CD28–B7 family members

CD28 is constitutively expressed on T cells and is upregulated upon T-cell activation. CD28 engagement by its ligands CD80 and CD86 (B7-1 and B7-2) amplifies signals through the TCR and stabilises the immunological synapse, thus lowering the threshold for naive T-cell activation, and facilitating entry into the cell cycle and secretion of IL-2 (Refs Reference Mondino and Mueller53, Reference Vincenti and Luggen54). Engagement of the TCR in the absence of CD28 costimulation may lead to cell death or induction of anergy (Refs Reference Lohr55). This maintains tolerance to self-antigens that are usually chronically encountered in the absence of costimulatory signals. The CD28-homologous molecule CTLA4 has much higher affinity for CD80 and CD86 and serves as a negative regulator of T-cell function. CTLA4, unlike CD28, signals through CD80 and CD86 to induce several regulatory molecules including indoleamine 2,3-dioxygenase (IDO), an enzyme that helps maintain tolerance through alterations in tryptophan metabolism (Refs Reference Mellor56, Reference Mellor57). A recently identified polymorphism of CTLA4 is associated with autoimmunity in humans (Ref. Reference Ueda58).

Blockade of CD28 is achieved by administration of the extracellular domain of CTLA4 fused to immunoglobulin heavy chain (CTLA4Ig, Abatacept) (Ref. Reference Vincenti and Luggen54). CTLA4Ig acts both as a competitive antagonist of B7–CD28 interactions and as an inducer of regulatory pathways in APCs. CTLA4Ig may also alter expression of adhesion molecules and chemokine receptors, resulting in inhibition of traffic of inflammatory cells to target organs (Ref. Reference Mirenda59), and prevents the upregulation of ICOS (Ref. Reference Akkerman36). However, under some circumstances, administration of CTLA4Ig might be detrimental. For example, although CTLA4Ig can induce T-cell deletion and anergy, it may prevent tolerance induction in circumstances where signalling through CTLA4 is necessary to establish active tolerance (Ref. Reference Davidson60).

CTLA4Ig and its higher-affinity mutant LEA29Y (Belatacept) have been successfully used in patients with psoriasis, rheumatoid arthritis and transplantation (Refs Reference Vincenti and Luggen54, Reference Moreland61). In mice, CTLA4Ig prevents the onset of SLE (Ref. Reference Finck, Linsley and Wofsy62) but does not reverse active inflammation (Ref. Reference Daikh63). For this reason, combination therapies have been tested. Administration of CTLA4Ig with concomitant CD40 or CD40L blockade to SLE-prone mice is a highly synergistic regimen that prevents the onset of SLE for many months after a short course of treatment. This is due to long-lasting blockade of T-cell help for autoreactive B cells (Ref. Reference Wang64). Dual treatment with a single dose of the alkylating agent cyclophosphamide and continuous CTLA4Ig is highly synergistic for the treatment of active SLE nephritis in mice and induces complete remission of histological changes in the kidney. The mechanism for this effect is not completely understood, but includes depletion of activated T and B cells, together with altered renal chemokine and chemokine receptor expression, and consequent dampening of the inflammatory response to glomerular immune-complex deposition (Ref. Reference Schiffer65). Clinical trials of Abatacept in active SLE and in combination with cyclophosphamide in SLE nephritis have been initiated.

B7–CD28 interactions are essential for activation of naive T cells, but have a lesser role in regulating effector responses and responses outside secondary lymphoid organs. Expression of other B7 family members allows regulation of effector T cells at peripheral sites (Ref. Reference Greenwald, Latchman and Sharpe66) (Fig. 2). ICOS is a CD28-like molecule that is expressed on activated and memory T cells and preferentially induces IL-4 and IL-10 production (Ref. Reference Hutloff67). ICOS is highly expressed on IL-10-expressing effector cells (Ref. Reference Lohning68) and on IL-21-secreting CXCR5-positive follicular T-helper (Th) cells located in the apical zone of germinal centres where somatic mutation occurs (Ref. Reference Rasheed69). Ligation of ICOS results in upregulation of CD40L on T cells, which costimulates immunoglobulin synthesis and differentiation of B cells to memory and plasma cells (Ref. Reference McAdam70). Dramatic defects of the humoral immune response occur in ICOS-deficient mice and humans as well as defects in IL-10 and IL-17 production (Refs Reference Dong71, Reference Shilling, Bandukwala and Sperling72). Antagonism of ICOS during the effector phase of Th2-mediated disease attenuates inflammation in murine models (Ref. Reference Shilling, Bandukwala and Sperling72). The ICOS ligand B7RP-1 is expressed on B cells and macrophages and in inflamed tissues. ICOS expression is increased on T cells from SLE-prone mouse strains, particularly mice that bear a Yaa mutation (Ref. Reference Akkerman36) and on T cells in SLE patients (Ref. Reference Hutloff73). The pathogenic role of ICOS has recently been highlighted by the development of B-cell autoimmunity in mice deficient in roquin, a repressor of ICOS expression (Ref. Reference Vinuesa74), and by studies showing that blockade of ICOS ligand prevents and treats disease in NZB/W SLE-prone mice (Ref. Reference Iwai75). Clinical trials of ICOS blockade in humans with SLE are now under way.

