Main

Apoptotic death of autoreactive lymphocytes constitutes a major mechanism of immunological tolerance. Defects in the deletion of autoreactive T and B lymphocytes, either in primary lymphoid organs1, 2, 3 or in the periphery,2 can lead to autoimmune disease. The ‘death receptor’ FAS (CD95, APO-1), and its ligand FASL (CD95L, APO-1L), has critical roles in the removal of autoreactive lymphocytes in the periphery,4 and in contraction of immune responses and lymphocyte homeostasis.5 Accordingly, loss-of-function mutations in these genes cause lymphadenopathy and systemic lupus erythematosus (SLE)-like multiorgan autoimmune disease in mice6 and humans.7

FAS, which is expressed on the surface of many cell types, allows FASL-mediated killing of pathogen-infected or damaged cells,4 whereas its ligand (FASL) is thought to be present only on activated T cells and NK cells.4 Upon deposition in the plasma membrane, membrane-bound FASL (mFASL) can also be processed and shed as secreted FASL (sFASL).4 The generation of mice that lack mFASL but express sFASL (FasLΔm/Δm) demonstrated unambiguously that only the former is essential for FAS-mediated cell killing, activation-induced T-cell death (AICD) and lymphocyte homeostasis.8 Although FasLΔm/Δm mice lack functional mFASL, they can still produce sFASL, which can engage FAS, an interaction postulated to trigger non-apoptotic signals,4, 9 which promote inflammation and cytokine production.8, 10 Consistent with this hypothesis, FasLΔm/Δm mice develop SLE-like kidney disease, autoimmune allergic dermatitis and histiocytic sarcoma with higher incidence and accelerated rate compared with FasLgld/gld mice,8 which carry a mutation that impairs the function of both mFASL and sFASL.

The onset of disease in FasLΔm/Δm mice is preceded by an elevation in immune cells with activated nuclear factor-κB (NF-κB) that coincides with elevated serum levels of NF-κB-regulated inflammatory cytokines.8 The NF-κB family of transcription factors, comprising homodimers and heterodimers of the five related proteins RELA, RELB, c-REL, NF-κB1 (p105/p50) and NF-κB2 (p100/p52), regulates the development, survival and function of innate and adaptive immune cells.11, 12 Different NF-κB dimers retained in a latent cytoplasmic state by various IκB inhibitory proteins are differentially activated by the canonical and non-canonical arms of the NF-κB pathway in response to distinct upstream signals (Supplementary Figure S1).13 Antigen receptors and inflammatory cytokines, such as TNF, IL-1β and IL-6, are among the many signals that activate the canonical NF-κB transcription factors (dimers of c-REL or RELA and p50/NF-κB111, 12) through IKKβ-dependent proteasome-mediated degradation of IκB inhibitory proteins (Supplementary Figure S1). Conversely, certain TNF receptor family members (e.g., BAFF-R) trigger the non-canonical NF-κB pathway through IKKα-induced proteolysis of p100/NF-κB2 bound to RELB, thereby generating dimers of p52/NF-κB2 and RelB (Supplementary Figure S1).11, 12 Canonical and non-canonical NF-κB transcription factors generally control different sets of genes,14 although a subset of genes is co-regulated by the two arms of the NF-κB pathway.

Abnormally elevated NF-κB activity features in various human autoimmune diseases, including SLE and rheumatoid arthritis,12, 15, 16 where it is thought to drive pathology by regulating the proliferation, survival and effector activities of immune cells.16, 17, 18, 19 The functions that NF-κB-regulated transcriptional programs exert in the immune system, combined with our findings that leukocytes in FasLΔm/Δm mice have abnormally elevated NF-κB activity before disease onset, led us to investigate whether NF-κB signaling is critical for this disease. Due to the profound deleterious consequences associated with systemic NF-κB inhibition,20 we focused on investigating the impact of selective loss of c-REL or NF-κB2 on autoimmune pathology in FasLΔm/Δm mice. These NF-κB transcription factors were chosen because of their selective function in the canonical or non-canonical NF-κB pathways, and because they are mainly required for immune cell function and not for embryogenesis or hematopoietic cell development. Mice lacking c-REL, which is mainly expressed in hematopoietic cells,21 have no major abnormality but exhibit defective activation of B and T lymphocytes, macrophages and DC that collectively contribute to reduced antibody and cytokine production.12, 22, 23 In nfkb2−/− mice, T-cell and myeloid cell development and functions are intact, but reduced mature B-cell numbers due to their impaired survival, combined with germinal center defects, result in reduced humoral immune responses.24, 25 Interestingly, with age, nfkb2−/− mice exhibit multiorgan leukocyte infiltrates and mild autoimmune disease.18, 26 Our study demonstrates that loss of c-REL prevents autoantibody production in FasLΔm/Δm mice and thereby substantially prolongs their lifespan. This suggests that selective targeting of c-REL may represent a promising strategy for the treatment of systemic autoimmune diseases.

Results

Loss of c-REL extensively prolongs the lifespan of FasLΔm/Δm mice, whereas loss of NF-κB2 has only a modest impact

To investigate the roles of the canonical or non-canonical NF-κB signaling pathways in the lymphadenopathy and autoimmune disease caused by defects in FASL, we intercrossed FasLΔm/Δm animals with c-rel−/− or nfkb2−/− mice. Both the lifespan and pathology of the FasLΔm/Δmc-rel−/− and FasLΔm/Δmnfkb2−/− compound mutants were compared with cohorts of wt, FasLΔm/Δm, c-rel−/− and nfkb2−/− mice for a period of 2 years. Although the absence of NF-κB2 or c-REL alone causes various immune defects,12, 18 these animals had lifespans comparable to wt mice (wt versus nfkb2−/− and wt versus c-rel−/− both P=NS, Figure 1a). Loss of NF-B2 only modestly (P=0.0344) prolonged the lifespan of FasLΔm/Δm mice, whereas this was greatly extended by loss of c-REL (P<0.0001, Figure 1a).

