Antagonistic Effects of TNF-α on TGF-β Signaling Through Down-Regulation of TGF-β Receptor Type II in Human Dermal Fibroblasts¹

Transforming growth factor-β is a multifunctional protein that plays an important role in regulating cellular growth, differentiation, adhesion, and apoptosis in many biological systems (1, 2, 3). TGF-β inhibits the growth of most cell types. In addition, TGF-β causes the deposition of the extracellular matrix (ECM)³ by simultaneously stimulating skin fibroblasts to increase the production of ECM proteins, such as collagen, fibronectin, and proteoglycan; decrease the production of matrix-degrading proteases; increase the production of inhibitors of these proteases; and modulate the expression of integrins (2, 3). TGF-β binds to transmembrane receptors that have intrinsic serine/threonine kinase activity (4). The type II TGF-β receptor (TβRII) binds TGF-β, and then the type I TGF-β receptor (TβRI) is recruited into the form of a heteromeric complex. TβRII transphosphorylates the glycine/serine-rich domain of TβRI kinase (5). Following the phosphorylation of Smad2 or Smad3 by the activated TβRI, a heteromeric complex is formed with Smad4, resulting in translocation of the complex into the nucleus (6, 7). The complex can act directly as a transcriptional activator and can also indirectly regulate gene transcription by interacting with other transcriptional factors (8, 9, 10, 11).

TNF-α is a potent proinflammatory cytokine implicated in the pathogenesis of degenerative diseases such as rheumatoid arthritis, osteoarthritis, graft-vs-host disease, scleroderma, and shock (12). Before its activation, the 26-kDa TNF-α propeptide is proteolytically converted to its active 17-kDa form. After subsequent trimerization, TNF-α binds and activates two distinct membrane-bound receptors, the 55-kDa type I receptor and the 75-kDa type II receptor. Most effects are transduced by the 55-kDa type I receptor, and the characterized transcription factor families activated by TNF-α are NF-κB and AP-1.

To date, despite a number of observations for the antagonistic activities of TNF-α and TGF-β, it is not clear whether TNF-α directly or indirectly interferes with TGF-β signaling. Conversely, TNF-α inhibits the activity of TGF-β in ECM production, such as type I collagen and elastin (13, 14, 15), and TGF-β signaling (16). These findings reflect the functionally antagonistic nature of these cytokines, and represent a useful paradigm to study the complex cellular signals that regulate ECM formation and remodeling in vivo. It has been reported that TNF-α through NF-κB activation can either induce or inhibit Smad 7 expression, a molecule that interferes with the Smad phosphorylation by TβRI and the subsequent translocation to the nucleus (17, 18). TNF-α alters inhibitory Smad7 expression in a cell-specific manner, as RelA translocation induces Smad7 expression and subsequent blockade of TGF-β signaling in mouse embryo fibroblasts (17), whereas in human embryonic kidney 293 cells, or human hepatoma HepG2 cells, NF-κB activation inhibits Smad7 gene expression (18). It has recently been demonstrated that TNF-α can inhibit Smad signaling through AP-1 activation (16). In contrast, Smad and AP-1 have a synergistic interplay in 12-O-tetradecanoyl-13-acetate-responsive gene promoter elements (19). These evidences indicate that the antagonistic effect on Smad signaling by TNF-α was dependent in a cell-specific or gene-specific manner. Like these phenomena, molecular mechanisms of early phase cell response, such as signaling cross talk between TNF-α and TGF-β, have recently been reported, whereas the TNF-α-mediated antagonistic influence of delayed phase on TGF-β signaling remains obscure.

For regulation of tissue homeostasis, the balance of TNF-α and TGF-β signaling on human dermal fibroblasts seems to be critical. In the present study, we examined the effect of TNF-α on TGF-β-mediated gene induction or TGF-β signaling, and on the regulation of TβR expression in human dermal fibroblasts. Because clarifying the mechanism of the regulation of TβR expression in human dermal fibroblasts could be instructive for elucidating the pathogenesis of progression of fibrotic diseases or malignancy, we also investigated the mechanism of the TNF-α-mediated regulation of TβR expression, and the effect of TNF-α on skin fibroblasts with systemic sclerosis (SSc).

