ARTHRITIS & RHEUMATISM Vol. 58, No. 8, August 2008, pp 2228–2235 DOI 10.1002/art.23645 © 2008, American College of Rheumatology REVIEW The Controversial Role of Tumor Necrosis Factor ␣ in Fibrotic Diseases Jörg H. W. Distler,1 Georg Schett,1 Steffen Gay,2 and Oliver Distler2 Introduction physiologic tissue structure, often causing severe functional impairment of the involved organs. Tumor necrosis factor ␣ (TNF␣) 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). TNF␣ exists in 2 forms, a membrane-bound form and a soluble form, both of which are functional and contribute to the effects of TNF␣ (5,6). Human TNF␣ 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, TNF␣ forms homotrimers, which interact and crosslink with the respective receptors. TNF␣ binds with high affinity to 2 transmembrane receptors, TNF receptor I (TNFRI; also referred to as TNFR␤, p55, or CD120a) and TNFRII (also referred to as TNFR␣ 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 TNF␣, 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 TNF␣ preferentially at low concentrations. TNF␣ is then passed to neighboring TNFRI molecules, which allows TNF␣ to exert its effects at lower concentrations (12). The efficacy of TNF␣ 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 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 ␤ (TGF␤), 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 Dr. Jörg H. W. Distler’s work was supported by the Interdisciplinary Center of Clinical Research (IZKF) in Erlangen (grant A20) and by the Career Support Award of Medicine of the Ernst Jung Foundation. 1 Jörg H. W. Distler, MD, Georg Schett, MD: University of Erlangen–Nuremberg, Erlangen, Germany; 2Steffen Gay, MD, Oliver Distler, MD: University Hospital Zurich, Zurich, Switzerland. Dr. Jörg H. W. Distler has received consulting fees, speaking fees, and/or honoraria from Abbott, Actelion, and Encysive (less than $10,000 each) and has received support from Novartis for a clinical trial studying systemic sclerosis. Dr. Gay has received consulting fees, speaking fees, and/or honoraria from Wyeth (less than $10,000). Address correspondence and reprint requests to Oliver Distler, MD, Center of Experimental Rheumatology and Center of Integrative Human Physiology, Department of Rheumatology, University Hospital Zurich, CH-8091 Zurich, Switzerland. E-mail: Oliver.Distler@usz.ch. Submitted for publication October 15, 2007; accepted in revised form April 18, 2008. 2228 TNF␣ IN FIBROTIC DISEASES 2229 Table 1. Overview of in vitro studies on the profibrotic and antifibrotic effects of tumor necrosis factor ␣* Cell type In vitro studies suggesting antifibrotic effects Dermal fibroblasts Dermal, gingival, and synovial fibroblasts Uterine fibroblasts Neonatal foreskin fibroblasts Dermal fibroblasts In vitro studies suggesting profibrotic effects Intestinal myofibroblasts Swiss 3T3 cells and primary mouse fibroblasts Fibroblasts Mechanism of action (ref.) Inhibition of transcription of fibronectin, type I collagen, and type III collagen (25,26) Induction of MMP-1 (31,32) Reduction of expression of TIMP-1 (33) Inhibition of TGF␤-mediated activation of Smad3 (30) Mediator of inhibitory effects of CD4 T cells on collagen production (27) Stimulation of collagen synthesis, increased expression of TIMP-1, and decreased activity of MMP-2 (35) Induction of TGF␤ via ERKdependent pathways (36) Inhibition of phagocytosis of collagen (38) * MMP-1 ⫽ matrix metalloproteinase 1; TIMP-1 ⫽ tissue inhibitor of metalloproteinases 1; TGF␤ ⫽ transforming growth factor ␤. lagen 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 TNF␣ antagonists might play a role in the treatment of fibrotic diseases. However, TNF␣ has long been considered an antifibrotic cytokine, and treatment with TNF␣ 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, TNF␣ antagonists have been associated with fibrosing alveolitis in patients with RA in a number of case reports (24). Thus, before TNF␣ 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 TNF␣ antagonists for the treatment of fibrotic disorders? Do molecular studies demonstrate a profibrotic role of TNF␣, or do they rather show