B. S. Kwon, Section of Clinical Immunology, Allergy and Rheumatology, Department of Medicine, Tulane University Health Sciences Center, New Orleans, LA, USA. E-mail: firstname.lastname@example.org
Autoimmune diseases are characterized by the body's ability to mount immune attacks on self. This results from recognition of self-proteins and leads to organ damage due to increased production of pathogenic inflammatory molecules and autoantibodies. Over the years, several new potential therapeutic targets have been identified in autoimmune diseases, notable among which are members of the tumour necrosis factor (TNF) superfamily. Here, we review the evidence that certain key members of this superfamily can augment/suppress autoimmune diseases.
Autoimmune diseases affect almost every human organ, including the nervous, gastrointestinal and endocrine systems, as well as skin and connective tissue, eyes, blood and blood vessels . There is a strong gender bias among individuals afflicted with autoimmune diseases; it is estimated that of 50 million Americans suffering from various forms of autoimmune diseases, 30 million are women. The current consensus is that autoimmune diseases are induced and orchestrated by autoreactive T (especially CD4+) and B cells that recognize self-proteins in the periphery [2,3]. Through a series of well-co-ordinated physiological events, the autoreactive T cells undergo antigen-specific clonal expansion and release pathogenic immune modulators culminating in tissue necrosis, organ failure and, in most cases, death. Autoantibody production by pathogenic B cells is required for full penetrance of the diseases . Interestingly, a majority of autoimmune diseases manifest late in life (around puberty). Why autoreactive cells remain dormant early in life, and what drives the sudden self-protein recognition process, and subsequent breach of immune tolerance, are still not completely understood [4–6].
The members of the tumour necrosis factor (TNF) superfamily are characterized by distinctive cytoplasmic death domains, and can induce apoptosis and activate receptors. There is no apparent homology between their cytoplasmic tails. The receptors that are activated are involved in gene expression and anti-apoptotic signalling . With only a few exceptions, TNF superfamily members are activation-induced, implying that they control late immune responses. Targeting members of the superfamily in various diseases, including autoimmune diseases, has met with significant success [8,9].
Because the subject matter of autoimmune diseases is vast and cannot be considered in detail here, we will restrict ourselves to an overview of the importance of certain key members of the TNF/TNF receptor (TNFR) superfamilies, such as CD27, CD30, CD40, CD134, CD137, Fas, TNFR1 and TNF-α-related apoptosis-inducing ligand; (TRAIL) in the development/suppression of certain prominent autoimmune diseases.
CD27, a type I disulphide-linked glycoprotein, was identified more than a decade ago on human resting peripheral blood T cells and medullary thymocytes. In both humans and mice, CD27 is expressed on naive and memory-type T cells, antigen-primed B cells and subsets of natural killer (NK) cells . The CD27 ligand, CD70, is expressed transiently and in a stimulation-dependent manner on T, B and dendritic cells (DCs) , whereas it is expressed constitutively on antigen-presenting cells (APCs) in the mouse intestine . CD27 co-stimulation of anti CD3-primed CD4+ T cells promotes cell division, increases BcLxL and promotes interferon (IFN)-γ induction . CD27–CD70 signals are important in the germinal differentiation of B cells into antibody-secreting plasma cells [14–16].
The importance of CD27–CD70 in autoimmune diseases has been underscored by a number of studies. CD27hi plasma B cells were shown to increase in humans afflicted with lupus and the increase was correlated with disease severity . That CD27hi B cells play critical roles in disease severity in patients with systemic lupus erythematosus (SLE) was confirmed by immunosuppressive therapy that resulted in a reduction of CD27hi plasma cells and concomitant disease remission . In addition, soluble CD27 was found to be elevated in the sera of patients with SLE . Furthermore, large numbers of human leucocyte antigen D-related (HLA-DR)hiCD27+ plasmablasts were found in patients with SLE, their numbers correlating with the extent of lupus activity and anti-dsDNA levels . Similarly, CD70 was overexpressed in aged CD4+ T cells in Murphy Roth Large (MRL)/lpr mice . Treatment of Swiss Jackson Laboratory (SJL) mice with anti-CD70 antobodies was found to prevent the development of experimental autoimmune encephalomyelitis (EAE) in a TNF-α-dependent manner, but this effect was independent of impairment of T and B cell effector functions . The mechanisms underlying these various effects are not clear. CD4+ T cells have been observed in synovia in rheumatoid arthritis, and psoriatic arthritis patients have been shown to express high levels of CD70 . Treatment with anti-CD70 antibody led to significant improvement in clinical symptoms, and marked reductions in autoantibody production, inflammation and bone and cartilage destruction  (Table 1, Fig. 1a). In chronic inflammatory disorders, B cells can contribute to tissue damage by producing autoantibodies and presenting antigens to T cells. B cells make important contributions to disease severity in autoimmune diseases such as rheumatoid arthritis (RA) . Thus, CD27+ memory B cells were found to be very abundant in the synovial fluid of patients with juvenile idiopathic arthritis and are believed to prime T cells as a result of their increased expression of CD86 .