Figure 1
figure 1

Loss of NF-κB2 or c-REL both prolonged the lifespan of FasLΔm/Δm mutant mice and delayed or even prevented fatal SLE-like GN in these animals. (a) Kaplan–Meyer survival curves for wt (n=54, mean survival 860 days, blue line), FasLΔm/Δm (n=174, mean survival 407 days, red line), nfkb2−/− (n=43, mean survival 750 days, dotted black line), FasLΔm/Δmnfkb2−/− (n=54, mean survival 434 days, solid black line), c-rel−/− (n=46, mean survival 883 days, dotted green line) and FasLΔm/Δmc-rel−/− (n=40, mean survival 703 days, solid green line) mice (wt versus FasLΔm/Δm: P<0.0001; FasLΔm/Δm versus FasLΔm/Δmnfkb2−/−: P=0.0344; FasLΔm/Δm versus FasLΔm/Δmc-Rel−/−: P<0.0001; wt versus nfkb2−/−: P=NS; wt versus FasLΔm/Δmc-Rel−/−/: P=NS). (b) Graphical representation of the incidence of severe autoimmune kidney disease (GN score ≥3) in mice of the indicated genotypes (FasLΔm/Δm versus FasLΔm/Δmnfkb2−/−: P<0.05, FasLΔm/Δm versus FasLΔm/Δmc-rel−/−: P=0.0002, lines colored as for (a), wt (n=4), FasLΔm/Δm (n=44), nfkb2−/− (n=22), FasLΔm/Δmnfkb2−/− (n=40), c-rel−/− (n=17), FasLΔm/Δmc-rel−/− (n=56)). (c) Representative H&E-stained sections of the kidneys from mice of the indicated genotypes were examined for pathological changes, such as hypercellularity, cellular crescents, dilated tubules or sclerotic glomeruli. Arrows indicate fibrinoid necrosis. Magnification × 40. (d) Representative confocal photomicrographs of frozen sections (eight kidneys per genotype analyzed) stained for the presence of IgA-, IgG- or IgM-containing immune complexes (green) in glomeruli. Nuclei are revealed by staining with DAPI (blue). Arrows indicate Ig deposition on the glomerular basement membrane (scale bars represent 25 μm)

Full size image

The morbidity and premature death of FasLΔm/Δm mice are due to lymphadenopathy and SLE-like autoimmune glomerulonephritis (GN, Figure 1b).8 Histological examination (Figure 1c) and immunostaining for glomerular immune complex deposition indicated that the incidence of severe autoimmune kidney damage was significantly reduced in both FasLΔm/Δmc-rel−/− and FasLΔm/Δmnfkb2−/− mice, and the damaging immunoglobulin deposition on the glomerular capillary loops (Figure 1d, arrows) in these animals was now mainly confined to mesangial cells (Figure 1d). Impairment of renal function (pathology grading 3–4) was modest and delayed in FasLΔm/Δmnfkb2−/− mice (FasLΔm/Δmnfkb2−/− versus FasLΔm/Δm, P<0.05). Loss of c-REL had a more profound impact on GN-associated autoimmune pathology (P=0.0002), manifest as substantial reductions in hypercellularity, capillary loop obliteration and fibrinoid necrosis (Figures 1b and c).

NF-κB2 and c-REL are required for inflammatory cytokine production in FasLΔm/Δm mice

Before the onset of SLE-like autoimmune disease (~5 months), FasLΔm/Δm mice harbor high levels of circulating inflammatory cytokines.8 We examined the levels of 31 cytokines and chemokines and found that TNF, IL-17A/F, KC (CXCL1) and MIP-1β levels were elevated in FasLΔm/Δm mice at 5 months of age (Figure 2a), whereas IFNβ remained largely undetectable by ELISA (Supplementary Figure S2), although type I interferons have been implicated previously in the GN in Faslpr/lpr mutant mice.27 NF-κB2 deficiency blunted the excess production of most cytokines (except KC) in FasLΔm/Δm mice (Figure 2a). As FasLΔm/Δm mice aged, IL-1β and IL-6 levels increased significantly, whereas in aged FasLΔm/Δmnfkb2−/− mice (283–683 days) all cytokines tested, with the exception of IL-17A/F and KC, remained at wt levels (Figure 2b). Remarkably, none of the cytokines or chemokines tested were elevated in FasLΔm/Δmc-rel−/− mice, even when aged (Figures 2a and b). This indicates that c-REL and, to a lesser extent, NF-B2 are required for accumulation of the cytokines and chemokines associated with autoimmune pathology in FasLΔm/Δm mice.

Figure 2
figure 2

Impact of loss of c-REL or NF-κB2 on the levels of pro-inflammatory cytokines and chemokines in FasLΔm/Δm mice. (a) The levels of 31 cytokines and chemokines (indicated in the Materials and Methods) were measured in the sera of mice of the indicated genotypes by the multiplex system at 5 months when animals still appeared healthy or (b) when they were terminally ill or aged. The levels of IL-1β, IL-6, 1L-17F, CCL4, TNF-α and CXCL1 are shown. Each dot represents a single mouse; the bar indicates the average; the error bars represent S.E.M.

