Tumor Necrosis Factor a Suppresses the Induction of Connective Tissue Growth Factor by Transforming Growth Factor- b in Normal and Scleroderma Fibroblasts*

Connective tissue growth factor (CTGF) is overexpressed in a variety of fibrotic disorders, presumably secondary to the activation and production of transforming growth factor- b (TGF- b ), a key inducer of fibroblast proliferation and matrix synthesis. The CTGF gene promoter has a TGF- b response element that regulates its expression in fibroblasts but not epithelial cells or lymphocytes. Recent studies have shown that the mac-rophage-produced cytokine tumor necrosis factor a (TNF a ) is necessary to promote inflammation and to induce genes, such as matrix metalloproteinases, involved with the early stages of wound healing. In this study, we examined the ability of TNF a to modulate CTGF gene expression. TNF a was found to suppress the TGF- b -induced expression of CTGF protein in cultured normal fibroblasts. The activity of TNF a was blocked by NF- k B inhibitors. We showed that sequences between 2 244 and 2 166 of the CTGF promoter were necessary for both TGF- b and TNF a to modulate CTGF expression. There was a constitutive expression of CTGF by scleroderma fibroblasts that was increased by TGF- b treatment. Although TNF a was able to repress TGF- b -in-duced CTGF and collagen synthesis both in normal and scleroderma skin fibroblasts, fibroblasts cultured from scleroderma patients were more resistant to TNF a as TNF a was unable to in combination, added further 24 At this point, media were removed, and the amount of type I collagen was assessed (26).

healing, being necessary for fibroblast proliferation, stimulation of granulation tissue formation, collagen deposition, and increasing the tensile strength of healing wounds (4,5). TGF-␤ is markedly up-regulated in normal wound healing where it elevates collagen synthesis until healing is complete. It is the persistent activation of the genes encoding extracellular matrix proteins that distinguishes controlled wound repair from uncontrolled connective tissue deposition leading to pathologic fibrosis. One such fibrotic disorder is the debilitating disease scleroderma (systemic sclerosis) which is characterized by progressive skin and internal organ scarring characterized by excessive collagen deposition. The fibrotic process results in the disruption of the normal architecture of the affected organs and ultimately leads to their dysfunction and failure.
TGF-␤ not only causes an increase in collagen expression in wound healing, but it also has diverse effects on other cell types; for example, it suppresses the growth of epithelial cells, inhibits keratinocyte proliferation, enhances neovascularization, and acts as a chemoattractant for monocytes and fibroblasts (6 -8). Connective tissue growth factor (CTGF) is a heparin-binding 38-kDa cysteine-rich peptide that promotes proliferation, collagen synthesis, and chemotaxis by mesenchymal cells (9 -12). CTGF protein is induced by TGF-␤ in connective tissue cells but not in other cell types (9,10,12,13). The regulation of CTGF appears to be controlled primarily at the level of transcription, and a brief exposure of fibroblasts to TGF-␤ is sufficient to induce a prolonged high level of CTGF expression (13). In human dermal and mouse NIH 3T3 fibroblasts, no CTGF protein or mRNA is observed without TGF-␤ treatment (12)(13)(14). This up-regulation is dependent on sequences present in the 5Ј upstream region of the CTGF promoter (13). CTGF mRNA is overexpressed in numerous fibrotic disorders both in skin and in internal organs, such as in scleroderma (12, 14, 16 -19). 2 Given the activity of CTGF (9 -12, 20, 21), understanding the regulatory mechanisms underlying CTGF gene expression in normal cells and how this goes awry in fibrosis might be informative in understanding the biology of CTGF and its role in development as well as in understanding how to modulate specific CTGF-dependent events in fibrotic disorders.
CTGF mRNA is constitutively expressed in bovine aortic endothelial cells, and TNF␣ suppresses this expression (22). In this report, we examined the role of TNF␣ in suppressing TGF-␤-induced CTGF gene and protein expression in normal fibroblasts. We found that TNF␣ suppressed the induction of CTGF by TGF-␤ in normal fibroblasts. Furthermore, we deter-mined that a sequence in the CTGF promoter was necessary both for the TGF-␤ induction and the TNF␣ suppression of CTGF gene expression. In contrast, scleroderma cells constitutively express CTGF protein, and TGF-␤ enhances this expression to approximately twice that of normal fibroblasts. However, scleroderma cells are more resistant to TNF␣ than normal dermal fibroblasts. We surmise that this defect in the TNF-mediated signaling pathway(s) might contribute to fibrosis in these patients.

