Cytokine disturbances in systemic lupus erythematosus

The pathogenesis of systemic lupus erythematosus (SLE) is complex, and the resulting disease manifestations are heterogeneous. Cytokine dysregulation is pervasive, and their protein and gene expression profiles may serve as markers of disease activity and severity. Importantly, biologic agents that target specific cytokines may represent novel therapies for SLE. Four cytokines (IL-6, TNFα, IFNα, and BLyS) are being evaluated as therapeutic targets in SLE. The present review will examine the roles of each of these cytokines in murine and human SLE, and will summarize results from clinical trials of agents that target these cytokines.


Introduction
Cytokines collectively play key roles in the regulation of systemic infl ammation, local tissue damage, and immuno modulation. Not surprisingly, cytokines often play direct roles in disease pathogenesis, including that of systemic lupus erythematosus (SLE). To the chagrin of both investigator and practitioner, the eff ects of cytokines are pleiotropic and include both synergistic and antagonistic eff ects on other cytokines -thereby introducing tremen dous complexity that has often led to diametrically oppos ing conclusions in diff erent studies. In the present review, we focus on four cytokines that have received great attention either as candidate biomarkers for disease activity and/or as candidate targets of novel biologic agents.

Interleukin-6
General biology of IL-6 IL-6 is a pleiotropic 26 kDa protein produced by a panoply of cell types, and aff ects the function of an equally broad spectrum of cell types. Of greatest relevance to SLE is the ability of IL-6 to promote activation and/or diff erentiation of cells central to the development of systemic autoimmunity and the attendant pathologic infl ammatory responses, including T cells, B cells, macrophages, and neutrophils [1].

IL-6 in murine systemic lupus erythematosus
Studies in multiple murine SLE models point to an essential role for the IL-6 pathway in SLE. Genetic defi ciency of IL-6 in MRL/lpr mice delays disease development, reduces CD4 + and CD8 + T-cell numbers, decreases glomerular IgG and C3 deposition, downregulates renal parenchymal vascular cell adhesion molecule-1 (VCAM-1) expression, and diminishes kidney macro phage infi ltration [2]. Th is last eff ect may be especially impor tant, in as much as a close association between activated renal macrophages and disease has been documented in (NZB x NZW)F1 (BWF) mice [3].
Additional important observations have been made with BWF mice. Exogenous IL-6 increases IgG anti-DNA autoantibody production by B cells isolated from clinically aff ected BWF mice, whereas neutralization of IL-6 via either addition of an anti-IL-6 mAb or macrophage depletion decreases production of such autoantibodies [4][5][6]. Building on these ex vivo fi ndings, administration of human IL-6 to 6-month-old female BWF mice promoted accelerated membranoproliferative glomerulonephritis associated with marked upregulation of mesangial MHC class II antigens and glomerular intercellular adhesion molecule-1 (ICAM-1) expression. Treatment with cyclosporin inhibited the development of glo merulo nephritis, distinguishing the eff ects of IL-6 on systemic infl ammation from its direct eff ect on renal mesangial cells [7]. More impressive were the signifi cant decreases in mortality, progression of proteinuria, and anti-dsDNA antibody levels in BWF mice chronically treated from 3 months of age with anti-IL-6 mAb or anti-IL6 receptor antibody [8,9].
IL-6 is also associated with SLE in other murine SLE models. In pristine-induced SLE, renal disease was milder, and high levels of IgG anti-single-strand DNA, anti-dsDNA, and anti-chromatin antibodies were absent in IL-6-defi cient mice [10]. In JunB Δep mice, development of an SLE-like phenotype (including skin lesions and immune complex glomerulonephritis) was linked to increased epidermal IL-6 secretion that arose from the specifi c loss of epidermal JunB. Intercrosses with IL-6defi cient mice abrogated the SLE phenotype [11].

