Article Text

Download PDFPDF

Tumour necrosis factor and other cytokines in murine lupus
  1. Argyrios N Theofilopoulos,
  2. Brian R Lawson
  1. Department of Immunology, The Scripps Research Institute, La Jolla, USA
  1. Professor A N Theofilopoulos, Immunology Department/IMM3, 10550 North Torrey Pines Road, La Jolla, CA 92037, USA.

Statistics from

Investigations into the structure, expression and functional status of cytokines, as well as on the possible utility of their agonists and antagonists as therapeutic agents in lupus, have received prominent attention. In fact, mouse strains predisposed to lupus (NZBxW, BXSB, MRL-lpr) have long constituted the primary resource to investigate the role of cytokines in autoimmunity. An additional impetus for such investigations in recent years has been provided by the discovery that T cells may be polarised during an ongoing immune response into the so called TH1 and TH2 subsets, which display distinct cytokine profiles and effector functions. Thus, following antigen recognition, cytokines present at the site of priming together with other factors, such as type of antigen presenting cell, amount of antigen, co-stimulatory molecules, affinity and duration of exposure direct the induction of either TH1 cells, which secrete IL2, IFNγ and TNFβ, or TH2 cells, which secrete IL4, IL5, IL6, IL10 and IL13.1 Another cytokine, IL12, produced by activated macrophages and dendritic cells, is a strong inducer of TH1 cells in which the β-subunit of its receptor is retained but lost on TH2 cells. The former cells then provide protection from intracellular pathogens, activate phagocytes, induce IgG2a antibodies, and promote DTH responses, whereas the latter cells provide protection from extracellular pathogens, activate eosinophils, induce IgE mediated allergic reactions, and generally promote humoral responses in which IgG1 predominates. The molecular events associated with this polarisation have not been fully eludicated, but certain protooncogenes, kinases and transcription factors seem to play a part.2-6

Based on this T cell division, it has been hypothesised that organ specific autoimmune diseases such as IDDM should be mediated by TH1 cells, whereas humorally mediated autoimmune diseases such as lupus should be mediated by TH2 cells. As summarised in tables 1, 2 and 3 and reviewed below, such a paradigm does not seem to apply to lupus because both TH1 and TH2 cytokines have been found to exert profound effects on spontaneous mouse models of this disease.

Table 1

In vitro and in vivo cytokine expression levels in lupus mice

Table 2

Effects of cytokine agonists and antagonists in murine lupus

Table 3

Cytokine gene knockout and transgenic lupus mice

TH1 cytokines

The earliest cytokine defect identified in all lupus strains was reduced production of, and response to, IL2.7 ,8 This defect appears at 4–6 weeks of age in MRL-lpr and BXSB mice, and somewhat later in the (NZBxNZW)F1 mice, wherein it becomes more pronounced with disease advancement. The cause(s) of this defect is unknown, but several possibilities have been considered, including impaired T cell receptor (TCR) signal transduction, IL2R structural defects, abnormalities in certain transcription factors, and exhaustion subsequent to excessive and repetitive activation in vivo.9-11 Among these possibilities, we favour that of exhaustion and attainment of a replicative senescence state. According to this hypothesis, because of continued autoantigen presence there is a repetitive interconversion of effector T cells into memory cells and vice versa. Upon a defined number of such interconversions, memory cells become refractory to further stimulation and entry into the cell cycle (fig 1). In support of this concept is the finding that the frequency of in vivo cycling T cells in lupus mice declines progressively with age, and this decline is associated with increased expression of cyclin dependent kinase inhibitors, a characteristic of cells that reach replicative senescence.12 ,13

Figure 1

Schematic representation of replicative senescence in repeatedly activated autoreactive memory T cells.

The relation of the IL2 defect to the disease process remains unclear. Several reports have shown correction of the in vitrolpr T cell defective proliferation and apoptosis by exogenous IL2.14-16 Moreover, an early study found reduced serological, cellular and histological abnormalities of the MRL-lpr mouse following infections with a vaccinia virus-IL2 construct.17 Similarly, in a very recent study,18 MRL-lpr mice infected orally by gavage with an attenuated strain ofSalmonella typhimurium transfected with the IL2 gene (administered at 6 weeks of age and repeated every three weeks to 15 weeks of age) were shown to have reduced double negative T cells, autoantibody levels, glomerulonephritis (GN) and vasculitis. In contrast, intramuscular injections of an IL2 encoding cDNA expression vector in MRL-lpr mice were reported to increase autoantibody production and disease.19 Another study20 reported no effect on disease progression and severity in (NZBxNZW)F1 mice treated with low or high doses of human recombinant IL2, while others found suppression of nephritis in (NZBxNZW)F1 mice treated with anti-IL2R mAb.21 The divergence of results in these studies may be attributable to differences in the vectors and lupus strains used as well as the route, timing and dose of IL2. Such differences may affect overall IL2 levels, its expression in the microenvironment of secondary lymphoid organs, and rates of T cell proliferation and apoptosis. Nevertheless, some studies have suggested dissociation of severe lupus disease and the in vitro IL2 deficiency.22-24 Overall, it seems that the role of IL2 in lupus disease has not yet been conclusively decided, and further investigation is required in view of the presence of autoimmunity in normal background mice with forced deletions of the IL225 or the IL2R26 ,27genes.