Table 1. An overview of tumour necrosis factor (TNF) superfamily members in autoimmune diseases.
Fas+ and FasL+ cells↑ on glial cells, macrophages, lymphocytes in MS brain [135,140]; acinar cells of salivary glands of Sjögren's syndrome , CD4+ and CD8+ T cells of SLE [142,143], RA [136,139], Sjögren's syndrome  and MS  patients
Fas- and FasL-deficient mice: autoimmune diseases↑[125,126]; adenovirus carrying FasL: apoptosis↑, RA↓; anti-Fas mAb: RA↓[117,145–147]; deletion of IFN-γ or IFN-γR in MRL/lpr mice:lupus↓[150,152]
sTNFR1↑ in SLE patients ; TNFR1+ cells↑ in skin lesions of MRL/lpr mice 
CD30 was identified originally in 1982 on tumour cells of Hodgkin's lymphoma . Also called Ki-1, it is a membrane glycoprotein consisting of two chains with molecular weights of 120 and 105 kDa. It is expressed by a subset of activated T cells (both CD4+ and CD8+), NK cells and B cells, and is expressed constitutively in decidual and exocrine pancreatic cells, with maximum expression on CD45RO+ memory T cells . The CD30 ligand (CD30L; CD153) is a 26–40 kDa protein cloned in 1993 and present on a variety of cells, including activated T cells, macrophages, resting B cells, granulocytes, eosinophils and neutrophils . In vitro, CD30 signalling has co-stimulatory effects on lymphoid cells . In vivo, its effects are varied and have been shown to play important roles in inflammatory conditions .
CD30 has been reported to function in regulating autoimmune diseases [32,33]. CD30 signalling protected against autoimmunity by preventing extensive expansion of autoreactive CD8+ effector T cells during secondary encounters with antigen in parenchymal tissues [32,33]. Also, elevated concentrations of the soluble form of CD153 were observed in the sera of RA patients together with increased levels of CD30 and CD153 in biopsies . There is also evidence that expression of CD153 in RA synovia contributes to mast cell activation . Savolainen et al.  and Okamoto et al.  have observed elevated concentrations of the soluble form of CD30 in RA patients, thus underlining the importance of this molecule in the development of RA. Okamoto et al.  have noted further that although CD4+ T cells from peripheral blood and synovial tissue expressed CD30 and produced interleukin (IL)-4 after in vitro stimulation, they underwent CD30-mediated cell death. In an analogous study, Gerli et al.  found that, in addition to IL-4 and IFN-γ, CD30+ T cells produced large amounts of inflammatory IL-10, and they suggested that synovial CD30+ T cells may play a role in the control of RA-induced inflammatory responses. Soluble forms of CD30 were found to be elevated in the sera and cerebrospinal fluids of multiple sclerosis (MS) patients, particularly during remission [38,39]. In addition, soluble forms of CD30 were elevated in patients with systemic lupus erythematosus and Sjögren's syndrome [40,41]. In non-obese diabetic (NOD) mice, expression of both CD30 and CD30L was elevated on peripheral lymph node CD4+ and CD8+ T cells . As a result, treatment of NOD mice with neutralizing anti-CD30L monoclonal antibodies (mAb) prevented the development of diabetes . Taken together, these observations underscore the importance of CD30/CD153 signalling in the development of autoimmune diseases (Table 1, Fig. 1b).