IL-6 in human systemic lupus erythematosus
As in murine SLE, serum levels of IL-6 are elevated in human SLE and have correlated with disease activity or anti-dsDNA levels in some studies [12][13][14]. Th e increased frequency of IL-6-producing peripheral blood mononuclear cells correlates with disease severity/activity and treatment response [15,16]. Of note, SLE B cells express IL-6R spontaneously and produce great amounts of IL-6 [17,18]. Moreover, autoreactive T-cell clones from SLE patients also produce large amounts of IL-6, and thereby promote B-cell activation and autoantibody pro duc tion [19]. Indeed, the copious spontaneous pro duc tion of immunoglobulin by SLE B cells can be enhanced by exogenous IL-6 and diminished by neutralizing antibodies to IL-6 [13].
Aside from its systemic eff ects, IL-6 is closely linked with specifi c disease manifestations. Elevated cerebral spinal fl uid levels of IL-6 are found in SLE patients with psychosis and may aff ord an eff ective measure of its diagnosis [20]. In SLE nephritis patients, urinary levels of IL-6 are elevated, correlate with titers of anti-dsDNA anti bodies, and decrease following treatment [14,21]. Local expression of IL-6 was documented in kidney tissue of SLE nephritis patients [22], arising at least in part from mesangial cells [23].

Therapeutic blockade of IL-6 in systemic lupus erythematosus
Given the considerable evidence pointing to a contri butory role for IL-6 in SLE pathogenesis, the a priori likelihood that therapeutic targeting of this pathway in SLE will be effi cacious is high. Tocilizumab, a mouse-human chimeric anti-human anti-IL-6R mAb, is already US Food and Drug Administration approved for the treatment of patients with rheumatoid arthritis. Tocilizu mab eff ectively inhibits IL-6 signal transduction by its ability to bind both mIL-6R (preventing IL-6 binding to cell-surface mIL-6R/gp130 complex) and sIL-6R (pre vent ing formation of IL-6/sIL-6R complex, which could bind to cell-surface gp130).
Although initial results suggest that tocilizumab can control SLE disease activity, an increased risk of infections may limit its ultimate clinical utility. In an open-label phase I dosage-escalation study (Table 1), 16 patients with mild to moderate disease activity (presence of chronic glomerulonephritis or extrarenal SLE Disease Activity Index -SELENA modifi cation score of 3 to 10) received one of three doses of tocilizumab (2, 4, or 8 mg/ kg) every other week for 12 weeks [24]. Disease activity showed improvement, with a decrease of ≥4 points in activity scores for eight of 15 evaluable patients. Additionally, arthritis improved in all seven patients who had arthritis at baseline, and completely resolved in four of them. Th ere were also signifi cant decreases in the frequency of circulating plasma cells and in levels of anti-dsDNA antibodies. Tocilizumab also led to a similar decline in levels of C3 and the complement activation products, iC3b and C5b-9 (terminal activation complex). Tocilizumab treatment led to dose-related decreases in the absolute neutrophil count, however, with two patients experiencing severe neutropenia (absolute neutrophil count of 500 to 1,000/μl). Th e decline in neutrophil counts probably had clinical consequences -16 infec tions developed in 11 patients, 10 of whom required systemic antibiotics or antivirals. Additional studies are clearly needed before tocilizumab or other IL-6 antagonists can be considered viable therapeutics in human SLE.

General biology of TNFα
TNFα is expressed as a 26 kDa transmembrane protein that can be cleaved by TNFα-converting enzyme to release a 17 kDa soluble protein [25]. Like IL-6, TNFα is produced by a variety of cells, including activated macrophages and monocytes as well as B cells, T cells, keratinocytes, renal cells, and fi broblasts.
Depending upon the conditions, TNFα can trigger either proinfl ammatory or anti-infl ammatory pathways by engaging one or both of two distinct transmembrane receptors: the type I, p55 tumor necrosis factor receptor (TNFR1), and the type II, p75 TNFR (TNFR2) [26]. TNFR1 may trigger apoptotic and anti-infl ammatory signals through Fas-associated death domain (FADD) and activa tion of the caspase cascade. Conversely, it may also trigger anti-apoptotic and proinfl ammatory responses through recruitment of TNF receptor-associated factor 2 (TRAF-2), which activates NF-κB, c-Jun NH2terminal kinase (JNK), and mitogen-activated protein (MAP) kinase. Binding of TNFα to TNFR2 also leads to recruitment of TRAF2 and promotes proinfl am matory responses. Th e TNF/TNFR system can thus serve dually both as a potent pro infl am matory mediator and as a key immune modulator as a function of the immuno logical setting [25,26].