Among the many cytokine abnormalities found in lupus mice, the most consistent has been high expression of IFNγ.28-30 The importance of this cytokine in murine lupus pathogenesis was initially suggested by the demonstration that (NZBxNZW)F1 mice treated with IFNγ showed accelerated disease and conversely, treatment with anti-IFNγ antibody31 or soluble IFNγR32 early in life significantly delayed disease progression. A study on a long lived substrain of MRL-lpr mice also showed reduced IFNγ levels compared with the early life severe disease developing parental strain concomitant with a shift of Ig isotypes from the C-fixing IgG2a and the cryogenic-nephritogenic IgG3 to the less pathogenic IgG1 isotype.33 Moreover, MRL-lprmice intercrossed with an IFNγ gene deleted mouse,34MRL-lpr mice rendered congenic for deletions in either the IFNγ35 or the IFNγR,36 ,37and (NZBxNZW)F1 mice congenic for the IFNγR deletion38 all showed significant reduction in humoral and histological characteristics of the disease. In one of these studies,35 three notable observations were made. Firstly, hypergammaglobulimaenia was maintained in IFNγ-/- mice with a switch from IgG2a to IgG1 predominance but, importantly, the dramatic decrease in levels of the dominant IgG2a anti-dsDNA autoantibodies was not associated with a compensatory increase in TH2 associated IgG subclasses. This finding suggested that therapeutic interventions with the goal of reducing IFNγ levels in lupus may selectively affect certain pathogenic autoimmune responses without significantly compromising the person's capacity to respond to exogenous antigens. This is a considerable advantage over other contemplated treatments, such as those with co-stimulation blockers wherein severe and indiscriminate immune compromise may result. Secondly, remarkably, early death and GN were also prevented in IFNγ+/- lpr mice (50% reduction in IFNγ levels) despite the fact that autoantibody levels and kidney immune complex deposits were equal to those in the wild type MRL-lpr mice. This result suggests that even partial reduction of IFNγ may curtail the deleterious effects it exerts locally on the afflicted organ. A similar uncoupling of the local inflammatory response from autoantibody production and immune complex kidney deposition has also been observed in (NZBxNZW)F1 mice deficient in the FcγRI and FcγRIII,39 that is, the FcγR-/-(NZBxNZW)F1 mice showed similar autoantibody levels and immune complex kidney deposits to the FcγR+/+ mice, yet they exhibited prolonged survival, reduced kidney disease and proteinuria. Thirdly, with regards to mechanisms, the major observation was severe reduction of MHC class I and class II expression by splenic macrophages in IFNγ-/- mice, but not IFNγ+/- mice, whereas reduction in MHC class II expression by tubular epithelial cells was observed in both homozygous and heterozygous IFNγ deleted mice. Thus, insufficient upregulation of MHC on antigen presenting cells, either systemically and/or locally, may be the major mechanism by which the beneficial effects of reduced IFNγ levels are mediated.

The role of IL12 in murine lupus has also been investigated. An intrinsic defect in the in vitro production of IL12 by endotoxin activated macrophages of MRL-+/+ and (NZBxNZW)F1 mice has been reported.40 Other studies, however, found that peritoneal macrophages of MRL-lpr mice hyperproduce IL12 after stimulation with IFNγ and/or LPS, and exhibit high concentrations of IL12 in the serum41 as well as kidney.42 Moreover, daily injections of recombinant IL12 led to increased serum levels of IFNγ and nitric oxide (NO) metabolites, and accelerated GN in this model.41 These findings, together with previous reports that NO synthase inhibitors can ameliorate autoimmune disease in MRL-lprmice,43 suggest that the high production and response to IL12 by these mice may be important in disease pathogenesis. Treatment of (NZBxNZW)F1 mice with anti-IL12, however, was ineffective in preventing the onset or severity of GN.44

