Transforming Growth Factor-β Repression of Matrix Metalloproteinase-1 in Dermal Fibroblasts Involves Smad3*

Enhanced production of matrix metalloproteinase-1 (MMP-1, collagenase-1) is implicated in pathological tissue destruction. Transforming growth factor-β (TGF-β) prevents cytokine-induced MMP-1 gene expression in fibroblasts. In these studies, we examined the hypothesis that repression of MMP-1 may be mediated through the Smad signaling pathway. The results showed that Smad3 and Smad4, but not Smad1 or Smad2, mimicked the inhibitory effect of TGF-β and abrogated interleukin-1β (IL-1β)-induced stimulation of MMP-1 promoter activity and NFκB-specific gene transcription in dermal fibroblasts. Experiments with truncation mutants indicated that both MH1 and MH2 domains of Smad3 were necessary for inhibitory activity. Dominant negative mutants of Smad3 or Smad4 and antagonistic Smad7, which disrupts ligand-induced Smad3 phosphorylation, abrogated the repression of MMP-1 transcription by TGF-β. Similar results were obtained using immunoblot and Northern analysis. Furthermore, TGF-β failed to repress MMP-1 promoter activity in Smad3-deficient murine embryonic fibroblasts. These results implicated cellular Smads in mediating the inhibitory effects of TGF-β. Overexpression of the transcriptional co-activator p300, but not its histone acetyltransferase (HAT)-deficient mutant, was able to relieve repression of MMP-1 gene expression, suggesting that Smad-dependent inhibition may be due to increased competition between Smad proteins and IL-1β signaling pathways for limiting amounts of cellular p300. Together, these results demonstrate that MMP-1 is a target for negative regulation by TGF-β through cellular Smad3 and Smad4. Smad-mediated repression of MMP-1 gene expression may be important for preventing excessive matrix degradation induced by inflammatory cytokines; disruption of Smad signaling, as occurs in certain cancer cells, may thus be causally linked to uncontrolled tissue destruction mediated through MMP-1.

Matrix metalloproteinases (MMPs) 1 cleave collagens and other components of the extracellular matrix and play important roles in physiological processes of tissue remodeling. Synthesis of MMP-1 (collagenase-1), the principal enzyme mediating the turnover of interstitial collagen in most human tissues, is markedly enhanced by pro-inflammatory cytokines such as interleukin-1 (IL-1␤) and tumor necrosis factor-␣ (1). Excessive matrix degradation is characteristic of rheumatoid arthritis and osteoarthritis, tumor invasion, and periodontitis. It is not surprising therefore that MMP-1 gene expression is under tight control through regulation of its promoter activity and mRNA stability. Interleukin-1␤, one of the most potent physiological inducers of MMP-1 production, stimulates MMP-1 gene expression via c-jun and c-fos, which recognize conserved AP-1 binding elements (2,3). A proximal AP-1 binding element appears to be essential for basal MMP-1 transcription but is not sufficient for full stimulation by IL-1␤ in fibroblasts (4,5). Members of the NFB family, including p65/relA, play a fundamental role in stimulation of MMP-1 transcription by inflammatory cytokines (6 -9). In contrast to extensively characterized positive modulation of MMP synthesis, to date relatively little is known about its repression by cytokines such as interferon-␥ (IFN-␥), despite the obvious significance of negative MMP regulation for maintaining tissue integrity (10,11).
Transforming growth factor-␤ (TGF-␤) plays a critical role in modulation of inflammatory responses, and TGF-␤1-null mice develop uncontrolled inflammation (12). The diverse cellular responses elicited by TGF-␤ are triggered by activation of serine/threonine kinase TGF-␤ receptors and mediated through multiple cellular signal transduction pathways. TGF-␤ receptors propagate signals downstream through direct interaction with cytoplasmic Smads, and possibly other proteins as well (13). Vertebrate Smads can be grouped into three structurally and functionally distinct classes (14). Receptor-activated Smads (Smad2 and Smad3) are directly phosphorylated by activated TGF-␤ receptors and form heteromeric complexes with Smad4 that translocate into the nucleus and modulate the transcription of target genes (15)(16)(17)(18)(19)(20). In many TGF-␤-regulated genes, Smad-binding sequences are located adjacent to AP-1 recognition sites (17,21). Although transcriptional responses can result from direct Smad binding to DNA, more commonly functional interaction of Smads with transcriptional co-factors and coactivators or co-repressors is required. In mink lung epithelial cells, a physical interaction of Smad3 with c-jun results in synergistic stimulation of MMP-1 transcription (21). Smad7, a structurally and functionally divergent member of the Smad family, forms stable association with the activated TGF-␤ receptor complex, thereby preventing phosphorylation of Smad3 and blocking downstream TGF-␤ signaling (22,23). We have shown that in primary fibroblasts, Smad7 abrogated TGF-␤ stimulation of COL1A2 promoter activity (20). Therefore, Smad7 appears to fulfill an important function in fibroblasts as an autocrine negative regulator of TGF-␤ signaling.