TNFα in murine systemic lupus erythematosus
Several studies point to a pathogenic role for TNFα in murine SLE. TNFα expression is increased in the sera and kidneys of MRL/lpr mice and correlates with disease activity [27,28]. Indeed, TNF antagonism/inhibition in such mice has clinical benefi t. Furthermore, in C3H.SW mice with induced SLE, treatment with anti-TNFα mAb results in decreased development of anti-DNA antibodies, proteinuria, and glomerular immune-complex deposition [29,30].
In sharp contrast, the administration of recombinant TNFα or TNFα-inducing agents to BWF mice is clearly benefi cial [31-33]. Moreover, TNFα defi ciency in other SLE mice has resulted in disease induction/exacerbation. Autoimmunity and fatal SLE nephritis develop in NZB mice when crossed with TNFα-defi cient mice [34]. Similarly, TNFα defi ciency in mice bearing a mixed B6/129 genetic background (H-2 b/b ) develop IgG 2b and IgG 3 anti-DNA antibodies, glomerular IgG deposition, and alterations in glomerular structure resembling SLE nephritis [35].
Th ese often confl icting eff ects of TNFα in diff erent murine SLE models underscore this cytokine's dual proinfl ammatory and immunoregulatory roles. Th ese confl icting eff ects are also disease stage specifi c and may refl ect the cell type expressing the receptors as well as diff erential expression of the receptors themselves by a given cell type. Although the underlying molecular and cellular mechanisms remain largely enigmatic, at least part of TNF's functional duality might segregate at the level of the two TNFα receptors [35,36]. Indeed, in anti-GBM antibody-induced glomerulonephritis, TNFR2 defi ciency is protective, whereas TNFR1-defi cient mice develop proteinuria and renal pathology similar to those in wild-type controls [37]. In addition, TNFR1 defi ciency greatly accelerates lymphoadenopathy, autoantibody produc tion, and mortality in C57BL/6-lpr mice [38]. In NZM2328 mice singly defi cient in one TNFR, however, signaling through the other receptor can compensate and permit persistence of the wild-type clinical and pathological phenotypes. Deletion of both receptors, with complete abrogation of TNFα signaling, results in signifi cantly accelerated disease [39].

TNFα in human systemic lupus erythematosus
As in animal models, the evidence has been confl icting regarding the role of TNFα in human SLE. Some studies have shown that serum TNFα levels are elevated in SLE patients and correlate with disease activity [40,41]. Other studies, however, have shown otherwise. One study demonstrated that elevated TNFα plasma levels do not correlate with SLE disease activity, and another study showed that TNFα levels were actually higher in patients with inactive disease, suggesting a protective role for TNFα in SLE [42,43]. Diff erences in patient charac teristics, assays for TNFα, and study designs probably each contribute to the somewhat inchoate picture that emerges, highlighting the pleiotropic and oftentimes unpredictable nature of TNFα.
In addition to systemic production of TNFα, local production of TNFα may contribute to SLE. Studies of local production of TNFα in the kidneys have documented upregulated TNFα protein and gene expression in renal biopsy tissue from 52% of SLE nephritis patients [22]. Local upregulated TNFα protein and gene expression within the resident glomerular, tubular, and mesangial cells has also been demonstrated [23].
Th e eff ects of systemically produced TNFα may be dissociated from the eff ects of locally produced TNFα. For example, DR3-positive or DR4-positive SLE patients, who genetically are high TNFα producers in response to appropriate induction, are not predisposed to nephritis. In contrast, DR2 and DQw1-positive SLE patients, who genetically are low TNFα producers following induction, have an increased incidence of SLE nephritis [44]. As another example, the TNFα adaptor proteins TNF receptor type 1-associated DEATH domain (TRADD), FADD, receptor interacting protein-1 (RIP-1), and LY2127399 Phase III planned [125] BLyS and APRIL Atacicept Phase II/III initiated [122] TRAF-2 are downregulated in SLE peripheral blood mono nuclear cells, and their decreased expres sion correlates with disease activity [45]. Th e down regulation of TRADD and FADD (which can recruit the caspase cascade) may thus promote an anti-apoptotic eff ect that leads to survival of autoreactive cells. Of note, TRADD, RIP-1, and TRAF-2 (which can activate the NF-κB pathway) are upregulated, rather than down regulated, in renal tissue from SLE nephritis patients, suggesting that TNFα may promote immunoregulatory functions systemically but may mediate infl ammation locally [46].

