Inhibition of transcription factor NF-κB reduces matrix metalloproteinase-1, -3 and -9 production by vascular smooth muscle cells

Abstract

Objective: Matrix metalloproteinases (MMPs) contribute to the destruction of the extracellular matrix at the shoulder regions of atherosclerotic plaques that leads to plaque destabilisation and triggers clinical cardiovascular disease. There is therefore considerable interest in establishing the mechanisms responsible for increased MMP production. MMPs-1, -3 and -9 are upregulated by inflammatory cytokines and growth factors that are produced by plaque resident macrophages and smooth muscle cells. Our present studies focused on NF-κB, which regulates numerous inflammatory genes, and is activated in plaque smooth muscle cells. Moreover, an NF-κB binding site is present in the promoter of the MMP-9 gene and an NF-κB-like element in the promoter of the MMP-1 gene. Methods: We used adenovirus mediated overexpression of its inhibitor, IκBα to investigate the role of NF-κB in regulation of MMP-1, -3 and -9 by isolated, cytokine stimulated rabbit aortic and human saphenous vein VSMC. Results: IL-1α potently activated NF-κB in VSMCs and acted synergistically with growth factors to upregulate expression of MMP-1, -3 and -9. Overexpression of IκBα, almost completely inhibited expression of MMP-1, -3 and -9 in response to IL-1α alone or in combination with bFGF and PDGF. Conclusion: NF-κB is required for cytokine upregulation of MMP-1, -3 and -9 in VSMCs, which suggests that NF-κB inhibition may promote plaque stabilisation.

Atherosclerosis, Cytokines, Extracellular matrix, Infection/inflammation, Smooth muscle

bFGF, basic fibroblast growth factor, EMSA, electromobility shift assay, GAPDH, glyceraldehyde-3-phosphate dehydrogenase, IL-1α, interleukin-1α, MMP, matrix metalloproteinase, PDGF, platelet derived growth factor, TNF-α, tumour necrosis factor-α

Time for primary review 23 days.

1 Introduction

The acute coronary syndromes of unstable angina and myocardial infarction most often result from thrombus formation at a site of atherosclerotic plaque rupture. A pathological feature characteristic of the unstable plaque is attenuation of the collagenous extra-cellular matrix, producing areas of structural weakness within the fibrous cap, the barrier between the highly thrombogenic lipid core and the circulating blood [1,2]. Collagen destruction is attributed to increased production of several matrix metalloproteinases (MMPs) by vascular smooth muscle cells (VSMCs) and inflammatory cells, which together can catalyse the complete destruction of interstitial collagen. Both human work and animal models have demonstrated MMP activity within atherosclerotic plaques [3–7]. In vitro studies have demonstrated regulation of MMP-1, -3 and -9 genes by growth factors and inflammatory cytokines produced by plaque resident cells [8,9]. In rabbit VSMCs we demonstrated that MMP-9 secretion is potently and synergistically induced by a combination of the inflammatory cytokine interleukin-1α (IL-1α) and the connective tissue mitogen platelet derived growth factor_BB (PDGF_BB) mimicking the in vivo environment of the atherosclerotic plaque where sources of both are present [9]. The transcription factor nuclear factor kappa-B (NF-κB) is involved in the regulation of several inflammation related genes and active NF-κB has been demonstrated within the atherosclerotic lesion [10–12]. Inflammatory stimuli known to stimulate MMP production from vascular smooth muscle cells including IL-1α and CD40-CD-40 ligand interactions are known to activate NF-κB [10,11]. Furthermore, other processes implicated in atherogenesis such as oxidative stress and vascular injury also result in NF-κB activation. The role of NF-κB in the regulation of MMP-1, -3 and -9 however remains unclear. The presence of an NF-κB binding site in the MMP-9 promoter [13,14] and a recently identified NF-κB-like element in the MMP-1 promoter [15] prompted us to investigate the role of NF-κB in the regulation of MMP-1 and -9 in both rabbit and human vascular smooth muscle cells. Using adenovirus mediated gene delivery of IκBα [16] the physiological inhibitor of NF-κB, we demonstrate that NF-κB plays an essential role in MMP-1 and -9 upregulation. Surprisingly, upregulation of MMP-3 production was profoundly inhibited despite lack of analogous NF-κB sites within the MMP-3 promoter region. The results obtained are important because they reveal potential for NF-κB inhibition as a therapy to promote plaque stabilisation.