In inflammation, mesenchymal cells are targeted by distinct cytokines acting in opposition or in concert. In particular, TGF-␤ elicits multiple biological responses in these cells that are opposite of those induced by inflammatory mediators. For instance, inflammation-induced expression of nitric oxide synthetase, E-selectin, MMP-1, and MMP-3 are abrogated in cell type-specific manner by TGF-␤ (24 -28). Repression of MMP-1 transcription by TGF-␤ in rabbit synovial fibroblasts was shown to be mediated through a sequence at Ϫ246 bp of the promoter that resembled a TGF-␤ inhibitory element previously identified in the rat stromelysin gene promoter (29). However, the level and mechanisms underlying the antagonistic regulation of MMP-1 expression by IL-1␤ and TGF-␤ remain uncertain. Because both TGF-␤ and IL-1␤ are released at sites of injury and play important roles in modulating inflammatory responses, we are interested in characterizing the functional interaction between these two critical cytokines.
In the present study, therefore, we examined the involvement of Smads in MMP-1 regulation by TGF-␤. We now report that TGF-␤ abrogated the stimulation of MMP-1 transcription and protein synthesis induced by IL-1␤, and overexpression of Smad3 or Smad4 mimicked this response. Inhibition of endogenous Smad signaling in the fibroblasts using dominant negative mutants of Smad3 or Smad4 prevented repression of MMP-1 by TGF-␤. In Smad3-deficient murine embryonic fibroblasts, TGF-␤ failed to repress MMP-1 activity. Furthermore, Smad7, an endogenous antagonist of Smad-mediated signaling, partially abolished the negative regulation of MMP-1 promoter activity by TGF-␤. Taken together, these results provide evidence that the Smad pathway of TGF-␤ signaling was necessary and sufficient for potent negative regulation of MMP-1 gene expression in dermal fibroblasts. Disruption of cellular Smad signaling could contribute to aberrant regulation of MMP-1 gene expression and uncontrolled matrix degradation characteristic of pathological conditions.

MATERIALS AND METHODS
Cell Culture and RNA Analysis-Primary cultures of human dermal fibroblasts were established from neonatal foreskin biopsies by previously described explant techniques (20). Embryonic fibroblasts were established from 14d pc Smad3Ϫ/Ϫ mice, as described previously (30). Media were obtained from Biowhittaker (Walkersville, MD); all other tissue culture reagents were from Life Technologies, Inc. (Gaithersburg, MD). Cells were grown at 37°C in a 5% CO 2 atmosphere in modified Eagle's medium supplemented with 10% fetal calf serum (FCS), 1% vitamins, 100 units/ml penicillin/streptomycin, and 2 mM L-glutamine and studied between passages 4 -8. When the cells reached early confluence, fresh medium containing TGF-␤1 (Amgen, Thousand Oaks, CA), IL-1␤ (Roche Molecular Biochemicals, Indianapolis, IN), or IFN-␥ (Genentech, South San Francisco, CA) was added to the cultures for the indicated periods. In some experiments, cycloheximide (from Sigma Chemical Co., St. Louis, MO) was added to the cultures at a final concentration of 5 g/ml for 1 h before TGF-␤ or IL-1␤. Viability of the cells estimated using trypan blue exclusion was Ͼ90%. For Northern analysis, total RNA was isolated from confluent fibroblasts using TRIzol reagent (Life Technologies, Inc.), and relative levels of mRNA were examined using a 32 P-labeled human MMP-1 cDNA probe. Following washing of the nitrocellulose membranes, the RNA-cDNA hybrids were visualized by autoradiography. The filters were scanned, and radioactivity was measured on a PhosphorImager.
Plasmids-The SBE4-luc construct contains four tandem repeats of a palindromic Smad-binding sequence (15). p3TP-lux contains the plasminogen activator inhibitor-1 TGF-␤ response element and three concatamerized repeats of the MMP-1 AP-1 site (31). Expression vectors for Smad1, Smad2, Smad3, Smad4, and Smad7 containing the CMV promoter have been previously described (21,23,32). Smad3A is a dominant negative mutant with three C-terminal serine phosphorylation sites changed to alanines (32). A dominant negative mutant Smad4 was created by deletion of the C-terminal 51 amino acids required for Smad4 interaction with other Smads (32). Terminal truncations of Smad3 were generated by restriction enzyme digestions that resulted in deletion of the first 114 (Smad3⌬N) or the last 148 (Smad3⌬C) amino acids, re-spectively (33). Expression vector for p300 contains FLAG-tagged fulllength p300 in pCI, whereas p300⌬HAT contains p300 lacking the HAT domain (amino acids 1472-1522) (34). The C/H1 and C/H3 p300 expression plasmids coding for p300 lacking amino acids 348 -412 (C/H1) or 1737-1836 (C/H3) were from Upstate Biotechnology Inc. (Lake Placid, NY). The 3.8MMP1/CAT and Ϫ72MMP1/CAT plasmids consist of the sequences Ϫ3.8 kb to ϩ37 bp, or Ϫ72 to ϩ36 bp of the human MMP-1 gene, respectively (35). The tk-renilla luciferase expression vector was used as standard for transfection efficiency. The p65 expression plasmid encodes the p65 subunit of human NFB. The B-luc plasmid consists of three tandem copies of the major histocompatibility complex class I gene NFB element in pGL2-Basic, and AP-1/CAT contains five tandem copies of the human MMP-1 gene AP-1 binding site (3).
