The controversial role of tumor necrosis factor α in fibrotic diseases

Fibrosis is characterized by an excessive deposition of extracellular matrix (ECM) components, predominantly of collagens. Fibrotic changes have been observed in various diseases affecting different organs, such as the skin, lung, liver, and kidney. Current concepts of fibrogenesis suggest a cascade of events similar to those in wound healing (1, 2). An initial trigger or injury activates resident cells, which produce proinflammatory mediators. The resulting gradient of chemotactic cytokines guides the infiltration of inflammatory cells, which in turn, produce profibrotic mediators such as transforming growth factor β (TGFβ), platelet-derived growth factor, and interleukin-4 (IL-4) (3). Depending on the organ involved, these cytokines activate fibroblasts, hepatic stellate cells, mesangial cells, or tubular epithelial cells, leading to an increased production of ECM proteins. In later stages of the fibrotic diseases, profibrotic cytokines are often produced by resident cells. In addition to an increased production of ECM proteins, an imbalance between matrix-degrading enzymes and their inhibitors might contribute to the development of fibrosis. The matrix components accumulate in the extracellular compartment and disrupt the physiologic tissue structure, often causing severe functional impairment of the involved organs. Tumor necrosis factor α (TNFα) is produced by a wide variety of hematopoietic and nonhematopoietic cells, such as macrophages, CD4+ and CD8+ T cells, B cells, natural killer cells, neutrophils, smooth muscle cells, endothelial cells, and fibroblasts (4). TNFα exists in 2 forms, a membrane-bound form and a soluble form, both of which are functional and contribute to the effects of TNFα (5, 6). Human TNFα is expressed as a precursor protein with a molecular mass of 26 kd, which is then enzymatically cleaved to yield a 17-kd active form. Under physiologic conditions, TNFα forms homotrimers, which interact and crosslink with the respective receptors. TNFα binds with high affinity to 2 transmembrane receptors, TNF receptor I (TNFRI; also referred to as TNFRβ, p55, or CD120a) and TNFRII (also referred to as TNFRα or p75) (7). The homology between the 2 receptors is limited, especially within their intracellular regions (7, 8). The expression of TNFRI and TNFRII is differentially regulated. The expression of TNFRI seems to be controlled by a housekeeping promoter with high basal activity and only a weak response to stimulation. In contrast, the expression of TNFRII is strongly influenced by external factors (9, 10). TNFRI can mediate almost all activities of TNFα, whereas TNFRII transduces signals at physiologic levels only in a few cell types such as T cells (11). However, TNFRII plays an important role in fine-tuning TNFRI signaling. Due to its higher affinity, TNFRII binds TNFα preferentially at low concentrations. TNFα is then passed to neighboring TNFRI molecules, which allows TNFα to exert its effects at lower concentrations (12). The efficacy of TNFα antagonists in inflammatory diseases such as rheumatoid arthritis (RA), spondylarthropathies, and Crohn's disease has been demonstrated in numerous clinical trials and plays an important role in the treatment of patients with these diseases (13-15). Recent reports suggest that TNFα antagonists might also be beneficial for the treatment of fibrotic disorders. In a retrospective case series, treatment with etanercept appeared to be efficacious in improving active inflammatory joint disease in patients with systemic sclerosis (SSc). Etanercept was generally well tolerated, and skin fibrosis was reduced under treatment in this uncontrolled study (16). Similarly, investigators in another case series reported on outcomes in 4 patients with SSc and erosive arthritis treated with TNFα antagonists (17). TNFα antagonists not only improved arthritis, but the modified Rodnan skin thickness score (MRSS) (18), used as a clinical score for the quantification of skin fibrosis in SSc patients, also decreased in all patients, with reductions of >50% in 3 patients. Another case report showed stabilization of lung fibrosis in a patient with SSc under treatment with infliximab (19). Three additional open-label studies with etanercept or infliximab have been reported in abstract form (20-22). Overall, treatment was well tolerated in these case series, including long-term application for a mean of 24 months in 1 study. The MRSS was stable or improved, and levels of N-terminal propeptide of type III procollagen as a biomarker for