2 Experimental procedures

2.1 Materials

Human recombinant PDGF_BB was purchased from Boehringer Mannheim Ltd, Lewes, East Sussex, UK. Basic FGF (bFGF) was purchased from Promega, Southampton, UK. Human recombinant IL-1-α was a generous gift from Roche Products, Welwyn Garden City, Herts., UK. Sheep antisera to rabbit MMP-1 and rabbit MMP-3 were generously given by Dr G. Murphy, School of Biological Sciences, University of East Anglia, Norwich, UK. Rabbit antisera to human MMP-1 and human MMP-3 were gifts from Pfizer, Sandwich, UK. Rabbit polyclonal antibody to IκBα purchased from Santa Cruz (Delaware Avenue, California). Plasmid pSP64 (human collagenase) was a gift from Dr Andrew Docherty, Celltech Ltd, Slough, UK. Recombinant adenovirus capable of overexpressing the porcine IκBα gene (rAd:IκBα) was a kind gift from Dr Rainer De Martin, Department of Vascular Biology and Thrombosis Research, University of Vienna, Austria [16]. Recombinant adenoviruses capable of overexpressing the bacterial LacZ gene (rAd:βGal) and a control virus with a silent expression cassette (rAd:null) were kind gifts from Dr G. Wilkinson, Department of Cardiology and Medicine, University of Wales College of Medicine, Heath Park, Cardiff, UK [17]. This investigation conforms with the Guide for the Care and Use of Laboratory Animals published by the US National Institute of Health (NIH Publication No. 85-23, revised 1996).

2.2 Methods

2.2.1 Cell culture

Primary cultures of human saphenous vein and rabbit aortic smooth muscle cells were prepared by modifications of the explant technique, as previously described in detail [18]. Explants were maintained in complete medium composed of DMEM containing penicillin–streptomycin (100 Units/ml and 100 μg/ml, respectively), 1 mM sodium pyruvate, 8 mM l-glutamine and 15% foetal bovine serum (FBS, Advanced Protein Products, UK). After 10–14 days, cells were subcultured by trypsin/EDTA treatment. Cells between passage one to three were plated at a density of 7×10⁴ cells/well into 12-well culture plates for zymography and western blotting or 8×10⁵ cells/75 cm² flasks for RNA studies. For all experiments, sub-confluent cells were rendered quiescent by incubation in serum-free DMEM supplemented with 0.25% (v/v) lactalbumin hydrolysate (Gibco BRL, Paisley, UK) for 3 days. Cultures were then exposed to fresh serum-free medium containing the appropriate concentration of the agent under investigation for varying time intervals.

2.2.2 Zymography

Gelatinase activity was detected in conditioned media as previously described [18]. Briefly, 10 μl aliquots of non-reduced conditioned media were electrophoresed at 4°C in 7.5% SDS–polyacrylamide gels containing 2 mg/ml gelatin derived from calf skin collagen (Sigma, type III). After removal of SDS gelatinase activity was revealed by overnight incubation at 37°C and staining with 0.1% Coomassie Brilliant Blue. Zymograms were quantified in the linear range by densitometry as described previously [19] using a Biorad GS 690 Image Analysis software system (Biorad Laboratories, Hemel Hempstead, Hertfordshire, UK) and were related to a standard mixture of MMP-2 and -9 run on each zymogram.

2.2.3 Western blotting

Western blots were performed on conditioned media samples concentrated 10-fold by ultrafiltration using Amicon 10 centrifugal concentrators (Amicon, Stonehouse, Gloucestershire, UK). Samples were subjected to SDS–polyacrylamide gel electrophoresis under reducing conditions and proteins were then electrophoretically transferred to Hybond-nitrocellulose membrane (Amersham International, Little Chalfont, Buckinghamshire, UK). Membranes were blocked in TBS–Tween (200 mM Tris/HCl, pH 7.4, 137 mM NaCl, 0.1% Tween 20) containing 5% skimmed milk powder followed by incubation with primary antibody. Following washing in TBS–Tween, blots were incubated with a horseradish peroxidase conjugated secondary antibody for 60 min and immunoreactive proteins visualised using the Enhanced Chemiluminescence (ECL) system (Amersham International). Bands were quantified by densitometry as described above.