PD98059, BAY 11-7082, MG132, and calpain inhibitor I (ALLN) were purchased from Calbiochem (La Jolla, CA). Recombinant human TNF-α and TGF-β1 were obtained from R&D Systems (Minneapolis, MN). The luciferase assay kit was purchased from Promega (Madison, WI). Abs specific for phospho-p38 mitogen-activated protein kinase (MAPK) were from New England Biolabs (Beverly, MA). Smad7 blocking peptide and Abs specific for TβRI, TβRII, c-Jun N-terminal kinase 1 (JNK1), p38 MAPK, phospho-JNK, Smad3, Smad2/3, and Smad7 were obtained from Santa Cruz Biotechnology (Santa Cruz, CA). The anti-phosphoserine Ab was purchased from Biomeda (Foster City, CA). Protein G-Sepharose was obtained from Zymed Laboratories (San Francisco, CA).

Human dermal fibroblasts were obtained by skin biopsy from the affected areas (dorsal forearm) of three randomly selected patients with diffuse cutaneous SSc of <2 years duration, and from the dorsal forearm of five healthy donors. All biopsies were obtained with informed consent and institutional approval. Primary explant cultures were established in 25-cm² culture flasks in DMEM supplemented with 10% FBS, 2 mM-glutamine, and 50 μg/ml gentamicin, as described previously (20, 21). Monolayer cultures were maintained at 37°C in 5% CO₂ in air. Fibroblasts between the third and sixth subpassages were used for the experiments.

Fibroblasts were washed with PBS at 4°C and solubilized in lysis buffer containing 50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 2 mM EDTA, 1% Nonidet P-40, 0.1% SDS, 50 mM sodium fluoride, and 1 mM PMSF, as described previously (22). Proteins were subjected to SDS-PAGE and transferred to nitrocellulose membranes. The membranes were incubated overnight with anti-TβRI (1/500 dilution), anti-TβRII (1/500 dilution), anti-phospho-JNK (1/500 dilution), or anti-phospho-p38 MAPK Abs, washed, and incubated with a secondary Ab against rabbit IgG for 60 min. After washing, visualization was performed by ECL (Amersham Pharmacia Biotech, Piscataway, NJ). After the analysis of phospho-JNK or phospho-p38 MAPK, blots were stripped and then incubated with anti-JNK (1/1000 dilution) or anti-p38 MAPK (1/1000 dilution) Abs, respectively. We performed immunoblot analysis of TβRII in human dermal fibroblasts using inhibitors of TNF-α-induced intracellular signaling: PD98059, 10 or 30 μM; BAY 11-7082, 1 or 5 μM. We also performed immunoblot analysis of TβRII using various inhibitors of TNF-α-induced proteases: EDTA, 1 mM; PMSF, 1 mM; calpain inhibitor I (ALLN), 5 μM; and, MG132, 10 μM.

Five hundred micrograms of total cellular protein were incubated with Abs to the Smad3 (2 μg/ml) at 4 °C overnight, followed by 2-h incubation with protein G-Sepharose at 4°C. After three washes in lysis buffer, the immunocomplexes were resolved by SDS-PAGE, transferred onto a nitrocellulose membrane, and incubated with anti-phosphoserine Ab (1/50 dilution). The membrane was washed and then incubated with a secondary Ab against goat IgG for 60 min. As a loading control, immunoblotting was also performed using Abs against Smad 2/3 (1/1000 dilution).

Total RNA was extracted using an acid guanidinium thiocyanate-phenol-chloroform method (20, 21, 22). Poly(A)⁺ RNA was extracted from total RNA using an oligotex-dT30 <Super> kit (Takara Shuzo, Otsu, Japan), and analyzed by Northern blotting, as described previously (20, 21, 22). A total of 2 μg of poly(A)⁺ RNA was subjected to electrophoresis on 1% agarose/formaldehyde gels and blotted onto nylon filters (Roche Diagnostics, Indianapolis, IN). The filters were UV cross-linked, prehybridized, hybridized, and sequentially hybridized with DNA probes for α2(I) collagen, tissue inhibitor of metalloproteinases 1 (TIMP-1), GAPDH, or RNA probes for TβRII. The membrane was then washed and exposed to x-ray film.