fibrosis decreased under treatment in 1 study. While definite conclusions cannot be drawn from open-label, uncontrolled case series, these data suggest that TNFα antagonists might play a role in the treatment of fibrotic diseases. However, TNFα has long been considered an antifibrotic cytokine, and treatment with TNFα antagonists might not be safe in SSc patients. Accordingly, another case report showed that treatment with adalimumab was associated with fatal exacerbation of fibrosing alveolitis in a patient with SSc (23). Similarly, TNFα antagonists have been associated with fibrosing alveolitis in patients with RA in a number of case reports (24). Thus, before TNFα antagonists can be considered for the treatment of patients with fibrotic diseases, the following open questions have to be addressed: What is the experimental evidence for TNFα antagonists for the treatment of fibrotic disorders? Do molecular studies demonstrate a profibrotic role of TNFα, or do they rather show antifibrotic effects of TNFα? In this review, we discuss in detail the results of in vitro and in vivo studies analyzing the role of TNFα signaling in organ fibrosis, with a focus on skin and pulmonary fibrosis, to provide a basis for further clinical studies in rheumatic diseases. The effects of TNFα on the production of collagen and the turnover of ECM have been analyzed in numerous studies (Table 1). Mauviel and coworkers (25, 26) demonstrated that TNFα inhibits the synthesis of type I collagen in cultured dermal fibroblasts on the transcriptional level, resulting in a dose-dependent reduction of the production of type I collagen as measured by radioimmunoassay and gel electrophoresis. In addition, TNFα inhibited the synthesis of type III collagen and fibronectin (25). TNFα also plays an important role in the inhibitory effects of CD4+ T cells on collagen production in dermal fibroblasts. Th1 and Th2 cells activated by CD3 crosslinking as well as preparations of Th1 and Th2 plasma membranes significantly decreased the production of collagen in dermal fibroblasts from healthy controls (27). Antagonism of TNFα by the addition of soluble TNFRI, but not neutralization of IL-4 and interferon-γ, significantly reduced the inhibitory activity of T cells on matrix production. Addition of Th2 membranes to fibroblasts stimulated with the profibrotic cytokines TGFβ and IL-4 abrogated the increased production of collagen, demonstrating that the inhibition by Th2 membranes was dominant over the stimulatory effects of TGFβ and IL-4. Interestingly, these antifibrotic mechanisms of TNFα via T cell membranes might not be fully active in fibrotic diseases, because SSc fibroblasts were resistant to inhibition by Th2 membranes and less sensitive to inhibition by Th1 membranes compared with healthy fibroblasts. The inhibitory effects of TNFα on the production of collagen are partially mediated by NF-κB. Using electrophoretic mobility shift assays (EMSAs) and supershift assays, Kouba and coworkers (28) demonstrated that RelA–NF-κB (p65–p50) complexes bind to a cis element adjacent to the promoter of COL1A2. Mutation of the NF-κB binding site almost completely prevented the inhibitory effects of TNFα. The crucial role of NF-κB in the down-regulation of COL1A2 synthesis was confirmed using fibroblasts deficient in the NF-κB–essential modulator (NEMO) (29). NEMO is activated in response to inflammatory stimuli and activates the associated IκB kinases (IKKs). In turn, IKKs phosphorylate IκB. Phosphorylated IκB dissociates from NF-κB, which can then enter the nucleus and activate transcription of NF-κB–dependent genes. NEMO deficiency prevented the inhibitory effects of TNFα on COL1A2 transcription. Similar results were also obtained in dermal fibroblasts transfected with dominant-negative forms of IKKα (29). In addition to direct effects on the transcription of collagens, TNFα might exert antifibrotic properties by interfering with TGFβ signaling cascades (30). Incubation of neonatal human foreskin fibroblasts with TNFα 1 hour prior to stimulation with TGFβ prevented Smad-specific gene transcription, as analyzed with reporter constructs of the Smad3/4-binding element (SBE). The inhibitory effect of TNFα on the activation of TGFβ-dependent Smad signaling was not mediated by induction of the inhibitory Smad7, but rather, was inhibited by activator protein 1 (AP-1)–dependent pathways. JunB and c-Jun, 2 members of the AP-1 family, are strongly induced in dermal fibroblasts upon stimulation with TNFα. Overexpression of JunB and c-Jun abrogated the Smad3-dependent transactivation of the SBE reporter construct