2.2.4 Preparation of RNA probes

Antisense riboprobes for MMP-9 and GAPDH have been described previously [20]. MMP-1 riboprobe template was generated by PCR from a plasmid containing a partial MMP-1 cDNA (pSP64 human collagenase) using primers designed using published sequence data [21]. MMP-3 riboprobe template was generated by RT-PCR from IL-1α stimulated rabbit smooth muscle cell total RNA. For human MMP-1, primer sequences were sense (5′-GGT GAT GAA GCA GCC CAG-3′) and antisense (5′-CAG TAG AAT GGG AGA GTC-3′) and for rabbit MMP-3 sense (5′-GC ATC AAG ACA GCA TAG AGC TGA GTA AAG C-3′) and antisense (5′-GAC AGG TTC CAT AGG CAC TTC AGA GTT ATC-3′). PCR products were cloned into PCR II vector (Invitrogen BV, NV Leek, Netherlands) that contains flanking SP6 and T7 promoters and their identity confirmed by diagnostic restriction fragment mapping and hybridization to mRNA transcripts of expected size. Antisense digoxygenin labelled riboprobes were generated from linearized plasmid using the DIG RNA labelling kit according to the manufacturers instructions (Boehringer Mannheim).

2.2.5 Northern analysis

Total cellular RNA was prepared from 1×10⁶ rabbit aortic SMC using Qiagen Rneasy total RNA extraction columns (Hybaid Ltd., Teddington, Middlesex, UK) according to the maunufacturer's instructions, and quantified by absorbance at 260 nm. First, 10 μg of RNA was fractionated on 1% agarose/2.2 M formaldehyde gels and transferred on to Hybond-N membranes (Amersham International) by capillary blotting overnight using 20× SSC (3 M sodium chloride, 300 mM sodium citrate). RNA was cross-linked to the membrane by UV irradiation (150 000 μJ/cm²) and stained with 0.03% methylene blue in 0.3 M sodium acetate, pH 5.2 to assess RNA integrity and equality of loading and transfer.

Prehybridization of the blots was carried out at 64–68°C in a buffer containing 50% (v/v) formamide, 5× SSC, 0.1% sodium lauroylsarcosine, 0.02% SDS, 1% blocking reagent (Boehringer Mannheim) for 1 h. Blots were subsequently hybridized in the same buffer overnight with 100 ng/ml DIG-labelled RNA probes. Post-hybridization was carried out at room temperature using 2× SSC, 0.1% SDS for 10 min followed by higher stringency washes with 0.2× SSC, 0.1% SDS at 64–68°C. Specifically hybridized probe was detected using the DIG-detection system (Boehringer Mannheim). Bands were quantified using a Biorad GS 690 Image Analysis software system.

2.2.6 Nuclear extracts and electrophoretic mobility shift assay

Nuclear proteins were extracted from approximately 1×10⁶ rabbit smooth muscle cell nuclei as previously described [22]. Binding reactions (20 μl) containing 2 μg Poly(dI-dC), 10 mM HEPES, pH 7.9; 50 mM NaCl, 0.5 DDT, 2.5 mM EDTA, 7 mM MgCl₂, 4% glycerol, and 4–6 μg nuclear extract were incubated on ice for 30 min with 20 000 cpm of ³²P-labelled oligonucleotide corresponding to the rabbit MMP-9 NF-κB element (5′-CCC CGG TGG AAT TCC CCA AAT CCT-3′) or the proximal AP-1 element (5′-CCG GCC CTG AGT CAG CAC TTG CCT G-3′). Complexes were separated on 6% nondenaturing polyacrylamide gels and visualised by autoradiography.

2.2.7 Adenovirus-driven overexpression of IκBα

A recombinant adenovirus carrying the porcine IκBα gene (rAd:IκBα) [16] was used to overexpress IκBα protein in cultures of rabbit and human vascular smooth muscle cells. The adenovirus titre required to produce a 100% infection efficiency (600 pfu/cell for rabbit VSMCs and 1000 pfu/cell for human VSMC) was determined histochemically using an adenovirus capable of expressing the bacterial LacZ gene. Rabbit and human smooth muscle cells were infected with 600 and 1000 pfu/cell of rAd:IκBα, respectively, for 16 h in serum free medium. IκBα transgene expression was analysed 3 days post-infection by western blotting as detailed above.