CD40 is the most extensively studied member of the TNF superfamily. First identified on B cells , it is present on a variety of cells including DCs, follicular DCs, monocytes, macrophages, mast cells, fibroblasts, vascular smooth muscle cells and endothelial cells , and as a functional molecule on CD4+ T cells . CD4–CD154 interactions generate one of the most effective APC-activating signals. Signalling via dendritic cell CD40 up-regulates expression of CD80 and CD86 and induces IL-12 secretion [46–48], and signalling via CD40 activates nuclear factor (NF)-κB [49,50] and rescues B cell receptor (BCR)-induced cell death . Moreover, studies using CD40−/− mice have shown that the CD40–CD154 pathway is central to germinal centre formation and immunoglobulin (Ig) isotype-switching .
Besides its importance in regulating/suppressing various diseases, the CD40 pathway plays critical roles in autoimmune diseases . Autoimmune thyroiditis, or Graves' disease, is due to increased infiltration of lymphocytes into the thyroid where they recognize the thyroid stimulating hormone receptor; this leads to autoantibody production, tissue necrosis and loss of thyroid function. The importance of CD40 signalling in Graves' disease was recognized after the discovery that CD40 is present on thyroid epithelial cells , where it interacts with CD40L (CD154)-expressing autoreactive T cells. In agreement with this observation, blockade of the CD40–CD40L interaction with anti-CD40L antibodies has been shown to prevent experimental thyroiditis .
Type 1 diabetes, or insulin-dependent diabetes, is caused by autoreactive T cells that recognize antigens such as insulin and glutamic acid decarboxylase on B cells in the islets of Langerhans. B cells also play important roles in disease pathogenesis, as revealed by B cell-deficient NOD mice  and treatment of NOD mice with CD40L antibodies . As the CD40 signal is critical for antibody production and Ig class-switching, depletion of CD40+ B cells, or deletion of endogenous B cells, lowers autoantibody production in these mice and decreases disease severity. In addition to CD40+ B, CD40+ T cells are important in the induction of diabetes in NOD mice .
The importance of CD40–CD40L has also been underscored in collagen-induced arthritis (CIA). Treatment of mice with collagen type II and anti-CD40L antibodies blocked joint inflammation, serum antibody titres to collagen, synovial infiltrates and erosion of cartilage and bone . Also, when treated with anti-CD40L antibodies, lupus-prone mice showed reduced glomerulonephritis . Similarly, in an open-label study in SLE patients treated with anti-CD40L, humanized mAb exhibited disease alleviation, including reduced anti-ds-DNA titres [61,62]. Blockade of CD40–CD40L interaction by anti-CD40L antibodies reduced the incidence and severity of T helper type 1 (Th1)-mediated experimental autoimmune uveoretinitis (EAU) in susceptible B10.RIII mice immunized with autoantigen interphotoreceptor retinoid binding protein (IRBP) in complete Freund's adjuvant (CFA) . Further analysis revealed that in anti-CD40L antibody-treated mice innate responses to autoantigen IRBP were reduced significantly, but no Th2 dominance was observed . The alleviation of EAE and MS by anti-CD40L therapy  further signifies the importance of CD40–CD40L axis in autoimmune diseases (Table 1, Fig. 1c).
CD134 (OX40), an inducible T cell co-stimulatory molecule, is one of the most extensively studied members of the TNF superfamily. OX40 expression is activation-induced and, once expressed, OX40 binds OX40L (CD134L) present on a variety of cells [65–67]. OX40 signalling promotes T cell activation, induction of cell survival genes and production of cytokines . OX40 signals play crucial roles in autoimmune and viral diseases, cancer and transplantation .
CD134+CD4+ T cells are elevated in SLE patients . Studies on collagen type II (CII)-induced arthritis in susceptible DBA/1 mice revealed that administration of anti-OX40L antibodies reduced the associated pathological lesions significantly; it did not inhibit the development of CII-reactive T cells, but suppressed IFN-γ and anti-CII IgG2a production . Similarly, the synovial fluid of patients with active RA contained increased numbers of OX40+ T cells . An important role of OX40 signalling in the progression of CII-induced RA has been demonstrated in studies with IL-1α/β−/−, mice where a reduced incidence of CII-induced RA was correlated with decreased expression of OX40 on T cells .