Transient Transfections-Neonatal dermal fibroblasts at near-confluency were transfected by the calcium phosphate/DNA co-precipitation method or using FUGENE (Roche Molecular Biochemicals, Indianapolis, IN), which permits Ͼ50% efficiency of transfection in primary fibroblasts, as described previously (20). The total amount of plasmid DNA within each experiment was kept constant by addition of appropriate empty vectors. 6 -12 h following transfection, cells were placed in media with 1% FCS and TGF-␤ or IFN-␥ together with IL-1␤. Cultures were harvested after a further 24-h incubation, and CAT and luciferase activities in aliquots containing equal amounts of protein were determined. The efficiency of transfections was monitored by measuring Renilla luciferase activity. The values shown are the means of triplicate determinations and are representative of multiple independent experiments.
Western Blot Analysis of MMP-1-Confluent fibroblasts were incubated with IL-1␤ and/or TGF-␤ in media containing 0.1% FCS. After 24 h, culture supernatants were harvested, and precipitated with 10% cold trichloroacetic acid. Equal amounts of protein were then fractionated by electrophoresis in 10% SDS-polyacrylamide gels under reducing conditions and subjected to Western blotting using 1:2000 dilution of a polyclonal antibody to human recombinant MMP-1 (AB806 from Oncogene Research Products, Cambridge, MA). To verify transfer efficiency, membranes were stained with Ponceau S. Immunoreactivity was visualized by chemiluminescence. The intensity of the bands was quantitated by densitometric analysis.
Statistical Analysis-Statistical differences between experimental groups were determined by analysis of variance, and values of p Ͻ 0.05 by unpaired two-tailed Student's t test were considered significant. Statistical analysis was performed using the Excel98 software program.

TGF-␤ Abrogates IL-1␤ Stimulation of MMP-1 Gene
Expression-Northern blot analysis indicated that levels of MMP-1 mRNA were very low in unstimulated dermal fibroblasts but were markedly increased by IL-1␤ treatment (Fig. 1A). Treatment of the cultures with TGF-␤ abrogated the induction of MMP-1 mRNA expression by IL-1␤, with maximal inhibition at a concentration of 12.5 ng/ml TGF-␤. Essentially identical results were observed with fibroblasts derived from four separate individuals. The regulation of MMP-1 synthesis was examined by Western immunoblot. The results showed that treatment of the fibroblasts with IL-1␤ induced a Ͼ5-fold increase in the secretion of MMP-1 into the culture media (Fig. 1B). Stimulation of MMP-1 secretion was completely abrogated in the presence of TGF-␤.
To examine the level of MMP-1 regulation by TGF-␤, transient transfections were performed. The 3.8MMP1/CAT construct contains 3.8 kb of the human MMP-1 promoter, including AP-1 binding sites that are considered essential for both basal and IL-1␤-induced promoter activity, as well as a consensus Smad binding site adjacent to the more proximal AP-1 site. The basal expression of the MMP-1 promoter in the transfected fibroblasts was relatively high, suggesting constitutive AP-1 activity (6), and was reproducibly increased by IL-1␤ (Fig. 1C, left panel). As noted previously, the increase in mRNA levels induced by IL-1␤ was consistently of greater magnitude than the increase in promoter activity noted in transient transfection assays (4), suggesting the involvement of upstream regulatory elements or stabilization of mRNA transcripts. Significantly, IL-1␤ stimulation of MMP-1 promoter activity was abrogated by TGF-␤ in a dose-dependent manner (Fig. 1C, left panel). In contrast to its inhibitory effect on MMP-1, TGF-␤ in parallel experiments caused stimulation of the SBE4-luc construct, as expected (Fig. 1C, right panel). These results demonstrate that TGF-␤ inhibition of 3.8MMP1/CAT did not reflect a nonspecific repressive effect on reporter activity and indicate the selectivity of TGF-␤ inhibitory activity for MMP-1 gene expression. Pretreatment of the cultures with cycloheximide at 5 g/ml, a concentration we previously determined to inhibit protein synthesis by Ͼ90% in fibroblasts, failed to prevent suppression of MMP-1, indicating that the TGF-␤ inhibitory response was not dependent on de novo protein synthesis (Fig. 1D).