The TβRII promoter (−1670/+35) fragment, derived as KpnI/HindIII fragment from a TβRII promoter luciferase construct (23), was inserted into the luciferase vector pA₃LUC (24) to construct TβRII promoter-pA₃LUC. The pA₃LUC vector included a trimerized SV40 poly(A) termination site, which reduced transcriptional read-through, and did not contain AP-1-responsive vector sequences (25). Plasmids used in transient transfection assays were purified twice on CsCl gradients, as described previously (21). At least two different plasmid preparations were used for each experiment.

For each transfection, 1 μg of TβRII promoter-pA₃LUC or p3TP-LUX, 1 μg of β-galactosidase (transfection efficiency control), and 4 μl of FuGENE 6 (Roche Diagnostics) were combined and added to the cells. Transfected cells were treated for 18 h with TNF-α in serum-free DMEM before cell lysis in 50 μl of Reporter Lysis Buffer (Promega). Luciferase activity was normalized by cotransfected β-galactosidase activity to correct for transfection efficiency. All transfections were repeated at least three times.

After treatment with DMEM lacking methionine (Life Technologies), we incubated human dermal fibroblasts with 0.1 mCi of [³⁵S]methionine for 24 h at 37°C. After incubation, human dermal fibroblasts were washed twice with cold PBS. Human dermal fibroblasts were lysed by scraping, and radioimmune precipitations were performed, as described above, using an anti-TβRII Ab, followed by adsorption to protein G plus agarose. SDS-PAGE was conducted on 7.5% gels. The gels were dried and exposed to x-ray film at −80°C for 1 wk.

Confluent human dermal fibroblasts grown in six-well plates were incubated for 3 h at 37°C in PBS to remove bound TGF-β, as described previously (26). Next, cells were incubated with ¹²⁵I-labeled TGF-β (¹²⁵I-TGF-β) (100 pM) for 3 h at 4°C in binding buffer (PBS containing 0.9 mM CaCl₂, 0.49 mM MgCl₂, and 1 mg/ml BSA). Unbound ¹²⁵I-TGF-β was then washed off. Cross-linking was performed by incubation with 30 mM disuccinimityl suberate for 30 min at 4°C in binding buffer. Cells were then washed once with PBS and lysed for 30 min in lysis buffer (20 mM Tris-HCl, pH 7.4, 150 mM NaCl, 2 mM EDTA, 1% Nonidet P-40, 1% aprotinin, and 1 mM PMSF). Cell lysates were centrifugated at 12,000 × g for 20 min. Supernatants were collected and incubated with an anti-TβRII Ab for 3 h at 4 °C. Immune complexes were bound to protein G-Sepharose for 30 min at 4°C. After three washes in lysis buffer, the immunocomplexes were resolved by SDS-PAGE. The gels were fixed, dried, and visualized by autoradiography.

First, to test the effect of TNF-α on TGF-β-mediated gene induction, we examined the α2(I) collagen or TIMP-1 expression using Northern blot analysis. Pretreatment of human dermal fibroblasts with TNF-α for 24 h inhibited TGF-β-dependent gene induction in a dose-dependent manner (Fig. 1, A and B). Cell viability was determined with trypan blue staining, which demonstrated that TNF-α did not cause cell death. Second, to test the influence of TNF-α on TGF-β signaling, we investigated the phosphorylation of TGF-β signaling, such as Smad3, JNK, and p38 MAPK, using immunoprecipitation or immunoblotting. TNF-α impeded the phosphorylation of the serine residue of Smad3, and the phosphorylation of JNK and p38 MAPK, in a dose-dependent manner under TGF-β stimulation (Fig. 1, C–H). We also examined the luciferase activities of the TGF-β-responsive reporter, p3TP-LUX. The luciferase activity of p3TP-LUX with a cotreatment of TNF-α and TGF-β in human dermal fibroblasts was lower than that with a treatment of only TGF-β (Fig. 1 I). The luciferase activity with a pretreatment of TGF-β was higher than that with a pretreatment of TNF-α. These results indicate that TNF-α might block TGF-β-dependent gene induction and signaling pathways in human dermal fibroblasts.

