal peptide, a propeptide, a catalytic domain, a hinge region and, in the majority of cases, a C-terminal domain. The propeptide contains an invariant Cys residue (in all but MMP-23), which ligates the active site zinc ion to maintain latency; the catalytic domain contains a HEXGHXXGXXH zinc-binding sequence characteristic of the metzincin superfamily of proteinases, followed by an invariant methionine which is involved in a structural feature called the ‘Met-turn’. In all family members except MMP-7(matrilysin-1), MMP-23 (CA-MMP) and MMP-26 (matrilysin-2), a hinge region links to a haemopexin-like C-terminal domain which is thought to be involved in substrate specificity and binding of inhibitors. Individual MMPs contain variations on this theme e.g.: membrane-type (MT)-MMPs (MMPs 14–17, -24 and -25) have either a transmembrane domain and cytoplasmic tail at the C-terminus (MMP 14–16 and -24) or are GPI-anchored (MMP-17 and -25). In common with MMP-11, -21, -23 and -28, the MT-MMPs also contain a potential furin-cleavage site within the propeptide allowing activation prior to secretion. The gelatinases (MMP-2 and -9) have an insert of three-fibronectin type II repeats in the catalytic domain and MMP-9 also has a collagen-like sequence at one end of the catalytic domain.

A family of four specific inhibitors, the TIMPs, has been described (Baker, Edwards, & Murphy, 2002). Whilst the ability of the four TIMPs to inhibit MMPs is largely promiscuous, a number of functional differences have been noted, e.g. TIMP-2, -3 and -4, but not TIMP-1, are effective inhibitors of the MT-MMP subclass and MMP-19.

Control over MMP and/or TIMP activity in vivo occurs at different levels and involves factors such as regulation of gene expression, activation of zymogens and inhibition of active enzymes by specific inhibitors. Many MMPs and TIMPs are regulated at the level of transcription by a variety of growth factors, cytokines and chemokines e.g. (Baker et al., 2002, Yan and Boyd, 2007).

In addition to showing characteristic patterns of tissue- and developmental stage-specific patterns of expression (Nuttall et al., 2004), particular cell types display signal-dependent activation and repression of MMP and TIMP gene transcription involving the mitogen-activated protein kinase (MAPK), nuclear factor-κB (NFκB) and Smad-dependent pathways (Baker et al., 2002, Borden and Heller, 1997; Hall et al., 2003, Overall and Lopez-Otin, 2002; Reunanen et al., 2002, Vincenti and Brinckerhoff, 2002; Westermarck, Li, Kallunki, Han, & Kahari, 2001). Most MMPs and TIMPs respond to stimuli at the transcriptional level with delayed kinetics over a timeframe of several hours and require ongoing translation (Sampieri et al., 2008). This suggests that they are components of genetic programmes such as the wound repair response, in which they are downstream targets of immediate-early (IE) response genes that are induced within minutes of cell stimulation and in the absence of new protein synthesis. Chief among these IE genes are the Fos and Jun genes that comprise activator protein-1 (AP-1) and at the level of the individual gene promoter, a promoter-proximal AP-1 site was the first cis-acting element to be implicated in the induction of MMP1 expression (Angel et al., 1987). A number of MMP promoters have now been at least partly characterized, revealing a variety of functional cis-acting elements (e.g. AP-1, PEA3, Sp1, Tcf/Lef-1, NFκB, RARE).

A recent review on MMP gene expression (Yan & Boyd, 2007) uses the basic promoter conformation to assign MMPs to one of three groups (i) those which contain TATA boxes at around −30 bp with AP-1 sites around −70 bp (MMP1, MMP3, MMP7, MMP9, MMP10, MMP12, MMP13, MMP19 and MMP26); (ii) those which contain a TATA box, but no promoter-proximal AP-1 site (MMP8, MMP11 and MMP21); and (iii) those with no TATA box nor proximal AP-1 (MMP2, MMP14 and MMP28).

A bioinformatic analysis across the MMP family proves problematic since many known functional sites, including e.g. the promoter-proximal PEA3 site in MMP1, are non-canonical. Furthermore, the transcription start point is unknown for many MMPs and can only be inferred by examination of the longest known mRNA. However, using such methodology to analyse 1 kb of upstream sequence for each MMP and TIMP gene (Fig. 1) adds MMP20 to group one (AP-1 and TATA containing); MMP15 and MMP27 to group two (TATA but no AP-1); and MMP16, MMP17, MMP23, MMP24 and MMP25 to group three (no TATA, no AP-1). Of these, the assignment of the promoter region for MMP23, MMP24 and MMP25 is most tenuous from the databases. As amongst most multi-gene families, the MMPs can be co-regulated under a variety of stimuli, and the above classification is certainly useful, with e.g. many of the group 1 genes (TATA box and proximal AP-1 site) induced by interleukin-1 (IL-1), tumour necrosis factor alpha (TNFα) or phorbol ester treatment in many cell types. However, it may disguise the inherent complexity which underlies the regulation of these genes and the subtlety by which differential regulation can be achieved.

Even accepting the promoter-proximal AP-1 site as a major mediator of regulation, specificity across genes or cell types remains possible. Firstly, the composition of the AP-1 complex itself may determine the response: early work indicated that c-Jun was able to induce MMP1 minimal promoter constructs (in F9 cells which lack endogenous AP-1) whilst JunB was reported as a repressive factor in the absence of c-Fos, but an inducing factor in the presence of c-Fos (Chiu, Angel, & Karin, 1989). The influence of the promoter-proximal AP-1 site may also change dependent on the construct under assay. Originally examined in the rabbit MMP1 promoter, a second, non-canonical AP-1 site was identified at −186 bp which altered the response of the promoter to phorbol ester and removed some of the dependency on the promoter-proximal (−77 bp) AP-1 site. The AP-1 complex at both sites contained c-Fos and JunD, but Fra-2 was only present at the −77 bp site (White & Brinckerhoff, 1995). Other proteins may interact with the AP-1 site. Nucleolin was recently identified as such a protein, and chromatin immunoprecipitation (ChIP) shows it binding in the vicinity of the AP-1 site within the MMP13 promoter. Moreover, overexpression of nucleolin can repress transactivation of the MMP13 promoter transiently transfected into HeLa cells (Samuel, Twizere, Beifuss, & Bernstein, 2008).