Smad3 and Smad4 Abrogate Stimulation of MMP-1 Gene
Expression-Smads function as mediators of several cellular responses elicited by TGF-␤ (reviewed in Ref. 14). To examine their involvement in down-regulation of MMP transcription, Smads were transiently overexpressed in confluent fibroblasts co-transfected with the 3.8-kb MMP-1 promoter construct. The results showed that ectopic Smad3 was able to abrogate IL-1␤ stimulation of promoter activity in a dose-dependent manner ( Fig. 2A, left panel). As a positive control, regulation of a TGF-␤-responsive minimal promoter was examined. As expected, 3TP-driven luciferase activity was markedly up-regulated by Smad3 ( Fig. 2A, right panel). The receptor-activated Smads Neonatal dermal fibroblasts were exposed to IL-1␤ (8 ng/ml) and/or TGF-␤ (12.5 ng/ml). Total RNA was isolated after 24-h incubation, and mRNA levels were analyzed by Northern analysis using human MMP-1 cDNA probe. A Northern blot representative of three independent experiments is shown. The intensities of the bands, quantitated by densitometry, are shown in the bottom panel in arbitrary units. B, Western blot analysis of MMP-1. Following 24-h incubation of confluent fibroblasts with IL-1␤ and/or TGF-␤, equal amounts of proteins from trichloroacetic acid-precipitated conditioned media were analyzed by Western blot, using polyclonal antisera to human recombinant MMP-1 as described under "Materials and Methods." A representative blot is shown. The levels of MMP-1 quantitated by densitometric scanning are shown below. C. 3.8 MMP1/CAT or SBE4-luc (3.5 g) was transiently transfected into fibroblasts, as described under "Materials and Methods." CAT (left panel) and luciferase (right panel) activities were determined following a 24-h incubation of transfected fibroblasts with media containing IL-1␤ (8 ng/ml) and/or increasing concentrations (0.1-12.5 ng/ml) of TGF-␤. The results represent the mean Ϯ S.E. of triplicate determinations from four independent experiments. *, statistical significance when compared with untreated controls (p Ͻ 0.05). **, statistical significance when compared with IL-1␤treated samples (p Ͻ 0.05). D, Northern blot analysis. Fibroblasts were preincubated with cycloheximide (5 g/ml) for 1 h prior to addition of IL-1␤ or TGF-␤ to the cultures for 24 h. Total RNA was then analyzed as described above.
share highly conserved and functionally distinct MH1 and MH2 domains. To characterize structural determinants of Smad3 inhibitory activity, we next determined the effects of truncated forms of Smad3 lacking the MH2 domain (Smad3⌬C), or the MH1 (Smad3⌬N). The results of transfection experiments showed that both terminal deletion mutants lost the ability to repress the MMP-1 promoter, indicating that the DNA-binding and protein interaction domains are both required for inhibitory activity of Smad3 (Fig. 2B). When the ability of the Smad3 mutants to stimulate minimal promoter activity was examined, we found that, consistent with other reports (33), overexpression of Smad3⌬N efficiently transactivated SBE4-luc, whereas deletion of the MH2 domain (Smad3⌬C) abrogated the ability of Smad3 to stimulate promoter activity (data not shown). The Smad signaling partner Smad4 by itself repressed the MMP-1 promoter and abrogated its activation by IL-1␤ (Fig. 2C), whereas Smad1, which is implicated in BMP but not TGF-␤ signaling (36), had no significant effect. Interestingly, Smad2, which shares a high degree of structural similarity to Smad3, also failed to inhibit MMP-1, although it was able to stimulate SBE4-luc (data not shown). The level of expression for each ectopically expressed Smad protein was comparable in transfected COS cells. Next, the regulation of MMP-1 mRNA by Smad3 was examined by Northern analysis. As shown in Fig. 2D, transient overexpression of Smad3 in the fibroblasts resulted in dose-dependent inhibition of cellular MMP-1 expression induced by IL-1␤.
To define the region of the MMP-1 gene mediating its transcriptional repression by TGF-␤/Smad3, a truncated promoter construct was used. The plasmid Ϫ72MMP1/CAT contains the Ϫ72 to ϩ36 sequence of the human MMP-1 gene, including a single AP-1 binding site and an overlapping Smad site that is identical to the proposed optimal Smad3 binding AGAC sequence (15). Treatment of transiently transfected fibroblasts with TGF-␤ caused a modest increase in basal or IL-1␤-stimulated Ϫ72MMP1/CAT activity (Fig. 2E). Overexpression of Smad3 in these cells greatly enhanced the activity of truncated MMP-1 promoter in the presence or absence of IL-1␤. These results indicate that TGF-␤/Smad3 enhances, rather than inhibits, basal or IL-1␤-stimulated activity of a truncated MMP-1 promoter, and is consistent with findings in Mv1Lu cells (49,59). Because optimal transactivation of a truncated MMP-1 promoter in Mv1Lu cells appeared to require DNA binding of both c-Jun and Smad3, maximal stimulation was attributed to ligand-induced interaction of DNA-bound c-Jun with DNAbound Smad3 (49). However, stimulation of MMP-1 short promoter activity by TGF-␤ was also seen in F9 cells devoid of c-Jun and c-Fos, suggesting that Smads were able to interact directly with the promoter even in the absence of AP-1 (59).