Figure 2.

Members of the CD28–B7 family mediate both costimulation and coinhibition. Two signals are required for T-cell activation. Signal 1 is mediated through interaction of the T-cell receptor (TCR) with MHC/peptide complex shown in blue. Signal 2 is mediated by non-antigen-specific interactions such as that of B7 with CD28. CD28-like receptors are shown on the surface of the T cell and B7-like ligands (blue) are shown on the surface of the antigen-presenting cell (APC). Receptors shown in green transduce a positive (costimulatory) signal to the T cell. Receptors shown in red transduce a negative (coinhibitory) signal to the T cell. Not all the receptors have been identified (indicated by ?). Abbreviations: BTLA, B and T lymphocyte attenuator; HVEM, herpesvirus entry mediator; ICOS, inducible costimulator; MHC, major histocompatibility complex.

The receptor programmed death 1 (PD-1), which is expressed on both B and T cells, binds to two ligands: PDL1, which is widely expressed on many tissues and PDL2, which is expressed mainly on activated monocytes. Like CTLA4, PD-1 transduces a negative signal into the cell and helps to maintain peripheral tolerance (Ref. Reference Sharpe76). Studies in mouse models of multiple sclerosis and diabetes have shown that this effect is mediated via the interaction of PD-1 with PDL1 but not with PDL2 (Refs Reference Ansari77, Reference Fife78). PD-1-deficient mice develop spontaneous strain-specific autoimmunity including SLE (Ref. Reference Okazaki, Iwai and Honjo79) and recently a PDCD1 polymorphism associated with human SLE has been identified (Ref. Reference Prokunina80).

Several other members of the CD28–B7 family have been characterised. B and T lymphocyte attenuator (BTLA), like CTLA4 and PD-1, is a negative regulator of T-cell responses and its deficiency results in increased susceptibility to induced autoimmunity. BTLA is unusual because it binds herpesvirus entry mediator (HVEM), a member of the tumour necrosis factor receptor family (Refs Reference Watanabe81, Reference Sedy82). B7-H3 and B7-H4 bind to unidentified receptors (Ref. Reference Greenwald, Freeman and Sharpe83). Little is currently known about the role of these molecules in autoimmune disease pathogenesis; however, these molecules appear to have their effects only at low antigen doses.

The TNF–TNFR family

Multiple members of this family mediate costimulation of T cells and B cells, with 4-1BB, CD40L and B-cell activating factor (BAFF) being most relevant to SLE therapies.

4-1BB

The engagement of the T-cell costimulatory receptor 4-1BB (CD137) paradoxically prevents germinal centre formation in a T-cell-dependent fashion. In SLE-prone mice, three doses of an agonistic anti-4-1BB antibody confers prolonged inhibitory effects on autoantibody production starting 1–2 weeks after antibody administration (Ref. Reference Foell84). In the chronic graft-versus-host (CGVH) model of SLE, administration of the anti-4-1BB antibody resulted in rapid depletion of donor T cells by activation-induced death (Ref. Reference Kim85). By contrast, in the NZB/W model, CD4⁺ T cells were not depleted, but cytokine secretion by these cells was markedly diminished (Ref. Reference Foell84). Transfer of either activated T cells or APCs from untreated mice overcame the T-cell anergy and precipitated disease. Another hypothesis for the activity of anti-4-1BB is that T-regulatory cells (T regs) are generated (Refs Reference Foell84, Reference Choi86). Although a regulatory population of CD8⁺ cells has been described, it is clear that the antibody still mediates tolerance in CD8-deficient mice (Ref. Reference Foell84). Finally, production of IFN-γ has been shown to enhance tolerance in the Murphy Roths Large (MRL)/lpr model (Ref. Reference Sun87) but not in the CGVH model, suggesting that IFN-γ is not essential for tolerance induction (Ref. Reference Kim85). This system is interesting because it shows that long-lasting restoration of tolerance can be achieved in mice with spontaneous SLE in which B and T cells are continuously being activated. Further studies in this system may reveal the mechanisms for maintenance, loss and restoration of tolerance in autoimmunity.