in a dose-dependent manner. The important role of AP-1 signaling for the inhibitory effect of TNFα was further supported by experiments demonstrating that expression of antisense c-Jun messenger RNA (mRNA) prevented the inhibition of TGFβ/Smad signaling by TNFα. The inhibitory effects of JunB and c-Jun on Smad signaling are mediated by 2 distinct mechanisms. Coprecipitation assays and EMSAs demonstrated that JunB and c-Jun form heterocomplexes with Smad3 and that JunB and c-Jun reduce the binding of Smad3 to its cognate cis DNA element. These findings suggest that Smad3–AP-1 interactions may sequester Smad3 and compete against Smad3 for binding to DNA. Furthermore, JunB and c-Jun bind p300, a cofactor essential for optimal transcription of Smad-regulated genes. Because the amount of p300 in the nucleus is limited, competition of JunB and c-Jun with Smad3 for p300 can efficiently prevent Smad signaling (30). In addition to its direct effects on the synthesis of ECM proteins, TNFα regulates the expression of matrix-degrading enzymes and their inhibitors. TNFα induced matrix metalloproteinase 1 (MMP-1) mRNA and protein in a dose-dependent manner in dermal, gingival, and synovial fibroblasts (31, 32). Furthermore, higher concentrations of TNFα reduced the expression of tissue inhibitor of metalloproteinases 1 (TIMP-1), thereby promoting the degradation of ECM proteins (33). The effects of TNFα on collagen synthesis and on the production of MMPs and TIMPs are not restricted to fibroblasts and are also found in other cell types (34). In contrast to the results discussed above, a recent study suggested that TNFα might promote a profibrotic phenotype in murine intestinal myofibroblasts in vitro (35). In vimentin- and α-smooth muscle actin–positive myofibroblasts isolated from wild-type (WT) mice, murine TNFα stimulated collagen synthesis, increased expression of TIMP-1, and decreased activity of MMP-2. Similar results were also obtained with myofibroblasts from mice homozygous for the disruption of the gene for TNFRI (TNFRI−/− mice). However, no induction of collagen was observed in myofibroblasts from TNFRII−/− mice and from TNFRI−/− and TNFRII−/− double-knockout mice (TNFRI−/−/II−/− mice), suggesting that the profibrotic effect of TNFα on myofibroblasts is mediated by signaling via TNFRII. It remains unclear why the effects of TNFα on the production of ECM in that study differed from those found in previous studies. The concentrations of TNFα used were within the range of those used in other studies, and species-specific differences can be excluded because murine fibroblasts respond to TNFα with an up-regulation of collagen and induction of MMPs similar to that found in human fibroblasts (29). However, the effects of TNFα might be cell type and organ specific, and intestinal myofibroblasts might react differently from other cell types. Investigators in another in vitro study also proposed a profibrotic effect of TNFα, indirectly via induction of TGFβ (36). Incubation of Swiss 3T3 cells (a mouse fibroblast cell line) and primary mouse fibroblasts with TNFα stimulated the expression of TGFβ mRNA and protein. The induction of TGFβ by TNFα was blocked by inhibitors of the ERK-specific MAPK pathway. However, it remains unclear whether the induction of TGFβ is sufficient to overcome the inhibitory effects of TNFα on collagen production, since the investigators did not analyze the expression of collagen. The induction of TGFβ upon stimulation with TNFα might therefore represent a counterregulatory mechanism to compensate for the inhibitory effects of TNFα on Smad signaling. Phagocytosis of collagen has been suggested to play a role in the homeostasis of the ECM, and inhibition of this phagocytosis might therefore lead to fibrosis (37). Although this mechanism is currently not considered to play a major role, it might further enhance fibrotic processes. TNFα has been suggested to inhibit the phagocytosis of collagen by fibroblasts, and it might also therefore enhance fibrotic processes via this mechanism. Incubation with TNFα increased slightly the expression of α1β1 integrin and α2β1 integrin, which bind collagen in human gingival fibroblasts, but the binding of collagen was significantly decreased, suggesting an inactivation of the binding sites for collagen by TNFα (38). Consistent with decreased binding, the percentage of phagocytosed collagen beads per cell and the proportion of phagocytically active gingival fibroblasts were reduced in a dose-dependent manner by TNFα. Results from animal studies of the overexpression of TNFα are inconsistent regarding