To determine the role of tumor necrosis factor receptor p55 (TNFRp55)–mediated signaling in the pathogenesis of scleroderma.
To determine the role of tumor necrosis factor receptor p55 (TNFRp55)–mediated signaling in the pathogenesis of scleroderma.
A murine model of scleroderma that closely resembles systemic sclerosis in humans was used. Wild-type and TNFRp55-deficient (TNFRp55−/−) mice received a subcutaneous injection of bleomycin each day. The extent of skin fibrosis was determined by measurements of the dermal thickness, as well as histologic examinations. Expression levels of fibrogenic cytokines, procollagen α1, and matrix metalloproteinase 1 (MMP-1), MMP-2, and MMP-9 messenger RNA (mRNA) were analyzed, both in vivo and in vitro, by reverse transcriptase–polymerase chain reaction assay or Western blotting.
TNFRp55−/− mice began to develop severe sclerotic changes of the dermis on day 3 of the subcutaneous injections of bleomycin, while wild-type mice did not. The expression levels of fibrogenic cytokines, procollagen α1, and MMP-2 and MMP-9 mRNA were unaffected in the skin of both wild-type and TNFRp55−/− mice, with or without bleomycin treatment. Induction of MMP-1 expression was significantly inhibited in the skin from bleomycin-treated TNFRp55−/− mice, and this phenomenon was also observed in vitro.
These results indicated that signaling mediated by TNFRp55 plays an essential role in MMP-1 expression and a key role in the collagen degradation process in this murine model. This study might provide a basis for understanding the pathogenesis of scleroderma and formulating therapeutic intervention.
Scleroderma is an autoimmune disease that is characterized by progressive fibrosis of the skin and internal organs, including the lungs and gastrointestinal tract. Skin fibrosis is caused by massive production of fibrous connective tissue in the dermis. Studies of growth factors (e.g., transforming growth factor β [TGFβ]) have suggested that they also have a role in the development of fibrosis. Later in the course of the response, TGFβ is primarily associated with extracellular matrix production and up-regulation of platelet-derived growth factor (PDGF) receptors in the scleroderma fibroblast (1). Interleukin-4 (IL-4) is also secreted by T cells and increases the synthesis of collagen by the fibroblast (2), and elevated levels of IL-4 are found in the blood (3, 4), bronchoalveolar lavage cells (5), and skin (6) of scleroderma patients. These cytokines might be involved in mechanisms that exacerbate the symptoms of this disease, but not in its onset.
Tumor necrosis factor (TNF) affects the growth, differentiation, and function of a multitude of cell types that are mediators of inflammation and cellular immune responses. Expression of TNF is detectable during the very early stages of scleroderma (7), and the serologic level of TNF increases with the clinical severity and biologic activity of the disease (8). Additionally, genetic analyses have revealed that microsatellite polymorphisms of the TNF and lymphotoxin α (LT-α) alleles are associated with scleroderma (9, 10). The serum concentration of the soluble TNF receptor p55 (TNFRp55) correlates with the severity of the disease (11–13). Because the soluble form of TNFRp55 neutralizes TNF in cytotoxic assays, and is functionally active as a TNF antagonist (12), alteration of the TNFRp55 signaling pathway might be a key factor in the pathogenesis of scleroderma. To evaluate the role of TNFRp55 signaling in the onset of skin sclerosis, TNFRp55-deficient (TNFRp55−/−) mice were used in a bleomycin-induced dermal fibrosis model. It has been shown that daily subcutaneous injections of bleomycin to mice gradually increase dermal sclerosis after 3–4 weeks (14). Surprisingly, in the present study the TNFRp55−/− mice exhibited severe sclerotic skin after only 3–5 days of daily subcutaneous injections of bleomycin. Alteration of the TNFRp55 signaling pathway might exacerbate this disease.
Mice deficient for the TNFRp55 gene were obtained as previously described (15). Animal care was in accordance with the institutional guidelines of Nagasaki University. TNFRp55−/− mice were backcrossed to B6 background 5 times. Wild-type and TNFRp55−/− mouse embryo fibroblasts (MEFs) were cultured in Dulbecco's modified Eagle's medium (Gibco BRL, Gaithersburg, MD) containing 10% fetal bovine serum (BioWhittaker, Walkersville, MD) and streptomycin at 37°C in a 5% CO2 atmosphere.