Further complexity can be generated by the juxtaposition of transcription factor binding sites or indeed the presence of composite sites. Fra-1 alone was recently shown to be unable to transactivate the human MMP9 promoter in MCF7 cells. Promoter sequence analysis showed a STAT3 binding site in juxtaposition with the AP-1 site and indeed, the co-transfection of Fra-1 and Stat3C strongly transactivated the promoter. A complex of c-Jun/Fra-1 and Stat3 was identified on this region of the promoter (Song et al., 2008). In the MMP1 promoter, the proximal AP-1 site has been reported to be within a composite element with at least a peroxisome proliferator-activated receptor gamma (PPARγ) consensus whereby PPARγ and AP-1 bind the element in a mutually exclusive manner (Fahmi et al., 2002, Francois et al., 2004). Burrage, Huntington, Sporn, and Brinckerhoff (2007) extended this observation to show that the promoter-proximal AP-1 site in the MMP1 and MMP13 promoters overlap a degenerate direct repeat (DR)1 nuclear hormone response element with similar elements found more distally in other MMP genes. Our own computer analysis in Fig. 1 shows such composite AP-1/RAR-RXR elements in the upstream 1 kb regions of at least MMP7, MMP9 and MMP19 genes.

Interestingly, MMP11 appears unique amongst characterised MMP promoters in having a functional retinoic acid response element (DR1-RARE) in its proximal promoter mediating its induction by retinoic acid. Two upstream DR2-RAREs are also present and appear to contribute to basal expression from the gene promoter (Ludwig, Basset, & Anglard, 2000). Bioinformatic analysis suggests several MMP and TIMP promoters contain canonical retinoic acid or retinoid X receptor response elements within the upstream 1 kb region (MMP1, MMP2, MMP7, MMP9, MMP11, MMP13, MMP14, MMP17, MMP19, MMP21, MMP23, MMP26, TIMP1, TIMP3 and TIMP4). Interestingly, these are almost all of the DR1 spacing, the most promiscuous of such sites, known to recruit at least RAR/RXR, RXR/RXR, and PPAR/RXR in vitro (Ijpenberg et al., 2004; Durand, Saunders, Leroy, Leid, & Chambon, 1992; Mader, Leroy, Chen, & Chambon, 1993; Mangelsdorf et al., 1991).

In many MMP promoters containing proximal AP-1 sites, a PEA3-binding consensus is found in close proximity upstream, and these sites act cooperatively. Again, since many of these sites are able to bind multiple Ets factors, they can mediate specificity. MMP1 expression was shown to be differentially regulated by different structural classes of Ets factors e.g. Ets1, ErgB/Fli-1 and Pu1. Ets1 increased basal activity of 3.8 kb MMP1 promoter, whilst the others had no effect. Ets1 augmented transactivation by both c-Jun and JunB, ErgB augmented transactivation only via JunB, whilst Pu1 repressed induction by both c-Jun and JunB. As above, all of these effects were dependent on the promoter-proximal AP-1 site (Westermarck, Seth, & Kahari, 1997).

Many MMPs have multiple GC boxes in their proximal promoters which bind Sp1 and Sp3 and potentially other GC-binding proteins. The MMPs without other obvious promoter features are generally expressed in a more constitutive fashion e.g. MMP2 and MMP14. Indeed, the promoter-proximal (−92 bp) Sp1 site in the MMP14 promoter is crucial in maintaining expression of this gene since its mutation reduces promoter activity by approximately 90% (Lohi, Lehti, Valtanen, Parks, & Keski-Oja, 2000). Our bioinformatic analysis would add MMP15, MMP16, MMP17, MMP23, MMP24, MMP25, TIMP2 and possibly MMP27 and TIMP3 to this list. However, these genes are still clearly regulated and a number of promoter dissections show responsive sites outside of the GC boxes.

In astroglioma cells, an Sp1 site at −91/−84 bp and an AP-2 site at −61/−53 bp mediate activation of the MMP2 gene by Sp1, Sp3 and AP-2 factors (Qin, Sun, & Benveniste, 1999). MMP2 was also shown to be a target of p53 transactivation via a p53 site at −1659/−1622 bp (Bian & Sun, 1997; Yan, Wang, & Boyd, 2002). Stat3 has also been linked with the induction of MMP2 in cells from metastatic melanoma (Xie et al., 2004) via a consensus site in the promoter at −617/−610 bp. Other Stat3 binding sites have been identified in the MMP2 promoter within the upstream enhancer sequence (denoted RE-1 in the rat promoter) (En-Nia et al., 2002). Interestingly, in an ischaemia model, Fos and Jun proteins were shown to bind to a non-canonical AP-1 site in the MMP2 promoter, with p53 binding to an adjacent enhancer site (RE-1) and NFAT-c2 binding within intron 1. Deletion of the former or substitution of the latter led to decreased expression in transgenic mice expressing beta-galactosidase from an MMP2 promoter/intron 1 construct (Lee et al., 2005). The upstream (−1394 bp) AP-1 site had previously been described as important for the induction of MMP2 expression by hypoxia in cardiac cells with binding of Fra1-JunB or FosB-JunB heterodimers (Bergman et al., 2003). This belies the idea that MMP2 is constitutively produced and not regulated by AP-1 complexes.

Egr1, like Sp1 and Sp3, binds GC-rich sequences, and both specificity of binding to cognate sites and cross-competition has also been demonstrated (Al-Sarraj, Day, & Thiel, 2005)). Egr1 has been implicated in the regulation of e.g. MMP14 in cells cultured in three-dimensional matrices (Barbolina, Adley, Ariztia, Liu, & Stack, 2007; Haas, Stitelman, Davis, Apte, & Madri, 1999) or MMP14 and MMP2 by cigarette smoke in lung fibroblasts (Ning et al., 2007).

In mesangial cells, MMP14 expression is regulated by proximal and overlapping Sp1- and Sp1/Egr1-binding sites as well as a more distal site for NFATc1 (Alfonso-Jaume, Mahimkar, & Lovett, 2004).

NFκB is known to regulate many MMP genes (Vincenti & Brinckerhoff, 2007). An NFκB site was originally identified in the MMP9 promoter (He, 1996) with a potential role in TNFα induction of the gene. NFκB was also shown as essential for the synergistic induction of MMP9 expression by growth factors and inflammatory cyokines (Bond, Fabunmi, Baker, & Newby, 1998). MMP1 is also regulated by NFκB, with a non-canonical site binding NFκB1 at −2886 bp in the rabbit promoter (Vincenti, Coon, & Brinckerhoff, 1998). In other MMP genes induced by NFκB, no functional NFκB sites have been identified: again, such sites could simply be non-canonical or some of these effects may be indirect. A number of MMP and TIMP proximal promoters show canonical NFκB sites on sequence analysis (MMP1, MMP2, MMP8, MMP9, MMP11, MMP13, MMP14, MMP15, MMP17, MMP19, MMP23, MMP25, MMP26, TIMP2 and TIMP4).