the effects on ECM synthesis (Table 2). A number of studies indicate profibrotic effects of TNFα. Overexpression of TNFα in the lungs has been linked to fibrosis. Using a replication-deficient adenovirus, a prolonged overexpression of TNFα for 7–10 days was achieved in the lungs of rats (39). This local overexpression resulted in severe infiltration of the lungs by neutrophils, macrophages, and lymphocytes. Later on, increased expression of TGFβ was detected and myofibroblasts appeared. After the inflammation subsided, a patchy fibrosis developed that persisted beyond day 64. These data are supported by the analysis of peripheral CD4+ T cells from patients with idiopathic pulmonary fibrosis. Compared with those from normal subjects, these CD4+ T cells synthesized higher levels of TNFα, as analyzed by intracellular fluorescence-activated cell sorting staining (40). The role of TNFα in pulmonary fibrosis has also been analyzed with transgenic mice overexpressing murine TNFα in the lungs under the control of the human surfactant protein C (SP-C) promoter. These mice spontaneously developed a chronic lymphocytic alveolitis, and its severity correlated with the expression of TNFα mRNA in the lungs (41). The transgenic mice also had greater lung volumes, pulmonary hypertension, and decreased elastic recoil compared with control mice. Histologically, the investigators observed thickened alveolar walls due to accumulation of desmin-containing fibroblasts and collagen fibers, a phenotype resembling idiopathic pulmonary fibrosis. Similar profibrotic effects of TNFα were observed in animal models of congestive heart failure using modified TNFα-transgenic mice (42, 43). Accordingly, continuous subcutaneous infusion of TNFα led to abundant infiltration with polymorphonuclear leukocytes and stimulated the growth of dermal fibroblasts, resulting in an increased local deposition of collagen (44). While these studies suggest that TNFα is a profibrotic cytokine, other studies in mice overexpressing TNFα indicate antifibrotic effects and a protective role of TNFα in models of experimental fibrosis. A more recent study by a different group on the same transgenic mice with lung-specific TNFα expression under the control of the human SP-C promoter led to different conclusions. Fujita and colleagues (45) confirmed the presence of chronic pulmonary inflammation and increased lung volumes with an increase in the total amount of hydroxyproline in the lungs. However, when the hydroxyproline content was normalized for the lung weight, no differences were observed between TNFα-transgenic mice and controls. Accordingly, Fujita et al observed only very little fibrosis in the lungs of transgenic animals. Furthermore, TNFα-transgenic mice were resistant to experimental pulmonary fibrosis induced by bleomycin or by overexpression of TGFβ. In contrast to WT mice, no increase in the hydroxyproline content and no fibrotic changes in histologic sections were observed in TNFα-transgenic mice. It needs to be stressed that the determination of the hydroxyproline content is often misleading, especially under inflammatory conditions, in which inflammatory cells produce C1q (which also contains hydroxyproline). The fact that type III collagen contains 30% more hydroxyproline than does type I collagen also contributes to this problem. The antifibrotic role of TNFα in experimental pulmonary fibrosis in these animals was further supported by experiments demonstrating that application of recombinant human TNFα attenuated bleomycin-induced pulmonary fibrosis in WT mice. An antifibrotic effect of TNFα was also reported by Buck and coworkers (46). They demonstrated that inoculation of nude mice with Chinese hamster ovary cells secreting TNFα resulted in a decreased collagen synthesis in the skin, which also led to impaired wound healing and was paralleled by decreased expression of TGFβ. Similar results in the liver were obtained by the same group by and coworkers demonstrated that TNFα antagonists can significantly reduce the accumulation of ECM in mouse models of and bleomycin-induced pulmonary fibrosis. and bleomycin the expression of TNFα and lead to a patchy fibrosis of or recombinant soluble TNFRI prevented the development of fibrosis in both of fibrosis with soluble TNFRI was also as demonstrated by experiments in which soluble TNFRI was days bleomycin and Accordingly, continuous infusion of recombinant TNFα strongly the fibrotic effects of TNFα inhibitors were also found in a from rats were disease with inflammatory and in of animals animals were treated with against TNFα or with