3 Results

3.1 Combinations of growth factors and cytokines synergistically stimulate MMP-9 expression in rabbit aortic and human saphenous vein smooth muscle cells

In agreement with our previous data [9] combinations of PDGF_BB and IL-1α synergistically enhanced secretion of MMP-9 from rabbit VSMC (Fig. 1A). Even greater effects were seen with combinations of bFGF and IL-1α (Fig. 1A). Similar results were found when TNF-α was substituted for IL-1α (results not shown).

Fig. 1

Synergistic regulation of MMP-1, -3 and -9 secretion by combinations of IL-1α with PDGF_BB or bFGF. Quiescent cultures of rabbit and human VSMC were stimulated with either 10 ng/ml IL-1α, 20 ng/ml bFGF, 20 ng/ml PDGF_BB or combination of IL-1α with either PDGF_BB or bFGF as indicated. Cell conditioned medium was collected after 18 h for rabbit cells and 48 h for human cells due to lower levels of secretion by HVSMC. VSMC analysed for (A) MMP-2 and MMP-9 secretion by gelatin zymography and (B) MMP-1 and MMP-3 secretion by Western blotting. The results are representative of three similar experiments.

A similar synergistic pattern of regulation was found in human saphenous vein smooth muscle cells despite lower maximal level of secretion (Fig. 1A). In the human cells, bFGF was only as effective as PDGF_BB (Fig. 1A), which may reflect differences in receptor levels between these cell types. No changes in MMP-2 levels were seen in either cell type (Fig. 1A).

3.2 Effect of combinations of growth factors and IL-1α on secretion of MMP-1 and MMP-3

In rabbit VSMC, secretion of MMP-1 and MMP-3 (Fig. 1B) was low in the presence of bFGF alone but was increased synergistically by combinations of bFGF with IL-1α. Responses to bFGF and IL-1α were comparable with those to phorbol myristate acetate (PMA, data not shown). In human cells MMP-1 secretion was submaximally upregulated by IL-1α or PDGF alone (Fig. 1B). Combinations of these agonists acted additively to increase MMP-1 secretion. MMP-3 secretion mirrored that observed in rabbit cells, being synergistically upregulated by a combination of PDGF_BB and IL-1α (Fig. 1B)

Using rabbit cells we investigated whether increased MMP secretion was mediated by increases in MMP mRNA levels. As shown in Fig. 2, mRNA levels of MMPs-1, -3 and -9 were all strongly and synergistically upregulated by combinations of bFGF with IL-1α. Levels of GAPDH mRNA, used as a housekeeping gene, did not alter under any conditions of stimulation (Fig. 2).

Fig. 2

Synergistic regulation of MMP-1, MMP-3 and MMP-9 mRNA by a combination of IL-1α with bFGF. Quiescent rabbit VSMC were stimulated for 4 h with 10 ng/ml IL-1α, 20 ng/ml bFGF, 10 ng/ml IL-1α plus 20 ng/ml bFGF or 10⁻⁷ M phorbol ester. Total RNA was extracted and analysed for MMP-1, -3 and -9 mRNA by Northern blotting. The house-keeping gene GAPDH was used as a control for RNA integrity and equal loading.

3.3 Activation of NF-κB and its inhibition by overexpression of IκBα

NF-κB activity, measured by electrophoretic mobility shift assay, was increased after 3 h stimulation by IL-1α, but not bFGF, with no synergy between them (Fig. 3A). EMSA were analysed by densitometric analysis and expressed as a percentage of the binding induced by bFGF+IL-1 (control: 12.6±10.4%; IL-1α: 105.4±18.7%; bFGF: 13.2±6.2%). Sub-maximal levels of NF-κB DNA binding were detected after prolonged (18 h) stimulation with bFGF (data not shown). The specificity of the binding was demonstrated by competition with specific ODNs and with supershift with antibodies to the p65/RelA subunit (Fig. 3C). Nuclear translocation of NF-κB measured by immunofluoresence demonstrated the same pattern of regulation by IL-1α but not bFGF (results not shown).