MMP7 was the first MMP shown to be regulated by Wnt signalling being the target of beta-catenin regulation via Tcf/Lef-1. The human MMP7 promoter has two upstream Tcf/Lef-1 binding sites at −109 bp and −194 bp which mediate transactivation of the promoter by beta-catenin (Gustavson, Crawford, Fingleton, & Matrisian, 2004). Tcf/Lef-1 synergises with other factors, particularly those of the PEA3 family, to drive expression of MMP7 in colorectal tumours (Crawford et al., 2001). MMP14 has also been identified as a target of beta-catenin signalling in colorectal cancers with a Tcf/Lef-1 site at −1169/−1163 bp mediating this effect (Takahashi, Tsunoda, Seiki, Nakamura, & Furukawa, 2002). MMP26, a poorly expressed gene in most cell types, is also reported to have a Tcf motif in its promoter (Marchenko, Marchenko, Leng, & Strongin, 2002). Bioinformatic analysis of the upstream 1 kb region reveals canonical Tcf/Lef sites in MMP1, MMP3, MMP7, MMP8, MMP9, MMP10, MMP14, MMP19, MMP21, MMP23, MMP26, MMP27, MMP28 and TIMP4. Thus, these are common sequences within these genes, though most are not functionally verified.

The MMP13 gene is regulated in part by a promoter-proximal Runx2 (aka Cbfa1, AML3) site which directs its expression in osteoblasts and hypertrophic chondrocytes (Jimenez et al., 1999, Porte et al., 1999). In stably transfected chondrocytic cell lines, IL-1 induces MMP13 expression via both promoter-proximal AP-1 and Runx2 sites (Mengshol, Vincenti, & Brinckerhoff, 2001). Transgenic mice expressing beta-galatosidase from −148 bp of the rat MMP13 promoter, containing both the AP-1 and Runx2 sites, display expression in bone, teeth and skin at levels that are elevated compared to transgenic lines in which these sites are mutated. This shows that the promoter-proximal AP-1 and Runx2 sites in the MMP13 promoter are necessary and sufficient to direct expression in these tissues in vivo, though may not be sufficient to recapitulate all aspects of endogenous MMP13 expression (Selvamurugan et al., 2006). Aside from MMP13, MMP9 expression is also regulated by Runx2 via a promoter-proximal consensus and three non-canonical upstream Runx2 sequences (Pratap et al., 2005). This fits with a role for MMP-9 in endochondral ossification revealed by the growth plate phenotype of the MMP9 null mouse (Ortega, Behonick, & Werb, 2004). Interestingly, bioinformatic analysis shows canonical Runx2 sites in the upstream 1 kb regions of MMP1, MMP7, MMP8, MMP13, MMP17, MMP21, MMP26, TIMP1, TIMP2 and TIMP3 though, of course, these await functional analysis. Runx2 may direct expression in hypertrophic chondrocytes and osteoblasts, but MMP13 and indeed most MMPs do not show true tissue specificity. MMP20 is the prime example, being expressed almost exclusively in tissues of the tooth. The bioinformatic analysis of 1 kb of upstream sequence (above) shows two canonical AP-1 sites and this has been backed up experimentally in a recent paper which shows four potential AP-1 sites within 1.5 kb of promoter with three of them binding c-Jun (Zhang, Li, Chi, Chen, & Denbesten, 2007).

Similar to the group 1 MMPs, TIMP1 has a promoter-proximal AP-1 site (though non-canonical) with a neighbouring PEA3 site as major mediators of transcription. Logan, Garabedian, Campbell, and Werb (1996) demonstrated that the proteins binding to the AP-1 site at −92/−86 bp may interact with those at the PEA-3 site at −79/−74 bp to enhance transcription driven from the whole element. Interestingly, Phillips, Sharma, Leco, and Edwards (1999) showed that the TIMP1 AP-1 site was also able to bind a single-stranded DNA-binding protein whereas the (canonical) MMP1 AP-1 site did not. TIMP1 is both constitutively expressed and inducible and this may explain its lack of a TATA box, multiple transcription start points and GC boxes. A number of other consensus sites have also been shown to be functional within the promoter (e.g. upstream TIMP element-1 which binds Runx1 and Runx2 proteins at −62/−52 bp (Bertrand-Philippe et al., 2004, Trim et al., 2000), a hypoxia response element at −27/−23 bp, an Egr1 site at −33 bp (Aicher et al., 2003). The TIMP1 gene differs from other TIMP family members in having a short first exon which is transcribed, but not translated, with the translation start site located on exon 2. There is evidence that regulatory sequences exist within the first intron of the Timp1 gene. Flenniken and Williams (1990) found that a construct containing around 1.3 kb of murine TIMP-1 5′ flanking sequence, exon 1 and most of intron 1 linked to a lacZ reporter in transgenic mice was sufficient to reproduce the spatial and temporal expression of the Timp1 gene in developing mouse embryos. In contrast to this, transgenic mice carrying lacZ linked to 2.7 kb of Timp1 5′ flanking sequence, but lacking intron 1, display a subset of the correct pattern of expression (e.g. appropriate expression in the developing vertebral column, and absence in the liver), but also inappropriate expression of the reporter in sites such as the spinal cord (D.R. Edwards, unpublished data). Thus, sequences within intron 1 are likely to repress Timp1 gene expression and this was demonstrated in vitro by Dean, Young, Edwards, and Clark (2000) who showed at least Sp1, Sp3 and Ets-factor binding to a repressive cis-acting sequence in the intron.

The TIMP2 gene has a TATA box at around −30 bp, though still displays multiple transcription start points. It contains several Sp1 sequences, characteristic of a housekeeping gene, two AP-2 sites, three PEA3 sites and an AP-1 site (at −281 bp) (Hammani et al., 1996). Interestingly, the AP-1 site does not lead to induction of the TIMP2 promoter by phorbol ester in the same way as does the more proximal site in the TIMP1 promoter. However, insertion of an AP-1 consensus at position −71 bp in the TIMP2 gene generates phorbol induction of the mutant. An Sp1 and Sp3-binding motif at −107/−98 bp cooperates with an inverted CCAAT box, binding NF-Y, at 73/−69 bp for both basal expression from the TIMP2 promoter and its induction by cAMP (Zhong, Hammani, Bae, & DeClerck, 2000).