Therapeutic blockade of TNFα in systemic lupus erythematosus
Use of TNFα antagonists is well established in the treatment of rheumatoid arthritis, juvenile idiopathic arthritis, psoriatic arthritis, chronic infl ammatory bowel disease, and ankylosing spondylitis [47]. Consistent with the competing proinfl ammatory and anti-infl ammatory eff ects of TNFα, use of TNFα antagonists in multiple sclerosis patients has led to immune activation and disease exacerbation [48]. Moreover, TNFα antagonists have triggered the development of autoantibodies, neuroinfl ammatory disease, or SLE-like features in some patients with rheumatoid arthritis or chronic infl ammatory bowel disease [49][50][51]. Indeed, depending on the disease, two-thirds of previously antinuclear antibody (ANA)-negative patients became ANA-positive after initiation of TNFα-antagonist therapy. A recent report described six rheumatoid arthritis patients who developed active SLE associated with major organ involve ment and life-threatening manifestations and required additional treatment beyond cessation of the TNFα-antagonist regimen [52]. Nevertheless, TNFα-antagonist therapy has been successful in some SLE patients. Six patients with moderately active SLE (three with nephritis, two with arthritis refractory to other therapies, and one with both joint and renal involvement) were given four infusions of 300 mg doses of infl iximab, in combination with immunosuppression with azathioprine or methotrexate [53]. Patients with joint involvement experienced remission of arthritis, which relapsed 8 to 11 weeks after the last infl iximab infusion. In the four patients with nephritis, proteinuria decreased by >60% within 8 weeks. Of note, anti-dsDNA and anti-cardiolipin antibodies increased in four patients. A number of long-term studies also revealed modest clinical benefi t, but there were several adverse events, including deep-vein thrombosis, infections, central nervous system lymphoma, and Legionella pneumonia [54,55].
Th e need to evaluate safety and effi cacy of TNFα blockade in SLE prompted two large randomized phase II and phase III trials with etanercept and infl iximab (Table 1), but both studies (NCT00447265 and NCT00368264) were terminated prematurely (although this may have had more to do with corporate business matters than with medical matters). As with neutralization of IL-6, additional studies with TNFα antagonists will be needed before use of such agents becomes a routinely viable option in human SLE.

General biology of type I interferon
Type I interferon represent a large family of cytokines that includes many IFNα subtypes, a single IFNβ, and multiple IFNω subtypes that exert their expansive biological properties through engagement of a common heterodimeric receptor (type I interferon receptor (IFNAR)) composed of IFNAR1 and IFNAR2 subunits. Th ese broad eff ects include activating dendritic cells; promoting prolifera tion, survival, and diff erentiation of monocytes into antigen-presenting cells and B cells into plasma cells; stimulating the Th 1 pathway and preventing apoptosis of activated cytotoxic T cells; and suppressing regulatory T cells, enhancing natural killer cell activity, and modulating each of these cells' respective cytokine production and signaling responses.
Within the type I interferon family, the IFNα subtypes are arguably the biologically most important in general, and this certainly is the case for SLE. Th e major producers of IFNα are plasmacytoid dendritic cells, although virtually any cell type can elaborate interferon [56]. IFNα is produced both in response to exogenous stimuli, such as bacterial and viral pathogens, and to endogenous stimuli, such as self-nucleic acids and nucleic acid-containing immune complexes, via Toll-like receptor (TLR)-dependent and TLR-independent pathways. Among the TLR-dependent pathways, engagement of TLR7 and TLR8 by single-strand RNA-containing complexes and engagement of TLR9 by dsDNA-containing complexes may be especially relevant to SLE pathogenesis [56].