Endogenous Smad Signaling Required for Inhibition of MMP-1 Transcription-To establish the functional involvement of cellular Smads in TGF-␤-induced repression, complementary loss-of-function approaches were employed. Fibroblasts were transfected with Smad3A, a mutant in which replacement of three C-terminal serine residues with alanine interferes with Smad3 phosphorylation by the activated TGF-␤ receptor (32). We hypothesized that overexpression of Smad3A would disrupt the inhibitory effect of TGF-␤. The results indicated that, although Smad3A by itself had no effect on basal or IL-1␤-stimulated activity of the MMP-1 promoter in transient transfection assays, in fibroblasts co-transfected with mutant Smad3 TGF-␤ failed to repress MMP-1 induction; indeed, IL-1␤-stimulated MMP-1 promoter activity was actually further enhanced by TGF-␤ in the presence of Smad3A (Fig. 3A, left  panel). These results were reproducible in several independent experiments. Transfection of dominant negative Smad4 like- wise prevented repression of MMP-1 promoter by TGF-␤ in the presence of IL-1␤; furthermore, mutant Smad4 abrogated repression of MMP-1 by Smad3 as well, indicating that a fully intact Smad signaling pathway was essential for the inhibitory response ( Fig. 3B and data not shown). It was important to confirm that the ability of dominant negative Smad3 to rescue stimulation of MMP-1 transcription in the presence of TGF-␤ was specifically due to disruption of cellular Smad signaling. To this end, we examined the effect of Smad3A on regulation of MMP-1 by IFN-␥, which has been previously shown to suppress MMP-1 expression in fibroblasts (10). The results showed that IFN-␥ (500 units/ml) inhibited basal, as well as IL-1␤-stimulated, activity of the MMP-1 promoter (Fig. 3A, left panel, lane  7). In marked contrast to its ability to prevent the repression of MMP-1 activity imposed by TGF-␤, however, dominant negative Smad3 could not abrogate the inhibitory effect imposed by IFN-␥. Taken together, these results indicate that Smad3 and Smad4 are critical effectors of TGF-␤-induced repression of MMP-1 in fibroblasts.
Smad7 forms stable interaction with the activated TGF-␤ receptor and thus blocks ligand-induced Smad3 phosphorylation (22,23). Transient overexpression of Smad7 in fibroblasts partially abrogated the repression of MMP-1 promoter activity by TGF-␤ (Fig. 4A, left panel). At the same concentrations, Smad7 prevented the stimulation of a Smad-regulated minimal construct by TGF-␤, indicating the specificity of Smad7 in blocking Smad-mediated transcriptional repression (Fig. 4A,  right panel). Next, the effect of Smad7 on TGF-␤ regulation of cellular MMP-1 mRNA was examined by Northern blot analysis. As shown in Fig. 4B, transient overexpression of Smad7 prevented down-regulation of MMP-1 mRNA expression in fibroblasts, indicating that the cellular Smad signaling pathway was responsible for the inhibitory response elicited by TGF-␤ (Fig. 4B, lane 5). To more directly examine the functional role that cellular Smads play in TGF-␤-induced repression of MMP-1 gene expression, embryonic fibroblasts derived from Smad3-null mice were used (30). As shown in Fig. 5, in Smad3deficient fibroblasts TGF-␤ failed to repress IL-1␤-induced MMP-1 promoter activity. As control, the regulation of SBE4luc was examined. As expected, TGF-␤ induced SBE4-luc activity in Smad3-null cells only in the presence of ectopically overexpressed Smad3. Together, these results firmly establish the essential functional requirement for cellular Smads in mediating transcriptional repression induced by TGF-␤.