CD40L

CD40L is an important costimulatory molecule, whose expression on activated T cells is regulated by CD28 and by ICOS (Ref. Reference Greenwald, Freeman and Sharpe83). Engagement of CD40 on APCs by CD40L synergises with TLR signalling and results in increased expression of MHC and B7 molecules and secretion of IL-12, thus enhancing their antigen-presenting function (Ref. Reference Salomon and Bluestone88). Engagement of CD40 on B cells delivers a survival signal and is essential for formation of the germinal centre and for reactivation of memory B cells (Refs Reference Castigli89, Reference Grewal and Flavell90). Absence of CD40 or CD40L in humans results in a profound humoral immunodeficiency (Ref. Reference Vogel and Noelle91). Antibodies to CD40L prevent onset and delay progression of SLE in mouse models (Ref. Reference Wang92), but two Phase II human clinical trials failed because of the unexpected development of thrombotic events in some of the treated patients (Ref. Reference Sidiropoulos and Boumpas93). Although one of the anti-CD40L trials failed to show clinical efficacy at the dose used (Ref. Reference Kalunian94), some of the SLE patients treated with a different anti-CD40L antibody showed clinical improvement (Ref. Reference Boumpas95). Mechanistic studies in these patients revealed that despite unaltered total B-cell numbers, treatment diminished the number of autoantibody-producing B cells (Refs Reference Boumpas95, Reference Huang96, Reference Grammer97). It has been suggested that the effect of anti-CD40L is due to specific B-cell unresponsiveness rather than T-cell anergy, because tolerance cannot be broken in mice by transfer of pathogenic T cells (Ref. Reference Mohan98). However, in humans with SLE, B cells producing anti-DNA antibody reappeared after cessation of anti-CD40L, indicating that tolerance had not been established (Ref. Reference Huang96). A striking therapeutic effect can be observed in mouse models when anti-CD40L is combined with CTLA4Ig. This combination, given for two weeks to SLE-prone mice, induces a prolonged effect on disease when given early in the course (Ref. Reference Daikh63), and, more importantly, maintains B-cell tolerance to autoantigens without preventing immune responses to exogenous antigens (Ref. Reference Wang64). However, this combination is not effective in the setting of active nephritis and requires the addition of a single dose of cyclophosphamide, which kills activated effector cells (Ref. Reference Schiffer65). Although anti-CD40L antibodies are not safe for use in humans, antibodies to CD40 have been recently developed for human use and appear to be safe for the treatment of B-cell tumours (Ref. Reference Drachman99). This may allow for a renewed attempt at CD40 blockade in human SLE.

BAFF

The TNF family member BAFF is a B-cell survival factor expressed by many different cell types that binds to three different BAFF receptors (TACI, BCMA and BAFF-R), which are variably expressed on the B-cell surface during development (Refs Reference Sutherland, Mackay and Mackay100, Reference Ramanujam and Davidson101). BAFF-R and TACI are the predominant receptors on most mature B cells, except for plasma cells, which express BCMA (Refs Reference Mackay102, Reference O'Connor103). APRIL, a closely associated ligand of BAFF, binds to TACI and BCMA and to proteoglycans such as CD138 on the plasma cell surface; CD138 serves as a coreceptor to enhance binding of APRIL to BCMA (Ref. Reference Ingold104). Maintenance of peritoneal cavity B1 B cells is BAFF independent (Refs Reference Schiemann105, Reference Ramanujam106) and germinal centre B cells are only partially dependent on BAFF. Thus BAFF-deficient mice can still develop autoimmunity, although they do so in a much delayed fashion (Refs Reference Vora107, Reference Jacob108). BAFF transgenic mice on a nonautoimmune genetic background develop a form of SLE that is due to production of autoantibodies by marginal zone and B1 cells and this depends on B-cell signalling through TLRs but does not require T cells (Ref. Reference Groom109). BAFF levels may control B-cell selection because competition for BAFF limits selection of autoreactive B cells into the mature compartment. Depletion of B cells results in excess production of BAFF, which may then decrease the stringency for B-cell selection (Refs Reference Lesley110, Reference Thien111). It remains unclear whether the increased sensitivity of autoreactive B cells to survival signals mediated through BAFF and APRIL is applicable to memory cells, plasmablasts or plasma cells (Ref. Reference Martin and Chan112). Patients with SLE and other autoimmune diseases have increased serum levels of BAFF (Refs Reference Zhang113, Reference Cheema114) and in murine SLE nephritis, BAFF is also found locally at the site of inflammation where it is made by activated macrophages and dendritic cells (Ref. Reference Schiffer207). Activated macrophages express BAFF receptors and thus may mediate their own survival in an autocrine fashion (Ref. Reference Chang115).