Full size image

NF-κB2 restrains lymphadenopathy in FasLΔm/Δm mice

Defects in AICD of T cells are a major cause of lymphadenopathy in mice and humans deficient in FASL/FAS function and are associated with the accumulation of ‘unusual’ TCRαβ+CD3+CD4−CD8−B220+ (hereafter called DN) T cells.6, 7 FasLΔm/Δm mice display lymphadenopathy and splenomegaly commencing from ~3 months of age (Figures 3a and b). Surprisingly, loss of NF-κB2 greatly accelerated lymphadenopathy in FasLΔm/Δm mice (Figures 3a and b) with a fourfold increase in DN T cells compared with FasLΔm/Δm controls (Figure 3c). By 3 months, these DN T cells comprised the majority of leukocytes in the lymph nodes (81.9±2.8%) and spleens of FasLΔm/Δmnfkb2−/− mice (76.7±1.1%, Figure 3c), resulting in a >10-fold increase in total lymph node and splenic cellularity compared with FasLΔm/Δm mice (Figure 3e). In contrast, loss of c-REL had no impact on the onset or magnitude of lymphadenopathy and DN T-cell numbers in FasLΔm/Δm mice (Figures 3a–e). To assess whether the enhanced lymphadenopathy of FasLΔm/Δmnfkb2−/− mice was intrinsic to the hematopoietic system, bone marrow chimeras were generated. C57BL/6-Ly5.1 (wt) mice reconstituted with hematopoietic stem/progenitor cells (HSPCs) from FasLΔm/Δmnfkb2−/− mice exhibited enhanced lymphadenopathy and elevated numbers of DN T cells compared with mice reconstituted with FasLΔm/Δm HSPCs (Supplementary Figure S3a and c). This indicates that loss of NF-κB2 in hematopoietic cells was sufficient to enhance the lymphadenopathy caused by defects in FASL.

Figure 3
figure 3

Loss of NF-κB2 accelerated and enhanced lymphadenopathy and splenomegaly in FasLΔm/Δm mice and loss of NF-κB2 and c-REL both reduced hypergammaglobulinaemia and ANA in these animals. (a) Weights of the lymph nodes (axillary, brachial, inguinal and mesenteric) and spleens from mice of the indicated genotypes at the ages indicated. (b) Representative photomicrographs of the lymph nodes (axillary, brachial, inguinal and mesenteric) and spleen from mice of the indicated genotypes, all at ~6 months of age. Scale bar, each line represents 1 mm. (c) The percentages of the ‘unusual’ DN T cells in the lymph nodes of mice of the indicated genotypes were measured by FACS analysis. In a and c, each dot represents a single mouse and the figure legend in c also applies to a. (d) Representative flow cytometric images of the DN T cells (numbers indicate percentage of cells in each quadrant). (e) Graph of total cell numbers for the indicated organs and the indicated genotypes of mice at 3 months of age (n=3–4 mice/genotype). (f) The levels of antibodies of the different Ig isotypes in the sera of the mice of the indicated genotypes (age 5 months). Data represent mean±S.E.M. (n=6/genotype). (g) ANA levels in the sera of mice of the indicated genotypes (5 months) were quantified by staining of human HEp-2 epithelial cells and scored according to a scale of fluorescence intensity on a scale of 0 (no fluorescence) to 3+ (maximal fluorescence). (h) Graphical representation of ANA levels in the sera of mice of the indicated genotypes quantified by ELISA (n=7–12 mice/genotype). (i) Pictorial examples of ANA quantification in the sera of mice of the indicated genotypes; arrows indicate staining of nuclear components in mitotic figures. Immunofluorescence intensity score is indicated in brackets; ANA (green) and nuclei (blue) are revealed by staining with DAPI. Positive control sera were from Fasgld/gld mice. Values in graphs represent mean±S.E.M., *P<0.05, **P<0.001, ***P<0.0001

Full size image

Loss of c-REL or NF-κB2 reduces total antibody and autoantibody levels in FasLΔm/Δm mice

FasLΔm/Δm mice have elevated levels of antibodies of all Ig isotypes (Figure 3f and O'Reilly et al.8). In contrast, mice lacking NF-κB2 or c-REL have reduced levels of Ig isotypes requiring Ig heavy chain class switching (Figure 3f and Franzoso et al.25, Köntgen et al.28). With the exception of IgA in FasLΔm/Δmnfkb2−/− mice, loss of either NF-κB2 or c-REL in FasLΔm/Δm mice reduced all Ig isotype levels to those of wt mice (Figure 3f). High titers of antinuclear autoantibodies (ANA) producing a homogenous, nuclear staining pattern are a hallmark of SLE and have been observed in FasLΔm/Δm mice (Figures 3g–i and O'Reilly et al.8). Although the absence of NF-κB2 significantly reduced ANA levels in FasLΔm/Δm mice (P<0.005; Figures 3g–i), there was no difference in the Ig isotypes among the ANA (Supplementary Figure S4). Remarkably, the sera of FasLΔm/Δmc-rel−/− mice remained ANA negative (Figures 3g–i).

To determine whether lymphocytic tissue infiltration, another feature of the autoimmune pathology in FasLΔm/Δm mice,8 was associated with the production of organ-specific autoantibodies, we assessed the reactivity of sera from sick animals against a panel of tissues from rag-1−/− mice. Sick FasLΔm/Δm mice harbored autoantibodies (IgM, IgA or IgG) against lung epithelia, thyroid follicles, gastric parietal cells, submandibular gland and pancreatic acinar cells (Figure 3f). No such autoantibodies were detected in the sera from similarly aged wt or c-rel−/− mice. Sera from nfkb2−/− mice stained certain organs at a low level, and, although FasLΔm/Δmnfkb2−/− animals generally had lower autoantibody levels than FasLΔm/Δm mice, the persistent elevation of gastric autoantibodies may be due to the sustained serum IgA levels (Figure 3f). FasLΔm/Δmc-rel−/− mice had a substantial reduction in IgM, IgA or IgG autoantibodies, particularly those directed against the submandibular gland and thyroid, compared with FasLΔm/Δm animals (Supplementary Figure S5,Table 1).