Type I interferon in murine lupus
Numerous studies have collectively demonstrated the complexity of the type I interferon/IFNAR axis in the pathogenesis of murine SLE. In several strains of mice with SLE-related autoimmune phenotypes, type I interferon/IFNAR signaling promotes autoantibody produc tion and development of renal disease. Indeed, genetic deletion of the Ifnar1 gene prevents the development of severe clinical and pathological disease in SLEprone NZB and NZM2328 mice [57,58]. Similarly, upregulation or administration of IFNα to BWF mice, (NZWxBXSB)F1 mice, B6.Sle123 mice, or NZM2328 mice markedly accelerates nephritis and death [59][60][61], and in BWF mice is associated with unabated expansion of short-lived plasma cells [62]. Nevertheless, type I interferon/IFNAR signaling is not always deleterious. IFNAR-defi cient MRL/lpr mice develop more severe disease than do their IFNAR-suffi cient counterparts [63], an observation that may relate to the ability of IFNβ to ameliorate disease manifestations in MRL/lpr mice. Importantly, these counterintuitive eff ects of type I interferon are not limited to MRL/lpr mice, in that blockade of type I interferon in B57BL/6 or B6.Sle2 mice led to increased, rather than decreased, levels of IgG autoantibodies [64].
In addition to studies that directly manipulated type I interferon and/or IFNAR expression, studies focused on TLR (especially TLR7 and TLR9) have also pointed to a vital role for IFNα in murine SLE. Interestingly, TLR7/9 doubly-defi cient MRL/lpr mice display signifi cantly reduced ANA titers, proteinuria, and kidney disease [65]. In agreement with these results, treatment of BWF mice with a dual inhibitor of TLR7 and TLR9 demonstrated signifi cant reductions in interferon produced by plasmacytoid dendritic cells; decreased proteinuria; reduced glomerulonephritis, serum autoantibody levels, and endorgan damage; and increased survival [66]. Indeed, adminis tration of bacterial or viral TLR ligands to SLEprone mice leads to increased IFNα production and disease exacerbation, whereas disease is not exacerbated in similarly treated IFNAR-defi cient SLE-prone mice [67].

Type I interferon and human systemic lupus erythematosus
Serum type I interferon activity is increased in SLE patients and their fi rst-degree relatives, and ages of peak type I interferon are associated with ages of peak SLE incidence [68,69]. Type I interferon levels have correlated with disease activity in a number of cross-sectional studies, although longitudinal correlations could not be established in other studies [70][71][72]. Th e use of elevated serum levels of type I interferon-regulated chemokines as biomarkers for disease activity was suggested in early studies [73] and was later replicated and validated by associations between elevated transcript levels of these chemokines with disease activity and organ damage [74,75]. By relying on expression of type I interferoninducible genes rather than on circulating type I interferon protein levels (which are often barely detectable or not detectable), most SLE patients have been demonstrated to unequivocally express type I interferoninducible genes in peripheral blood mononuclear cells. Th is so-called interferon signature has been convincingly correlated with active disease, renal manifestations, and increased damage index [70,76,77]. Furthermore, the interferon gene signature has been found in glomerular and synovial tissue, suggesting local organ involvement of type I interferon [78,79]. Indeed, plasmacytoid dendritic cells -the main producers of IFNα -accumulate in the glomeruli of SLE nephritis patients and can also be found in cutaneous lesions where they promote continuous IFNα release [80,81].
In terms of potential interferon contribution to lymphocyte autoreactivity in SLE, interferon not only promotes B-cell activation, antibody production, and class switching, but can also decrease B-cell selectivity for CpG-rich DNA, thereby activating TLR9, and can also allow stimulation of B cells by non-CpG DNA [82,83]. Moreover, interferon can potentially promote survival of autoreactive B cells, since it can prevent B-cell apoptosis and enhance proliferation even in the absence of mitogenic stimuli [84].
T-cell function is also modulated by interferon in SLE. Ample literature has demonstrated dysfunction of regula tory T cells in SLE, with SLE regulatory T cells being ineffi cient suppressors of infl ammation and T-cell prolifera tion [85,86]. Part of this dysfunction appears to be due to interferon production by antigen-presenting cells [87].
Th e ex vivo and in vitro fi ndings that suggest a role for type I interferon in SLE pathogenesis are supported by the development of SLE or SLE-like syndromes (malar rash, oral ulcers, photosensitivity, renal involvement, and anti-Sm and anti-dsDNA antibodies) following administration of IFNα [88,89]. Occasionally, IFNα-induced SLE has led to life-threatening multiorgan involvement [90].
Recent investigations into the genetics of SLE have also revealed a strong connection between the type I interferon pathway and SLE. Th e gene coding for the transcription factor IRF5 was the fi rst gene involved in IFNα production to be associated with SLE susceptibility [91]. Further evidence of genetic association between SLE and IRF5 and for interaction between IRF5 and TYK2, a Janus kinase involved in cytokine signaling that binds to IFNAR, has also been demonstrated [92]. Moreover, IRF5 risk loci are also associated with high serum type I interferon activity and the development of autoantibodies to dsDNA and RNA-binding proteins. [93]. Subsequently, the related IRF7 has also been associated with SLE [94]. Furthermore, the STAT4 gene, which codes for a protein that interacts with the cytoplasmic portion of IFNAR, is strongly associated with SLE. STAT4 risk loci increase sensitivity to IFNα, are associated with a more severe phenotype that includes anti-dsDNA antibodies and renal involvement, and interact with IRF5 susceptibility loci [95][96][97]. Another SLE susceptibility gene codes for IRAK1, which is involved in TLR-triggered signal transduction [98].
IFNα, independent of its eff ects on the immune system, may be especially important in promoting atherosclerotic disease in SLE. Th e dysfunction of circulating endothelial progenitor cells and myelomonocytic circulating angio genic cells in SLE appears to be mediated by IFNα, in as much as neutralization of IFNα restores a normal endo thelial progenitor cell/circulating angiogenic cell pheno type [99]. Indeed, SLE patients with robust interferon signatures have decreased endothelial function [100], and interferon signatures correlate with carotid intima media thickness [101].