BAFF blockade has been achieved either with a monoclonal anti-BAFF antibody or with soluble BAFF receptors. BAFF-R-Ig selectively blocks BAFF, whereas TACI-Ig blocks both BAFF and APRIL, and therefore is able to deplete plasma cells that are dependent on APRIL–BCMA interactions. Despite these differences, both selective and nonselective BAFF blockers prevent SLE in mouse models (Ref. Reference Ramanujam106), indicating that BAFF is more important than APRIL in SLE pathogenesis. BAFF blockade depletes B cells by 50% but does not significantly decrease the titres or affinity of autoantibodies, nor does it prevent T-cell activation. By significantly decreasing B-cell numbers and spleen size, BAFF blockade results in a decrease in the total number of activated T cells and monocytes, thus decreasing the total inflammatory load and the capacity of inflammatory cells to migrate to the kidney. The combination of BAFF blockade and CTLA4Ig, but not either drug alone, can induce remission of SLE in NZB/W mice by mechanisms that are not fully delineated (Refs Reference Ramanujam106, Reference Ramanujam116). A human clinical trial of the anti-BAFF antibody Belimumab in moderately active SLE showed only modest clinical effects, with failure to meet initial efficacy end points. However, long-term follow up and post-hoc analysis of patients that were serologically active at initiation has shown a greater sustained improvement in treated patients than in placebo-treated controls (Ref. Reference Ginzler117). More extensive trials of this agent are in progress. TACI-Ig is also entering clinical trials in SLE.

B-cell modulation and depletion

SLE is characterised by the production by B cells of high-affinity pathogenic autoantibodies that mediate tissue injury. In addition, B cells have other important functions that can contribute to disease pathogenesis. Activated autoreactive B cells that take up autoantigen through their antigen receptor can function as APCs and serve to diversify the epitopes of self-antigen presented to autoreactive T cells. B cells also produce cytokines involved both in lymphoid regulation and in inflammatory processes (Ref. Reference Grammer and Lipsky118). Because autoreactive B cells play a role in both the inductive and the effector arms of autoimmune disease, there has been much interest in B-cell depletion or modulation as a treatment strategy for autoimmune disease. One such strategy has been the use of anti-CD20 antibodies.

CD20, a transmembrane protein with four subunits, is upregulated on most B cells past the pre-B-cell stage but is absent on plasma cells (Refs Reference Riley and Sliwkowski119, Reference Uchida120). Anti-CD20 monoclonal antibodies successfully deplete peripheral human B cells for periods ranging from three months to more than one year through mechanisms involving Fc-receptor dependent killing, complement-dependent killing and antibody-dependent apoptosis (Refs Reference Clynes121, Reference Uchida122). Studies in mice show that the kinetics of B-cell depletion varies between different B-cell compartments with 90% follicular B cells in the spleen being depleted within two days of anti-CD20 monoclonal antibody administration but less effect on marginal zone B cells. The peritoneal cavity (particularly B1) and the germinal centre B-cell compartments demonstrate the greatest relative resistance to anti-CD20 monoclonal antibody treatment although marginal zone, germinal centre and peritoneal B1 B cells express comparable levels of CD20 and appear to have access to the antibody. These differential sensitivities depend partly on the number and localisation of mononuclear phagocytes within tissues that mediate antibody-dependent cell-mediated cytotoxicity (Refs Reference Uchida122, Reference Gong123, Reference Hamaguchi124). Hence, location and microenvironment influence the extent of B-cell depletion (Ref. Reference Gong123). Because B-cell depletion causes a compensatory increase in BAFF levels, blockade of BAFF synergises with anti-CD20 monoclonal antibody to enhance B-cell depletion (Ref. Reference Gong123). A new approach that may avoid the need for two drugs is the use of a depleting anti-BAFF-R antibody. This antibody blocks the effects of BAFF and also depletes BAFF-R-bearing cells. However, like anti-CD20, this antibody does not deplete long-lived plasma cells that no longer express BAFF-R (Ref. Reference Lin125).