Perivascular infiltrates of the central nervous system (CNS) of mice treated with myelin oligodendrocyte glycoprotein (MOG)35–55 peptide, and of patients with multiple sclerosis, contain a large number of CD134+ cells . That CD134 signalling is important in the resolution of EAE was confirmed by showing that induction of EAE in CD134−/− mice yielded in clinical evidence of reduced severity, and decreased inflammatory infiltrates markedly within the CNS . Moreover, the resistance to EAE of CD134−/− mice was found to be associated with a marked reduction in the number of pathogenic IFN-γ-producing T cells infiltrating the CNS . Conversely, triggering OX40 signalling exacerbated EAE [74,75]. In accordance, blockade of CD134–CD134L interaction by soluble CD134 at the onset of disease reduced disease symptoms . Increased OX40 expression on the CD4+ T cells of patients suffering from myasthenia gravis, a protoypic antibody-mediated organ-specific autoimmune disease, has also been reported . Pakala et al.  have demonstrated that administration of blocking anti-CD134L mAb to NOD mice had reduced glucose levels and islet infiltrating leucocytes and reduced the incidence of diabetes significantly. The significance of CD134–CD134L in autoimmune diseases is highlighted in Table 1 and Fig. 1d.
CD137 (4-1BB), an important T cell co-stimulatory molecule , exists as both a 30-kDa monomer and 55-kDa homodimer . Its expression is activation-induced [79,80] and it is expressed primarily on activated CD4+ and CD8+ T cells  and on activated NK and NK T cells . In contrast, 4-1BB is expressed constitutively on primary human monocytes, DCs, blood vessel endothelial cells and human follicular DCs, as well as CD4+CD25+ regulatory T cells (Tregs) [82–86]. In vitro and in vivo studies indicate that signalling via 4-1BB preferentially activates CD8+ T over CD4+ T cells . Soluble forms of CD137 (sCD137) and sCD137L have been observed in sera of RA and MS patients, where levels of sCD137 and sCD137L correlated with disease severity [88–91]. The precise role of sCD137 and sCD137L in autoimmune diseases is, however, not understood completely.
The first evidence that in vivo administration of 4-1BB antibodies is an effective treatment for autoimmune diseases was provided by Sun et al. . These authors observed that treatment of lupus-prone lpr mice with agonistic anti-4-1BB antibodies increased induction of IFN-γ and affected CD4+ T and B cells number and function, leading to reduced autoantibody production and significant reversal of the associated clinical symptoms . In an analogous study, Foell et al.  demonstrated that treatment of New Zealand black (NZB) × NZ white (NZW) F1 mice with agonistic anti-4-1BB antibodies reversed acute lupus disease in these mice by suppressing B cell function, but without affecting CD4+ T cell function. Although the two studies [92,93] point to a common mechanism of B cell impairment, due perhaps to increased IFN-γ production, the difference between them in the effect on CD4+ T cells may have been due to the use of different strains. That 4-1BB signalling plays important roles in the regulation of lupus disease was confirmed by using lpr mice deficient in endogenous 4-1BB. The lpr/4-1BB−/− mice displayed early onset of clinical symptoms, increased autoantibody production, skin lesions, increases lacrimal gland dysfunction and early mortality, compared to lpr/4-1BB+/+ mice [94,95].
In experimental autoimmune encephalomyelitis (EAE), treatment of C57BL/6 mice with MOG35–55 peptide (an EAE-inducing agent) and anti-41BB antibodies reduced symptoms without affecting total CD4+ T cell numbers, but it increased the probability that the CD4+ T cells underwent subsequent activation-induced cell death . Interestingly, adoptive transfer of T cells obtained from mice treated previously with anti-4-1BB failed to prevent EAE even after boosting their function by administering anti-4-1BB, suggesting that anti-4-1BB treatment is only effective during the induction phase of autoreactive T cell immune responses .
Seo et al.  made the interesting observation that in collagen type II-treated DBA/1 mice, anti-4-1BB antibody therapy resulted in an increase of a novel subset of CD8+ T cells co-expressing CD11c. The expansion of the CD11c+ CD8+ T cells correlated with amelioration of the clinical symptoms of RA . This was confirmed by observing reversal of the clinical lesions in collagen II-treated DBA/1 mice upon adoptive transfer of CD11c+CD8+ T cells from arthritic mice exposed previously to anti-4-1BB . The anti-4-1BB-expanded CD11c+CD8+ T cells expressed high levels of IFN-γ which, in turn, induced macrophages and DCs to up-regulate IDO. The IDO+ cells then provoked deletion of the pathogenic CD4+ T cells by interacting with them and depleting tryptophan levels . Increased levels of CD11c+CD8+ T cells were also found in the blood of patients with RA . In addition, the increases in levels of circulating soluble 4-1BB and 4-1BBL in patients with RA were correlated with disease severity .