antifibrotic effects of TNF␣? In this review, we discuss in detail the results of in vitro and in vivo studies analyzing the role of TNF␣ signaling in organ fibrosis, with a focus on skin and pulmonary fibrosis, to provide a basis for further clinical studies in rheumatic diseases. Effects of TNF␣ on ECM remodeling in vitro (13–15). Recent reports suggest that TNF␣ 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 TNF␣ antagonists (17). TNF␣ 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 procol- Evidence for antifibrotic effects of TNF␣. The effects of TNF␣ 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 TNF␣ 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, TNF␣ inhibited the synthesis of type III collagen and fibronectin (25). TNF␣ 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 TNF␣ by the addition of soluble TNFRI, but not neutralization of IL-4 and interferon-␥, significantly reduced the inhibitory activity of T cells on matrix production. Addition of Th2 membranes to fibroblasts stimulated with the profibrotic cytokines TGF␤ and IL-4 abrogated the increased production of collagen, demonstrating that the 2230 inhibition by Th2 membranes was dominant over the stimulatory effects of TGF␤ and IL-4. Interestingly, these antifibrotic mechanisms of TNF␣ 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 TNF␣ on the production of collagen are partially mediated by NF-B. Using electrophoretic mobility shift assays (EMSAs) and supershift assays, Kouba and coworkers (28) demonstrated that RelA–NF-B (p65–p50) complexes bind to a cis element adjacent to the promoter of COL1A2. Mutation of the NF-B binding site almost completely prevented the inhibitory effects of TNF␣. The crucial role of NF-B in the down-regulation of COL1A2 synthesis was confirmed using fibroblasts deficient in the NF-B– essential modulator (NEMO) (29). NEMO is activated in response to inflammatory stimuli and activates the associated IB kinases (IKKs). In turn, IKKs phosphorylate IB. Phosphorylated IB dissociates from NF-B, which can then enter the nucleus and activate transcription of NF-B–dependent genes. NEMO deficiency prevented the inhibitory effects of TNF␣ on COL1A2 transcription. Similar results were also obtained in dermal fibroblasts transfected with dominant-negative forms of IKK␣ (29). In addition to direct effects on the transcription of collagens, TNF␣ might exert antifibrotic properties by interfering with TGF␤ signaling cascades (30). Incubation of neonatal human foreskin fibroblasts with TNF␣ 1 hour prior to stimulation with TGF␤ prevented Smadspecific gene transcription, as analyzed with reporter constructs of the Smad3/4-binding element (SBE). The inhibitory effect of TNF␣ on the activation of TGF␤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 TNF␣. 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 TNF␣ was further supported by experiments demonstrating that expression of antisense c-Jun messenger RNA (mRNA) prevented the inhibition of TGF␤/Smad signaling by TNF␣. The inhibitory effects of JunB and c-Jun on Smad signaling are mediated by 2 distinct mechanisms. Coprecipitation assays and EMSAs demonstrated that JunB DISTLER ET AL 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, TNF␣ regulates the expression of matrixdegrading enzymes and their inhibitors. TNF␣ 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 TNF␣ reduced the expression of tissue inhibitor of metalloproteinases 1 (TIMP-1), thereby promoting the degradation of ECM proteins (33). The effects of TNF␣ 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). Evidence for profibrotic effects of TNF␣. In contrast to the results discussed above, a recent study suggested that TNF␣ might promote a profibrotic phenotype in murine intestinal myofibroblasts in vitro (35). In vimentin- and ␣-smooth muscle actin–positive myofibroblasts isolated from wild-type (WT) mice, murine TNF␣ 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–/– doubleknockout mice (TNFRI⫺/⫺/II⫺/⫺ mice), suggesting that the profibrotic effect of TNF␣ on myofibroblasts is mediated by signaling via TNFRII. It remains unclear why the effects of TNF␣ on