p300 Rescues Stimulation of MMP-1 in Presence of TGF-␤-The direct interaction of the co-activator p300 with NF-B/p65/ relA and with Smad3 plays functionally significant roles in p65-and Smad3-mediated transcriptional responses (37)(38)(39)(40)(41)(42)(43). One possible mechanism of TGF-␤/Smad3 interference with MMP-1 stimulation may be competition for co-factors that are required for IL-1␤-induced full transcriptional responses. Because the levels of p300 in most cells are limiting relative to those of transcription factors, competition among transcription factors for cellular p300 may serve as the locus of integration for antagonistic regulatory signaling (44 -47) and has been implicated in repression of the MMP-12 and E-selectin genes by TGF-␤/Smad3 (28,48). We hypothesized that similar competition between activated Smads and IL-1␤-induced transcription factors for limiting amount of cellular p300 may underlie antagonistic regulation of MMP-1 promoter by the two cytokines. In that case, overexpression of p300 should overcome the inhibitory effect of TGF-␤/Smad3 on IL-1␤-stimulated MMP-1 expression. Therefore, we sought to examine the effect of p300 on MMP-1 expression in the presence of TGF-␤ and IL-1␤. For this purpose, wild-type p300 was overexpressed in fibroblasts transiently transfected with 3.8MMP1/CAT, followed by incubation with IL-1␤ and/or TGF-␤. The results showed that TGF-␤ repression of MMP-1 promoter activity in the presence of IL-1␤ was relieved by p300 in a dose-dependent manner (Fig. 6A). By itself, overexpression of p300 caused a 2-fold increase in MMP-1 promoter activity and enhanced the stimulation induced by IL-1␤. Identical results were found when p300 was used to overcome Smad3-induced inhibition of MMP-1 (data not shown). Treatment of the fibroblasts with TGF-␤ did not reduce cellular levels of p300 (42). Together, these results suggest that recruitment of limiting amounts of cellular p300 by activated Smad3 may have been important for antagonistic regulation of IL-1␤ responses by TGF-␤. p300 possesses intrinsic histone acetylase activity, which is believed to play a key role in transcriptional regulation by altering chromatin structure (34). To further characterize the mechanism of p300-mediated rescue of MMP-1 stimulation, a mutant p300 deficient in the HAT domain was used. In contrast to wild-type p300, co-transfection of the HAT-deficient p300 in fibroblasts was unable to prevent repression of MMP-1 promoter activity by TGF-␤ (Fig. 6A), suggesting a link between p300-associated protein acetylation and IL-1␤-induced stimulation of MMP-1 transcription. Next, truncation mutants of p300 deficient in the NFB/p65 binding domain (C/H1) or the E1A/c-Fos binding domain (C/H3) were used. In contrast to p300⌬HAT, both C/H1 and C/H3 truncation mutants of p300 retained the ability to rescue MMP-1 stimulation in the presence of TGF-␤ (Fig. 6B). This suggests that both truncation mutants may have interacted with activated Smad3, thereby liberating endogenous p300 for transactivation of MMP-1 via NFB/p65. Smad3 Modulation of NFB-dependent Transcriptional Activity-Both AP-1 and NFB/p65 are implicated in mediating full transcriptional stimulation of the MMP-1 gene induced by IL-1␤ (4,7,8). To further characterize the mechanism underlying inhibition of MMP-1 activity by TGF-␤/Smad3, minimal promoters consisting of multimerized AP-1 or NFB binding site sequences in front of the CAT gene were used in co-transfection experiments. Treatment of transiently transfected fibroblasts with IL-1␤ caused a ϳ2-fold increase in AP-1/CAT activity, and treatment with TGF-␤ by itself, or Smad3 overexpression in these cells, was strongly stimulatory (Fig. 7A). IL-1␤ was unable to synergize with TGF-␤ or Smad3 in stimulating AP-1/CAT activity. These results, comparable to those with the truncated MMP-1 promoter (Fig. 2E) and consistent with findings in Mv1Lu cells (49,59), indicate that, in contrast to their effect on the cellular MMP-1 promoter, TGF-␤/Smad3 fails to inhibit basal or IL-1␤-induced stimulation of the activity of the multimerized AP-1 sequence artificial construct.