The mice were killed and back skin was removed on the day after the final injection. The skin pieces were fixed with 10% formalin, embedded in paraffin, and sectioned using a microtome. They were then stained with hematoxylin and eosin (H&E).
Glycosaminoglycans, which were obtained by β-elimination and pronase digestion from proteoglycans, were treated with chondroitinase ABC, chondroitinase AC, Streptomyces hyaluronidase, or NaNO3, as previously described (16, 17). The resultant reaction mixtures were analyzed by electrophoresis on cellulose acetate membranes.
Six-millimeter circles of biopsied skin were homogenized in acetic acid at 4°C to extract collagen. One milligram of pepsin was added to each homogenized sample, which was incubated at 4°C for 24 hours with shaking. The pepsin-solubilized material was collected after removal of the insoluble residue by centrifugation at 35,000g for 60 minutes at 4°C. The extracted collagen was analyzed using 5% polyacrylamide gel, and the gels were stained with Coomassie brilliant blue to identify the pepsin-resistant collagen.
Total RNA was extracted using an RNeasy mini kit (Qiagen, Hilden, Germany), according to the protocol provided by the manufacturer. First-strand complementary DNA (cDNA) was synthesized using an RT-PCR kit (Stratagene, La Jolla, CA) with oligo dT primers. Thereafter, the cDNA was amplified for 25 cycles for fibrogenic cytokines, procollagen α1 (proα1), and matrix metalloproteinases (MMPs). As an internal control, β-actin was amplified under the same conditions. The oligonucleotide primers used for RT-PCR were as follows: for TNF sense 5′-GCGACGTGGAACTGGCAGAAG-3′, antisense 5′-GGTACAACCCATCGGCTGGA-3′; for TGFβ sense 5′-GCTAATGGTGGACCGCAACAACG-3′, antisense 5′-CTTGCTGTACTGTGTGTCCAGGC-3′; for PDGF sense 5′-CACATCGGCCAACTTCCT-3′, antisense 5′-TCACACGCCACGCACATC-3′; for IL-4 sense 5′CGAAGAACACCACAGAGAGTGAGCT-3′, antisense 5′-GACTCATTCATGGTGCAGCTTATCG-3′; for proα1 sense 5′-AGAGCGGAGAGTACTGGGATCGAC-3′, antisense 5′-TCTGAGCTCGATCTCGTTGGAT-3′; for MMP-2 sense 5′-TGCAACCACAACCAACTACG-3′, antisense 5′-TCTGCGATGAGCTTAGGGAA-3′; for MMP-9 sense 5′-CTCTGAATAAAGACGACATAGACGGC-3′, antisense 5′-TCTGAGCTCGATCTCGTTGGAT-3′, for LT-α sense 5′-TGGCTGGGAACAGGGGAAGGTTGAC-3′, antisense 5′-CGTGCTTTCTTCTAGAACCCCTTGG-3′, for β-actin sense 5′-GTGGGCCGCTCTAGGCACCAA-3′, antisense 5′-CTCTTTGATGTCACGCACGATTTC-3′.
Skin samples were frozen in liquid nitrogen and were homogenized in 0.1M NaCl, 0.01M Tris HCl (pH 7.6), 1 mM EDTA (pH 8.0), and 0.02 mg/ml complete protease inhibitor (Roche Diagnostics, Mannheim, Germany). For cultured cell sampling, ∼5 × 105 cells were solubilized at 4°C in a lysis buffer (0.5% sodium deoxycholate, 1% Nonidet P40, 0.1% sodium dodecyl sulfate, 100 μg/ml phenylmethylsulfonyl fluoride, and 1 mM sodium orthovanadate) for 30 minutes at 4°C. The protein extracts (20 μg of each) were analyzed with an anti–MMP-1 antibody (Sigma, St. Louis, MO), an anti–MMP-2 antibody (Fuji Yakuhin, Saitama, Japan), an anti-phosphorylated c-Jun N-terminal kinase/stress-activated protein kinase (JNK/SAPK) antibody (New England Biolabs, Beverly, MA), and an anti-JNK/SAPK antibody (New England Biolabs), as previously described (18). An antiactin antibody (Chemicon, Temecula, CA) was used as a control.