the production of ECM in that study differed from those found in previous studies. The concentrations of TNF␣ used were within the range of those used in other studies, and species-specific differences can be excluded because murine fibroblasts respond to TNF␣ with an up-regulation of collagen and induction of MMPs similar to that found in human fibroblasts (29). However, the effects of TNF␣ 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 TNF␣, indirectly via induction of TGF␤ (36). Incubation of Swiss 3T3 cells (a TNF␣ IN FIBROTIC DISEASES mouse fibroblast cell line) and primary mouse fibroblasts with TNF␣ stimulated the expression of TGF␤ mRNA and protein. The induction of TGF␤ by TNF␣ was blocked by inhibitors of the ERK-specific MAPK pathway. However, it remains unclear whether the induction of TGF␤ is sufficient to overcome the inhibitory effects of TNF␣ on collagen production, since the investigators did not analyze the expression of collagen. The induction of TGF␤ upon stimulation with TNF␣ might therefore represent a counterregulatory mechanism to compensate for the inhibitory effects of TNF␣ 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. TNF␣ has been suggested to inhibit the phagocytosis of collagen by fibroblasts, and it might also therefore enhance fibrotic processes via this mechanism. Incubation with TNF␣ increased slightly the expression of ␣1␤1 integrin and ␣2␤1 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 TNF␣ (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 TNF␣. The role of TNF␣ in different in vivo models of tissue fibrosis Effects of overexpression of TNF␣ on the turnover of ECM. Results from animal studies of the overexpression of TNF␣ are inconsistent regarding the effects on ECM synthesis (Table 2). A number of studies indicate profibrotic effects of TNF␣. Overexpression of TNF␣ in the lungs has been linked to fibrosis. Using a replication-deficient adenovirus, a prolonged overexpression of TNF␣ 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 TGF␤ 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 2231 Table 2. Overview of studies on the role of TNF␣ in animal models of skin and pulmonary fibrosis* Organ In vivo studies suggesting antifibrotic effects Lungs Lungs Trachea Skin In vivo studies suggesting profibrotic effects Lungs Heart Skin Model (ref.) Reduction by TNF␣ antagonists of accumulation of extracellular matrix in mouse models of silica- and bleomycin-induced pulmonary fibrosis (48,49) Protection from experimental pulmonary fibrosis induced by silica, asbestos, and bleomycin in TNFRI⫺/⫺/II⫺/⫺ mice (52–54) Reduced fibroproliferative lesions after allogeneic transplantation of rat tracheas by treatment with TNF␣ antagonists (50) Decreased collagen synthesis in the skin of nude mice injected with Chinese hamster ovary cells overexpressing TNF␣ (47) Induction of pulmonary fibrosis in mice overexpressing TNF␣ (39,41) Induction of interstitial myocardial fibrosis by overexpression of TNF␣ (42) Increased deposition of collagen after subcutaneous infusion of TNF␣ (44) * TNF␣ ⫽ tumor necrosis factor ␣. subjects, these CD4⫹ T cells synthesized higher levels of TNF␣, as analyzed by intracellular fluorescenceactivated cell sorting staining (40). The role of TNF␣ in pulmonary fibrosis has also been analyzed with transgenic mice overexpressing murine TNF␣ 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 TNF␣ 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 TNF␣ were observed in animal models of congestive heart failure using modified TNF␣-transgenic mice (42,43). Accordingly, 2232 continuous subcutaneous infusion of TNF␣ 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 TNF␣ is a profibrotic cytokine, other studies in mice overexpressing TNF␣ indicate antifibrotic effects and a protective role of TNF␣ in models of experimental fibrosis. A more recent study by a different group on the same transgenic mice with lung-specific TNF␣ 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 TNF␣transgenic mice and controls. Accordingly, Fujita et al observed only very little fibrosis in the lungs of transgenic animals. Furthermore, TNF␣-transgenic