The murine Timp3 gene has a TATA box and single transcription start point. The proximal (approximately 500 bp) promoter has three GC boxes, and more distally (−600 to −2000 bp) contains six AP-1 consensus sequences, two NFκB sites and two p53 sites (Sun et al., 1995). This structure is reflected in the activity of promoter fragments in transient transfection experiments where a −2846/ + 58 bp construct is inducible by TNFα or phorbol ester, but a −491/ + 23 bp construct is not inducible by these factors. Interestingly, this group also found that Timp3 is poorly expressed in a number of neoplastic cell lines and that the promoter in these lines can be hypermethylated or hypomethylated. Treatment with a DNA methyl transferase inhibitor leads to reexpression of Timp3 only in the line displaying promoter hypermethylation. Similarly, the human TIMP3 promoter displays a promoter-proximal GC-rich region which is capable of mediating basal expression, with regions further upstream mediating serum induction (Wick et al., 1995).

Amongst the TIMPs, TIMP4 displays tight tissue specificity, predominantly in the heart and brain (Leco et al., 1997). The TIMP4 promoter has no TATA box, but does contain an initiator sequence and the gene displays multiple transcription start points (Young et al., 2002). The proximal promoter contains an inverted CCAAT box and an Sp1 site. Mutation of the former causes an increase in reporter expression. More significantly, mutation of either the initiator sequence or the Sp1 site abolishes reporter expression completely.

Acetylation is a post-translational covalent protein modification that is strongly implicated in transcriptional regulation. Histones were the first proteins identified as showing variable acetylation status. Simplistically, acetylation weakens the histone:DNA interaction, allowing access to transcription factors and therefore generally associated with gene activation. This has been followed by a plethora of molecules ranging from structural proteins, intracellular signaling molecules, nuclear membrane receptors and transcription factors that have been shown to be acetylated. Acetylation, like phosphorylation, is a reversible modification with acetyl groups added by a family of histone acetyl transferase enzymes (HATs) and removed by histone deacetylases (HDACs). Inhibitors of the so-called classical HDACs (HDACs 1–11) are in development as cancer therapeutics as they have potent anti-proliferative and pro-apoptotic activities in cancer cells.

The role of acetylation in the expression of MMPs has been probed for a number of genes. An early report of acetylation impacting upon MMP gene expression was by Pender, Quinn, Sanderson, and MacDonald (2000) working with human fetal mucosal mesenchymal cells. In this paper, HDAC inhibitors (HDACi) were shown to enhance IL-1β or TNFα induction of MMP3 and to repress IL-1β or TNFα induction of MMP1 and MMP9 both at the mRNA and protein level. No effect of these HDACi was seen on unstimulated levels of any MMP examined, and no effect was seen on MMP2 or TIMP1, neither of which was induced by IL-1β or TNFα. From these data, it is likely that the effect of HDACi is on the signaling pathways induced by these proinflammatory cytokines, rather than on the MMP genes themselves, though this is currently unknown. It is interesting that the same HDACi can lead to enhancement or repression of an induced MMP (compare MMP3 to MMP1 and MMP9 above): this might suggest that there are differences in the pathways by which IL-1 or TNFα induce MMP3 compared to MMP1 and MMP9 or that there are subtle differences in the impact of HDACi on e.g. NFκB that are gene specific.

Ailenberg and Silverman, 2002, Ailenberg and Silverman, 2003 then described that HDAC inhibitors could repress MMP2 gene expression at the mRNA level in NIH 3T3 cells. In this cell line, TSA did not alter expression of either MMP14 or TIMP2, factors known to be involved in proMMP-2 activation. MMP2 expression in the human fibrosarcoma cell line HT1080 was much less sensitive to HDACi treatment, even at high concentrations. Similarly, Kaneko et al. (2004) demonstrate both reduced invasion through Matrigel, and reduced MMP expression and activity, in response to HDACi. In human liver cancer cell lines, treatment with HDACi reduced activity of MMP-2 and MMP-9 and expression of MMP1. The expression of TIMP1 and TIMP2 was unaffected. Since the effect of HDACi on MMP levels was similar to that of interferon-α, the authors speculate on cross-talk or commonality between the signalling pathways activated by these factors.

These papers do not dissect in detail the mechanism(s) by which HDACi impact upon MMP gene expression in terms of the HDAC involved, nor in terms of the number of MMPs affected by an HDACi. Yan, Wang, Toh, and Boyd (2003) showed that the metastasis-associated gene MTA1, a component of the NuRD repressor complex binds to −650/−450 bp of the MMP9 promoter, recruiting HDAC2 and decreasing histone acetylation and thereby gene expression. An HDAC independent mechanism involving the Mi-2 nucleosomal remodelling activity was also postulated. Martens, Verlaan, Kalkhoven, and Zantema (2003) considered the induction of the MMP1 gene in T98G human glioblastoma cells by a combination of serum and phorbol ester. This combination gives a potent induction of MMP1 expression at the mRNA level between 90 min and 2 h with a maximal induction at 4–6 h. These authors used ChIP to identify factors recruited to the MMP1 promoter in this system and examined the kinetics of histone modifications occurring with induction of the gene. Acetylation per se is insufficient to induce MMP1 expression in these cells since HDACi had no effect on MMP1 expression despite an increase in local H3 acetylation at the MMP1 promoter. It appears that upon activation with serum and phorbol ester, c-Jun, c-Fos, TBP, RNAPII and SET9 assemble on the MMP1 promoter and histones at this location are dimethylated (and eventually trimethylated). p300 and RSK2 are then recruited which correlates with an increase in acetylation and phosphorylation of histones at the MMP1 promoter. Finally, Swi/Snf, an ATP-dependent nucleosome remodeling complex is recruiting allowing initiation of transcription.

In a similarly detailed dissection of events at the promoter, Ma, Shah, Chang, and Benveniste (2004) examined PMA induction of the MMP9 gene in HeLa cells. Increased MMP9 mRNA is apparent by 2 h of PMA treatment and becomes maximal at 6 h. An NFκB site, two AP-1 sites and an Sp1 site in the MMP9 promoter are all involved in induction. Micrococcal nuclease digestion demonstrates that the MMP9 promoter is in a regular nucleosomal array, and chromatin remodeling is necessary for MMP9 transcription. This latter is elegantly proven by using a Brg1-deficient cell line (SW-13 cells); Brg1 is the ATPase subunit of the Swi/Snf complex. In these cells, MMP9 is not expressed on PMA stimulation, but transfection with Brg1 (but not an ATPase-null mutant) rescued responsiveness of MMP9 expression to PMA. ChIP experiments demonstrated recruitment of AP-1 factors, NFκB factors and Sp1 at appropriate time points, and modifications in histone acetylation, phosphorylation and methylation. MTA, a specific protein methyltransferase inhibitor, suppressed PMA-induced MMP9 expression, whilst HDACi enhance PMA-induced MMP9 expression. These authors also demonstrated that both HDAC1- and HDAC3-containing complexes occupy the MMP9 promoter in unstimulated HeLa cells, but that these are removed upon stimulation with PMA. This was reinforced by the repressive effects of HDAC1 or HDAC3 (but not HDAC2 or HDAC4) on PMA-induced MMP9. Both of these papers (Ma, Shah et al., 2004; Martens et al., 2003) demonstrate a coordinated cascade of cell signaling, histone modifications, nucleosome remodeling and recruitment of transcription factors. This may be cell type, species, or stimulus-specific. Ma, Chang et al. (2004) also showed that Brg-1, the ATPase subunit of the Swi/SNF complex impacts upon MMP2 gene expression. Using SW-13 cells and adding back Brg-1 demonstrated that this factor is required for recruiting Sp1, AP-2 and RNA polymerase II to the MMP2 promoter, with Sp3-binding decreased.