MATERIALS AND METHODS
Cells and Cell Culture-Human scleroderma (SSc) lesional dermal fibroblasts and normal fibroblast cell lines were established from skin biopsies of lesional (fibrotic) areas of the skin of patients with diffuse cutaneous SSc, and age, sex, and anatomically site-matched healthy volunteers, respectively. All patients fulfilled the criteria of the American College of Rheumatology for the diagnosis of SSc. Informed consent and ethical approval were obtained for all procedures. Fibroblasts were maintained in Eagle's minimal essential medium supplemented with 10% fetal calf serum, 100 units/ml penicillin, 100 mg/ml streptomycin, and 2 mM L-glutamine (minimal essential medium) (Life Technologies, Inc.) and cultured at 37°C in a humidified atmosphere of 5% CO 2 in air. At confluence, brief trypsin treatment and re-culture at a fibroblast cell dilution of one in three passaged fibroblasts. All cells lines were used between passages two and five (23,24).
Western Blot Analysis-Human foreskin fibroblast cells (Clonetics) were grown in DMEM and 10% fetal bovine serum. Cells were grown to confluence and switched to DMEM supplemented with insulin, transferrin, and selenium. Cells were grown for an additional 18 h, and then 10 ng/ml TNF␣ (Roche Molecular Biochemicals) was added. Two hours later, 25 ng/ml TGF-␤2 (Celtrix) was added. No difference was observed between the abilities of TGF-␤1 or TGF-␤2 to induce CTGF. For anti-CTGF blots, 24 h later, media were removed, and 25 l was electrophoresed through a 12% SDS/polyacrylamide gel (Novex) and blotted to nitrocellulose (Bio-Rad). Alternatively, cells were harvested 24 h post-TGF-␤ addition and lysed in 2% SDS. 25 g of total protein was separated by SDS-polyacrylamide gel electrophoresis under non-reducing conditions on a 4 -20% Tris/glycine gel (Novex) and then electrophoretically blotted to polyvinylidene difluoride (Millipore). Filters were blocked overnight at 4°C in 5% nonfat dry milk in TBS, 0.1% Tween 20. CTGF protein was detected by a 1-h incubation of a 1:1000 dilution of rabbit anti-CTGF antibody in 1% milk, TBS, 0.1% Tween 20, followed by a 1-h incubation with a 1:5000 dilution of a horseradish peroxidase-conjugated anti-rabbit antibody (Jackson ImmunoResearch). Proteins were detected by a chemiluminescence kit (Pierce). Vimentin was detected similarly, except 25 g of cell layer, a mouse anti-vimentin antibody (1:1000 dilution; Dako), and a horseradish peroxidase-conjugated anti-mouse antibody (Jackson ImmunoResearch) was used. Scleroderma and normal dermal fibroblasts were similarly analyzed, except TNF␣ was from R & D Systems. Sodium salicylate (Calbiochem) was added 45 min prior to addition of factor and was used at concentrations shown in the figures. Cell-permeable peptide SN50 (Calbiochem) which blocks the migration of NF-B into the nucleus was used as described (25).
Immunofluorescence-Human foreskin fibroblasts were grown on glass coverslips to approximately 80% confluence in DMEM plus 10% fetal bovine serum. Cells were switched to serum-free DMEM for 10 h, and 10 ng/ml TNF␣ was added. Twelve hours later, 25 ng/ml TGF-␤2 was added. Twenty four hours later, cells were washed in PBS and were fixed in absolute methanol for 10 min at Ϫ20°C. After blocking in PBS with 3% bovine serum albumin and 0.1% Triton X-100, a 1:400 dilution of rabbit anti-CTGF antibody was added for 30 min. After washing in PBS, a 1:300 dilution of fluorescein isothiocyanate-conjugated antirabbit antibody was added for 30 min (Jackson ImmunoResearch). After washing, nuclei were stained in 4,6-diamidino-2-phenylidone (DAPI; Molecular Probes). Cells were examined using a Nikon E800 microscope and documented by photography.