Our results with the truncated MMP-1 promoter and artificial AP-1 reporter constructs suggested that the AP-1 binding site was not responsible for mediating TGF-␤/Smad3-induced repression of MMP-1. In addition to AP-1, NFB is strongly implicated in IL-1␤ stimulation of MMP-1 transcription. In dermal fibroblasts, IL-1 was shown to induce activation of NFB, and disruption of this activation abolished MMP-1 stimulation (8). Using gel mobility shift assays with consensus NFB binding site oligonucleotides as probes, we confirmed that IL-1␤ treatment of fibroblasts for 30 min induced efficient NFB activation, and found that NFB binding was not altered by TGF-␤ (data not shown). Therefore, we examined the modulation of NFB minimal promoter activity by the two cytokines. IL-1␤ caused a significant increase in NFB-driven transcriptional activity, which was repressed by TGF-␤, or by overexpression of Smad3 in the fibroblasts (Fig. 7B and data not shown). The effect of Smad3 on the minimal NFB promoter was comparable to its effect on the 3.8-kb MMP-1 promoter construct (Fig. 2A). These results suggest that TGF-␤/ Smad3 may modulate the transcriptional induction of MMP-1 through an NFB-dependent mechanism. We next examined the effect of Smad3 on p65-mediated transactivation of the NFB construct. As shown in Fig. 7C, overexpression of Smad3 reduced p65-dependent NFB transactivation in the fibroblasts. Repression by Smad3 was relieved by p300 in a dosedependent manner, suggesting that Smad3 competed with p65/ NFB for interaction with limiting amounts of cellular p300. DISCUSSION By its ability to modulate a wide range of cellular responses, TGF-␤ plays a major role in controlling inflammation. TGF-␤ is a potent inhibitor of MMP gene expression in mesenchymal cells (50 -52). Matrix degrading activity is constitutively elevated in mice with genetic disruption of TGF-␤ signaling, highlighting the physiological significance of negative MMP regulation by TGF-␤ (53). Previous studies investigating the mechanistic basis of MMP-1 inhibition demonstrated that IL-1 receptor expression was reduced by TGF-␤ treatment in a variety of cell types (25,54). However, in skin fibroblasts we found that TGF-␤ repressed MMP-1 selectively without altering IL-1␤-induced responses such as NFB activation (data not shown). In the present study, we demonstrate that Smad3 and Smad4 play essential roles in mediating the inhibitory effect of TGF-␤ on MMP-1 transcription. Ectopic expression of Smad3 or Smad4 in the fibroblasts mimicked the effects of TGF-␤ and completely abrogated MMP-1 induction by IL-1␤. Despite their close structural similarity, Smad2 and Smad3 display distinct patterns of transcriptional responses (20,33,48). Although they are more commonly implicated in positive modulation of TGF-␤-regulated genes, Smads have been shown to repress transcription of other genes beside MMP-1. For instance, Smad3 mediates TGF-␤-dependent transcriptional repression of MMP-12 (27,48). In contrast, Smad2 is responsible for repression of E-selectin expression (28).
The functional requirement for cellular Smads in mediating TGF-␤ repression of MMP-1 was examined using dominant negative Smad mutants that have been previously shown to selectively interfere with TGF-␤-induced transcriptional responses in fibroblasts (55). The results indicated that dominant negative Smad4 blocked the transcriptional repression of MMP-1 induced by TGF-␤. Furthermore, in fibroblasts transfected with an activation-defective Smad3 mutant, TGF-␤ not only failed to repress stimulation of MMP-1 expression but actually enhanced MMP-1 levels above those induced by IL-1␤ alone. These observations suggest that TGF-␤ simultaneously exerts both positive and negative effects on the regulation of MMP gene expression in normal fibroblasts. Although the inhibitory effects of TGF-␤ mediated through Smads predominate in normal cells, disruption of cellular Smad signal transduction may "unmask" potential stimulatory effects of TGF-␤ that are mediated through Smad-independent signaling pathways. TGF-␤ is known to activate c-Jun N-terminal kinase, a member of the mitogen-activated protein kinase family, in a bimodal manner, involving both Smad-independent and Smaddependent mechanisms (56). In addition, both the p38 and ERK1/2 MAPK pathways are also induced by TGF-␤ in fibroblasts and may potentially contribute to enhanced MMP-1 stimulation by IL-1␤ when Smad signaling is disrupted (57,58).
Smad3-mediated MMP-1 repression may involve direct Smad-DNA interaction or Smad modulation of IL-1␤-induced transcription factor function through protein-protein interactions (21,49). Smad3 could bind to positive regulatory elements of the MMP-1 promoter, potentially displacing IL-1␤-induced transcriptional activators. Such mutually exclusive promoter interaction by AP-1 and Smad was previously suggested to be a mechanism for promoter-specific functional interaction between AP-1 and Smads (59). A Smad-binding CAGA box is located adjacent to and partially overlapping the proximal FIG. 6. p300 expression rescues MMP-1 stimulation in the presence of TGF-␤. Fibroblasts were transfected with 3.8MMP1/CAT (2.5 g) along with: A, expression plasmid for 1 g of wild-type p300 or p300⌬HAT; or B, p300C/H1 or C/H3 mutants (0.5 or 1 g) or pCMV empty vector. Following a 24-h incubation of the cells in media with IL-1␤ (8 ng/ml) and/or TGF-␤ (12.5 ng/ml), CAT activities were determined. The results, corrected for minor variations in transfection efficiencies, are expressed as relative CAT activities and represent the mean Ϯ S.E. of triplicates from two independent experiments. AP-1 site in the human MMP-1 promoter. However, Smad and AP-1 binding sites are often adjacent to each other in TGF-␤regulated promoters (17,60). Smad3 may interact with ciselements distinct from well-characterized positive regulatory sequences of the MMP-1 gene. In this regard, a TGF-␤ inhibitory element at Ϫ246 bp was implicated in TGF-␤ repression of phorbol 12-myristate 13-acetate-stimulated MMP-1 transcription in rabbit synovial fibroblasts (29). The putative TGF-␤ inhibitory element does not appear to harbor a consensus Smad recognition sequence and fails to be recognized by cellular Smads (data not shown). Nevertheless, the Smad complex could interact with and activate proteins binding to this site.