Using HDAC7 null mice, Chang et al. (2006) show that HDAC7 maintains vascular integrity in early embryogenesis via repression of MMP10. HDAC7 (or indeed HDAC5) binds to MEF2, shown to be a transcriptional activator of MMP10 to repress expression of the latter.

Young, Lakey et al. (2005) show that HDACi block IL-1/OSM-induced cartilage resorption in an explant model in a dose-dependent manner at the level of both proteoglycan and collagen release. This was accompanied by a reduction in activity and activation of procollagenases in the conditioned medium and a similar reduction in gelatinase activity. In cultured primary chondrocytes, HDACi repressed the IL-1/OSM-induction of collagenolytic MMPs (MMP1 and 13). In SW1353 chondrosarcoma cells, the majority of metalloproteinase genes that were robustly induced by the IL-1/OSM combination, were then repressed by the HDACi; these were MMP1, MMP3, MMP7, MMP8, MMP10, MMP12 and MMP13. The basal expression of three metalloproteinase genes (MMP17, MMP23 and MMP28) was also induced by HDACi. These data suggest that HDACi impact upon IL-1/OSM signaling to repress induced levels of metalloproteinases and that the basal expression of a number of metalloproteinases is responsive to HDACi. There is no overlap between these groups, which may point to mechanisms/pathways by which groups of genes are coexpressed in these enzyme families.

At the simplest level, acetylation of histones at a gene promoter may increase access to cis-acting sequences by their cognate transcription factors. However, acetylation and deacetylation of signaling pathway components and transcription factors themselves makes the situation complex. This complexity is undoubtedly increased by cell specific differences in some of these events.

Young, Billingham, Sampieri, Edwards, and Clark (2005) detail the differential effects of HDACi on the TIMP1 gene dependent on the stimulus used to induce the gene. HDACi enhance PMA-induced TIMP1 expression but repress TGFβ-induced TIMP1 expression. Interestingly, the dose (of HDACi)-response curves for these two effects are markedly different. This strongly suggests that different HDACs are the target of HDACi in each case. Furthermore, the effect of HDACi on the endogenous TIMP1 gene can be reiterated at the level of transient transfection of promoter-reporter constructs. This perhaps suggests that their effects are more likely mediated by acetylation of signaling molecules or transcription factors rather than histones.

As discussed above, many inducible MMP genes contain promoter-proximal AP-1 sites that are key features of their inducibility. In order for c-Jun to activate target gene transcription, it requires phosphorylation by the Jun-N-terminal kinase (JNK) at serines 63/73 and threonines 91/93. An ‘activation by de-repression’ model has been proposed for the mechanism by which phosphorylation activates c-Jun. Thus, c-Jun phosphorylation mediates dissociation of an inhibitory complex which is associated with HDAC3. Subsequent to this dissociation, c-Jun can go on to activate target gene expression whether or not it is phosphorylated. c-Jun can also be activated by an increase in cellular levels of c-Jun protein thereby titrating out limiting components of the repressor complex (Weiss et al., 2003). c-Jun can itself be the target of acetylation, at least under specific circumstances. The MMP1 promoter can be activated by c-Jun, and repressed by the adenoviral E1A protein. p300 binds to E1A and repression of c-Jun induction of MMP1 expression is dependent upon acetylation of c-Jun at Lys271 (Vries et al., 2001) A further report also demonstrates that HDACi can suppress the expression of c-jun and therefore the level of c-Jun protein. This in turn leads to decreased binding of c-Jun to promoters of cognate genes such as COX2 or MMP1 and therefore HDACi suppress the phorbol ester-induced expression of these genes (Yamaguchi, Lantowski, Dannenberg, & Subbaramaiah, 2005). Data from the Clark laboratory align with this where TSA represses PMA-induced MMP1 expression in a number of cell lines (MRC5, HeLa, SW1353) (unpublished).

A recent paper suggests that Stats undergo acetylation at Lys 685 and that this is essential for their dimerisation and nuclear translocation (Yuan, Guan, Chatterjee, & Chin, 2005). In this case, the action of HDACi would increase acetylation and therefore potentiate Stat signaling. Whilst there are instances of this in the literature (e.g. the IL-4 induction of the 15-lipoxygenase-1 gene, (42), more recent data appear to show the opposite effect in the majority of cases. Thus, HDACi have revealed an essential role for HDACs in the transcription of interferon-responsive genes (Chang et al., 2004, Genin et al., 2003; Nusinzon & Horvath, 2003; Rascle, Johnston, & Amati, 2003; Sakamoto, Potla, & Larner, 2004) since such inhibitors repress IFN-stimulated gene expression.

As above, several MMP genes respond to proinflammtory stimuli, at least in part, via the NFκB pathway. Acetylation can impact upon this signaling pathway at a number of levels. Firstly, NFκB interacts with a number of HATs (including CBP, p300, P/CAF) and HDACs to impact upon gene expression. More specifically, HDACi delay postinduction repression of NFκB via prolonged activity of IKK and therefore persistent degradation of IκBα and delayed build up of cytoplasmic IκBα after an inflammatory stimulus. The mechanism of this enhanced IKK activity is unknown (Quivy & Van Lint, 2004). Further, both p50 and p65, the most common components of the NFκB dimer can be acetylated at multiple lysine residues. p50 acetylation increases DNA-binding affinity and this correlates with induction of genes such as COX2 and iNOS (Quivy & Van Lint, 2004). There are opposing views of the outcome of p65 acetylation. Kiernan et al. (2003) show that p65 acetylation reduces DNA-binding affinity and enhances NFκB removal from the nucleus by IκBα and therefore abrogating NFκB action. Chen, Fischle, Verdin, and Greene (2001) and Chen, Mu, and Greene (2002) show that p65 acetylation diminishes binding to IκBα, allowing increased nuclear translocation of NFκB and potentiation of signaling. They further show that HDAC3 deacetylates p65 to abrogate signaling and this fits with the ability of HDACi to potentiate or prolong NFκB signaling induced by TNFα.