Preliminary clinical experience with a mouse–human chimaeric anti-CD20 antibody (Rituximab) in patients with SLE (Ref. Reference Sfikakis, Boletis and Tsokos126) and other autoimmune diseases has been encouraging (Refs Reference Looney127, Reference Leandro128, Reference van Vollenhoven129), with rapid resolution of symptoms in some patients that depends on B-cell depletion but not on depletion of autoantibodies. Autoantibodies continue to be produced because plasma cells do not express CD20. These studies confirm the role of B cells as important effector cells independently of their ability to produce antibodies. The degree of B-cell depletion is affected by FcRIIIA genotype; there is less depletion in patients bearing the low-affinity allele, which is also a SLE-susceptibility allele (Ref. Reference Anolik130). In addition, the rapid induction of human antichimaeric antibodies (HACA) has prevented B-cell depletion in some SLE patients. A fully human anti-CD20 is now available and may help solve this problem. In some patients, remissions are long-lived and returning B cells are predominantly of the transitional phenotype with low numbers of memory B cells (Ref. Reference Anolik131). The reappearance of memory B cells is also associated with earlier relapse in rheumatoid arthritis patients treated with anti-CD20 (Ref. Reference Leandro132). Long-term remissions have been associated with the absence of antibodies to extractable nuclear antigens (Ref. Reference Ng133). Progressive multifocal leukoencephalopathy has been observed in two SLE patients that received more than one course of anti-CD20 together with other immunosuppressive therapy. Most of the reports of the efficacy of anti-CD20 in SLE have been uncontrolled case series in small numbers of treatment-resistant patients in which the drug has been combined with other therapies including high doses of corticosteroids and cyclophosphamide. A placebo-controlled Phase III trial of anti-CD20 in SLE patients is now in progress.

CD22 is a member of the sialoadhesin (siglec) family of adhesion molecules that is coexpressed with the B-cell receptor on mature B cells and up to the late germinal centre stage (Refs Reference van Vollenhoven129, Reference Tedder134, Reference Crocker, Paulson and Varki135). CD22 has both immunoreceptor tyrosine activation motif (ITAM) and immunoreceptor tyrosine inhibitory motif (ITIM) domains in its intracellular domain and can act both as a negative regulator of BCR signalling and as a costimulatory molecule, depending on the context in which it is activated (Ref. Reference Otipoby136). Because of this complexity, the precise effect of CD22 upon BCR signalling in B cells at the various developmental stages is not fully elucidated. B cells from CD22-deficient mice have a shortened lifespan and the mice have reduced circulating B-cell numbers; however, these mice also have hyper-reactive B cells and may develop autoantibodies with age (Ref. Reference O'Keefe137). Epratuzumab, a humanised monoclonal anti-CD22 antibody, can mediate antibody-dependent cellular cytotoxicity and has been used to treat human B-cell malignancies (Ref. Reference Leonard138). The rationale for using anti-CD22 in human autoimmunity is based on the notion that modulating B-cell function may be safer than therapies that deplete B cells. However, given the complexity of CD22 signalling in B cells and the ability of this molecule to mediate functions both in a ligand-dependent and a ligand-independent fashion, it is difficult to predict what the effect of an anti-CD22 antibody will be in humans with autoimmune disease. The first clinical trial of epratuzumab in 14 SLE patients with mild to moderately active SLE has just been reported (Ref. Reference Dorner139). Treatment with anti-CD22 appeared to be safe and resulted in modest B-cell depletion in some of the patients. Serum levels of IgG and C3 were not altered, and autoantibody levels either increased or stayed the same. Although clinical efficacy could not be assessed in this study, improvements in BILAG (British Isles lupus assessment group) scores were reported. Mechanistic studies of patients in this clinical trial showed preferential targeting of CD27-negative naive or transitional B cells that have high levels of cell surface CD22 but no decrease in memory B cells or plasma cells (Ref. Reference Jacobi140). Cell-surface CD22 levels were downregulated by epratuzumab. Cell culture experiments further showed that epratuzumab modulated the abnormal proliferative response of SLE B cells to TLR or CD40 ligation. An antiproliferative effect was also observed in vitro when human B-cell lines were immobilised and stimulated with anti-IgM (Ref. Reference Carnahan141). Further studies will be needed to determine whether these modest B-cell effects result in a measurable clinical effect in controlled trials.