Table 1 Percentage of mice with organ-specific autoantibodies against the tissues indicated
Full size table

The roles of NF-κB2 and c-REL in autoimmune pathology in FasLΔm/Δm mice

To further understand how NF-κB2 and c-REL influence the early phase of autoimmune pathology in FasLΔm/Δm mice, we examined the lungs, liver, salivary glands and pancreas at 5 months. Although lung disease was not apparent in FasLΔm/Δm mice, severe peribronchial and perivascular lymphocytic infiltrates were prominent in FasLΔm/Δmnfkb2−/− mice and constituted the main pathology (Figures 4a and b). The infiltrates were predominantly composed of DN T cells with a polyclonal TCRβ variable chain repertoire (Supplementary Figure S6), demonstrating their nonmalignant status.

Figure 4
figure 4

FasLΔm/Δmnfkb2−/− mice developed more severe lymphoproliferative disease compared with FasLΔm/Δm mice. (a) Representative photomicrographs of the lung, pancreas and salivary glands from mice of the indicated genotypes (5 months); arrows indicate areas of infiltration; n=4/genotype. (b) Summary (mean±S.E.M.) of the incidence of lymphocytic infiltration into the indicated organs of mice of the indicated genotypes (5 months of age), graded 0–3+, 3–5 mice/genotype; ***P<0.0001. (c) Graphical depiction (mean±S.E.M.) of the combined scores for lymphocytic infiltration (lung, liver, pancreas, salivary and stomach), each scored 0–3+ from the indicated mice when sick/aged, from wt (555±176 days), FasLΔm/Δm (436±98 days), nfkb2−/− (627±182 days), FasLΔm/Δmnfkb2−/− (405±181 days), c-rel−/− (790±84 days) and FasLΔm/Δmc-rel−/− (603±116 days) mice, ***P<0.0001, n=number of mice analyzed. (d) Expression of the activation marker CD69 on CD4+ T cells (percentage of positive cells indicated) from the indicated organs from FasLΔm/Δmnfkb2−/− mice (red line) and wt mice (black line); n=3/genotype. (e) Representative photomicrographs depicting necrotizing vasculitis in mice of the indicated genotypes. Arrows indicate areas of necrotizing vasculitis; *indicates vasculature. Magnification × 20. (f) Graphical representation of the percentages of mice of the indicated genotypes with autoimmune dermatitis (ears and neck); n= number of mice per genotype analyzed. (g) Serum IgE levels measured by ELISA when the animals were afflicted with dermatitis or otherwise sick. (h) Representative images, depicting autoimmune dermatitis either macroscopically or microscopically. Magnification × 10 or × 40; scale bar 128 μm

Full size image

Unlike sick FasLΔm/Δmnfkb2−/− mice, which exhibited extensive lymphoid infiltrates and damage in the lungs, liver, pancreas and salivary glands, similarly aged sick FasLΔm/Δm (P<0.0001) and FasLΔm/Δmc-rel−/− mice (P<0.0001) displayed milder pathology (Figure 4c). Aged or sick nfkb2−/− mice only exhibited mild organ leukocyte infiltration (nfkb2−/− versus wt, P<0.0001; nfkb2−/− versus c-rel−/− mice, P<0.05). Many of the CD4+8− and CD4−8+ T cells from the organs of aged or sick FasLΔm/Δmnfkb2−/− mice were CD69+ (Figure 4d), indicating recent activation, possibly by self-antigens. Sick FasLΔm/Δmnfkb2−/− mice (but none of the animals of the other genotypes) also suffered from multiorgan necrotizing vasculitis, an inflammatory disease characterized by damaged vessel walls, involving the kidney and occasionally other organs (Figure 4e,Supplementary Table S1).

FasLΔm/Δm mice develop autoimmune dermatitis (32%), which presents as reddening, thickening and scabbing of the ears and neck (Figure 4h and O'Reilly et al.8). Interestingly, FasLΔm/Δmnfkb2−/− mice did not develop autoimmune dermatitis, whereas FasLΔm/Δmc-rel−/− mice showed a similar incidence of this disease (39%), albeit with delayed onset compared with the FasLΔm/Δm animals (660±134 days versus 435±86 days) (Figures 4f and h). Dermatitis in FasLΔm/Δm and FasLΔm/Δmc-rel−/− mice was associated with elevated serum IgE levels (Figure 4g and O'Reilly et al.8), a feature shared with humans suffering from this autoimmune pathology.29 These findings show that loss of c-REL merely delayed autoimmune dermatitis in FasLΔm/Δm mice, whereas loss of NF-κB2 abrogated this pathology.

Regulatory T cells (Tregs) have a key role in the suppression of autoimmunity and modulation of the intensity and quality of immune responses to foreign antigens.30 Of relevance to this study, reductions in Treg numbers or impaired Treg function have been implicated in murine and human autoimmune dermatitis.31, 32 To ascertain whether this was the case in FasLΔm/Δm mice and whether this was influenced by the absence of NF-κB2 or c-REL, Tregs (CD4+CD25+Foxp3+) were enumerated in lymphoid tissues. Treg numbers in FasLΔm/Δm and nfkb2−/−26 mice were normal (Figures 5a and b) and even elevated in the FasLΔm/Δmnfkb2−/− mice (Figure 5b). However, consistent with a previous report33 showing that loss of c-REL resulted in a substantial reduction in Tregs, both the percentages and absolute numbers of Tregs in FasLΔm/Δmc-rel−/− mice were markedly reduced compared with FasLΔm/Δm and wt animals (Figures 5a and b). These results, combined with the impact that NF-κB2 or c-REL loss had on the phenotype in FasLΔm/Δm mice, indicate that while Treg defects are not likely to be critical for the development of autoimmune pathology in these animals reduction in Treg numbers may explain in part the dermatitis seen in the FasLΔm/Δmc-rel−/− mice.