A number of other transcription factors relevant to MMP gene expression are also subject to regulation at the level of acetylation, including Sp1 and Sp3 and Ets family members (Ammanamanchi, Freeman, & Brattain, 2003; Braun, Koop, Ertmer, Nacht, & Suske, 2001; Czuwara-Ladykowska, Sementchenko, Watson, & Trojanowska, 2002; Huang et al., 2005, Yang and Sharrocks, 2004).

In multicellular eukaryotes, DNA methylation is restricted to cytosine bases with a number of known DNA methyltransferases acting to methylate cytosines within CpG dinucleotides. Methylation is usually associated with a repressive chromatin state and inhibition of gene expression. Methylation may block the binding of transcriptional activators and/or methyl binding proteins may recruit transcriptional repressors including HDACs (Klose & Bird, 2006).

In lymphoma cells an inverse correlation was noted between MMP9 promoter methylation and the level of MMP9 expression (Chicoine et al., 2002). This was confirmed functionally with in vitro experiments. Similarly, a colon cancer cell line in which the DNA methyltransferases Dnmt-1 and Dnmt-3b were knocked out, show increased expression of MMP3 (but not MMP1 or MMP2). Treatment of wildtype cells with DNA methyl transferase inhibitors recapitulated this effect whilst in vitro methylation of the MMP3 promoter suppressed its activity (Couillard, Demers, Lavoie, & St-Pierre, 2006). The induction of MMP3 by hypomethylation was also shown to be cell-specific since in a lymphoma cell line the methylase inhibitors showed no induction of MMP3, but induction of MMP10. A further group (Shukeir, Pakneshan, Chen, Szyf, & Rabbani, 2006) used the invasive prostate cancer cell line PC-3, treated with a methyl donor or antisense against methyl DNA-binding domain protein 2. Both treatments led to the repression of MMP2 expression and decrease in tumour volume in vivo. Bisulfite sequencing was used to show that the 5′ region of the MMP2 gene was methylated in response to the above treatments.

In osteoarthritis, chondrocytes expressed increased levels of MMP3, MMP9 and MMP13. Methylated CpG sites were decreased across the promoters of these genes with specific sites showing significantly higher demethylation in each gene (Roach et al., 2005).

As mentioned above for TIMP3, a number of studies demonstrate that silencing of TIMP genes via promoter methylation as a feature of cancer cells e.g. (Ivanova et al., 2004, Sun et al., 1995, Yuan et al., 2004). The TIMP1 gene is on the X chromosome and is thus subject to X chromosome inactivation in females. Not all genes on the inactive X chromosome are completely silenced and TIMP1 is reported to display variable inactivation via both methylation status and changes in chromatin structure (Anderson & Brown, 2005).

It is interesting to note that whilst the MMP genes are distributed widely among the human chromosomes, there is a cluster at 11q22 containing nine gene (MMP1, MMP3, MMP7, MMP8, MMP10, MMP12, MMP13, MMP20 and MMP27). The regulation of these genes may require particular epigenetic features (chromatin marks and/or structure) that are currently not understood.

At the post-transcriptional level, gene expression can be regulated via the stability of mRNA in the cytoplasm. This is mediated via a variety of trans-acting factors including both RNA-binding proteins and microRNAs (miR) that interact with cis-elements located at many sites in the mRNA. The most commonly described cis element is the AUUUA sequence which is often found in multiple copies within the 3′ UTR of mRNAs. Binding of protein factors to these elements can stabilize (e.g. HuR) or destabilise (e.g. AUF1) such mRNAs (Garneau, Wilusz, & Wilusz, 2007).

MMP and TIMP expression may be regulated (or perhaps fine-tuned) at this level. Vincenti, White, Schroen, Benbow, and Brinckerhoff (1996) showed that the induction of MMP1 by IL-1β in rabbit synovial fibroblasts required both transcriptional and post-transcriptional components. In terms of the latter, the 3′ UTR of the rabbit gene was able to destabilize a constitutively expressed reporter. IL-1 treatment stabilized a cloned human MMP1 transcript, as did mutation of ATTTA motifs in the 3′ UTR. This demonstrates that, aside from transcriptional activation, IL-1 increases MMP1 steady state mRNA by countering the destabilizing effects of the 3′UTR.

Similarly, cortisol induces MMP13 steady state mRNA in osteoblasts by stabilizing transcripts (Rydziel, Delany, & Canalis, 2004). Cortisol was shown to increase protein binding to AU-rich elements in the MMP13 3′UTR. Studies using transgene reporters show that the MMP13 3′UTR stabilizes a c-fos mRNA, with cortisol further increasing mRNA stability in this system. Mutation of the MMP13 3′UTR AU-rich elements destabilizes c-fos transcripts compared to wild-type and blocks the effects of cortisol. Both vinculin and far upstream element (FUSE) binding protein 2 were shown to interact with the MMP13 3′UTR and knockdown of these proteins impacted upon MMP13 mRNA decay.

MMP9 expression can also be controlled at the level of mRNA stability. In mesangial cells, the ATP analog ATPγS potentiates the ability of IL-1beta to induce steady-state MMP9 mRNA. This effect is via three AU-rich elements in the 3′ UTR of the MMP9 gene which are constitutively bound in these cells by the RNA stabilizing factor HuR. The binding of HuR-containing complexes to these sites was increased by ATPγS, and the ATP-dependent effect on MMP9 UTR was abolished by mutation in the three AURE (though this had no effect on IL-1 induction itself) (Huwiler et al., 2003). The same group identified a similar mechanism for the ability of nitric oxide to reduce MMP9 steady state mRNA. Recombinant HuR stabilized, whilst anti-HuR antibody destabilized, the MMP9 mRNA. Nitric oxide was shown to attenuate the expression of HuR and its binding to the MMP9 3′UTR AU-rich elements via a cGMP-dependent mechanism (Akool el et al., 2003).

The TIMP3 3′ UTR was recently identified as a target for HuR binding in a screen of such mRNAs (Lal et al., 2004). There are other reports of regulation of TIMP mRNA stability, e.g. in astrocytes, regulation of mRNA decay is reported to contribute to the downregulation of TIMP1 expression by IL-1 (Gardner et al., 2006), however, molecular detail is yet to be described.