Smad7 was recently identified as a direct Smad3/4 target downstream of TGF-β (27, 28, 29, 30). A recent report has shown that Smad7 induction by TNF-α via NF-κB activation may be responsible for its ability to interfere with the Smad signaling pathway in mouse fibroblasts, Mv1Lu, COS, and National Institutes of Health-3T3 (17). However, Smad7 was not induced by TNF-α in human dermal fibroblasts, and HUVEC (16, 31). In HepG2 and human embryonic kidney 293 cells, NF-κB decreased Smad7 (18). Therefore, we tested whether Smad7 was rapidly or slowly induced by TNF-α in our experimental system and could therefore account for the inhibitory activity of TNF-α on TGF-β-induced gene trans activation and TGF-β signaling. For this purpose, confluent fibroblast cultures were treated for the indicated times with TNF-α, and with TGF-β as a positive control. Total cell extracts were subjected to Western blot analysis using an anti-Smad7 Ab. TNF-α did not induce Smad7 (Fig. 2,A), which was consistent with previous reports (16), whereas TGF-β induced Smad7 in human dermal fibroblasts. In addition, these bands were confirmed to be Smad7 protein by competition assay using Smad7 blocking peptide (Fig. 2,B). Next, we investigated the effects of TNF-α on the expression of TβRs in human dermal fibroblasts. Notably, TNF-α down-regulated the protein level of TβRII dose dependently (Fig. 2,C). In contrast, TNF-α did not change the TβRI protein. To determine whether the TNF-α-mediated down-regulation of TβRII protein expression correlated with a decrease in the mRNA level, human dermal fibroblasts were incubated with the indicated doses of TNF-α, and the mRNA were analyzed by Northern blotting. TNF-α did not change the TβRII mRNA levels in comparison with the levels in the control cells (Fig. 2,D). We then examined the time-dependent effect of TNF-α on TβRII. The TβRII protein level was slightly decreased after stimulation with TNF-α for 12 h, and has markedly decreased after 24 and 48 h in comparison with the level in the control (Fig. 2,E). We then examined whether the cotreatment of TNF-α and TGF-β decreased the level of TβRII. TβRII was decreased with the cotreatment of TNF-α and TGF-β to the same degree as with the treatment of TNF-α (Fig. 2,F). ¹²⁵I-TGF-β-binding experiments showed that pretreatment with TNF-α for 24 h decreased the ¹²⁵I-TGF-β-ΤβRII complex in a dose-dependent manner (Fig. 2,G). We also examined TβRII promoter activity treated by TNF-α. TNF-α did not change TβRII promoter activity (Fig. 2,H). Metabolic labeling experiments revealed that TNF-α reduced de novo synthesized TβRII (Fig. 2 I). These results suggest that TNF-α had no effect on the expression of Smad7 or TβRI protein levels, but down-regulated the TβRII protein presumably due to degradation in human dermal fibroblasts. Moreover, TNF-α did not down-regulate the expression of TβRII mRNA or TβRII promoter activity.

To determine which signaling pathway contributes to TNF-α-mediated receptor down-regulation, we used pharmacological reagents to examine the mechanism of this phenomenon. Pretreatment of fibroblasts with PD98059, a specific mitogen-activated protein/extracellular signal-regulated kinase kinase (MEK) inhibitor, or BAY 11-7082, a selective inhibitor of TNF-α-inducible I-κB-α phosphorylation, did not change the TNF-α-reduced TβRII protein level (Fig. 3,A). These results indicate that the activity of the MEK/extracellular signal-regulated kinase or NF-κB signaling pathways was not involved in the TNF-α-mediated decrease of TβRII expression. To clarify the mechanism by which TNF-α down-regulates TβRII protein in human dermal fibroblasts, we tested whether protease inhibitors affected the TNF-α-mediated decrease of the TβRII protein. EDTA is an inhibitor of matrix metalloprotease; PMSF inhibits a serine protease; calpain inhibitor I (ALLN) inhibits thiol proteases, including cathepsin B and L and calpain; and MG132 inhibits the ubiquitin/proteasome pathway. Among these protease inhibitors, ALLN hampered the TNF-α-induced down-regulation of TβRII (Fig. 3 B). Proteasome inhibitors such as MG132 failed to affect the degradation of TβRII by TNF-α. Taken together, TNF-α might block the TGF-β-dependent signaling pathway and TGF-β-mediated gene induction in human dermal fibroblasts through the protease-mediated degradation of the TβRII protein.