Figure 5
figure 5

CD4+ Tregs were not a limiting factor in the autoimmune disease of FasLΔm/Δm mice. (a) The percentages of Treg (CD4+CD25+Foxp3+) were determined by flow cytometric analysis of the thymus, spleen and lymph nodes of mice of the indicated genotypes. Each dot represents a single mouse (n=3–4/genotype); bars indicate the average. (b) Absolute numbers of Tregs (CD4+CD25+Foxp3+) determined by flow cytometry (n=4/genotype at 3 months of age). The legend in b also applies to a. Error bars represent S.E.M., *P<0.05, **P<0.001, ***P<0.0001

Full size image

NF-κB2 but not mFASL or c-REL is required for the development of AIRE+ mTECs

The epithelial components of the thymic stroma have critical roles in promoting central tolerance to tissue-specific antigens.34 A reduction in the number or function of medullary thymic epithelial cells (mTECs) is linked to T-cell-dependent inflammatory diseases triggered by autoreactive T cells that escape negative selection.34, 35 Both the canonical and non-canonical NF-κB pathways have been implicated in tolerogenic processes mediated by the thymic stroma,36 with the non-canonical pathway reported to be important for the differentiation and/or survival of mTECs.34 NF-κB2 deficiency was associated with a loss of CD80+ mTEC, but not the reduction of mTECs that express the autoimmune regulator (AIRE).18 We conducted a detailed analysis of the thymic microenvironment to examine whether mTEC defects might be responsible for the organ-specific autoimmunity observed in FasLΔm/Δm mice or for the alterations in the FasLΔm/Δm disease profile exerted by loss of NF-κB2 or c-REL (summarized in Table 2).

Table 2 Summary of the effects that loss of NF-κB2 or c-REL exerts on abnormalities in FasLΔm/Δm mice
Full size table

Thymi from 6- or 20-week-old FasLΔm/Δm mice stained for mTEC markers (keratin-5 (K5), UEA-1 and AIRE) displayed a medullary architecture comparable to that of age-matched wt mice, although the levels of UEA-1 may be lower (Figures 6a and b). The expression of mTEC markers in the thymi of c-rel−/− and FasLΔm/Δmc-rel−/− mice was also normal (Figures 6a and b). In contrast, thymi from nfkb2−/− and FasLΔm/Δmnfkb2−/− mice exhibited reduced UEA-1 expression (from 6 weeks) and an abnormal distribution of K5+ mTEC (from 20 weeks, Figures 6a and b). Flow cytometric analysis indicated that the total numbers of epithelial cells were similar across all genotypes (Figure 6c) and the total numbers of cortical TEC (cTEC) and mTEC subsets expressing high or low levels of MHCII (mTEChi and mTEClow) were comparable between wt, FasLΔm/Δm, c-rel−/− and FasLΔm/Δmc-rel−/− mice (Figure 6d). However, the numbers of mTEChi cells were substantially diminished in thymi of nfkb2−/− and FasLΔm/Δmnfkb2−/− mice (Figure 6e), reflected by an almost complete absence of AIRE+ TECs (~10% of wt numbers). Interestingly, FasLΔm/Δm mice also displayed elevated CD4-CD8+ thymocyte numbers (Figure 6f). In summary, although FasLΔm/Δm mice display many features of autoimmune disease, they have a normal TEC composition, indicating that mFASL has no major role in TEC development and survival. Loss of AIRE expressing mTECs, however, correlates with impaired negative selection and multiorgan autoimmunity34 and this may provide an explanation for the mild age-related autoimmunity in nfkb2−/− mice18, 26 and the exacerbated organ-specific lymphocyte infiltration observed in FasLΔm/Δmnfkb2−/− animals.

Figure 6
figure 6

FasLΔm/Δm mice had normal mTEC numbers with normal AIRE expression but loss of NF-κB2 caused abnormalities in the thymic stroma. (a) Representative confocal images of thymi from mice of the indicated genotypes at 6 weeks. Sections were stained for UEA-1 (green), AIRE (red) and keratin-5 (blue); n=3/genotype/time point. (b) Representative confocal images of thymi from mice of the indicated genotypes at 20 weeks as for (a). (c) Graphical representation of total thymic cellularity from mice of the indicated genotypes at 20 weeks of age; n=4/genotype. (d) Graphical depiction of flow cytometric analysis of TECs from mice of the indicated genotypes at 20 weeks of age, as defined; mTEChi (CD45− Ep-CAM+class II MHChi Ly51−), mTEClow (CD45− Ep-CAM+class II MHClow Ly51−), cTEC (CD45− Ep-CAM+class II MHC+ Ly51+); n=4/genotype. (e) Graphical depiction of flow cytometric analysis of mTEChiAIRE+ mTECS (CD45− Ep-CAM+class II MHChi Ly51− AIRE+) from mice of the indicated genotypes at 20 weeks of age. (f) Percentages of thymocytes that are CD4 single positive (CD4+ CD8−) or CD8 single positive (CD4−CD8+) from mice of the indicated genotypes at 3 months of age; n=4/genotype. *P<0.05, **P<0.005