Fig. 3

EMSA analysis of NF-κB activation and its inhibition by IκBα. (A) Quiescent rabbit VSMC were stimulated for 3 h with 10 ng/ml IL-1α, 20 ng/ml bFGF or a combination of IL-1α plus bFGF. Nuclear extracts were prepared and analysed for NF-κB and AP-1 DNA binding by EMSA. Binding specificity was demonstrated by competition with 100-fold excess of unlabelled specific or non-specific oligonucleotide. (B) Rabbit VSMC were infected with rAd:null of rAd:IκBα and rendered quiescent by 72 h serum deprivation. Cells were then stimulated for 3 h with 10 ng/ml IL-1α plus 20 ng/ml bFGF. Nuclear extracts were prepared and analysed for NF-κB and AP-1 DNA binding. (C) Identity of the binding complex was confirmed by supershift with anti-p65/NF-κB antibody.

To establish whether activation of NF-κB was essential for upregulation of MMP-9 we used adenovirus mediated overexpression of nuclear localised IκBα, the inhibitory subunit NF-κB. Infection conditions were optimised using adenovirus carrying the bacterial LacZ β-galactosidase reporter gene. X-gal staining of infected cells demonstrated a multiplicity of infection of 600 and 1000 virus particles per cell yielded 100% infection in rabbit VSMC and HSV SMC, respectively (results not shown). Infection of rabbit VSMC with rAd:IκBα but not the control adenovirus rAd:null increased expression of IκBα detected 3 days later by Western blotting (Fig. 4A). Moreover, upon stimulation with IL-1α, endogenous IκBα was degraded, as expected (Fig. 4B), but high levels of exogenous IκBα remained in the cells infected with rAd:IκBα, possibly due to the high levels of expression achieved and the nuclear localisation of this protein. Infection of HSV SMC with rAd:IκBα but not rAd:null caused a similar overexpression of IκBα as seen in rabbit cells (Fig. 4A). Infection with rAd:IκBα almost completely suppressed NF-κB DNA binding (Fig. 3B). As a control for the DNA binding the specificity of IκBα overexpression we measured the effect of infection with rAd:IκBα on AP-1 DNA binding. AP-1 DNA binding activity was increased more than additively by a combination of IL-1α and bFGF (Fig. 3A). AP-1 binding was analysed by densitometric analysis and expressed as a percentage of the binding induced by bFGF+IL-1 (control: 10.3±4.2%; IL-1α: 31±0.4%; bFGF: 27.3±5.6%). This binding was competed by specific ODNs (Fig. 3A). Unlike the results for NF-κB binding, overexpression of IκBα did not affect AP-1 binding stimulated by IL-1α and bFGF (Fig. 3B).

Fig. 4

Adenovirus-mediated overexpression of IκBα. Rabbit and human VSMC were infected with rAd:IκBα or rAd:null and rendered quiescent by serum deprivation for 72 h. (A) Cell lysates were collected and analysed for IκBα expression by Western blotting. (B) Cells were stimulated with 10 ng/ml IL-1α for times indicated before cell lysates were collected and analysed for IκBα expression by Western blotting. Top panel: endogenous IκBα levels; lower panel: adenovirus mediated expression IκBα (ns indicates a non-specific band).

As further controls for the specificity of the effects of IκBα overexpression we measured its effect on cell viability and proliferation in rabbit VSMC under serum-free conditions similar to those used for the MMP secretion studies. Levels of cellular ATP (nmol/μg DNA, n=3), which were 0.48±0.03 in uninfected cells, were unaltered 3 days after infection with rAd:IκBα (0.55±0.04) or rAd:null (0.51±0.03), demonstrating that virus infection and IκBα overexpression had no adverse effect on cell viability. Cellular proliferation, measured as [³H]thymidine incorporation 24 h after addition of PDGF or bFGF (1000 dpm/μg DNA) was not significantly different in rAd:IκBα (38±8 for PDGF and 61±4 for bFGF) and rAd:null (56±5 for PDGF and 52±7 for bFGF) infected cells. Hence overexpression of IκBα had no effect on the various transcription factor events that led to proliferation. Taken together, these data all support the conclusion that overexpression of IκBα had a specific effect on NF-κB activity.