Therapeutic blockade of type I interferon in systemic lupus erythematosus
Th e fact that type I interferon exacerbates SLE in some mouse strains but ameliorates SLE in others [59][60][61]63] raises the possibility that markedly divergent clinical responses to type I interferon antagonists might emerge among individual SLE patients. Experience to date with antagonists against type I interferon in SLE is very limited ( Table 1). Treatment of SLE patients (n = 62) in a phase I trial with anti-IFNα mAb sifalimumab (MEDI-545) in single escalating intravenous doses of 0.3, 1.0, 3.0, 10.0, or 30.0 mg/kg led to dose-dependent reversal of the interferon signature in both the blood and skin, down regulation of several cytokines, and reduced disease activity without any increase in serious infections [102]. Results from a phase I trial with the anti-IFNα mAb rontalizumab (rhuMab IFN-α) in SLE patients (n = 32) replicated this dose-dependent reduction in the interferon signature [103]. Two phase II trials are currently ongoing to evaluate the eff ects of anti-IFNα in SLE (NCT01031836 and NCT00657189), but as yet it will be some time before the utility of type I interferon antagonists in SLE is known.

General biology of BLyS
BLyS (also known as BAFF) is a cytokine that is essential for survival of most B cells beyond the transitional 1 stage [104]. BLyS binds to three receptors: BCMA, TACI, and BR3 (also known as BAFFR). Two of these receptors (BCMA and TACI), but not the third (BR3), also bind APRIL, a cytokine closely related to BLyS that shares some, but not all, of its biological properties. Whereas survival of plasma cells, which express TACI and BCMA, is supported by either BLyS or APRIL, the survival of preplasmablast mature B cells, which express much BR3 but little TACI and essentially no BCMA, is supported only by BLyS. Of note, memory B cells appear to be independent of both BLyS and APRIL.
In addition to aff ecting B-cell survival/function, BLyS can aff ect other cell types that express BLyS receptors. Specifi cally, BR3 is expressed on T cells, although its role in T-cell signaling/stimulation is controversial [104]. Moreover, dendritic cells also express BLyS receptors, and BLyS-stimulated dendritic cells upregulate co-stimulatory molecules and produce infl ammatory cytokines and chemokines such as IL-1, IL-6, CCL2, and CCL5.