Full size image

Discussion

The many roles of canonical and non-canonical NF-κB signaling in the immune system identify constituents of these pathways as potential therapeutic targets for inflammatory and autoimmune diseases. Indeed, the immunosuppressive and anti-inflammatory activity of glucocorticoids, a treatment of choice for many autoimmune diseases, is at least in part due to attenuated NF-κB signaling.20 However, as NF-κB signaling regulates a wide variety of biological processes both within and outside the immune system, prolonged blockade of overall NF-κB activity, not surprisingly, can cause severe adverse effects.37 Accordingly, we assessed whether targeting individual NF-κB transcription factors would be a superior strategy for treating autoimmune disease. FasLΔm/Δm mice were used as a model to determine the impact that the loss of c-REL (canonical pathway) or NF-κB2 (non-canonical pathway) had on their multifaceted autoimmune pathology. The largely hematopoietic restricted expression of c-REL21 offers the added advantage of primarily targeting immune functions.22 Notably, targeting the canonical pathway in T cells using a proteasome degradation resistant form of IκBα(IκB) afforded only minor protection from autoimmune disease in FasLgld/gld mice.38 However, given that NF-κB signaling was only diminished, not abrogated, combined with the essential loss of FAS function in B cells in this autoimmune disease,39 the full impact of completely inhibiting a particular NF-κB transcription factor on autoimmune disease caused by defects in FASL/FAS remains unclear. Our study is the first detailed, long-term analysis of the roles that specific NF-κB family members have in the autoimmune disease caused by defects in FASL/FAS.

With respect to non-canonical NF-κB signaling, loss of NF-κB2 had contrasting impacts on different aspects of the FasLΔm/Δm autoimmune disease. Autoantibody-mediated pathologies (dermatitis and severe GN) were reduced but the lymphoproliferative disease was enhanced, causing respiratory difficulties and multiorgan vasculitis. This establishes that NF-κB2-dependent signals restrain proliferation and/or survival of the lymphocytes that accumulate to excess as a consequence of defective FASL/FAS-mediated apoptosis. Our studies using chimeric mice indicate that NF-κB2 activity within the hematopoietic system is mainly responsible for restraining the proliferation and/or survival of these T lymphocytes, although non-hematopoietic functions regulated by NF-κB2 may contribute. The NF-κB2 precursor p100, a protein with IκB-like activity,40 limits NF-κB-dependent T-cell proliferation by preventing the nuclear activation of p50-NF-κB1/p65 heterodimers.41 The absence of this inhibitory p100-NF-κB2 protein may be a mechanism that contributes to the increased accumulation of T cells in FasLΔm/Δmnfkb2−/− mice. The escape of T cells with higher affinity autoreactive TCRs due to impaired negative selection as a consequence of reduced numbers of AIRE+ mTECS might also contribute to the enhanced accumulation of DN T cells in FasLΔm/Δmnfkb2−/− mice. Finally, the reduction in autoantibody-mediated aspects of FasLΔm/Δm autoimmune disease associated with NF-κB2 loss can be explained by reduced BAFF-R survival signals in B cells and/or germinal center defects, both of which involve p52-NF-κB2/RELB dimers.19 Overall, these findings suggest that, while systemic therapeutic targeting of NF-κB2 could alleviate autoantibody-induced pathology, this benefit would be negated by enhanced lymphoproliferative disease.

Our study also provides insight into the role of NF-κB2 in autoimmune dermatitis. Unlike FasLΔm/Δm mice that develop IgE-dependent autoimmune dermatitis, the FasLΔm/Δmnfkb2−/− mice were protected. Consistent with NIK, the apical kinase in the non-canonical NF-κB2-dependent pathway42 being critical for isotype switching to IgE,42 our aged nfkb2−/− mice had ~100-fold lower serum IgE levels compared with wt controls, and even in FasLΔm/Δmnfkb2−/− mice the serum IgE levels were still below those found in wt mice. Given that topical delivery of NF-κB decoy oligodeoxynuclotides alleviates murine atopic dermatitis,43 our findings indicate that topical NF-κB2 inhibitors may represent a promising strategy for treating IgE-dependent autoimmune dermatitis. In contrast to a prior report, which showed that isotype switching to IgE was c-REL dependent,44, 45 we found that IgE levels were not markedly reduced by loss of c-REL on either a wt or a FasLΔm/Δm background. The reason for these differences remains to be determined.

This report demonstrates that c-REL loss substantially prolongs the lifespan of FasLΔm/Δm mice by preventing autoantibody-mediated pathology. Although c-REL is critical for normal T-cell activation through a combination of T-cell intrinsic and extrinsic processes,12, 22, 28 lymphadenopathy was unaffected by loss of c-REL. This indicates that the accumulation of DN T cells is driven by other NF-κB proteins or is NF-κB independent. This disjunction between extended lifespan despite persistence of the lymphoproliferative component of the autoimmune disease elicited by defects in FASL/FAS is consistent with studies of gene-targeted mice lacking FAS in select cell populations.39 Loss of c-REL has also been shown to prevent pathology in mouse models of collagen-induced arthritis,46 streptozotocin-induced diabetes47 and autoimmune encephalomyelitis.48 These outcomes can be explained by the involvement of Th1 cytokines and/or IL-17 in these diseases, because their production and the differentiation of Th1,49 as well as Th1748 effector T cells, are c-REL-dependent. The fact that c-REL is also critical for cytokine production by macrophages, dendritic cells and NKT cells12 is also likely to contribute to the reduced pathology in FasLΔm/Δmc-rel−/− mice. Although we were unable to detect IFNβ in the sera of sick FasLΔm/Δm, FasLΔm/Δmnfkb2−/− and FasLΔm/Δmc-rel−/− mice, others have reported that introduction of a loss-of-function mutation of the type I interferon receptor decreased glomerular immune complex deposition in Faslpr/lpr mutant mice.27 However, no data on animal lifespan were provided and these mice were on a mixed genetic background; notably, genetic background greatly affects pathology in Faslpr/lpr mice. Thus, the role of type I interferons in SLE-like disease in mice with defects in FASL or FAS remains unclear.