Increased IFN-γ is required to suppress the development of experimental autoimmune uveoretinits (EAU) , and neutralization of IFN-γ and IFN-γ deficiency were shown to lead to severe EAU [100,101]. Choi et al.  have exploited the finding that increased production of IFN-γ is the hallmark of in vivo anti-4-1BB administration  to treat EAU: treatment of C57BL/6 mice with IRBP peptide (an EAU-inducing agent) and anti-4-1BB led to expansion of IFN-γ+ CD11c+CD8+ T cells and indoleamine 2,3-dioxygenase (IDO)+ DCs and these, in combination, led to deletion of autoreactive CD4+ T cells . Taken together, these various findings indicate that targeting CD137 is an attractive strategy for preventing the symptoms associated with various autoimmune diseases (Table 1, Fig. 1e).
The Fas (Apo-1/CD95) and Fas ligand (FasL) are one of the extensively studied TNF superfamily members. The Fas was described originally as a cell surface molecule capable of inducing apoptosis when stimulated by Fas ligand (FasL) or agonistic anti-Fas mAb [104–106]. However, there are reports that ligation of Fas on freshly isolated T cells co-stimulates cellular activation and proliferation , an attribute that is somewhat conflicting with its proposed role in apoptosis. The Fas is expressed in most tissues  and is up-regulated further during inflammation [109,110]. At the cellular level, Fas expression is low on freshly isolated lymphocytes but is up-regulated on activated T cells . Also, proportions of Fas-positive cells in peripheral T and B cells have been reported to increase in humans with advancing age . Conversely, the expression of FasL is governed tightly and is expressed, among others, by activated T cells .
The Fas and FasL have been shown to play critical roles in various diseases including fulminant hepatitis [114,115], graft-versus-host disease  and tissue-specific autoimmune disease . Fas–FasL interactions also are important in T cell-mediated cytotoxicity , immune privilege tissues [119–121], activation-induced cell death (AICD) [122,123] and transplant tolerance . The Fas- and FasL-deficient mice develop autoimmune diseases and lymphadenopathy due to the inability to delete the autoreactive T and B lymphocytes [125,126].
The importance of the Fas–FasL pathway has been underscored in a number of autoimmune diseases, including lupus , SLE , autoimmune lymphoproliferative syndrome (ALPS) [128,129], Canale–Smith syndrome , type 2 autoimmune hepatitis , Hashimoto's syndrome , insulin-dependent diabetes mellitus [133,134], MS , Sjögren's syndrome , myasthenia gravis , EAE  and RA . Increased Fas+ and FasL+ cells were observed on the glial cells, macrophages and infiltrating lymphocytes in the white matter of MS brains [135,140]. Also, acinar cells of salivary glands of Sjögren's syndrome patients show high expression of Fas and FasL and were shown to die by apoptosis . While patients with Hashimoto's disease showed decreased sFas, increased levels were noted in Graves' thyroiditis and SLE patients [142,143]. Increased Fas expression was also noted on CD4+ and CD8+ T cells of human SLE patients , RA patients, primary Sjögren's syndrome  and MS patients .
Given its importance in autoimmune diseases, targeting of the Fas–FasL pathway has been attempted by a number of investigators. It has been demonstrated that in RA high levels of Fas have been found expressed on activated synovial cells and infiltrating leucocytes in the inflamed joints . In contrast, FasL expression was found to be extremely low in arthritic joints and as a result most synovial cells survive despite high levels of Fas . To correct this, Zhang et al.  have developed a strategy wherein arthritic DBA/1 mice were treated with an adenovirus carrying FasL resulting in increased apoptosis and alleviation of RA symptoms. These authors have also found that reversal of RA in FasL-injected mice was associated with reduced production of IFN-γ by collagen-specific T cells . Using a severe combined immune deficient (SCID) mouse model, Odani-Kawabata et al. have demonstrated that treatment with anti-human Fas mouse/human chimeric monoclonal IgM antibody ARG098 suppressed synovial hyperplasia by up-regulating apoptosis and prevented cartilage destruction . Similarly, administration of humanized anti-human Fas mAb (R-125224) to SCID mice suppressed osteloclastogenesis via induction of apoptosis in CD4+ T cells . In line with these observations, Nishimura-Morita et al. have also observed that administration of anti-Fas mAb clone RK-8 but not Jo2 increased apoptosis and arrested the development of autoimmune diseases, including arthritis [117,147].