3.4 Effect of overexpression of IκBα on MMP-1, -3 and -9 production rabbit and human vascular smooth muscle cells

Infection of rabbit VSMC with rAd:IκBα but not rAd:null completely suppressed upregulation of MMP-9 in response to IL-1α alone, and the much greater response to IL-1α in the presence of PDGF or bFGF. Upregulation of MMP-9 secretion in response to phorbol ester stimulation was also inhibited by overexpression of IκBα (Fig. 5A). Infection with rAd:null or rAd:IκBα did not affect the constitutive expression of MMP-2, demonstrating that the effect was selective and providing further evidence against and effect on cell viability. Infection with rAd:IκBα but not rAd:null also greatly suppressed the upregulation of MMPs-1 and -3 (Fig. 5B). To establish whether activation of NF-κB was also essential in for upregulation of MMP-1, -3 and -9 by human VSMCs in response to stimulation with IL-1α and PDGF_BB we again employed adenovirus mediated overexpression of the inhibitory subunit IκBα. Infection of human VSMC with rAd:null significantly stimulated expression of MMP-1 and MMP-9 in response to IL-1α and PDGF (Fig. 6A,B) and had no effect on MMP-3 (Fig. 6A,B). Similar effects of adenovirus infection alone have been reported for other NF-κB regulated genes and most likely reflects activation of NF-κB by adenovirus infection [23–25]. Infection with rAd:IκBα but not rAd:null suppressed upregulation of MMP-9 by 77±12% (P<0.0001, n=8) but again had no effect on constitutive expression of MMP-2 (Fig. 6B). Infection with rAd:IκBα but not rAd:null also resulted in an extremely significant reduction in both MMP-1 upregulation by 97±4% (P<0.0001, n=5) and MMP-3 upregulation by 94±7% (P<0.0001, n=5), (Fig. 6A,B). These data clearly indicate that activation of NF-κB is essential to upregulate MMPs-1, -3 and -9 in rabbit and human vascular smooth muscle cells.

Fig. 6

Adenovirus mediated expression of IκBα inhibits secretion of MMP-1, -3 and -9 in human VSMC. Human VSMC were infected with rAd:null or rAd:IκBα and rendered quiescent by serum deprivation for 72 h. Cells were then incubated in serum free media or serum free media supplemented with 10 ng/ml IL-1α plus 20 ng/ml PDGF_BB for 48 h. Cell conditioned media was collected and concentrated 10-fold before being analysed for (B) MMP-9 secretion by gelatin zymography and MMP-1 and -3 secretion by western blotting. (A) Gelatin zymograms and Western blots were analysed by densitometry using Molecular Analyst software. Values are mean±S.E.M. of three separate experiments.

Fig. 5

Adenovirus mediated expression of IκBα inhibits secretion of MMP-1, -3 and -9 in rabbit VSMC. Rabbit VSMC were infected with rAd:null or rAd:IκBα and rendered quiescent by serum deprivation for 72 h. (A) Cells were stimulated for 18 h with either 10 ng/ml IL-1α, 20 ng/ml bFGF. 20 ng/ml PDGF_BB or a combination of IL-1α with either PDGF_BB or bFGF. While 10⁻⁷ M PMA (phorbol ester) was used as a positive control. Cell conditioned medium was collected and analysed for MMP-2 and MMP-9 secretion by gelatin zymography. (B) Quiescent VSMC infected with rAd:null or rAd:IκBα (as above) were stimulated with IL-1α plus bFGF for 18 h. Cell conditioned medium was collected and analysed for MMP-1 and -3 secretion by Western blotting. The results are representative of three separate experiments.

4 Discussion

We confirmed our previous data showing synergistic regulation of MMP-9 by PDGF and IL-1α in rabbit smooth muscle cells [9] and extended this to human smooth muscle cells. We showed furthermore that additive or synergistic interactions between growth factors and cytokines apply also to MMP-1, and -3 secretion by VSMC from both species. The effects on MMP secretion in rabbit VSMC were accompanied by parallel changes in steady-state mRNA levels.

Upregulation of MMP secretion by IL-1α implies an association between increased matrix turnover and inflammation. Indeed, Lee and co-workers [26,27] demonstrated that macrophages stimulate VSMC to produce MMP-1 and -3 via secretion of IL-1β. Increased secretion of MMPs in response to IL-1α in the presence of growth factors further implies an association with inflammation and cell injury, which can be sources of PDGF and bFGF [2]. For example, in atherosclerosis, expression of PDGF, IL-1α and TNF-α is known to occur together [2], and this may help to explain the presence of MMP-1, -3 and -9 in lesions [28–30]. The preponderant location of MMP-9, MMP-1 and MMP-3 at the shoulder regions of plaques, where there are abundant inflammatory cells is also consistent with an influence of locally released growth factors and inflammatory cytokines [28,31,32]. The consequences of excessive matrix remodelling in vascular pathology may be plaque rupture leading to myocardial infarction [33,34] or disruption of the vascular media leading to aneurysm formation [35,36].