Reporter Gene Assays-A stable NIH 3T3 cell line containing the CTGF promoter coupled to the luciferase reporter gene was generated. First, a CTGF promoter containing 805 base pairs of 5Ј upstream region coupled to the luciferase reporter gene (13) was digested with BamHI and SmaI and was subcloned into the BamHI and EcoRV sites of pcDNAneo 3.1 (Invitrogen). This resulted in a plasmid containing the CTGF promoter driving luciferase and a gene encoding G418 resistance. This plasmid was transfected into NIH 3T3 (American Type Culture Collection) cells using LipofectAMINE Plus (Life Technologies, Inc.), and drug-resistant colonies were selected by adding 800 g/ml G418 (Life Technologies, Inc.). For luciferase assays involving this cell line, cells were seeded in 6-well plates at a density of approximately 1.25 ϫ 10 5 /well and grown in DMEM with 5% calf serum and 800 g/ml G418. The next day, cells, at 80% confluence, were changed to DMEM with no serum. Then TNF␣ was added, and 1 h later 10 ng/ml TGF-␤2 was added. After 24 h, cells were washed in PBS, and 350 l of reporter lysis buffer was added for 30 min (Promega), and the cell layer were scraped and pelleted for 10 min at 4°C. Protein concentration was determined using a BCA kit (Pierce) and a microplate reader (Molecular Devices). Equal amounts of protein (5 g) were analyzed using a TR717 microplate luminometer (Tropix) and a luciferase assay kit as described in the manufacturer's protocol (Promega).
Promoter Constructs-By using the CTGF luciferase plasmid as template, promoter fragments of various lengths were amplified using polymerase chain reaction (Perkin-Elmer) with vent polymerase (New England Biolabs) according to the manufacturer's instructions. Primers were from Life Technologies, Inc. PCR fragments were designed so that they could be digested with KpnI and XhoI and subcloned into the SEAP basic vector (CLONTECH). The resultant constructs were sequenced using an ABI model 373 sequencer as described by the manufacturer. Transfections were performed with LipofectAMINE Plus, and media were assayed (SEAP Detection Kit, CLONTECH). To control for variations in transfection efficiency, CTGF promoter plasmid DNAs were cotransfected with cytomegalovirus ␤-galactosidase (CLONTECH), and cell layers were assayed by the Galscreen kit (Tropix). For NF-B assays, we used pNFBSEAP, a commercially available SEAP reporter construct driven by four copies of a site that confers NF-B responsiveness to promoters and thus is a total readout of NF-B and IB activity in a cell (CLONTECH).
Measurement of Collagen Type I-Collagen type I levels were measured using an inhibition enzyme-linked immunosorbent assay as described previously (22). Briefly, when fibroblast cultures reached confluence, the media were removed, and fresh minimal essential medium without fetal calf serum, but supplemented 1% bovine serum albumin and ascorbate acid (50 g/ml), was added. Twenty four hours later, TNF␣ and TGF-␤1 (R & D Systems), alone or in combination, were then added for a further 24 h. At this point, media were removed, and the amount of type I collagen was assessed (26).

TNF␣ Blocks the TGF-␤-mediated Induction of CTGF in
Fibroblasts-Normal foreskin fibroblasts did not produce appreciable levels of CTGF in media, as judged by Western analysis unless incubated with TGF-␤2 (Fig. 1A), as expected from past studies (9). TNF␣ did not increase CTGF production (Fig.  1A). However, preincubation with 25 ng/ml TNF␣ for 12 h before addition of TGF-␤ completely abolished the ability of TGF-␤ to induce CTGF protein in media (Fig. 1A). Similarly, TNF␣ suppressed the induction of CTGF in the cell layer (Fig.  1B). This effect was not due to a general inhibition of protein expression, as neither the addition of TGF-␤ nor TNF␣ affected expression of vimentin protein in the cell layer (Fig. 1C), nor did they affect total protein expression, as visualized by Coomassie stain of total cell extracts (Fig. 1D).
To verify these results, we used a polyclonal CTGF antibody to stain methanol-fixed fibroblasts with or without exposure to TGF-␤ or TNF␣ plus TGF-␤. These studies showed that the addition of TGF-␤ to fibroblasts caused an increase in CTGF protein, which was largely localized to the extracellular matrix and cell surface (Fig. 2C). Conversely, addition of TNF␣ blocked CTGF protein accumulation in TGF-␤-treated cells (Fig. 2D).
Blocking NF-B Inhibits the Ability of TNF␣ to Suppress CTGF Induction by TGF-␤-TNF␣ often exerts its effects on cells through the NF-B pathway, which is involved in the pathogenesis of the inflammatory response (27). Recently, it has been shown that sodium salicylate selectively inhibits NF-B by blocking the activity of IB kinase-␤, a component of IKK (28) that blocks the induction of some TNF␣-inducible genes (29,30). To determine if TNF␣ suppressed CTGF gene induction through NF-B, we first transfected pNFBSEAP into NIH 3T3 cells and incubated cells with or without TNF␣.