The role of Fas and FasL is exemplified further in studies dealing with MRL/lpr and MRL-gld/gld mouse models in which lack of Fas/FasL expression leads to reduced apoptosis, abnormal lymphoproliferation and development of autoimmune diseases, including lupus and Sjögren's syndrome . When MRL-gld/gld strain mice were given anti-Fas mAb (clone RK8) to correct the defective apoptosis, it was observed that RK8-treated mice had reduced splenomegaly and lymphadenopathy . These authors have also observed that RK8-treated MRL-gld/gld mice had reduced salivary gland damage and reduced incidence of Sjögren's syndrome . As increased IFN-γ has been implicated in lupus severity and as IL-12 drives IFN-γ induction , MRL-Faslpr mice with IFN-γ or IFN-γR deletion have a reduced incidence of lupus nephritis [150,151]. Collectively, these data demonstrate the importance of Fas-mediated apoptosis in the development of autoimmune diseases and highlight further the beneficial effects of anti-Fas mAbs in disease alleviation (Table 1, Fig. 1f).
TNF-α, a pleiotropic cytokine with both beneficial and lethal effects, is one of the extensively studied cytokines . The significance of TNF-α in the pathogenesis has been well proven by clinical efficacy of its blockade in a number of diseases including autoimmune diseases [152,153]. TNF-α exerts its biological functions following interaction with its cognate membrane receptor p55TNFR (TNFR1) and p75TNFR (TNFR2) . TNFR1/2 are expressed in all cell types and activate both cellular responses [155–157] and mediate anti-apoptotic and inflammatory responses through the recruitment of TNF receptor-associated factor (TRAF) 2 and receptor-associated protein (RIP)-1, which are critical in the activation of NF-kB, c-Jun NH2 terminal kinase (JNK) and mitogen-activated protein kinase (MAPK) . Although the two receptors have similar extracellular sequences that are rich in cysteine, the hallmark of the TNF superfamily, TNFR1 alone possesses a cytoplasmic death domain, an 80 amino acid sequence that rapidly engages the apoptotic signalling pathway of the cells .
Because dysregulated TNF-α secretion has been implicated in several autoimmune diseases, blocking TNF-α production has therefore been shown to have beneficial effects against various autoimmune diseases . However, the timing of TNF-α therapy is critical for its therapeutic outcome. For example, administration of dimerized TNFR1 (TNFR1-IgG) to block TNF-α-TNFR interaction after the onset of experimental autoimmune neuritis (EAN) failed to alter the course of disease . However, TNFR1-IgG therapy when administered at the onset of disease delayed EAN and was accompanied by inhibition of blood–nerve barrier permeability and inflammation . Interestingly, blockade of TNF-α–TNFR interaction by specific fusion proteins during CIA in DBA/1 strain versus endogenous TNFR1 gene deletion yielded mixed results. Treatment of CIA mice with TNFR1–IgG1 fusion protein to neutralize systemic TNF-α before the onset of clinical disease showed inhibition of clinical disease in these mice . In contrast, induction of CIA in TNFR1−/− mice on a DBA/1 background showed an initial milder form of disease but, with time, the severity of joint disease was comparable among wild-type and TNFR1−/− mice . The importance of TNFR1 gene deletion and increased severity of CIA was suggested further by the observation that mainly TNFR1 gene deletion caused development and exacerbation of inflammation [162,163]. In contrast to the effect of TNFR1 gene deletion, which showed severe arthritis [162,163], suppression of MOG-induced EAE is less severe in TNFR1−/− mice . Also, TNFR1−/− mice who have defective IFN-γ-dependent nitric oxide (NO) production from macrophages and significantly reduced CD113+ and CD4+ cells within the target organ are resistant to the induction of EAU [165,166]. In accordance, blockade of TNF-TNFR by soluble TNFR1–Ig fusion protein was shown to inhibit clinical symptoms associated with EAE [167,168]. To understand further the relative roles of TNFRI and TNFRII in MOG-induced EAE, Suvannavejh et al. observed that disease was reduced significantly in TNFR1−/−/2−/− double knock-out and TNFR1−/− but not in TNFR2−/− mice. It was also noted that amelioration of EAE in TNFR1−/− mice was associated with significantly increased IL-5 levels and decreased proliferation of MOG-specific delayed-type hypersensitivity (DTH) responses .