mice were resistant to experimental pulmonary fibrosis induced by bleomycin or by overexpression of TGF␤. In contrast to WT mice, no increase in the hydroxyproline content and no fibrotic changes in histologic sections were observed in TNF␣-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 TNF␣ in experimental pulmonary fibrosis in these animals was further supported by experiments demonstrating that application of recombinant human TNF␣ attenuated bleomycin-induced pulmonary fibrosis in WT mice. An antifibrotic effect of TNF␣ was also reported by Buck and coworkers (46). They demonstrated that inoculation of nude mice with Chinese hamster ovary cells secreting TNF␣ resulted in a decreased collagen synthesis in the skin, which also led to impaired wound healing and was paralleled by decreased expression of TGF␤. Similar results in the liver were obtained by the same group (47). Studies with inhibitors of TNF␣. Studies by Piguet and coworkers (48,49) demonstrated that TNF␣ antagonists can significantly reduce the accumulation of ECM in mouse models of silica- and bleomycin-induced pulmonary fibrosis. Silica and bleomycin stimulate the expression of TNF␣ and lead to a patchy fibrosis after 15 DISTLER ET AL days. Infusion of anti-TNF␣ antibodies or recombinant soluble TNFRI prevented the development of fibrosis in both models. Treatment of established fibrosis with soluble TNFRI was also effective, as demonstrated by experiments in which soluble TNFRI was administered 25 days after bleomycin and silica. Accordingly, continuous infusion of recombinant TNF␣ strongly augmented the fibrotic process (48). Beneficial effects of TNF␣ inhibitors were also found in a transplantation model (50). When tracheas from Brown-Norway rats were transplanted into Lewis rats, obliterative airway disease with inflammatory infiltrates and fibroproliferative lesions developed. Luminal occlusion occurred in 58% of untreated animals after 14 days. When animals were treated with antibodies against TNF␣ or with RDP-58, a translational inhibitor of TNF␣, fibroproliferative lesions were strongly reduced and the percentage of luminal occlusion decreased from 58% to 32%. Similar antifibrotic effects with anti-TNF␣ antibodies were observed in experimental crescentic glomerulonephritis as a model for kidney fibrosis (51). Lessons from studies of mice deficient in TNF receptors. Numerous studies have analyzed outcomes in mice deficient in TNFRI (TNFRI–/– mice), TNFRII (TNFRII–/– mice), or both receptors (TNFRI⫺/⫺/II–/– mice) in different models of organ fibrosis. In contrast to experiments with TNF␣-transgenic mice, these studies demonstrated uniformly that inhibition of signaling via TNFRI and TNFRII prevents the development of fibrosis. For instance, TNFRI⫺/⫺/II–/– mice were protected from experimental pulmonary fibrosis induced by silica, asbestos, and bleomycin (52–54). In contrast to WT mice, TNFRI⫺/⫺/II–/– mice did not demonstrate significant inflammation or increased proliferation in the lungs after exposure to profibrotic agents, the induction of TNF␣ was diminished, and there was no increased deposition of ECM. Similar results were obtained with TNFRI–/– and TNFRI⫺/⫺/II–/– mice in different models of liver and kidney fibrosis, while TNFRII–/– mice were not protected from kidney and liver fibrosis (55–59). Summary and conclusion Results of in vitro and in vivo experiments concerning the role of TNF␣ in fibrosis are in part contradictory and do not allow definite conclusions about the role of TNF␣ as a profibrotic mediator or an antifibrotic cytokine. The majority of in vitro studies show antifibrotic effects of TNF␣, in that it suppresses the production of collagen, reduces the expression of TIMPs, and stimulates the release of MMPs, thereby preventing the TNF␣ IN FIBROTIC DISEASES Figure 1. Tumor necrosis factor ␣ (TNF␣) exerts prominent proinflammatory properties and activates inflammatory cells in the initial phases of experimental models of fibrosis. Inflammatory cells secrete profibrotic mediators, which activate fibroblasts. Activated fibroblasts produce increased amounts of extracellular matrix, resulting in tissue fibrosis