TH2 cytokines

The part played by the TH2 cytokines IL6, IL4 and IL10 in murine lupus has also been the subject of intense study. Unaltered or increased levels of IL6 have been reported.45-47Macrophage depletion in in vitro cultures of splenic cells from (NZBxNZW)F1 mice yielded reduced IL6 levels concomittant to decreased IgG anti-DNA levels.48 Furthermore, administration of recombinant IL6 to (NZBxNZW)F1 mice caused accelerated GN that correlated with marked upregulation of mesangial MHC class II antigen and glomerular ICAM-1 expression.49 Chronic administration of a rat anti-IL6 mAb had no effect on the development of GN in (NZBxNZW)F1 mice because of the elicitation of an anti-rat response, but tolerance induced against the heterologous Ig by the concurrent administration of anti-CD4 resulted in prevention of autoantibody production, reduced proteinuria and prolonged survival.50 Blockade of the IL6R by a neutralising mAb to IL6R was also reported to be beneficial in MRL-lpr mice.51

Reduced levels of IL4 have been found in MRL-lpr and (NZBxNZW)F1 mice, resulting in an increased IFNγ to IL4 ratio.30Interestingly, in contrast with the expected paradigm of systemic autoimmunity being causally related to TH2 response, GN development was completely abrogated in (NZWxC57BL/6.Yaa) F1 mice rendered transgenic for the IL4 gene under the control of the IgH enhancer,52 and this protection was associated with retention of overall titres of IgG anti-DNA autoantibodies, but a severe reduction in the nephritogenic IgG3 isotype. Others, in contrast, found that transfer of IL4 (or IL12) stimulated splenocytes from 5 month old (NZBxNZW)F1 mice into syngeneic recipients increased the production of IgG anti-dsDNA antibodies, while administration of anti-IL4 before disease onset inhibited this production.44 Moreover, anti-IL4 treatment alone prevented GN, while anti-IL12 alone was ineffectual. It is noteworthy that combined treatment with both antibodies abrogated the beneficial effect of anti-IL4.44 Finally, IL4 gene deletion34 as well as recombinant mouse IL4R or anti-IL4 mAb treatments53 led to significantly reduced lymphadenopathy and end organ disease in MRL-lpr mice. BXSB male mice congenic for the IL4 deletion, however, develop disease equally as severe as that seen in the wild type mice (D Kono, A N Theofilopoulos, in preparation), indicating that this cytokine is not an obligatory participant in the autoimmune process.

All lupus strains also show increased IL10 levels,29 and (NZBxNZW)F1 mice repeatedly injected (from birth) with anti-IL10 mAb showed substantially delayed onset of autoimmunity,54 an effect apparently mediated by upregulation of endogenous TNFα, as the anti-IL10 protection was abolished when anti-TNFα was introduced along with the anti-IL10 treatment. Conversely, administration of IL10 accelerated the onset of autoimmunity in these mice, but similar injections into normal mice had no adverse effects. The important role of this cytokine in lupus has also been suggested by the reduced autoantibody levels and kidney disease in (NZBxNZW)F1 mice treated with an IL10 inhibiting immunomodulator (AS101),55 as well as by increased Ig and autoantibody production in IL10 treated cultured peripheral blood lymphocytes (PBL) from systemic lupus erythematous (SLE) patients, and inhibition of these manifestations in anti-IL10 treated SCID mice transplanted with PBL from such patients.56 The inference has been made that IL10 promotes systemic autoimmunity by increasing Fas/FasL mediated apoptosis.57

Other cytokines

Additional cytokines have been implicated in lupus pathogenesis, including TNFα, IL1 and TGFβ. TNFα, (a product of macrophages, TH1 and TH2 cells as well as B cells) exerts wide ranging effects on immune responses and inflammation and, therefore, its role in autoimmunity has been extensively investigated. It was initially shown by Jacob and McDevitt58 that LPS activated macrophages of NZW mice produce 5-fold to 10-fold lower levels of TNFα than activated macrophages from non-autoimmune strains, and this low production was associated with a unique BamHI RFLP in the NZW TNFα gene. Studies by these and other investigators, however, showed that the same RFLP defined allele was present in MRL and BXSB lupus strains as well as several normal strains and wild mice that produced intermediate to high levels of TNFα.58 ,59More recent studies by Jacob and associates60 identified diverse mutations in the 5' and 3' untranslated region (UTR) of the TNFα gene in several strains of mice. Among them, a three base pair insertion disrupting the AU rich motif of the 3'UTR was detectable in TNFα low producing NZW, B10.KPA44, SM/J strains, and inMus spretus. Transient expression experiments in a macrophage cell line using the luciferase reporter system indicated that presence of the NZW 3'-UTR, indeed, leads to reduced TNFα production.61