Depletion of autoantibodies or specifically of autoantibody-producing B cells is another approach that is currently in development. Abetimus sodium is a compound consisting of four double-stranded 20-mer ODNs that binds to anti-dsDNA antibodies and clears them from the serum. In mice, this drug has been shown to decrease splenic autoantibody-producing B cells, but it is not clear whether this also occurs in humans. Although the drug depletes anti-DNA antibodies in SLE patients by 30%, two Phase III clinical trials have failed to reach their primary endpoints of decreased time to renal flare and clinical benefit was demonstrated only using post-hoc analysis of a patient subset with high-affinity abetimus-binding antibodies at trial initiation (Ref. Reference Alarcon-Segovia142). Further studies of this agent are in progress. Other compounds directed specifically at autoantibody-producing B cells have been developed and are described in a recent review (Ref. Reference Wiesendanger143).

IL-6

IL-6 is a multifunctional cytokine that is critical for the development and functioning of many different cells and plays a crucial role in immune responses and inflammation (Refs Reference Liang146, Reference Kishimoto147). In the setting of activation of the innate immune system, IL-6 dampens T reg function (Ref. Reference Pasare and Medzhitov26). Conversely, IL-6 has a regulatory role in the resting immune system as it has been shown to maintain anergy of resting B cells chronically exposed to self-antigens (Ref. Reference Wiesendanger208). Excessive secretion of IL-6 is found in many autoimmune diseases (Ref. Reference Mihara, Nishimoto and Ohsugi148) including SLE (Ref. Reference Linker-Israeli149), where it induces plasma cell and effector T-cell differentiation (Refs Reference Pasare and Medzhitov26, Reference Tackey, Lipsky and Illei150). A critical role for IL-6 in the pathogenesis of SLE was demonstrated by the beneficial effects of IL-6 receptor blockade and the exacerbating effect of IL-6 in NZB/W F1 mice (Refs Reference Liang146, Reference Mihara151, Reference Finck, Chan and Wofsy152). A Phase I clinical study using an anti-IL6R monoclonal antibody (MRA) in SLE has been completed but the development of dose-dependent neutropaenia may complicate the use of this agent in human SLE (Ref. Reference Illei153).

TNF-α

TNF-α is a pleiotropic cytokine that can mediate both cell survival and cell death (Ref. Reference Hill and Lunec154). The importance of this cytokine in maintaining the balance of the immune system is highlighted by acceleration of autoimmunity in a TNF-α-deficient background (Refs Reference Jacob and McDevitt155, Reference Kontoyiannis and Kollias156) and the triggering or maintenance of autoimmunity in the setting of excess TNF-α (Refs Reference Mountz157, Reference Romas, Gillespie and Martin158, Reference Feldmann and Maini159). TNF-α antagonists have a remarkable therapeutic effect in a number of autoimmune diseases including rheumatoid arthritis and Crohn disease (Refs Reference Sandborn and Hanauer160, Reference Feldmann and Maini161, Reference Tobin and Kirby162). However, TNF-α blockade appears to exacerbate certain autoimmune diseases (Ref. Reference Kollias163) and has resulted in the emergence of new autoimmunity in up to 50% of treated patients with the development of antinuclear antibodies and even clinical SLE (Ref. Reference Swale164). By contrast, TNF-α is known to be pathogenic in mouse models of nephritis and is produced by intrinsic renal cells as a result of exposure to inflammatory cytokines including IL-1 (Ref. Reference Tipping and Holdsworth165). Based on these findings, successful anecdotal use of TNF-α blockade has been reported in a small number of patients with refractory SLE nephritis (Ref. Reference Aringer166). Not surprisingly, autoantibody titres increased in the treated SLE patients, confirming that some features of autoimmunity are exacerbated by this approach. One hypothesis for this effect is that TNF-α blockade increases the load of apoptotic cells (Ref. Reference Aringer167). One group found increased expression of TNFRII on CD4⁺ CD25⁺ T regs of SLE patients. Signalling through this receptor resulted in loss of suppressive function of T regs that could be reversed by IL-2 supplementation (Ref. Reference Valencia168). Other hypotheses include the induction of IFN-α following TNF blockade and the requirement for TNF-α in negative regulation of activated T and B cells. Since TNF-α is made as a soluble and a membrane protein and can bind to two different receptors with different functions (Refs Reference Chan and Lenardo169, Reference Locksley, Killeen and Lenardo170, Reference Fu and Chaplin171), there has been recent interest in selective TNF-α blockade as a strategy for a safer and more specific therapeutic response (Refs Reference Kollias163, Reference Kassiotis and Kollias172).