BLyS in murine systemic lupus erythematosus
Th ere is an irrefutable link between BLyS and murine SLE. Constitutive overexpression of BLyS in BLyS-transgenic nonautoimmune-prone mice leads to SLE-like features, including elevated levels of multiple auto antibodies (including anti-dsDNA), circulating immune complexes, and glomerular immunoglobulin deposition [105]. Moreover, BLyS overexpression accelerates the development of SLE-like features in mice that bear an autoimmune diathesis but otherwise do not develop overt SLE [106]. On the other hand, SLE-prone NZM2328 mice genetically defi cient in BLyS are largely spared from overt disease (severe proteinuria and premature death), although the lifelong absence of BLyS does not protect them from ultimately developing serological autoimmunity and renal pathology [107]. Most importantly, treatment of BWF mice, (NZMxBXSB)F1 mice, MRL/lpr mice, or NZM2410 mice with either TACI-Ig (which neutralizes both BLyS and APRIL) or BR3-Ig (which selectively neutralizes BLyS) is eff ective at preventing clinical disease and ameliorating renal injury [108]. Intriguingly, IFNα-driven exaggerated disease in several SLE-prone mouse strains is associated with increases in serum BLyS levels [59][60][61], and IFNα-driven exaggerated disease is completely blocked in BLySdefi cient NZM2328 mice [109], indicating that BLyS is a vital contributor to the IFNα-driven pathogenic pathway in SLE.

BLyS in human systemic lupus erythematosus
Not only is BLyS associated with murine SLE, but is also associated with human SLE. Circulating BLyS levels are elevated in as many as 50% of SLE patients [110], and disease activity correlates with blood leukocyte expres sion of BLyS mRNA [111]. Aberrant expression of BLyS on multiple immune cells has been observed in SLE. Although BLyS is largely expressed in myeloid lineage cells, activated B cells can also express both BLyS and APRIL. In SLE patients, B cells and plasma cells express high levels of BLyS and APRIL mRNA, which correlate with disease activity and levels of anti-dsDNA antibodies [112].
In addition to the biologic studies summarized above, investigations into genetic susceptibility loci in SLE have revealed associations between polymorphisms in BLyS and APRIL genes with human SLE [113,114]. In light of the results from murine SLE models, from ex vivo studies of human SLE leukocytes, from BLyS serum analyses, and from the genetics of SLE, BLyS has emerged as a highly attractive cytokine target in SLE.