The expression of TNFRp55 on wild-type MEFs was analyzed by flow cytometry with fluorescein isothiocyanate–labeled TNFRp55 monoclonal antibody, as previously described (35).
After a daily subcutaneous injection of 100 μg of bleomycin, wild-type mice began to exhibit sclerosis on day 14 of the injections. In contrast, on day 3 of bleomycin treatment, the skin of the TNFRp55−/− mice exhibited severe sclerosis. Skin sections stained with H&E from bleomycin-injected TNFRp55−/− mice were histologically characterized by thickened and homogeneous collagen bundles and inflammatory infiltrates, whereas the skin from bleomycin-injected wild-type mice developed severe inflammatory infiltrates (Figure 1B), but neither sclerosis nor fibrosis (Figure 1A). The thickness of the skin of bleomycin-treated TNFRp55−/− mice underwent a time-dependent increase and exhibited a significant difference as compared with that of the skin of bleomycin-treated wild-type mice during days 3–7 of injections (Figure 2). On day 14, when the skin of wild-type mice started to thicken, the skin of TNFRp55−/− mice instead started to become atrophic. These results indicate that disruption of the TNFRp55 gene enhanced the effect of bleomycin to cause the development of thickened skin.
To investigate the etiology of bleomycin-induced thickened skin in TNFRp55−/− mice, the contents of the extracellular matrix in skin of wild-type and TNFRp55−/− mice with or without bleomycin treatment were studied (Figures 3A–C). An azan-stained section of the skin of TNFRp55−/− mice obtained after 3 days of bleomycin treatment exhibited thickened collagen bundles, while that of the skin of wild-type mice did not (Figure 3A). In the sclerotic skin of TNFRp55−/− mice after 1–5 days of bleomycin injections, the content of both the α1 and α2 collagen chains was significantly increased as compared with that in normal, untreated mice and the skin of bleomycin-treated wild-type mice (Figure 3B). The collagen content at the site of bleomycin injection was almost 3-fold that in the skin of the untreated mice and the bleomycin-treated wild-type mice. In contrast, no sclerotic skin or accumulation of collagen was detected at sites far from the bleomycin injection site (Figure 3B). This clearly indicates that sclerotic skin and collagen accumulation were induced only at bleomycin injection sites. Furthermore, the expression level of another component of the extracellular matrix, proteoglycan, was also analyzed (Figure 3C). Hyaluronic acid and dermatan sulfate were evenly expressed in both genotypes. Taken together, these results might suggest that bleomycin-induced thickened skin is caused by an accumulation of collagen.
Because collagen turnover is regulated by 2 reciprocal pathways, collagen synthesis and collagen degradation, the induction of expression of some fibrogenic cytokines in the skin of bleomycin-treated mice was examined using RT-PCR analysis (Figure 4). The expression levels of IL-4 and PDGF were not influenced by the bleomycin treatment. TGFβ messenger RNA (mRNA) was reduced, but TNF mRNA was induced in a time- and dose-dependent manner related to the bleomycin injections. There were no significant changes between phenotypes in the expression levels of these cytokines. The LT-α transcript peaked 5 days after bleomycin treatment in wild-type mice, whereas it peaked at 12 hours in TNFRp55−/− mice. However, at the protein level, there was no difference between the 2 genotypes (data not shown). Although we currently do not know the reason, it seems that LT-α is not involved in this pathology. The expression level of proα1 mRNA was not altered in the skin of TNFRp55−/− mice as compared with the skin of wild-type mice (Figures 5A and B). These results indicate that the collagen synthesis pathway was not affected in the skin of TNFRp55−/− mice.
To determine whether the collagen degradation pathway was affected, the expression levels of MMPs in the skin of bleomycin-treated mice (Figures 5 and 6) were examined. First, the expression of MMP-2 and MMP-9, which are representative of gelatinases, were analyzed by RT-PCR (Figures 5A and B). Their expressions were not altered in the skin of bleomycin-treated TNFRp55−/− mice. Although the expression of MMP-1 (collagenase) was increased in the skin of wild-type mice following bleomycin treatment, it was significantly reduced in the skin of TNFRp55−/− mice (Figure 6). A discrepancy was found in the low MMP-2 transcript measured 3 and 5 days after bleomycin treatment and the high protein concentration of MMP-2 detected at the protein level. It has been reported that bleomycin can induce MMP-2 expression in alveolar epithelial cells (19, 20), and some shedded MMP-2 was stored in endosomal vesicles for recycling (21). Thus, it is possible that we detected not only newly produced MMP-2 but also endosomestored MMP-2 at the protein level. These results raised the hypothesis that aberrant MMP-1 expression in TNFRp55−/− mice might cause severe accumulation of collagen in bleomycin-treated skin, and that TNFRp55 might play an important role in the regulation of MMP-1 expression.