A requirement for c-REL in antigen receptor-induced T- and B-cell activation23 is the most likely explanation for why the reduction in Treg numbers, caused by the loss of c-REL, 33 neither exacerbates autoimmune disease in FasLΔm/Δmc-rel−/− mice nor instigates autoimmunity in c-rel−/− mice.23 Interestingly, mice deficient for PELI1,50 a unbiquitin E3-ligase that controls c-REL levels, develop an autoimmune phenotype similar to those observed in FasLΔm/Δm mice, accompanied by nuclear accumulation of c-REL and excessive canonical NF-κB activation. As PELI1 appears to be critical for the maintenance of T-cell tolerance, it would be interesting to test whether c-REL dysregulation, specifically in T cells, is responsible for the SLE-like autoimmune disease in FasLΔm/Δm mice.

In conclusion, our studies using FasLΔm/Δm mice demonstrate that the canonical and non-canonical NF-κB pathways have critical, yet distinct roles in autoimmune and lymphoproliferative pathology. Although autoimmune diseases caused by inherited mutations in FASL or FAS (ALPS patients) are rare, subtle abnormalities in this apoptotic pathway arising from gene polymorphisms are likely to contribute to autoimmune disease predisposition more generally.51 Combined with the recent discovery of links between polymorphisms in c-Rel and increased susceptibility to certain human autoimmune diseases,23 our findings suggest that targeting c-REL constitutes an attractive strategy to alleviate or prevent autoimmune pathology, while avoiding the major side effects associated with pan-NF-κB inhibition.

Materials and Methods

Mice

All animal experiments were approved by the Walter and Eliza Hall Institute Animal Ethics Committee. FasLΔm/Δm mice were generated on a C56BL/6 genetic background.8 nfkb2−/−25 and c-rel−/− mice22,28 were generated on a mixed C57BL/6x129SV background and backcrossed onto a C57BL/6 background for >10 generations. FasLΔm/Δm mice were intercrossed with nfkb2−/− or c-rel−/− mice. The resultant FasLΔm/wtnfkb2+/− or FasLΔm/wtc-rel+/− mice were intercrossed to produce FasLΔm/Δmnfkb2−/− or FasLΔm/Δmc-rel−/− mice. Control wt, FasLΔm/Δm, nfkb2−/− and c-rel−/− mice used in this study were maintained within the same colony.

Histological analysis and scoring

Mice were killed by CO2 asphyxiation. Organs were removed and fixed in 80% Histochoice (Amresco, Solon, OH, USA) or 10% buffered formalin and paraffin embedded. Histological examination was performed by staining with hematoxylin and eosin (H&E). H&E-stained sections of the kidneys were examined for evidence of GN and scored on a scale of 0–4 (0=normal; 1=minor mesangial hypercellularity; 2=moderate hypercellularity; 3=severe hypercellularity; 4=severe hypercellularity with obliteration and thickening of all capillary loops or the presence of fibrinoid necrosis or glomerular crescents). Lymphoid infiltration into organs (lung, liver, pancreas, submandibular gland and stomach) was assessed by examining H&E-stained sections and graded 0–3 (0=none; 1=occasional small perivascular foci (age related); 2=more dense, well-defined perivascular and periductal foci; 3=extensive infiltrate with parenchymal destruction). Organs were also examined and scored for the presence of necrotizing vasculitis. All photomicrographs were acquired using a × 10/NA 0.3, × 20/NA 0.50 or × 40/NA 0.75 objective lens attached to an Axioplan 2 microscope (Carl Zeiss, North Ryde, NSW, Australia).

Immunofluorescent staining and confocal microscopy

The presence of autoantibodies against classical target tissues in autoimmune diseases (salivary gland, lacrimal gland, eye, kidney, liver, lung, pancreas, stomach and thyroid) was assessed by staining frozen tissue sections (5 μm, air dried and fixed in acetone) of rag-1−/− mice (to eliminate endogenous Ig, a source of background staining) with sera (1/100 dilution) from aged or sick mice. Staining was detected with FITC conjugated goat antibodies against mouse IgG, IgA or IgM (Cappel (MP Biochemicals), Santa Ana, CA, USA). To stain for immune complex deposits, the kidneys from aged or sick mice were snap-frozen in isopentane, sectioned, acetone fixed, and blocked with PBS/2% FCS, followed by staining with FITC-coupled goat antibodies specific to mouse IgM, IgG or IgA (Southern Biotechnology, Birmingham, AL, USA). All immunofluorescent staining was analyzed using a Leica laser scanning (SP2) confocal microscope.

Frozen sections of thymi from 6- or 20-week-old mice were cut at −20 °C on a cryostat (5 μm), air dried and fixed in acetone. Sections were incubated with primary mAbs in 0.5%FCS/PBS for 30 min in a moist chamber, washed three times in PBS and incubated, while protected from light, with the appropriate secondary Abs plus DAPI, followed by washing. Primary Abs included biotin-conjugated anti-UEA-1 (Vector Laboratories, Burlingame, CA, USA), rabbit anti-keratin-5 (MK5, Covance, Biolegend, San Deigo, CA, USA) and Alexa 405-conjugated anti-AIRE. Detection was achieved using Alexa Fluor 555-conjugated goat anti-rabbit IgG antibodies or FITC-conjugated Streptavidin (both from Invitrogen, Carlsbad, CA, USA). Staining was analyzed using a Zeiss LSM5 line scanning confocal microscope (Carl Zeiss).