This construct contains the SEAP reporter driven by four copies of a NF-B response element, and thus is a net readout of NF-B and IB activity in a cell. Adding TNF␣ for 24 h resulted in an approximately 3-fold increase in reporter activity, which was reduced by sodium salicylate in a dose-responsive fashion (Fig. 3A). To determine if blocking NF-B activity with sodium salicylate blocked the ability of TNF␣ to suppress CTGF induction by TGF-␤, we then preincubated human foreskin fibroblasts with sodium salicylate for 1 h prior to TNF␣ addition. We found that sodium salicylate blocked the ability of TNF␣ to suppress CTGF gene induction (Fig. 3B). Similar results were observed when adult dermal fibroblasts were used (not shown). Finally, we showed that peptide SN50, which blocks the migration of NF-B into the nucleus, blocked the ability of TNF␣ to suppress CTGF induction by TGF-␤ (Fig. 4). These results suggest that TNF␣ suppresses CTGF gene induction by a NF-B-dependent mechanism.
TNF␣ Suppresses the CTGF Gene via cis-Elements Located between Ϫ244 and Ϫ166 of the CTGF Promoter-To determine if TNF␣ suppressed the TGF-␤-mediated increase in CTGF by inhibiting its transcription, we utilized a cell line that contained a stably integrated luciferase reporter molecule driven by a fragment of the CTGF 5Ј upstream promoter (nucleotides Ϫ805 to ϩ74). These studies showed that preincubation of cells for 2 h with TNF␣ before addition of TGF-␤ suppressed luciferase gene expression, in a dose-dependent fashion (Fig. 5).
To identify cis-sequences necessary for the CTGF promoter to respond to TGF-␤ and TNF␣, we generated a clone containing a truncated version of the CTGF promoter comprising nucleotides between Ϫ805 and ϩ17 driving expression of the SEAP reporter (CTGFSEAP(Ϫ805); Fig. 6). This construct was transfected into NIH 3T3 fibroblasts. This proximal promoter region of the CTGF promoter conferred responsiveness to the SEAP reporter to TGF-␤ (Fig. 7). We also generated additional constructs containing truncated versions of the CTGF promoter (Fig. 6). We found that removal of the sequences between Ϫ244 and Ϫ166 abolished the ability of the CTGF promoter to confer responsiveness to both TGF-␤ and TNF␣. That is, both the TNF␣ and TGF-␤ responsive elements in the CTGF promoter lay between nucleotides Ϫ244 and Ϫ166 (Fig. 7).

Dermal Fibroblasts from Lesional Areas of Scleroderma Patients Are Less Responsive to TNF␣ Than Normal Fibroblasts-
To determine if TNF␣ could suppress CTGF in scleroderma cells, we studied skin fibroblasts cultured from affected areas of four scleroderma patients and from four normal individuals. In the absence of TGF-␤, we found no CTGF protein in normal dermal fibroblasts. However, abundant CTGF expression was observed 24 h after addition of TGF-␤ (Fig. 8, top panel) that was decreased by TNF␣. Conversely, CTGF protein was found to be constitutively expressed by the fibroblasts from the scleroderma lesions, even without addition of TGF-␤ (Fig.  8, top panel). The normal and scleroderma fibroblasts showed further elevation of CTGF protein expression upon TGF-␤ treatment. Addition of TNF␣ was less effective at suppressing CTGF induction in scleroderma fibroblasts than in normal fibroblasts (Fig. 8, bottom panel). TNF␣ (500 units) did not reduce basal levels of CTGF in scleroderma fibroblasts (Fig. 8).
To determine if scleroderma fibroblasts were generally less responsive to the antifibrotic effects of TNF␣, we assessed the effect of TNF␣ on collagen synthesis in the dermal fibroblasts used in the study described above. We found that addition of TNF␣ (500 units) did not significantly alter the basal levels of collagen in normal dermal fibroblasts (Table I), while completely abolishing the stimulation of collagen synthesis by TGF-␤ (Table  I). In contrast, 500 units of TNF␣ only reduced TGF-␤-induced collagen levels by 50% in scleroderma fibroblasts (Table I). These results were consistent with our data concerning the lower sensitivity of scleroderma fibroblasts to the anti-fibrotic effects of TNF␣ on CTGF production. Furthermore, these results suggest that an overproduction of CTGF due to a decreased response to inhibitory factors may contribute to the overproduction of collagen by scleroderma fibroblasts. DISCUSSION Collagen accumulation is the hallmark of fibrosis. The production or activation of TGF-␤ is thought to be up-regulated in normal wound healing and fibrotic lesions and is believed to elevate collagen synthesis. Thus agents that neutralize TGF-␤ have been shown to block or attenuate experimentally induced fibrosis (31)(32)(33). This factor represents a potential therapeutic target, although TGF-␤ knockout mice die soon after birth due to defects in blood vessel formation and their immune system that results in general organ failure (34,35).