To confirm the above hypothesis, wild-type and TNFRp55−/− MEFs were treated with TNF in vitro (Figure 7). The induction of MMP-1 expression was impaired in TNFRp55−/− MEFs, while wild-type MEFs were able to induce MMP-1 expression following TNF treatment. Because TNFRp55−/− MEFs express functional TNFRp75, this indicated that signaling mediated by TNFRp55, but not TNFRp75, is essential for the induction of MMP-1 expression. Next, we examined whether the nuclear factor κB (NF-κB) pathway in TNFRp55−/− MEFs was functional by using another activator of NF-κB, IL-1α. IL-1α induced the translocation of p65 to the nucleus (data not shown), and MMP-1 was induced in a manner comparable with that in wild-type MEFs. Because TNF also activates activator protein 1 (AP-1), and the induction of MMP-1 was demonstrated by the binding of AP-1 to TPA responsive element (TRE) in MMP-1 promoter (22), JNK activities in TNF-treated wild-type and TNFRp55−/− MEFs were examined (Figure 7B). The phosphorylation of JNK was detected in TNFRp55−/− MEFs as in wild-type MEFs, although its appearance was delayed until 60 minutes after treatment. The delayed kinetics of JNK activation in TNFRp55−/− MEFs was not seen when IL-1α was used (Figure 7B). Although the activation pathways for NF-κB and AP-1 were functional with the stimulation of IL-1α, MMP-1 induction was largely dependent on the TNF/TNFRp55 pathway in vivo. Alternatively, some negative regulatory pathway(s) in MMP-1 induction may have been aberrantly dominated in the absence of TNF/TNFRp55 signaling.
To examine the effect of bleomycin on the expression of TNFRp55, the cell surface expression of the TNFRp55 protein was analyzed by flow cytometry (Figure 8). The up-regulation of TNFRp55 protein on bleomycin-treated wild-type MEFs was noted. Taken together, the data indicated that bleomycin induces TNF and TNFRp55 expression in fibroblasts, and then TNFRp55-mediated signaling activated JNK and NF-κB. The subsequent production of MMP-1 protein led to the degradation of collagen. The absence of this cascade sequence might lead to the development of the severe sclerotic skin in TNFRp55−/− mice following bleomycin treatment.
While there has been much research into the effects of the TNF/TNFR system on scleroderma, little is known about its effects on the development of skin sclerosis. The experimental data presented here show that TNFRp55 is a key regulator in the induction of sclerotic skin. TNFRp55−/− mice developed severe skin sclerosis and accumulation of collagen on day 3 of bleomycin treatment, whereas wild-type mice did not. As described above, there is a possibility that the mutual antagonism of TNF and soluble TNFRp55 is involved in the mechanism of scleroderma pathogenesis. This murine model shows that impaired TNFRp55 signaling might also occur in scleroderma patients. Until now, it was believed that the increase in the serum concentration of TNF reflected the inflammatory stages and also the extent of internal involvement (23, 24). The results of this study suggest that it may in fact function to improve the clinical condition.
In the course of scleroderma, a gradual increase of soluble TNFRp55 in serum occurs and may lead to the exacerbation of disease symptoms. However, an increase of soluble TNFR in serum has been described in patients with other autoimmune diseases, e.g., systemic lupus erythematosus, mixed connective tissue disease, and rheumatoid arthritis (RA) (12, 25, 26). Perhaps not only TNFRp55 signaling, but also other abnormalities, are required for exhibiting specific symptoms of these diseases.