Type I interferons

Type I IFNs made by plasmacytoid dendritic cells are implicated in the pathogenesis of SLE (Refs Reference Pascual, Farkas and Banchereau173, Reference Ronnblom and Alm174). Their therapeutic use can induce SLE-like syndromes, and peripheral blood mononuclear cells from a subset of patients with SLE display increased expression of IFN-α-regulated genes (Refs Reference Bennett175, Reference Baechler176). IFN-α promotes the differentiation of monocytes into dendritic cells, induces differentiation of CD8⁺ T cells and, together with IL-6, induces differentiation of B cells into plasma cells. However, the role of IFN-α in autoimmunity is pleiotropic and absence of IFN-α or its receptor has had opposite effects in different mouse models of SLE (Ref. Reference Koutouzov, Mathian and Dalloul177). The explanation for this might lie in genetic differences and/or in the different effects of IFN-α on resting and activated dendritic cells. In vitro experiments have shown that triggering of TLR9 in naive peripheral blood mononuclear cell cultures results in an IFN-α-dependent inhibition of inflammatory Th1 cytokine secretion upon subsequent activation (Ref. Reference Meyers178). For this reason, type I IFN therapy has been used to treat autoimmune diseases such as multiple sclerosis. However, if type I interferons are administered to activated dendritic cells, proinflammatory cytokines are generated (Ref. Reference Nagai179). This suggests that the outcome of treatment with IFN-α antagonists in patients may be stage dependent and may not always be predictable. In addition, the recent finding that the interferon signature can be elicited in human monocytes in the absence of IFN-α by ligation of activating Fc receptors, together with blockade of inhibitory Fc receptors, suggests that there are multiple pathways for induction of inflammation by immune complexes (Ref. Reference Dhodapkar180). Early clinical trials of anti-IFN antibodies are in process and should more definitively define the pathogenic role of this group of cytokines in SLE.

IL-10

IL-10 is increased in patients with active SLE and correlates with disease activity (Refs Reference Park181, Reference Hagiwara182). Although IL-10 is classically considered an anti-inflammatory cytokine, its role in SLE appears to be an inflammatory one. In the presence of IFN-γ and immune complexes, there is attenuation of the suppressive functions of IL-10 mediated by a decrease in expression of the IL-10 receptor and decreased activation of JAK-1 (Ref. Reference Ji183). In the presence of IFN-α there is a gain of proinflammatory function of IL-10 leading to STAT-1-dependent expression of genes that are normally induced by IFN-γ. In NZB/W mice, continuous administration of IL-10 accelerated onset of renal disease and treatment with an anti-IL-10 monoclonal antibody delayed disease onset (Ref. Reference Ishida184). By contrast, in NZM2410 mice, IL-10 was protective when administered early in disease (Ref. Reference Blenman185). Similarly, in MRL/lpr mice, IL-10 deficiency resulted in a more severe disease (Ref. Reference Yin186). The differences between these models may be related to the amount of concomitant inflammation at the time of IL-10 administration or depletion. In a pilot open-label short-term study of a mouse anti-IL-10 monoclonal antibody in a small number of active SLE patients, disease improvement was evident at 2 months, with continued responses over 3–6 months (Ref. Reference Llorente187).

Interferon γ

IFN-γ accelerates disease in SLE-prone NZB/W mice whereas treatment with anti-IFN-γ is protective (Refs Reference Jacob, van der Meide and McDevitt188, Reference Ozmen189). IFN-γ^–/– MRL/lpr mice were also protected from early death with a reduction in the severity of glomerulonephritis. These studies highlight the importance of IFN-γ in accelerating SLE development, presumably by increasing MHC expression and autoantigen presentation to otherwise quiescent nontolerant anti-self T cells, and also by promoting local immune and inflammatory processes (Ref. Reference Balomenos, Rumold and Theofilopoulos190). By contrast, results in STAT-4- and STAT-6-deficient SLE-prone mice showed surprisingly that a decrease in IFN-γ and an increase in IL-4 accelerated SLE nephritis even though autoantibody titres were diminished (Ref. Reference Singh191). This may be due to early regulatory effects of IFN-γ on the development of Th17 effector cells. IL-17-secreting cells appear to be critical effector cells in rheumatoid arthritis and multiple sclerosis (Ref. Reference Weaver192), but very little is known as yet about the role of this cytokine in SLE.