In response to vascular injury, release of bFGF from intracellular compartments is believed to be an important immediate event [37], with PDGF production a more sustained consequence [38,39]. Induction of MMP-9 occurs early after injury to the rat [40,41] and pig [42] carotid artery and is sustained for several days, so that both bFGF and PDGF may play an important part in this response. Our data imply that there is a simultaneous requirement for an inflammatory cytokine, perhaps IL-1α or TNFα. The similar pattern of regulation of MMP-1, -3 and -9 protein secretion implies that concerted turnover of all extracellular matrix components may occur after injury and inflammation. This may be important, in part, as a regulatory cascade, since MMP-3 can activate both MMP-1 and -9 [43,44].

Much further work, which is beyond the scope of the present study, is necessary to elucidate the mechanisms underlying the synergistic regulation of MMP secretion by growth factors and cytokines. We focused instead on defining the role played by the NF-κB transcription factor in the upregulation of MMPs in VSMC. A functional NF-κB site occurs in the proximal stimulatory region of the MMP-9 promoter [45,46] and deletion of this site reduces upregulation of reporter gene constructs in response to phorbol ester and TNF-α. Until now, however it has been unclear what contribution NF-κB plays in the upregulation of the endogenous MMP-9 gene. For example, transient overexpression IκBα in mesangial cells only partially impaired upregulation of MMP-9, suggesting that NF-κB simply plays a permissive role in the upregulation of MMP-9 [47]. Our present results demonstrate, however, an absolute requirement for NF-κB activity, in addition to the essential role played by the AP-1 transcription factor, for MMP-9 production in rabbit and human VSMC in response to a wide variety of stimuli including cytokines, growth factors and phorbol esters [45,46]. Our parallel studies in rabbit dermal fibroblasts also demonstrated an absolute requirement for NF-κB in MMP-9 secretion in this cell type [19].

Our present results further demonstrate that NF-κB activity is essential for upregulation of MMP-1 and -3 in rabbit and human vascular smooth muscle cells. In the case of MMP-1, this is consistent with the observations by Vincenti et al. [15], who used reporter gene analysis of the MMP-1 promoter sequence. An NF-κB-like element was identified at nucleotides −3030 to −3000 that was required, in addition to the proximal AP-1 binding site at –70, for IL-1β stimulated MMP-1 promoter activity in synovial fibroblasts [50]. Furthermore, Xu et al. [48] demonstrated that a peptide inhibitor targeted at nuclear translocation of NF-κB inhibited MMP-1 expression by human fibroblasts grown in a three-dimensional collagen lattice. The importance of NF-κB for increased secretion of MMP-3 is more surprising because no consensus NF-κB element has yet been identified in its promoter [49]. However, it is possible that NF-κB regulates this gene via an, as yet unidentified NF-κB element or via interactions with other transcription factors that regulate MMP-3 expression. For example, Stein et al. [51] demonstrated that NF-κB can potentiate transcription from promoters lacking NF-κB elements via interactions with AP-1 transcription factor. Furthermore, a requirement for NF-κB in the upregulation of α2 integrin expression in fibroblasts also occurs in the absence of known NF-κB promoter elements. Future studies will address the presumably indirect mechanism underlying the regulation of MMP-3 secretion by NF-κB, which also occurs in rabbit dermal fibroblasts [52].

Our demonstration that NF-κB activity is essential for MMP-1, -3 and -9 secretion probably explains previous data showing that the anti-oxidant, N-acetyl-cysteine, inhibits MMP production [53]. N-acetyl-cysteine is a known inhibitor of NF-κB activity. Furthermore, inhibition of NF-κB activity may play a role in the inhibition of MMP secretion by HMG-CoA reductase inhibitors [54,55]. Our data also provides further impetus to pursue the use of specific pharmacological inhibitors of NF-κB activation or decoy NF-κB oligonucleotides [56], and evaluate them in the context of unstable angina pectoris and myocardial infarction.

Acknowledgements

This work was supported by the Biotechnology and Biological Science Research Council (BBSRC) and the British Heart Foundation (BHF).