in later stages. However, TNF␣ directly inhibits the synthesis of extracellular matrix proteins in fibroblasts in vitro in most studies. We do not yet have a clear picture of the net result of these contrary effects in human fibrotic diseases, in which inflammation is often less pronounced than in experimental models. accumulation of ECM. The intracellular signaling pathways that mediate these effects have been identified and include activation of JunB and c-Jun. However, under certain circumstances, TNF␣ might have profibrotic effects in vitro. Myofibroblasts in particular, which are thought to play a central role in fibrotic disorders such as SSc by releasing large amounts of ECM proteins, might react to TNF␣ differently from resting fibroblasts. While the in vitro studies overall favor TNF␣ as an antifibrotic cytokine, different in vivo animal studies with antagonists of TNF␣ and mice lacking TNFRI or both TNF␣ receptors demonstrated that inhibition of TNF␣ signaling can prevent fibrosis. However, studies on TNF␣-transgenic mice yielded different results, with most suggesting a profibrotic role of TNF␣. In our opinion, these differences in the results between in vitro and in vivo studies might be explained by the inflammatory component in animal models of experimental fibrosis (Figure 1). Most of these models depend critically on a strong inflammatory component at the initial stages as a trigger for the later development of fibrosis. As a major proinflammatory stimulus, TNF␣ has been shown to play an important role in the activation of inflammatory cells. Inflammatory conditions and cellular inflammatory infiltrates are absent in vitro, and 2233 TNF␣ can exert its direct effects on the matrixproducing cells. In our view, a likely scenario is that the direct antifibrotic effects of TNF␣ on fibroblasts might be outweighed in experimental models of fibrosis by its important role in driving inflammation. Thus, to predict the effects of anti-TNF␣ treatments in human fibrotic diseases, the key question is whether inflammation triggers and perpetuates the development of fibrosis in humans. The role of inflammation in human fibrotic diseases is less clear than was initially thought; it might differ between the specific organs and might also depend on other parameters such as disease duration. For example, in SSc skin fibrosis, inflammatory infiltrates are thought to play an important role in very early disease stages, but are rarely seen in later stages of the disease. This might predict that anti-TNF␣ treatments could be a promising antifibrotic strategy for very early inflammatory skin fibrosis in SSc, while it might even be deleterious for later noninflammatory stages of skin fibrosis. Accordingly, the effects on lung fibrosis might be different from those on skin fibrosis. Inflammation is thought to play a more important role in interstitial pulmonary fibrosis associated with SSc, with recurrent alveolitis triggering progression of fibrosis during the course of the disease. If this concept holds true, patients with pulmonary fibrosis and alveolitis might benefit from anti-TNF␣ treatments, while the effects on lung fibrosis might again be deleterious in patients with no evidence of inflammation/alveolitis. However, it must be emphasized that this concept is theoretical, because so far, no controlled study has looked at the effects of TNF␣ in inflammatory phases of SSc. Taken together, the results of available molecular and cellular studies do not allow definite conclusions about the role of TNF␣ in fibrotic diseases. More experimental studies are necessary, with a particular focus on fibrotic animal models that are independent of inflammation as a trigger for the fibrotic process. Cellular studies have lacked a complete description of the transcriptom after TNF or anti-TNF treatment, which would provide a more comprehensive analysis and, together with a proteomics study, could elucidate the contradictory findings reported so far instead of analyzing only selective and expected pathways. Open-label uncontrolled studies with TNF␣ inhibitors in SSc appear to be promising. However, from the existing literature, there