CTGF was originally detected in conditioned media of cultured endothelial cells and induces proliferation, cell migration, and collagen production (9 -12). Furthermore, CTGF protein is induced by TGF-␤ through a TGF-␤ response element located in its promoter (9 -13). Since CTGF is not expressed in

FIG. 3. TNF␣ induces NF-B activity in fibroblasts, and blockage of NF-B activity by sodium salicylate blocks the suppression of TGF-␤-mediated CTGF induction by TNF␣.
A, TNF␣ induces expression of an NF-B-responsive promoter in fibroblasts that is blocked by sodium salicylate. The plasmid pNFBSEAP (CLONTECH) was transfected into fibroblasts, and 16 h later 100 ng/ml TNF␣ was added, with or without pretreatment with different amounts of sodium salicylate (1, 5, and 25 mM, as shown). B, sodium salicylate blocks the ability of TNF␣ to suppress CTGF induction by TGF-␤. Western blots were performed as described, with or without a 1-h preincubation with sodium salicylate before addition of TGF-␤ or TNF␣. epithelial cells or lymphocytes, other factors must control its differential expression. As shown in this report, TNF␣ suppresses the TGF-␤-induced expression of CTGF acting at the transcriptional level. We found that the TGF-␤-mediated in-duction of CTGF required a sequence of the CTGF promoter binding between Ϫ244 and Ϫ166 site. Removing this sequence also reduced the ability of the CTGF promoter to respond to TNF␣. This sequence is upstream from the previously identified TGF-␤RE (17) and suggests that factors binding to the CTGF promoter between Ϫ244 and Ϫ166 act with the TGF-␤RE to activate gene expression. 3 The identity of these factors is under investigation.
Our results are supported by a recent report that showed that TNF␣ suppressed the basal levels of CTGF mRNA in bovine aortic endothelial cells (22). Furthermore, another report showed that TNF␣ suppressed the dexamethasone-mediated induction of CTGF mRNA in fibroblasts (36). Added to previous observations that CTGF can induce collagen gene expression, these results suggest that, at least in part, TNF␣ may suppress collagen gene expression by inhibiting CTGF induction by TGF-␤. 3 A. Leask and S. Sa, unpublished data.  Previously, CTGF mRNA was detected in situ in fibrotic lesions of scleroderma patients (14) and was detected in a differential display analysis of normal and scleroderma fibroblasts. 2 In this report, we found that scleroderma fibroblasts showed constitutive expression of CTGF protein. We also showed that these cells were less sensitive in their response to TNF␣ than normal fibroblasts in terms of both CTGF and collagen type I synthesis. TNF␣ levels are up-regulated in scleroderma (37). We surmise that TNF␣ in addition to initiating inflammatory responses, may also attenuate fibrosis by suppressing CTGF and collagen synthesis. Furthermore, our data suggest that TNF␣ does not act on the constitutive production of CTGF seen in fibroblasts from scleroderma lesions. We hypothesize that the excessive production of CTGF and the inability to suppress its production contribute to the fibrotic phenotype of this disease by allowing excessive scarring.
These results suggest that investigating the molecular mechanism underlying the TNF␣-mediated suppression of CTGF expression may yield valuable clues as to how the wound healing process is normally regulated and how this process goes awry in fibrosis. Normally, NF-B is sequestered in the cytosol by inhibitory proteins, IB, which may be phosphorylated by a cellular kinase complex termed IKK. TNF␣ stimulates this phosphorylation resulting in the degradation of IB and the translocation of NF-B into the nucleus where it activates gene transcription (38 -40). To this end, we have found that the IKK inhibitor sodium salicylate (28 -33) blocks the ability of TNF␣ to suppress CTGF gene expression. In fact, using this inhibitor to block NF-B reduces inflammatory responses and tissue destruction in rheumatoid arthritis (15). NF-B may act either by directly binding to the CTGF promoter or by inducing genes that are involved with the suppression of CTGF. Further experiments will differentiate between these hypotheses. The molecular difference between normal and scleroderma fibroblasts that results in the latter's inability to respond as effectively to TNF␣ by suppressing CTGF and collagen gene expression may provide a molecular mechanism by which CTGF is constitutively expressed in scleroderma. These approaches should provide novel approaches to developing effective therapeutics for scleroderma, for which currently there are no effective treatments.