Preclinical testing of novel therapies in murine models

As discussed above, immune activation pathways involved in initiation of SLE may not always be the same as those that are pathogenic during the effector phase. Furthermore, inflammatory environments can alter signalling cascades or provide redundant survival signals. Because the clinical effects of new therapeutics may not always be predictable from in vitro studies or studies in non-autoimmune mice, it is important to test these drugs in preclinical models. Many different mouse models of SLE now exist (Refs Reference Fairhurst, Wandstrat and Wakeland3, Reference Theofilopoulos and Kono4, Reference Lauwerys and Wakeland5) and several of these have been extensively characterised and used for therapeutic studies (Ref. Reference Hahn, Wallace and Hahn200) (Table 1). Most have been used for studies of prevention or treatment of SLE nephritis or of the effects of treatment on autoantibody production and immune cell activation. In addition, the MRL/lpr mouse can be used for study of vasculitis, arthritis and skin disease, whereas the NZW/BXSB mouse can be used to study the effects of therapy on antiphospholipid syndrome (Ref. Reference Akkerman201). It is important to recognise that murine SLE models have a number of limitations. First, mouse immune pathways are not always the same as human pathways. Second, not all clinical SLE manifestations are present in each mouse strain. In particular, there are no good mouse models for the treatment of the central nervous system manifestations of SLE. Third, responses to some therapies are strain dependent and therapies that improve disease in one model may worsen it in another. In particular, a number of differences have been observed in responses to therapy of MRL/lpr mice and NZB/W mice. Fourth, most mouse models do not undergo spontaneous disease remissions, although it is now possible to study disease relapses after remissions have been induced with biological therapies. Nevertheless, animal models allow analysis of basic immune mechanisms of disease and therapy that are not possible in human studies and can help direct appropriate clinical use of new therapeutics.

Table 1.

Common mouse models of systemic lupus erythematosus used for therapeutic studies

Abbreviations: ANAs, antinuclear antibodies; CNS, central nervous system; GVHD, graft-versus-host disease; MHC, major histocompatability complex; TLR, Toll-like receptor; RBC, red blood cell; RNP, ribonucleoprotein; Sm, Smith antigen.

Clinical trials in human SLE

New agents for which the results of clinical trials have been published are shown in Table 2. Most of these trials were Phase II studies and had insufficient statistical power to assess clinical efficacy. However, several large Phase III studies are now in progress and results can be expected in the next few years. The design of SLE clinical trials will be of crucial importance in determining optimal therapies for induction of disease remission and for maintenance of disease remission. Much thought is being given to recruitment of patient populations, disease activity measurements, disease response measurements, primary outcome measures and mechanistic analyses of immunological responses in treated patients. Most clinical trials now employ the BILAG, a scoring system that measures the severity of disease in individual organ systems. Some drugs are being tested in remission induction studies whereas others are being tested in ‘time-to-flare’ studies. The choice of design will depend on the expected mechanism of action of each drug. For example, blockade of inflammatory cytokines or depletion of whole cell populations would be expected to induce remission of active disease whereas therapies that prevent naive cell activation or modulate B-cell selection would be expected to have a preventive rather than a remission induction effect. Testing in preclinical models may help direct trial design for each new agent.

Table 2.

Therapeutic trials of novel agents for systemic lupus erythematosus

^aPublished in abstract form only but included here because it is a large Phase III study. Abbreviations: BAFF, B-cell-activating factor belonging to the TNF family (also known as BLyS, B lymphocyte activator); BILAG, British Isles lupus assessment; SLE, systemic lupus erythematosus; TNF, tumour necrosis factor.

The range of new drugs being developed reflects the global nature of the immune defects in SLE (Fig. 5). Approaches include measures to decrease antigen load (DNase), measures to dampen the innate immune system (ODN and IFN-α blockade), drugs that antagonise activation of the acquired immune system (costimulatory blockade, modulation of BCR signals), blockade of the soluble mediators derived from the effector arm of the immune response (cytokine antagonists), protection of target organs from damage (chemokine inhibitors, complement inhibitors) and even depletion of whole cell populations (B-cell depletion, stem cell transplant). In addition, there are agents designed to specifically target autoreactive B cells through induction of anti-idiotypic antibodies. The scope of this article does not allow discussion of all the new drugs being developed, but for a comprehensive review see Ref. 143. The challenge will now be to determine at which stage of disease each reagent is most likely to be effective and which reagents will work together. Optimal therapy will also require continued research that will eventually allow identification of the underlying immune defect in each patient. The ultimate goal is to selectively correct deficits that contribute to autoimmunity without impacting protective immunity.

Figure 5.

Interactions of the innate and acquired immune system amplify immune responses in systemic lupus erythematosus. Therapeutic interventions for SLE (shown in red) target multiple arms of the immune system (see Ref. 143 for review). Abbreviations: APRIL, A proliferation inducing ligand; BAFF, B-cell-activating factor belonging to the TNF family; DC, dendritic cell; ICOS, inducible costimulator; IFN-α, interferon α; ODN, oligodeoxynucleotide; TLR, Toll-like receptor.