are concerns that treatment with TNF␣ antagonists could lead to progression of fibrosis. Therefore, TNF␣ antagonists should not be used in daily clinical practice for the treatment of 2234 DISTLER ET AL patients with fibrotic diseases until these concerns are cleared. To elucidate the definitive clinical effects of TNF ␣ antagonists in fibrotic diseases, placebocontrolled trials need to be performed. Patients with inflammatory stages of the fibrotic disease are most likely to respond, while patients with noninflammatory stages of fibrosis might even show deleterious effects. Regarding safety issues, these trials need to focus on fibrosis outcomes in addition to the conventional safety issues such as infection and malignancy, and they need to be designed with sufficient statistical power to exclude potentially harmful effects. REFERENCES 1. Friedman SL, Bansal MB. Reversal of hepatic fibrosis—fact or fantasy? Hepatology 2006;43 Suppl 1:S82–8. 2. Liu Y. Renal fibrosis: new insights into the pathogenesis and therapeutics. Kidney Int 2006;69:213–7. 3. Varga J, Abraham D. Systemic sclerosis: a prototypic multisystem fibrotic disorder. J Clin Invest 2007;117:557–67. 4. Idriss HT, Naismith JH. TNF␣ and the TNF receptor superfamily: structure-function relationship(s). Microsc Res Tech 2000;50: 184–95. 5. Beyaert R, Fiers W. Molecular mechanisms of tumor necrosis factor-induced cytotoxicity: what we do understand and what we do not. FEBS Lett 1994;340:9–16. 6. Watts AD, Hunt NH, Hambly BD, Chaudhri G. Separation of tumor necrosis factor ␣ isoforms by two-dimensional polyacrylamide gel electrophoresis. Electrophoresis 1997;18:1086–91. 7. Loetscher H, Steinmetz M, Lesslauer W. Tumor necrosis factor: receptors and inhibitors. Cancer Cells 1991;3:221–6. 8. Beutler B, Bazzoni F. TNF, apoptosis and autoimmunity: a common thread? Blood Cells Mol Dis 1998;24:216–30. 9. Kalthoff H, Roeder C, Brockhaus M, Thiele HG, Schmiegel W. Tumor necrosis factor (TNF) up-regulates the expression of p75 but not p55 TNF receptors, and both receptors mediate, independently of each other, up-regulation of transforming growth factor ␣ and epidermal growth factor receptor mRNA. J Biol Chem 1993;268:2762–6. 10. Rothe J, Bluethmann H, Gentz R, Lesslauer W, Steinmetz M. Genomic organization and promoter function of the murine tumor necrosis factor receptor ␤ gene. Mol Immunol 1993;30:165–75. 11. Vandenabeele P, Declercq W, Beyaert R, Fiers W. Two tumour necrosis factor receptors: structure and function. Trends Cell Biol 1995;5:392–9. 12. Barbara JA, Smith WB, Gamble JR, Van Ostade X, Vandenabeele P, Tavernier J, et al. Dissociation of TNF-␣ cytotoxic and proinflammatory activities by p55 receptor- and p75 receptor-selective TNF-␣ mutants. EMBO J 1994;13:843–50. 13. Furst DE, Breedveld FC, Kalden JR, Smolen JS, Burmester GR, Bijlsma JW, et al. Updated consensus statement on biological agents, specifically tumour necrosis factor ␣ (TNF␣) blocking agents and interleukin-1 receptor antagonist (IL-1ra), for the treatment of rheumatic diseases, 2004. Ann Rheum Dis 2004;63 Suppl 2:ii2–12. 14. Keyser FD, Mielants H, Veys EM. Current use of biologicals for the treatment of spondyloarthropathies. Expert Opin Pharmacother 2001;2:85–93. 15. Olsen NJ, Stein CM. New drugs for rheumatoid arthritis. N Engl J Med 2004;350:2167–79. 16. Lam GK, Hummers LK, Woods A, Wigley FM. Efficacy and safety 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. of etanercept in the treatment of scleroderma-associated joint disease. J Rheumatol 2007;34:1636–7. Bosello S, De Santis M, Tolusso B, Zoli A, Ferraccioli G. Tumor necrosis factor-␣ inhibitor therapy in erosive polyarthritis secondary to systemic sclerosis. Ann Intern Med 2005;143:918–20. Clements P, Lachenbruch P, Seibold J, White B, Weiner S, Martin R, et al. Inter and intraobserver variability of total skin thickness score (modified Rodnan TSS) in systemic sclerosis. J Rheumatol 1995;22:1281–5. Antoniou KM, Mamoulaki M, Malagari K, Kritikos HD, Bouros D, Siafakas NM, et al. Infliximab therapy in pulmonary fibrosis associated with collagen vascular disease. Clin Exp Rheumatol 2007;25:23–8. Denton CP, Engelhart