In this study, a bleomycin-induced dermal fibrosis model was used. One of the characteristic histologic features of bleomycin-induced sclerosis is the presence of infiltrating mononuclear cells in the dermis in its early stages. However, it has been suggested that participation of T cells or B cells, as well as mast cells, is not essential for the development of bleomycin-induced dermal sclerosis, since similar pathology was induced in both SCID mice and mast cell–deficient WBB6F1-W/Wv mice (27). It was shown that bleomycin up-regulates type I collagen and fibronectin mRNA in cultured normal skin fibroblasts (28). Exposure of rat lung fibroblast cultures to bleomycin results in elevated TGFβ mRNA synthesis and TGFβ protein, which is followed by increased procollagen gene transcription (29). A recent study showed that TGFβ is a mediator of the fibrotic effect of bleomycin at the transcriptional level and that the TGFβ response element is required for bleomycin stimulation of the proα1(I) collagen promoter (30). These previously published results (27–30) indicate that bleomycin increases extracellular matrix production even without the involvement of immune cells.
We show here that TNFRp55-mediated signaling is essential for bleomycin-induced MMP-1 expression. The role of TNF in extracellular matrix regulation is not limited to MMP-1. Indeed, a TNFRp55-specific form of mutant TNF markedly induces collagenase and stromelysin 1 gene expression in dermal fibroblasts (31). In addition, TNFRp55-specific TNF suppresses type I collagen mRNA levels as potently as wild-type TNF (31). Therefore, the abnormality in TNF/TNFRp55-mediated signaling might be involved in the aberrant regulation of fibrosis seen in scleroderma patients. However, an intrinsic abnormality was noticed in scleroderma fibroblasts. It was reported that the connective tissue growth factor (CTGF) protein, which promotes collagen synthesis, was constitutively expressed in scleroderma fibroblasts, but not in the normal counterpart (32). The CTGF protein is induced by TGFβ exclusively in connective tissue cells (33–35). Although TNF was able to repress TGFβ-induced CTGF and collagen synthesis in skin fibroblasts, the basal level of CTGF expression in scleroderma fibroblasts was unaffected (32). Therefore, we propose that at least 2 mechanisms are operative in progressive fibrosis: the inhibition of TNF/TNFRp55-mediated signaling, and the TNFRp55-independent, intrinsically aberrant regulation of fibrogenic molecules, such as CTGF.
It is interesting that MMP-1 is not so much induced in control mice as it is suppressed in TNFRp55−/− mice after bleomycin treatment. Therefore, the suppression, but not the inability of induction, of MMP-1 seems to cause sclerodermic skin. Unexpectedly, bleomycin stimulation of TNFRp55−/− MEFs did not cause the suppression of MMP-1 mRNA transcription (data not shown). In the absence of TNF/TNFRp55 signaling, it is possible that some cytokines aberrantly regulate the expression of MMP-1 in vivo. TGFβ and TNF are known to antagonize each other's function. A molecular mechanism of this reciprocal inhibition is at least partly attributable to the competition between TGFβ-activated Smad-3 and TNF-activated AP-1 for limiting the amount of the transcriptional coactivator p300 (36–38). Because TGFβ expression in the lesional skin is similar between the 2 genotypes and MMP-1 is under the control of TGFβ, it is reasonable to speculate that the inhibition of TGFβ in the expression of MMP-1 is exaggerated without TNF/TNFRp55 signaling.
Regarding MMP-2, the expression level was maintained much longer in TNFRp55−/− mice (Figure 6). A recent study showed that synthesis and/or secretion of MMP-2 was increased in cultured fibroblasts from a scleroderma patient (39), and MMP-2 was proposed to be involved in the pathophysiology of the disease by initiating microvascular damage, and leading to fibroblast activation. Interestingly, TGFβ increases the expression of MMP-2 in cultured fibroblasts (40). From this point of view, TGFβ might exert its positive regulatory activity in MMP-2 expression in the absence of TNF/TNFRp55 signaling. Studies to clarify of the role of TGFβ signaling in TNFRp55−/− mice are now under way.
In recent clinical trials on RA, TNF has been the target of immunotherapy using anti-human TNF chimeric monoclonal antibodies (41, 42). It has also been reported that serum MMP-1 and MMP-3 were reduced in RA patients following anti-TNF therapy (43). Based on the present results, it should be kept in mind that these patients may be at risk for the side effect of scleroderma-like symptoms.
The authors thank Drs. A. Koda, H. Ichinose, and M. Miyazaki for encouragement, and Ms Fumiyo Tujita for secretarial work.