a inhibitor of TNFα, were strongly reduced and the percentage of decreased from to Similar antifibrotic effects with were observed in experimental as a for fibrosis studies have analyzed outcomes in mice deficient in TNFRI (TNFRI−/− mice), TNFRII mice), or both receptors (TNFRI−/−/II−/− in different models of organ fibrosis. In contrast to experiments with TNFα-transgenic mice, these studies demonstrated that inhibition of signaling via TNFRI and TNFRII the development of fibrosis. mice were from experimental pulmonary fibrosis induced by and bleomycin In contrast to WT mice, mice did not demonstrate inflammation or increased in the lungs to profibrotic the induction of TNFα was and was no increased deposition of Similar results were obtained with TNFRI−/− and mice in different models of liver and fibrosis, TNFRII−/− mice were not from and liver fibrosis Results of in vitro and in vivo experiments the role of TNFα in fibrosis are in and do not definite conclusions the role of TNFα as a profibrotic or an antifibrotic The of in vitro studies show antifibrotic effects of TNFα, in that it the production of collagen, the expression of and the of thereby the accumulation of The intracellular signaling that mediate these effects have been and activation of JunB and However, under TNFα might have profibrotic effects in in which are to play a role in fibrotic such as SSc by of ECM proteins, might react to TNFα differently from fibroblasts. While the in vitro studies TNFα as an antifibrotic cytokine, different in vivo animal studies with antagonists of TNFα and mice TNFRI or both TNFα receptors demonstrated that inhibition of TNFα signaling can prevent fibrosis. However, studies on TNFα-transgenic mice different with suggesting a profibrotic role of TNFα. In these differences in the results between in vitro and in vivo studies might be by the inflammatory in animal models of experimental fibrosis 1). of these models on a inflammatory at the initial stages as a trigger for the later development of fibrosis. a major proinflammatory TNFα has been to play an important role in the activation of inflammatory cells. and inflammatory are in and TNFα can exert its direct effects on the cells. In a is that the direct antifibrotic effects of TNFα on fibroblasts might be in experimental models of fibrosis by its important role in Tumor necrosis factor α (TNFα) proinflammatory properties and activates inflammatory cells in the initial of experimental models of fibrosis. cells profibrotic which activate fibroblasts. fibroblasts produce increased of extracellular resulting in tissue fibrosis in later However, TNFα inhibits the synthesis of extracellular matrix proteins in fibroblasts in vitro in studies. do not have a of the of these effects in human fibrotic diseases, in which inflammation is often less than in experimental Thus, to the effects of in human fibrotic diseases, the is whether inflammation and the development of fibrosis in The role of inflammation in human fibrotic diseases is less than was it might between the and might also on other such as disease in SSc skin fibrosis, inflammatory are to play an important role in very disease but are in later stages of the This might that be a antifibrotic for very inflammatory skin fibrosis in it might be for later stages of skin fibrosis. Accordingly, the effects on lung fibrosis might be different from those on skin fibrosis. is to play a more important role in pulmonary fibrosis associated with with alveolitis of fibrosis the of the this patients with pulmonary fibrosis and alveolitis might from the effects on lung fibrosis might be in patients with no evidence of However, it be that this is because no controlled study has at the effects of TNFα in inflammatory of the results of molecular and studies do not definite conclusions the role of TNFα in fibrotic diseases. experimental studies are with a focus on fibrotic animal models that are of inflammation as a trigger for the fibrotic studies have a of the TNF or which provide a more analysis with a the findings reported of analyzing only and pathways. uncontrolled studies with TNFα inhibitors in SSc to be However, from the are that treatment with TNFα antagonists lead to of fibrosis. TNFα antagonists not be used in clinical for the treatment of patients with fibrotic diseases these are the clinical effects of TNFα antagonists in fibrotic diseases, trials to be with inflammatory stages of the fibrotic disease are to patients with stages of fibrosis might show these trials to focus on fibrosis outcomes in addition to the such as and and they to be with sufficient to