Micro RNAs (miRs) are small 21–25 nucleotide RNAs that negatively regulate gene expression at the post-transcriptional level, causing either inhibition of translation or mRNA degradation. The numbers of identified miRs are growing rapidly, and there is speculation that as much as 30% of the human genome may be subject to regulation in this fashion (Lewis, Shih, Jones-Rhoades, Bartel, & Burge, 2003). In cancer, miR profiles both identify the tissue of origin of tumours (Lu et al., 2005), and are prognostic (Volinia et al., 2006). Individual miRs function as tumour suppressor genes by regulating the expression of proto-oncogenes such as Ras, or conversely examples such as miR-21 act as oncogenes when over-expressed, by down-regulating expression of pro-apoptotic genes (Chan, Krichevsky, & Kosik, 2005). To date little is known of the role of miRs in regulation of cellular protease networks, but bioinformatic analysis has indicated that several MMP and TIMP genes contain miR binding sites in their 3′UTRs, including MMP2 (miR-29), MMP14 (miR-24, miR-26 and miR-181), TIMP2 (miR-30) and TIMP3 (miR-21, miR-1/206 and miR-181) (Dalmay & Edwards, 2006). Since TIMP3 is considered to be a tumour suppressor function that is often down-regulated in cancers, its possible regulation by miRs is intriguing and merits further investigation.

Single nucleotide polymorphisms are variations in DNA sequence common in the population which may exist anywhere across a gene. Within gene promoters, such differences in genotype may alter promoter function and thus gene regulation.

The first single nucleotide polymorphism to be described in an MMP gene promoter was the 5A/6A SNP at −1612 bp in MMP3 (Ye et al., 1996). This SNP is within an interleukin-1 responsive element, with the 5A allele driving greater expression in reporter assays. The transcription factor ZBP89/ZNF148 and p65 containing dimers of NFκB all bind with similar affinity to either 5A or 6A allele, though p50 homodimers binds more effectively to the latter. This is hypothesized to act as a transcriptional repressor by competing for other, activating, NFκB variants, though this has not been experimentally verified. This SNP associates with a number of cardiovascular conditions (Ye, 2006).

Rutter et al. (1998) reported a SNP in the promoter of the MMP1 gene at −1607 bp where an additional G residue creates an Ets binding site (5′GGAT3′ compared to 5′GAT3′) adjacent to an AP-1 site at −1602 bp. The 2G allele leads to higher levels of MMP1 expression (e.g. in A2058 melanoma cells) with evidence pointing to at least Fra-1 preferentially targeting transcription from this allele (Tower, Coon, Belguise, Chalbos, & Brinckerhoff, 2003). A variety of reports show association with either favourable (e.g. (Hettiaratchi et al., 2007)) or unfavourable e.g. (Cao & Li, 2006; Ju et al., 2005, Zhu et al., 2001) facets of cancer, though it should be noted that a number of studies show no association e.g. (Fong et al., 2004). Other pathologies have also been examined, with e.g. Malik, Jury, Bayat, Ollier, and Kay (2007) reporting that the 2G MMP1 genotype was highly associated with aseptic loosening after total hip replacement compared to controls.

A C to T SNP at −1562 bp in the MMP9 gene impacts upon transcription of the gene. An unidentified nuclear protein has higher affinity to the T-allele than the C-allele with the former exhibiting higher activity in reporter assays (Zhang et al., 1999). Studies in tissues from individuals suggest that this is replicated in vivo. The MMP9 promoter also contains a CA repeat region at approx −131/−90 bp which can vary between 14 and 23 repeats. Shorter repeat regions display reduced MMP9 promoter activity, and this also correlates with expression of the gene both lung and oesophageal cancer cell lines (Huang et al., 2003, Shimajiri et al., 1999).

Functional SNPs in the MMP2 promoter include a −1575 bp G to A polymorphism altering binding of oestrogen receptor (Harendza et al., 2003) and a −1306 bp C to T polymorphism altering binding of Sp1 (Price, Greaves, & Watkins, 2001). Other SNPs have been described in the promoters of MMP7, MMP12 and MMP13 (Ye, 2006).

Similarly to the MMP2 SNP described above, a −418 bp G to C polymorphism abolishes an Sp1 site in the TIMP2 promoter and this shows association with COPD (Hirano et al., 2001) and some association with breast cancer risk (Zhou et al., 2004). Two promoter variants in the TIMP3 gene, a −915 bp A to G and −1296 bp T to C polymorphism, appear to contribute to susceptibility to the chronic lung disorder pigeon breeders disease (Hill et al., 2004).

Section snippets

Conclusions

Unsurprisingly the regulation of the MMP and TIMP gene families is complex. These genes are regulated across development, in adult physiology and in disease. In many situations, a number of genes within these families are being transcribed simultaneously, though such subsets will differ depending on e.g. cell type and stimulus. Hence, the combinatorial control of gene expression at the promoter level can be coupled with epigenetic control at the level of chromatin structure and

Acknowledgements

T.E.S. is funded by the Dunhill Medical Trust; D.R.E. would like to acknowledge the support of the European Union Framework 6 Cancerdegradome project (LSHC-CT-2003-503297).

References (134)

  • J. Czuwara-Ladykowska et al.

    Ets1 is an effector of the transforming growth factor beta (TGF-beta) signaling pathway and an antagonist of the profibrotic effects of TGF-beta

    J. Biol. Chem.

    (2002)
  • G. Dean et al.

    The human tissue inhibitor of metalloproteinases (TIMP)-1 gene contains repressive elements within the promoter and intron 1

    J. Biol. Chem.

    (2000)
  • B. Durand et al.

    All-trans and 9-cis retinoic acid induction of CRABPII transcription is mediated by RAR-RXR heterodimers bound to DR1 and DR2 repeated motifs

    Cell

    (1992)
  • H. Fahmi et al.

    Peroxisome proliferator-activated receptor gamma activators inhibit MMP-1 production in human synovial fibroblasts likely by reducing the binding of the activator protein 1

    Osteoarthritis Cartilage

    (2002)
  • M. Francois et al.

    Peroxisome proliferator-activated receptor-gamma down-regulates chondrocyte matrix metalloproteinase-1 via a novel composite element

    J. Biol. Chem.

    (2004)
  • T.L. Haas et al.

    Egr-1 mediates extracellular matrix-driven transcription of membrane type 1 matrix metalloproteinase in endothelium

    J. Biol. Chem.

    (1999)
  • M.C. Hall et al.

    The comparative role of activator protein 1 and Smad factors in the regulation of Timp-1 and MMP-1 gene expression by transforming growth factor-beta 1

    J. Biol. Chem.

    (2003)
  • K. Hammani et al.