Obviously low TNFα production is not required nor sufficient for induction of lupus. This is the case even in the New Zealand mice, as documented by the presence of delayed but still severe disease in (NZBxNZW.PL)F1 mice in which the TNFα haplotype is d/d—that is, without the NZW low TNFα defect.62Moreover, expression of TNFα is increased in the diseased kidneys of (NZBxNZW)F1 63 ,64 as well as MRL-lpr mice.65 ,66 In the latter strain, a biphasic increase in circulating TNFα was recorded with an initial peak in neonatal mice 703±208 pg/ml followed by normalised levels by 2 months of age 87±13 pg/ml and progressive increases thereafter that were proportional to the severity of renal disease (non-proteinuric, 570±87; proteinuric, 1255±135 pg/ml).67 While early in life TNFα was detected only in tubular epithelial cells (TEC), in adult MRL-lpr mice expression was more ubiquitous and was detected in glomeruli, perivascular infiltrating cells as well as TEC.67 Although these findings firmly establish the high in vivo TNFα levels in MRL-lpr mice, it should be noted that Beller and associates40 ,68observed defective in vitro induction of TNFα in LPS activated macrophages of the congenic non-Fas defective, mild disease manifesting MRL-+/+ mouse. This defect was noted in other lupus strains as well, and was thought to be the basis for defects in in vitro induction of other cytokines, such as IL1 and IL6. Notwithstanding these in vitro findings, the relevance of which is unclear, the fact remains that, in contrast with the NZW and (NZBxNZW)F1 mice, other lupus strains are characterised by increased TNFα levels. Several studies have also shown that human lupus is characterised by high serum levels of TNFα and soluble TNFR that parallel disease activity.69-72

Regardless of variances in TNFα levels, the initial finding of low TNFα production by macrophages of NZW mice prompted Jacob and associates to assess its possible therapeutic effects in the (NZBxNZW)F1 mouse.58 ,73 They found that TNFα replacement therapy at relatively high doses (10 μg/3 × weekly) started up to 4 months of age effectively delayed disease, but was ineffective when started at 6.5 months of age (near the 50% mortality range for these mice). Gordon et al,74 using a similar injection and dose schedule, also reported that TNFα treatment started after the onset of clinical disease led to improved survival relative to controls (92% survival rate in treated versus 42% of control at 10 months of age). They further indicated that while administration of TNFα delayed disease progression, sustained treatment did not prevent the eventual development of severe renal disease. As noted earlier, an additional finding by Ishida et al 54 was that sustained anti-IL10 treatment of (NZBxNZW)F1 mice was beneficial, and that this effect was mediated by the concurrent TNFα upregulation, as simultaneous administration of anti-TNFα antibody abolished the anti-IL10 antibody effect. Overall, the above findings indicated clearly that TNFα replacement therapy is beneficial to the (NZBxW)F1 mouse.

Additional findings, however, have indicated that the picture is much more complicated than was originally thought. Thus, Kelly and associates,63 prompted by their finding of increased TNFα expression in the diseased kidneys of the (NZBxW)F1mouse, revisited the issue of TNFα therapy for lupus and reported that (NZBxNZW)F1 female mice treated from 4 months of age with a low dose of recombinant TNFα (0.2 μg/3 × weekly) exhibited accelerated disease, while those receiving an intermediate dose (2 μg/3 × weekly) showed minor or no acceleration. Yet when the low dose treatment was started at 2 months and continued up to 4 months of age, there was no acceleration of renal disease. This latter finding was interpreted to indicate that for TNFα to cause renal injury, it must interact with other pathological features in these animals that appear after 4 months of age. Additional findings have provided further support for the notion that TNFα may, under certain circumstances, exert detrimental effects in lupus. Firstly, treatment of MRL-lpr mice with an IgG anti-TNFα antibody was reported to prevent development of pulmonary inflammatory lesions such as lung fibrosis and alveolitis.75 Secondly, mice rendered defective for the zfp36 gene, which encodes tristetraprolin, develop systemic autoimmunity apparently mediated by an attendant upregulation of TNFα, as antibodies to TNFα abrogated this syndrome76; tristetraprolin was recently shown to be a component of a negative feedback loop that interferes with TNFα production by destabilising its mRNA.77 Thirdly, treatment with soluble, dimeric TNFR led to reduction of disease in systemic autoimmunity manifesting motheaten mice,78 and treatment with a novel transcriptional inhibitor of TNFα reduced superantigen induced inflammatory arthritis in MRL-lpr mice.79