SSc is a systemic fibrotic disease involved in the skin, lung, and other organs, in which excessive ECM is deposited. Our recent reports suggested that elevated levels of TβRI and TβRII correlate with elevated levels of α2(I) collagen expression (22, 26). First, we examined whether TNF-α could down-regulate the up-regulation of TβRII in SSc fibroblasts. Western blot analysis revealed that TβRII protein was down-regulated by TNF-α in SSc fibroblasts, as well as in human dermal fibroblasts (Fig. 4,A). SSc fibroblasts were reported to produce several ECM components, mainly of type I collagen (32, 33, 34), and TIMP-1 (35, 36). Next, we examined whether TNF-α affects the elevated levels of α2(I) collagen or TIMP-1 mRNA in SSc fibroblasts. The TNF-α-mediated decrease of α2(I) collagen and TIMP-1 mRNA was demonstrated by Northern blotting, whereas TNF-α did not change TβRII mRNA levels (Fig. 4, B and C). These results indicate that TNF-α has a potential to down-regulate the overexpression of the α2(I) collagen or TIMP-1 mRNA through the down-regulation of TβRII in SSc fibroblasts.

Accumulating evidence indicates that TNF-α counteracts against TGF-β-mediated gene induction or the signaling pathway (13, 14, 15, 16). Our current study provides a new insight into the pathophysiological role of TNF-α. In the present study, we showed that TNF-α impaired TGF-β responsiveness. This impairment results from the reduction of TβRII expression. However, TNF-α did not change the mRNA levels of TβRII or TβRII promoter activity. This phenomenon is complementary to a recently uncovered mechanism by which TNF-α may block TGF-β signaling. Recent studies have shown that, in certain situations, RelA/NF-κB can induce the expression of the inhibitory Smad7, and stabilize the association of the latter with activated TβRI (17). Thus, Smad7 induction by TNF-α via RelA activation could represent a mechanism by which TNF-α antagonizes TGF-β signaling at the level of TβRI function by preventing Smad2 and Smad3 phosphorylation and subsequent translocation into the nucleus. This is, however, not a universal mechanism because other reports indicate that TNF-α did not induce Smad7 expression in human dermal fibroblasts or HUVEC (16, 18). Our results were consistent with the previous report (16). It appears, therefore, that the control of Smad7 expression by cytokines is cell type specific. In addition, it has recently been reported that AP-1 expression in response to TNF-α represents an inhibitory influence counteracting Smad-driven gene trans activation (16). In their report, the mechanism of suppression exerted by AP-1 proteins appears to occur directly at the level of Smad/DNA interactions. However, there have been other reports providing evidence for a synergistic interplay between Smad and AP-1 (19). Taken together, this antagonistic mechanism of interaction between AP-1 and Smad also seems to be in a gene- or cell-specific manner. Our current study showed that inhibiting TGF-β signaling by TNF-α might be the attenuation of a TGF-β-stimulated cellular response through the down-regulation of TβRII. However, such a phenomenon was observed after the long-term exposure of human dermal fibroblasts to TNF-α. These results indicate that the antagonistic effect of TNF-α against TGF-β might be dependent on the expression level of TβRII in a late phase cell response, whereas, in the early phase of cell reaction, Smad7 or AP-1, induced by TNF-α, executes an inhibitory effect on TGF-β signaling in a gene- or cell-specific manner.