M, Tvede N, Khan K, Carreira PE, Gonzalez FD, et al. An open-label pilot study of infliximab therapy in diffuse cutaneous systemic sclerosis [abstract]. Arthritis Rheum 2006;54 Suppl 9:S524. Ellman MH, MacDonald PA, Hayes FA. Etanercept as treatment for diffuse scleroderma: a pilot study [abstract]. Arthritis Rheum 2000;43 Suppl 9:S392. Ellman MH, MacDonald PA, Katz RS. Open label use of etanercept in eight scleroderma patients [abstract]. Ann Rheum Dis 2003;62 Supp I:229. Allanore Y, Devos-Francois G, Caramella C, Boumier P, Jounieaux V, Kahan A. Fatal exacerbation of fibrosing alveolitis associated with systemic sclerosis in a patient treated with adalimumab. Ann Rheum Dis 2006;65:834–5. Ostor AJ, Crisp AJ, Somerville MF, Scott DG. Fatal exacerbation of rheumatoid arthritis associated fibrosing alveolitis in patients given infliximab. BMJ 2004;329:1266. Mauviel A, Daireaux M, Redini F, Galera P, Loyau G, Pujol JP. Tumor necrosis factor inhibits collagen and fibronectin synthesis in human dermal fibroblasts. FEBS Lett 1988;236:47–52. Mauviel A, Heino J, Kahari VM, Hartmann DJ, Loyau G, Pujol JP, et al. Comparative effects of interleukin-1 and tumor necrosis factor-␣ on collagen production and corresponding procollagen mRNA levels in human dermal fibroblasts. J Invest Dermatol 1991;96:243–9. Chizzolini C, Parel Y, De Luca C, Tyndall A, Akesson A, Scheja A, et al. Systemic sclerosis Th2 cells inhibit collagen production by dermal fibroblasts via membrane-associated tumor necrosis factor ␣. Arthritis Rheum 2003;48:2593–604. Kouba DJ, Chung KY, Nishiyama T, Vindevoghel L, Kon A, Klement JF, et al. Nuclear factor-B mediates TNF-␣ inhibitory effect on ␣2(I) collagen (COL1A2) gene transcription in human dermal fibroblasts. J Immunol 1999;162:4226–34. Verrecchia F, Wagner EF, Mauviel A. Distinct involvement of the Jun-N-terminal kinase and NF-B pathways in the repression of the human COL1A2 gene by TNF-␣. EMBO Rep 2002;3:1069–74. Verrecchia F, Pessah M, Atfi A, Mauviel A. Tumor necrosis factor-␣ inhibits transforming growth factor-␤/Smad signaling in human dermal fibroblasts via AP-1 activation. J Biol Chem 2000;275:30226–31. Dayer JM, Beutler B, Cerami A. Cachectin/tumor necrosis factor stimulates collagenase and prostaglandin E2 production by human synovial cells and dermal fibroblasts. J Exp Med 1985;162:2163–8. Meikle MC, Atkinson SJ, Ward RV, Murphy G, Reynolds JJ. Gingival fibroblasts degrade type I collagen films when stimulated with tumor necrosis factor and interleukin 1: evidence that breakdown is mediated by metalloproteinases. J Periodontal Res 1989; 24:207–13. Ito A, Sato T, Iga T, Mori Y. Tumor necrosis factor bifunctionally regulates matrix metalloproteinases and tissue inhibitor of metalloproteinases (TIMP) production by human fibroblasts. FEBS Lett 1990;269:93–5. Armendariz-Borunda J, Katayama K, Seyer JM. Transcriptional mechanisms of type I collagen gene expression are differentially TNF␣ IN FIBROTIC DISEASES 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. regulated by interleukin-1␤, tumor necrosis factor ␣, and transforming growth factor ␤ in Ito cells. J Biol Chem 1992;267: 14316–21. Theiss AL, Simmons JG, Jobin C, Lund PK. Tumor necrosis factor (TNF) ␣ increases collagen accumulation and proliferation in intestinal myofibroblasts via TNF receptor 2. J Biol Chem 2005; 280:36099–109. Sullivan DE, Ferris M, Pociask D, Brody AR. Tumor necrosis factor-␣ induces transforming growth factor-␤1 expression in lung fibroblasts through the extracellular signal-regulated kinase pathway. Am J Respir Cell Mol Biol 2005;32:342–9. McCulloch CA, Knowles GC. Deficiencies in collagen phagocytosis by human fibroblasts in vitro: a mechanism for fibrosis? J Cell Physiol 1993;155:461–71. Chou DH, Lee W, McCulloch CA. TNF-␣ inactivation of collagen receptors: implications for fibroblast function and fibrosis. J Immunol 1996;156:4354–62. Sime PJ, Marr RA, Gauldie D, Xing Z, Hewlett BR, Graham FL, et al. Transfer of tumor necrosis factor-␣ to rat lung induces severe pulmonary inflammation and patchy interstitial fibrogenesis with induction of transforming growth factor-␤1 and myofibroblasts. Am J Pathol 