How can one reconcile the apparent dichotomous results reviewed above in which some studies reported beneficial and others detrimental effects of TNFα in murine lupus? On one hand, it could be argued that low doses of TNFα or short duration of treatment may promote autorecognition by, for example, upregulating MHC80 ,81 or by increasing membrane expression of autoantigens.82 In contrast, high doses or chronic exposure to TNFα may inhibit autoimmunity by a variety of specific and non-specific immunosuppressive effects. With regard to non-specific immunosuppressive effects, studies by Gordon and Wofsy83have shown that mice treated with high doses of TNFα exhibit severe lymphopenia, while, with regard to specific effects studies by Copeet al 84 ,85 in TCR transgenic animals showed that prolonged exposure to TNFα caused severe suppression in a broad range of T cell responses, including proliferation and cytokine production. The possible relevance of this immunosuppression has been suggested by the finding that a small, but significant, percentage (6–7%) of RA patients treated with anti-TNFα mAb developed anti-dsDNA autoantibodies.86Diverse effects of TNFα are obviously not confined to lupus, but seem to apply to other autoimmune disorders. Thus, TNFα treatment has been shown to inhibit or promote diabetes in NOD mice depending on the time of treatment initiation,81 while anti-TNFα mAb treatment81 as well as high transgenic expression of soluble TNFRp55-FcIgG3 fusion molecules87 ,88 have been shown to avert disease. Apparently, because of reported beneficial and detrimental effects of TNFα in autoimmunity, terms such as “Trojan horse”89 and “pretty girl or old witch”90 have been coined to illustrate the unpredictable effects of this molecule.

Increased IL1 levels have been described for all lupus strains.29 ,91 ,92 Recombinant IL1 given to (NZBxNZW)F1 mice increased nephritis,63 while recombinant IL1R given to MRL-lpr mice inhibited GN, splenomegaly/lymphadenopathy and autoantibody levels in one study,93 while in another study, an IL1R antagonist had no effect.94 Surprisingly, in vitro experiments showed that IL1 and IL1R antagonists both induced significant suppression in IgG production by B cells derived from diseased MRL-lpr mice, but they had no effect on B cells from young animals.92 ,95 ,96

Male BXSB and MRL-lpr mice show increased levels of TGFβ,29 which was shown in MRL-lpr mice to adversely affect host defence against both Gram negative and positive bacterial infections because of the failure of initial polymorphonuclear leucocyte migration to the infection site.97 ,98 The role of increased TGFβ levels in this defect was directly demonstrated by the fact that this abnormality was duplicated in MRL-+/+ mice injected with TGFβ at the time of bacterial infection, and lethality of infected MRL-lpr mice was ameliorated by administration of anti-TGFβ mAb.97 Such findings suggest that increased TGFβ production in lupus may suppress host defence mechanisms against bacterial infection, thereby providing an explanation for the increased risk of such infections in SLE patients. Nevertheless, direct injections into the skeletal muscle of a TGFβ cDNA expression vector was reported to reduce autoantibody levels in MRL-lpr mice,19 while infection with a non-pathogenic strain of Salmonella typhimurium carrying the TGFβ gene was without effect.18


The summarised findings on the role of cytokines in murine lupus cast doubt on the widely held view that TH2 cytokines play the primary part in this humorally mediated autoimmune disease. It would appear that both TH1 and TH2 cytokines are involved, and that agonists and antagonists can disturb these cross talking classes of cytokines to exert disease inhibiting or promoting effects without any dogmatic predictability and in a far more complex manner than the simple TH1 versus TH2 dualism dictates. It is also evident that the effects of a given cytokine agonist or antagonist in the lupus models frequently differ from one study to another. These differences can be accounted for on the basis of variables in study design, including administration, dose, and timing of treatment initiation. Therefore, caution should be exercised in generalisations derived from one or another experiment wherein beneficial or detrimental effects are observed. Obviously, further work is needed before we fully comprehend the complex interplay between these molecules and confidently design cytokine treatments for human lupus.


This is Publication No 12324-IMM from the Department of Immunology, The Scripps Research Institute, La Jolla, California. The work of the authors herein was supported, in part, by NIH grants AR31203, AR39555 and AG15061.


Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.