TβR expression is regulated by a plethora of external factors, including cytokines and growth factors. It has been shown that 1,25-dihydroxyvitamin D3 and prostaglandin E₂ down-regulate TβRII expression in human osteoblastic cells and human fibroblasts, respectively (37, 38). Furthermore, another study revealed down-regulation of TβRI in human monocytes by IFN-γ (39). In human lung fibroblasts (40) and human corpus cavernosum smooth muscle cells (41), TGFβ1 increases the steady state level of TβRI mRNA, possibly by increasing TβRI promoter activity (40). Regarding the effects of integrins, α₂β₁ integrin interaction with type I collagen down-regulates TβRs (42), whereas α₅β₁ integrin binding to fibronectin up-regulates TβRII (43). Recently, several reports have shown the mechanism of down-regulation of TβRI. Ebisawa et al. (44) described that Smurf1 and Smad7 complexes functioned to regulate the turnover of TβRI by a ubiquitin-proteasome pathway. Also, Kavsak et al. (45) showed that Smad7 binds to Smurf2 to form an E3 ubiquitin ligase that targets TβRI for degradation. In the present study, we demonstrated for the first time that TNF-α down-regulated TβRII protein levels, but not TβRI. Moreover, the mechanism of down-regulation might result from TNF-α-mediated proteolysis of TβRII.

TGF-β exerts its biological effects by interacting with specific cell surface receptors. The modulation of the level of TβRI and TβRII expression plays important roles, both in the mechanism of wound healing and in the progression of malignancy. Disorders of TβR expression lead to various diseases. For example, the reduction of TβR levels contributes to the resistance of tumor cells to TGF-β. Several lines of evidence suggest that transcriptional repression of the TβR gene may be a major mechanism in inactivating TβR in tumor cells (46). In contrast, up-regulation of TβR expression has been demonstrated in fibrotic diseases, such as SSc, localized scleroderma, hepatic fibrosis, idiopathic hypertrophic obstructive cardiomyopathy, and atherosclerosis (22, 26, 47, 48, 49, 50). Up-regulation of TβR expression would result in the deposition of ECM components, which are the chief pathologic features of fibrotic disorders. With respect to SSc, in which progressive fibrosis in the skin is the major cause of the disease, it was reported that scleroderma fibroblasts in the involved area secrete similar levels of TGF-β1 to normal fibroblasts (22, 51). The mechanism of tissue fibrosis in such diseases remains to be determined. We reported the overexpression of TβRI and TβRII in scleroderma fibroblasts compared with normal human dermal fibroblasts, indicating one possible mechanism of autocrine TGF-β activity by the overexpression of TβRI or TβRII (22, 26). In addition, we previously reported that the blockade of TGF-β signaling with anti-TGF-β Abs, or a TGF-β1 antisense oligonucleotide, abolished the increased mRNA expression, as well as the up-regulated transcriptional activity of the human α2(I) collagen gene in SSc fibroblasts (22). In the present study, we have shown that TNF-α-mediated inhibition of up-regulated TβRII expression in SSc fibroblasts led to the down-regulation of α2(I) collagen and TIMP-1 gene overexpression. TβRII is essential for the binding of TGF-β to the receptor complex, and TβRI is necessary for the downstream signal transduction induced by TGF-β binding to TβRII (5). The binding of TGF-β to TβRII is the first critical step in the TGF-β signaling cascade, because TβRI does not bind to TGF-β in the absence of TβRII (52, 53, 54). Hence, we assumed that the TNF-α-mediated down-regulation of TβRII might contribute to the suppression of overexpression of α2(I) collagen and TIMP-1 in SSc fibroblasts.

In conclusion, we showed that TNF-α down-regulated the TβRII protein level through proteolysis in human dermal fibroblasts. We demonstrated that TNF-α weakened the response of the cells to TGF-β by regulating the turnover of TβRII, which exhibited the antagonistic effects on TGF-β-induced gene, such as α2(I) collagen and TIMP-I, and on TGF-β signaling consisting of Smad3, JNK, and p38 MAPK. In a future study, identification of which protease mediated down-regulation of TβRII may be of therapeutic importance as a target for the inhibition of overexpression of TβRII in fibrotic diseases, such as SSc.

We thank Dr. S. J. Kim for kindly providing the TβRII promoter luciferase constructs. We thank Dr. W. M. Wood for kindly providing the pA₃LUC luciferase constructs. We thank Dr. J Massagué for kindly providing the p3TP-LUX luciferase constructs.