ANA in sera from aged or sick mice was detected by staining (1/100 serum dilution) of slides coated with HEp-2 human epithelial cells according to the manufacturer’s instructions (Immuno Concepts, Sacramento, CA, USA) and detected with FITC-conjugated rabbit antibodies against all mouse Igs (Cappel (MP Biochemicals)) or FITC-conjugated rabbit antibodies selective for mouse IgM, IgG or IgA (Southern Biotechnology). ANA levels were quantified according to the brightness of fluorescence intensity on a scale of 0 (no fluorescence) to 3+ (maximum fluorescence intensity). Serum ANA levels were also measured by ELISA (The Binding, Site, Birmingham, UK) according to the manufacturer’s instructions.

Hematopoietic reconstitution

Ablation of the hematopoietic compartment was achieved by exposing mice to two doses of 5.5 Gy of γ-irradiation from a 60Co source, 2 h apart. Irradiated mice were reconstituted by intravenous (i.v.) injection of 106 total bone marrow cells in a 100 μl volume of PBS. Reconstituted mice were maintained on 2 mg/ml neomycin-supplemented water for 21 days.

Flow cytometric analysis of thymic stromal cell populations

This procedure was adapted from previously described protocols.52, 53 Thymi were processed individually for digestion as described.53 First, thymi were cleaned of fat and connective tissue, the lobes separated and the capsule nicked with fine scissors. Thymi were then individually agitated in 10 ml RPMI-1640 medium for several minutes to gently flush out thymocytes using a wide-bore pipette. The remaining thymic tissue was then incubated at 37 °C for 15 min in 1 ml of Liberase/DNAase I (0.5 units and 0.1%, respectively, both from Roche, Germany) in RPMI-1640 medium with regular agitation, as detailed.54 Fragments were allowed to settle and the supernatant was collected (fraction 1), kept on ice and the digestion process repeated with the settled thymic fragments. After a second digestion (fraction 2), the remaining aggregates were centrifuged, and resuspended in PBS with 5 mM EDTA, 2% FCS and 0.02% NaN3. Cells were then filtered through a 100 mm mesh and counted using a Countless automated cell counter (Invitrogen). For phenotypic analysis, 1.5 × 107 cells were stained.

Flow cytometric analysis

Single-cell suspensions of the spleen and lymph nodes (pooled axillary, brachial, inguinal and mesenteric) were stained as described1 with FITC-, R-PE- or APC-conjugated mAbs specific for mouse cell surface markers: RB6-8C5: anti-Gr-1, MI/70: anti-Mac-1, M3/84.6.34: anti-Mac-2, Ter119, F4/80, T24.31.2: anti-Thy-1, GK1.5: anti-CD4, 53.6.72: anti-CD8, RA3-6B2: anti-CD45R-B220 and PK136: anti-NK1.1 (BD Biosciences, Franklin Lakes, NJ, USA). The vital dye PI (1 μg/ml) was included and data were acquired on a FACScan (Becton Dickinson, Franklin Lakes, NJ, USA) or LSR II (Becton Dickinson). Tregs were detected after fixation and permeabilization (eBiosciences) by staining with anti-CD25-R-PE, anti-CD4-FITC and anti-Foxp3-APC antibodies, followed by data acquisition on a LSR II (Becton Dickson). Single-cell suspensions were prepared from the lungs by digestion for 20 min at room temperature with collagenase/DNAse (7 mg collagenase (Worthington, Lakewood, NJ, USA) in 1 ml volume of 0.1% w/v DNAse1 (Roche)) treated with EDTA as described,55 and then stained with an R-PE-coupled antibody to Thy-1 plus a panel of biotinylated antibodies specific to different TCRVβ regions followed by staining with FITC-coupled streptavidin (BD-Pharmingen).

Phenotypic analysis of thymic stromal cells by flow cytometry was performed using the following antibody conjugates: PerCP-cy5.5 anti-CD45.2 (clone 30F11, Biolegend, San Diego, CA, USA) and PE anti-Ly51 (Biolegend). APC anti-AIRE, FITC anti-MHC class II (H-2A/E, clone M5 114/15.2) and APC-cy7 anti-EpCAM (clone G8.8a) were produced in-house.

Measurement of serum immunoglobulin levels

Serum immunoglobulin levels were determined by ELISA using sheep anti-mouse Ig antibodies (Silenus Laboratories, Australia) as a capture reagent and developed with mouse Ig isotype-specific goat antibodies conjugated to horseradish peroxidase (Southern Biotechnology). Purified myeloma proteins were used as standards (Sigma, MO, USA).

Analysis of serum cytokine and chemokine levels

Cytokine and chemokine levels in sera were measured by using Bio-Plex Pro mouse cytokine 23-plex (Bio-Rad, Gladesville, NSW, Australia) and Bio-Plex Pro mouse cytokine Th17-plex magnetic bead immunoassays on a Bio-Rad Bio-Plex instrument, following the manufacturer’s instructions. The cytokines measured included the following: CD40L, IL-1α, IL-β, IL-2, IL-3, IL-4, IL-5, IL-6, IL-9, IL-10, IL-12(P40), IL-12(p70), IL-13, IL-17A/F, IL-21, IL-22, IL-23(p19), IL-31, IL-33, eotaxin, G-CSF, GM-CSF, IFN-γ, KC (CXCL1), MCP-1, MIP-1α, MIP-1β, MIP-3α, RANTES and TNFα. Serum IFNβ levels were measured using the VeriKine-HS (PBL Assay Science, Piscataway, NJ, USA) Mouse IFN-β serum ELISA kit, according to the manufacturer’s instructions.

Statistical analysis

Statistical analysis was performed using Student’s t-test, log rank (Mantel–Cox) test for animal survival curves, one-way analysis of variance using Tukey’s comparison test to compare multiple groups or the Mann–Whitney test where appropriate.