1998;153:825—32. Feghali-Bostwick CA, Tsai CG, Valentine VG, Kantrow S, Stoner MW, Pilewski JM, et al. Cellular and humoral autoreactivity in idiopathic pulmonary fibrosis. J Immunol 2007;179:2592–9. Miyazaki Y, Araki K, Vesin C, Garcia I, Kapanci Y, Whitsett JA, et al. Expression of a tumor necrosis factor-␣ transgene in murine lung causes lymphocytic and fibrosing alveolitis: a mouse model of progressive pulmonary fibrosis. J Clin Invest 1995;96:250–9. Kubota T, McTiernan CF, Frye CS, Slawson SE, Lemster BH, Koretsky AP, et al. Dilated cardiomyopathy in transgenic mice with cardiac-specific overexpression of tumor necrosis factor-␣. Circ Res 1997;81:627–35. Li YY, Feng YQ, Kadokami T, McTiernan CF, Draviam R, Watkins SC, et al. Myocardial extracellular matrix remodeling in transgenic mice overexpressing tumor necrosis factor ␣ can be modulated by anti-tumor necrosis factor ␣ therapy. Proc Natl Acad Sci U S A 2000;97:12746–51. Piguet PF, Grau GE, Vassalli P. Subcutaneous perfusion of tumor necrosis factor induces local proliferation of fibroblasts, capillaries, and epidermal cells, or massive tissue necrosis. Am J Pathol 1990;136:103–10. Fujita M, Shannon JM, Morikawa O, Gauldie J, Hara N, Mason RJ. Overexpression of tumor necrosis factor-␣ diminishes pulmonary fibrosis induced by bleomycin or transforming growth factor-␤. Am J Respir Cell Mol Biol 2003;29:669–76. Buck M, Houglum K, Chojkier M. Tumor necrosis factor-␣ inhibits collagen ␣1(I) gene expression and wound healing in a murine model of cachexia. Am J Pathol 1996;149:195–204. 2235 47. Houglum K, Buck M, Kim DJ, Chojkier M. TNF-␣ inhibits liver collagen-␣1(I) gene expression through a tissue-specific regulatory region. Am J Physiol 1998;274:G840–7. 48. Piguet PF, Collart MA, Grau GE, Sappino AP, Vassalli P. Requirement of tumour necrosis factor for development of silicainduced pulmonary fibrosis. Nature 1990;344:245–7. 49. Piguet PF, Vesin C. Treatment by human recombinant soluble TNF receptor of pulmonary fibrosis induced by bleomycin or silica in mice. Eur Respir J 1994;7:515–8. 50. Farivar AS, Mackinnon-Patterson B, McCourtie AS, Namkung J, Ward PA, Mulligan MS. Obliterative airway disease in rat tracheal allografts requires tumor necrosis factor ␣. Exp Mol Pathol 2005;78:190–7. 51. Khan SB, Cook HT, Bhangal G, Smith J, Tam FW, Pusey CD. Antibody blockade of TNF-␣ reduces inflammation and scarring in experimental crescentic glomerulonephritis. Kidney Int 2005;67: 1812–20. 52. Liu JY, Brass DM, Hoyle GW, Brody AR. TNF-␣ receptor knockout mice are protected from the fibroproliferative effects of inhaled asbestos fibers. Am J Pathol 1998;153:1839–47. 53. Ortiz LA, Lasky J, Hamilton RF Jr, Holian A, Hoyle GW, Banks W, et al. Expression of TNF and the necessity of TNF receptors in bleomycin-induced lung injury in mice. Exp Lung Res 1998;24: 721–43. 54. Ortiz LA, Lasky J, Lungarella G, Cavarra E, Martorana P, Banks WA, et al. Upregulation of the p75 but not the p55 TNF-␣ receptor mRNA after silica and bleomycin exposure and protection from lung injury in double receptor knockout mice. Am J Respir Cell Mol Biol 1999;20:825–33. 55. Guo G, Morrissey J, McCracken R, Tolley T, Klahr S. Role of TNFR1 and TNFR2 receptors in tubulointerstitial fibrosis of obstructive nephropathy. Am J Physiol 1999;277:F766–72. 56. Guo G, Morrissey J, McCracken R, Tolley T, Liapis H, Klahr S. Contributions of angiotensin II and tumor necrosis factor-␣ to the development of renal fibrosis. Am J Physiol Renal Physiol 2001; 280:F777–85. 57. Kitamura K, Nakamoto Y, Akiyama M, Fujii C, Kondo T, Kobayashi K, et al. Pathogenic roles of tumor necrosis factor receptor p55-mediated signals in dimethylnitrosamine-induced murine liver fibrosis. Lab Invest 2002;82:571–83. 58. Sudo K, Yamada Y, Moriwaki H, Saito K, Seishima M. Lack of tumor necrosis factor receptor type 1 inhibits liver fibrosis induced by carbon tetrachloride in mice. Cytokine 2005;29:236–44. 59. Tomita K, Tamiya G, Ando S, Ohsumi K, Chiyo T, Mizutani A, et al. Tumour necrosis factor ␣ signalling through activation of Kupffer cells plays an essential role in liver fibrosis of nonalcoholic steatohepatitis in mice. Gut 2006;55:415–24.