The significance TNF-α-TNFR1 interaction is also underscored in SLE. Zhu et al. have observed that SLE patients have increased levels of TNFRI, TNFRII and TRAF2 and decreased levels of RIP . However, no correlation was found among soluble TNFR1/2 and serum TNF-α levels or their RNA expression . It is important to note that lupus-prone NZB/F1 mice deficient in both TNFR1 and TNFR2 showed accelerated course of disease . Conversely, NZB/F1 mice deficient in TNFR1 or TNFR2 had a comparable phenotype . TNFR1, but not TNFR2, was expressed dominantly in skin lesions of MRL/lpr mice . Taken together, these data indicate that TNF-α is a critical parameter of several autoimmune diseases and its blockade ameliorates as well as exacerbates autoimmune disease pathology (Table 1, Fig. 1g).
The TNF-α-related apoptosis-inducing ligand (TRAIL; Apo2L) is a type II membrane protein and plays an important role in immune regulation [173,174]. In humans, TRAIL expression is inducible on IFN-γ activated fibroblasts , peripheral blood monocytes , monocyte-derived DCs , immature NK cells , T cells [179–181] and NK T cells . In the case of mice, TRAIL is expressed by activated NK  and liver NK cells [184,185]. TRAIL binds to two death receptors: death receptor (DR) 4 and DR5 and two decoy receptors: decoy receptor (DcR1) 1 and DcR2, and following binding to its death receptors DR4 and DR5 TRAIL can induce apoptosis, as they contain intracellular death domains [186–188]. Incidentally, the binding of TRAIL to DR5 can also activate the transcription factor NF-κB, which is known to control cell proliferation . Thus, depending on the cellular system, TRAIL is capable of initiating apoptosis or cell survival.
Importance of the TRAIL pathway in autoimmune diseases is revealed by a number of studies. Chronic in vivo blockade of TRAIL–DR5 interaction by soluble DR5 has been shown to induce hyperproliferation of synovial cells and arthritogenic lymphocytes, resulting in increased production of proinflammatory cytokines and autoantibodies leading to exacerbation of arthritis . That the TRAIL pathway plays critical roles in arthritis is also corroborated by amelioration of disease by intra-articular transfer of the TRAIL gene [190,191] and by intraarticular transfer of recombinant TRAIL . Further proof that the TRAIL signal is important in arthritis pathogenesis came from gene knock-out studies which showed that TRAIL deficiency increases the susceptibility of mice to autoimmune arthritis . Interestingly, Liu et al. have reported that adoptive transfer of TRAIL-transfected DCs pulsed with collagen into susceptible mice suppressed disease pathology . DR5 expression is noted on synovial fibroblasts, and ligation of DR5 on these cells by anti-DR5 antibodies induces apoptosis . Administration of soluble TRAIL receptor to block TRAIL–DR interaction exacerbated MOG-induced EAE . In these mice the degree of apoptosis of inflammatory cells in the CNS was not affected by sTRAIL treatment, but rather involved significant increases in MOG-specific Th1/Th2 responses .
The importance of the TRAIL–DR interaction is also exemplified in autoimmune diabetes. Lamhamedi-Cherradi et al. have demonstrated that treatment of NOD mice with soluble TRAIL enhanced autoimmune inflammation significantly in pancreatic islets and salivary glands, increased glutamic acid decarboxylase 65 (GAD65)-specific immune responses and, in turn, diabetes . These authors also observed that in a streptozoticin-induced diabetes model, treatment of TRAIL−/− mice with soluble TRAIL significantly enhanced the incidence and the degree of diabetes , suggesting the importance TRAIL signalling in autoimmune diabetes (Table 1, Fig. 1h).
In summary, the last few years have seen rapid growth in the number of known members of the TNF/TNFR superfamily. Exploitation of the various unique biological functions of these proteins for therapeutic purposes has shown promise. Further research in this area will undoubtedly point the way to effective therapeutic interventions in autoimmunity.
This study was supported by grants from the National Cancer Center, Korea (NCC-0890830-2 and NCC-0810720-2), the Korean Science and Engineering Foundation (Stem Cell-M10641000040 and Discovery of Global New Drug-M10870060009), the Korean Research Foundation (KRF-2005-084-E00001) and Korea Health 21 R&D (A050260).
The authors have no conflicts of interest to declare.