Inhibition of IL-1␤-induced MMP-1 transcription by TGF-␤/ Smad3 may involve an intermediate "relay" protein functioning as a transcriptional repressor. Expression of junB is induced rapidly by TGF-␤ in several cell types, including skin fibroblasts (51). The junB promoter harbors a consensus CAGACA Smad binding site, and its transcription is directly stimulated by Smad3 (18). Interestingly, in this regard, TGF-␤-induced junB expression was previously shown to abrogate stimulation of MMP-1 promoter activity in fibroblasts (51). However, our present results with cycloheximide suggest that MMP-1 repression by TGF-␤ is not dependent on de novo protein synthesis, strongly arguing against a role for an endogenous mediator of the inhibitory response.
In cells treated with TGF-␤, activated Smad3 competes with other transcription factors for physical interaction with limiting amounts of co-activators such as p300, which is implicated in Smad signaling (37,38). We showed previously that transcriptional responses elicited by TGF-␤/Smad3 in fibroblasts were absolutely dependent on the interaction between p300 and Smad3 (42). Similarly, p300 also functions as an essential transcriptional coactivator for NFB/p65 (28,40,43). Because its level within the nucleus is limited, sequestration of p300 by Smad3 would reduce p300 availability for interaction with NFB/p65, resulting in suppression of NFB-driven transcrip-tional responses. Such p300 competition may provide the molecular basis for the antagonism between IL-1␤/NFB/p65 and TGF-␤/Smad3 in regulation of MMP-1 transcription and other cellular responses. The present results indicate that overexpression of p300 rescued the stimulation of MMP-1 in the presence of TGF-␤. Thus, ligand-activated Smad3 could squelch the induction of MMP-1 promoter activity through sequestration/depletion of cellular p300. Our studies with Smad3 deletion mutants indicated that, in addition to the MH1 DNA binding domain, the MH2 domain is also essential for the inhibitory function of Smad3. The MH2 domain of Smad3 can bind directly to p300 (38). NFB/p65 and Smad3-driven promoters require p300 for optimal transcriptional activation and are inhibited by competitive recruitment of p300 to other promoters (44 -47). Based on these observations and our results, we hypothesize that NFB/p65 competes with Smad3 for limiting amounts of p300. Because the activity of NFB in fibroblasts is rapidly induced by IL-1␤, and NFB mediates IL-1␤induced stimulation of MMP-1 transcription (7,8), loss of p300 availability for p65 interactions would result in reduced NFB/ p65-driven gene transcription. The failure of TGF-␤ to prevent IL-1 induction of p65 DNA binding activity or to reduce IL-1␤induced nuclear accumulation of p65 (data not shown) suggests that TGF-␤1/Smad3 repression of NFB-mediated MMP-1 expression involves steps downstream from IL-1␤-induced NFB activation and is consistent with the competitive p300 interaction model. It was previously shown that, in cells stimulated with IL-1␤ and TGF-␤1, Smads are able to effectively recruit p300/CBP whereas p65 is not (28).
In conclusion, the present results demonstrate that Smads are essential mediators of TGF-␤ repression of inducible MMP-1 gene expression in fibroblasts. The TGF-␤/Smad pathway interferes with IL-1␤-triggered cellular signaling at the level of NFB/p65-mediated gene transcription. Inhibition of MMP-1 expression occurs as a result of the competition between Smad proteins activated by TGF-␤ and NFB proteins FIG. 7. Smad3 regulation of minimal promoters containing AP-1 or NFB binding sequences. Fibroblasts were transiently transfected with 3.5 g of AP-1/CAT or NFB-luc constructs, along with 0.4 g of expression plasmid for Smad3, p65 (0.4 g) or p300 (0.4 or 1 g), or pCMV empty vector, as indicated. Following a 24-h incubation of the cells in media with IL-1␤ (8 ng/ml) and/or TGF-␤ (12.5 ng/ml), cells were harvested, and reporter activities were determined as described above. The results, corrected for minor variations in transfection efficiencies, are expressed as change in CAT or luciferase activity relative to untreated fibroblasts (1.0), and represent the mean Ϯ S.E. of triplicates from two independent experiments. *, statistically significant (p Ͻ 0.05) compared with IL-1␤ (B) or p65 alone (C). activated by IL-1␤ for transcriptional coactivator p300. Activated Smads appear to recruit p300, reducing its availability for IL-1␤-induced responses. Thus, our results suggest a mechanism of interaction between IL-1␤/-activated NFB and TGF-␤-activated Smads in the regulation of MMP-1 expression in the inflammatory milieu.