    Structure and characterization of the human tissue inhibitor of metalloproteinases-2 gene

    J. Biol. Chem.

    (1996)
  • S. Harendza et al.

    Linked common polymorphisms in the gelatinase a promoter are associated with diminished transcriptional response to estrogen and genetic fitness

    J. Biol. Chem.

    (2003)
  • C. He

    Molecular mechanism of transcriptional activation of human gelatinase B by proximal promoter

    Cancer Lett.

    (1996)
  • T.S. Huang et al.

    Shortening of microsatellite deoxy(CA) repeats involved in GL331-induced down-regulation of matrix metalloproteinase-9 gene expression

    Biochem. Biophys. Res. Commun.

    (2003)
  • W. Huang et al.

    Trichostatin A induces transforming growth factor beta type II receptor promoter activity and acetylation of Sp1 by recruitment of PCAF/p300 to a Sp1. NF-Y complex

    J. Biol. Chem.

    (2005)
  • A. Huwiler et al.

    ATP potentiates interleukin-1 beta-induced MMP-9 expression in mesangial cells via recruitment of the ELAV protein HuR

    J. Biol. Chem.

    (2003)
  • W. Ju et al.

    Promoter polymorphism in the matrix metalloproteinase-1 and risk of cervical cancer in Korean women

    Cancer Lett.

    (2005)
  • R. Kiernan et al.

    Post-activation turn-off of NF-kappa B-dependent transcription is regulated by acetylation of p65

    J. Biol. Chem.

    (2003)
  • R.J. Klose et al.

    Genomic DNA methylation: The mark and its mediators

    Trends Biochem. Sci.

    (2006)
  • K.J. Leco et al.

    Murine tissue inhibitor of metalloproteinases-4 (Timp-4): cDNA isolation and expression in adult mouse tissues

    FEBS Lett.

    (1997)
  • B.P. Lewis et al.

    Prediction of mammalian microRNA targets

    Cell

    (2003)
  • S.K. Logan et al.

    Synergistic transcriptional activation of the tissue inhibitor of metalloproteinases-1 promoter via functional interaction of AP-1 and Ets-1 transcription factors

    J. Biol. Chem.

    (1996)
  • J. Lohi et al.

    Structural analysis and promoter characterization of the human membrane-type matrix metalloproteinase-1 (MT1-MMP) gene

    Gene

    (2000)
  • M.G. Ludwig et al.

    Multiple regulatory elements in the murine stromelysin-3 promoter. Evidence for direct control by CCAAT/enhancer-binding protein beta and thyroid and retinoid receptors

    J. Biol. Chem.

    (2000)
  • Z. Ma et al.

    Brg-1 is required for maximal transcription of the human matrix metalloproteinase-2 gene

    J. Biol. Chem.

    (2004)
  • S. Mader et al.

    Multiple parameters control the selectivity of nuclear receptors for their response elements. Selectivity and promiscuity in response element recognition by retinoic acid receptors and retinoid X receptors

    J. Biol. Chem.

    (1993)
  • D.J. Mangelsdorf et al.

    A direct repeat in the cellular retinol-binding protein type II gene confers differential regulation by RXR and RAR

    Cell

    (1991)
  • R.K. Nuttall et al.

    Expression analysis of the entire MMP and TIMP gene families during mouse tissue development

    FEBS Lett.

    (2004)
  • N. Ortega et al.

    Matrix remodeling during endochondral ossification

    Trends Cell. Biol.

    (2004)
  • B.W. Phillips et al.

    A sequence-selective single-strand DNA-binding protein regulates basal transcription of the murine tissue inhibitor of metalloproteinases-1 (Timp-1) gene

    J. Biol. Chem.

    (1999)
  • S.J. Price et al.

    Identification of novel, functional genetic variants in the human matrix metalloproteinase-2 gene: Role of Sp1 in allele-specific transcriptional regulation

    J. Biol. Chem.

    (2001)
  • H. Qin et al.

    The transcription factors Sp1, Sp3, and AP-2 are required for constitutive matrix metalloproteinase-2 gene expression in astroglioma cells

    J. Biol. Chem.

    (1999)
  • V. Quivy et al.

    Regulation at multiple levels of NF-kappaB-mediated transactivation by protein acetylation

    Biochem. Pharmacol.

    (2004)
  • N. Reunanen et al.

    Activation of p38 alpha MAPK enhances collagenase-1 (matrix metalloproteinase (MMP)-1) and stromelysin-1 (MMP-3) expression by mRNA stabilization

    J. Biol. Chem.

    (2002)
  • S. Rydziel et al.

    AU-rich elements in the collagenase 3 mRNA mediate stabilization of the transcript by cortisol in osteoblasts

    J. Biol. Chem.

    (2004)
  • S. Sakamoto et al.

    Histone deacetylase activity is required to recruit RNA polymerase II to the promoters of selected interferon-stimulated early response genes

    J. Biol. Chem.

    (2004)
  • C.L. Sampieri et al.

    Activation of p38 and JNK MAP kinase pathways abrogates requirement for new protein synthesis for phorbol ester mediated induction of select MMP and TIMP genes

    Matrix Biol.

    (2008)
  • S. Shimajiri et al.

    Shortened microsatellite d(CA)21 sequence down-regulates promoter activity of matrix metalloproteinase 9 gene

    FEBS Lett.

    (1999)
  • W.K. Aicher et al.

    Transcription factor early growth response 1 activity up-regulates expression of tissue inhibitor of metalloproteinases 1 in human synovial fibroblasts

    Arthritis Rheum.

    (2003)
  • S. Akool el et al.

    Nitric oxide increases the decay of matrix metalloproteinase 9 mRNA by inhibiting the expression of mRNA-stabilizing factor HuR

    Mol. Cell. Biol.

    (2003)
  • M.A. Alfonso-Jaume et al.

    Co-operative interactions between NFAT (nuclear factor of activated T cells) c1 and the zinc finger transcription factors Sp1/Sp3 and Egr-1 regulate MT1-MMP (membrane type 1 matrix metalloproteinase) transcription by glomerular mesangial cells

    Biochem. J.

    (2004)
  • A. Al-Sarraj et al.

    Specificity of transcriptional regulation by the zinc finger transcription factors Sp1, Sp3, and Egr-1

    J. Cell. Biochem.

    (2005)
  • C.L. Anderson et al.

    Epigenetic predisposition to expression of TIMP1 from the human inactive X chromosome

    BMC Genet.

    (2005)
  • Cited by (0)

    1

    Present address: Instituto de Salud Pública, Universidad Veracruzana, Xalapa, Veracruz, México.

    View full text