Therapeutic blockade of BLyS in human systemic lupus erythematosus
Th e greatest experience to date with BLyS antagonists (Table 1) has accrued with belimumab, a fully human IgG 1λ mAb that binds and neutralizes soluble BLyS [115]. Belimumab was shown to be safe in a randomized, double-blind, placebo-controlled phase I trial of SLE, in which the prevalence of adverse events was no diff erent between belimumab-treated and placebo-treated patients [116]. Of note, only modest reductions in peripheral blood B cells were observed among belimumab-treated patients. No clinical effi cacy was demonstrated in this phase I trial, but the small number of patients (n = 70) and the very brief treatment schedules (single infusion or two infusions 3 weeks apart) and follow-up period (12 weeks after fi nal infusion) precluded demonstration of clinical benefi t.
Disappointingly, the subsequent phase II trial (n = 449) failed to meet its co-primary endpoints [117]. Extensive post hoc analysis, however, led to a novel composite index of clinical response (SLE responder index) [118] and demonstrated signifi cantly increased clinical response among belimumab-treated patients at 52 weeks among the patients who were seropositive (ANA titer ≥1:80 and/ or positive for anti-dsDNA antibodies) at entry. Using this novel SLE responder index, two separate large randomized, double-blind, placebo-controlled phase III trials (n = 865 and n = 819, respectively) of belimumab in seropositive SLE patients each met their primary endpoints (increased percentage of responders at 52 weeks) [119,120]. Importantly, analysis of the combined 1,864 SLE patients in both trials pointed to reductions in disease activity and prevention of worsening across vital internal organ systems [121]. Although questions remain regarding the durability of the clinical response, a US Food and Drug Administration advisory panel in November 2010 recommended approval for belimumab in the treatment of SLE, and fi nal approval by the US Food and Drug Administration was given on 9 March 2011.
Concern has been raised regarding the ostensibly lower clinical effi cacy of belimumab in human SLE in comparison with the ostensibly more robust clinical effi cacy of BLyS antagonists in murine SLE. Th is diff erence in clinical response between human and murine SLE may be more apparent than real. First, the clinical response in the phase III human trials was based on a composite of several instruments each rooted in multiple organ systems [118]. In contrast, clinical response in the murine trials was simply the absence of pre-moribund proteinuria and death. (Belimumab in all likelihood will prove to be very eff ective in preventing pre-moribund proteinuria and death; however, it will take many years of follow-up to formally prove this point.) Second, human SLE patients received standard-of-care therapy in addition to belimumab in the human clinical trials, whereas murine SLE patients in the murine clinical trials received no treatment other than a BLyS antagonist. (Human SLE patients treated with belimumab alone would undoubtedly do much better than untreated patients, but such a human trial would be entirely unethical.) In addition to belimumab, several other BLyS antagonists are undergoing clinical evaluation in SLE. Th e one furthest advanced in clinical evaluation is atacicept, a fusion protein between one of the BLyS receptors (TACI) and the Fc portion of IgG. Favorable safety and tolerability were demonstrated in a randomized, doubleblind, placebo-controlled phase I trial (n = 49) [122]. Dose-dependent reductions in peripheral blood B cells and in circulating immunoglobulin levels were noted, but clinical effi cacy could not be demonstrated due to the limited treatment and limited follow-up period. Of concern, an increased risk of severe infections was observed in a subsequent trial involving patients with SLE nephritis who were concurrently taking mycophenolate mofetil and corticosteroids (NCT00573157). As a consequence, this trial was prematurely terminated. Nevertheless, a separate phase II/III trial of atacicept in SLE has recently been initiated (NCT00624338). Whether atacicept achieves clinical success from effi cacy and safety standpoints remains to be determined.
A third BLyS antagonist being tested in clinical trials is A-623 (previously known as AMG 623), a fusion between the Fc portion of IgG and a peptide sequence selected for its ability to bind with high affi nity to BLyS. In a doubleblind, placebo-controlled phase I trial, SLE patients received a single dose (n = 54) or four weekly doses (n = 63) of escalating doses of AMG 623 or matching placebo [123]. A dose-independent decrease in naive and total peripheral blood B cells was accompanied by an increase in memory B cells -an observation that has now also been made in patients treated with atacicept or belimumab [122,124]. Clinical responses were not reported, so the relevance of the disparate changes among B-cell subsets to clinical parameters remains unknown. A phase II trial of A-623 in SLE had been initiated but was suspended due to 'structural failure identifi ed in some product vials' , but the trial was recently resumed (NCT01162681).
A fourth BLyS antagonist in clinical development for SLE is LY2127399, a mAb that binds both soluble and membrane BLyS [125]. Two phase III trials in SLE are planned (NCT01205438 and NCT01196091), but neither has yet started recruiting patients. It remains to be determined whether neutralization of soluble plus membrane BLyS (as with LY2127399) will have greater therapeutic effi cacy than neutralization of soluble BLyS alone (as with belimumab).

Concluding remarks
Cytokine-targeted therapy may prove eff ective in the treatment of SLE and off er less toxic options in com parison with 'conventional' therapies such as gluco corticoids or cytotoxics. In addition to the four cytokines reviewed above, there are many more cytokines -includ ing IFNγ, IL-1, and IL-17 -that have been implicated in SLE. As has become painfully evident, cytokine biology is extremely complex -and increasingly so in a complex disease such as SLE. Continued investigations of cytokine pathways in vivo in animal models, ex vivo in human SLE tissues, and in the genetics of SLE will reveal the roles of additional cytokines in disease pathogenesis and should off er additional novel targets for treatment.

Competing interests
WS has conducted clinical trials in systemic lupus erythematosus with belimumab (sponsored by Human Genome Sciences). NJ has no competing interests.