To investigate the mechanism by which interferon-α (IFNα) accelerates systemic lupus erythematosus (SLE) in (NZB × NZW)F1 (NZB/NZW) mice.
To investigate the mechanism by which interferon-α (IFNα) accelerates systemic lupus erythematosus (SLE) in (NZB × NZW)F1 (NZB/NZW) mice.
NZB/NZW mice were treated with an adenovirus expressing IFNα. In some mice, T cells were depleted with an anti-CD4 antibody. The production of anti–double-stranded DNA (anti-dsDNA) antibodies was measured by enzyme-linked immunosorbent assay and enzyme-linked immunospot assay. Germinal centers and antibody-secreting cells (ASCs) in spleens and IgG deposition and leukocyte infiltrates in kidneys were visualized by immunofluorescence staining. The phenotype of splenic cells was determined by flow cytometry. Finally, somatic hypermutation and gene usage in VH regions of IgG2a and IgG3 were studied by single-cell polymerase chain reaction.
IFNα-accelerated lupus in NZB/NZW mice was associated with elevated serum levels of IgG2 and IgG3 anti-dsDNA antibodies and accumulation of many IgG ASCs in the spleen, which did not develop into long-lived plasma cells. Furthermore, IgG2a and IgG3 antibodies in the mice were highly somatically mutated and used distinct repertoires of VH genes. The induction of SLE in the mice was associated with an increase in B cell Toll-like receptor 7 expression, increased serum levels of BAFF, interleukin-6 (IL-6), and tumor necrosis factor α, and induction of T cells expressing IL-21. Although IFNα drove a T cell–independent increase in serum levels of IgG, autoantibody induction and the development of nephritis were both completely dependent on CD4+ T cell help.
These findings demonstrate that, although IFNα activates both innate and adaptive immune responses in NZB/NZW mice, CD4+ T cells are necessary for IFNα-driven induction of anti-dsDNA antibodies and clinical SLE.
Systemic lupus erythematosus (SLE) is an autoimmune disease characterized by the production of pathogenic autoantibodies specific for nuclear components. Immune complexes (ICs) containing nucleic acids are endocytosed by B cells and dendritic cells (DCs) that express intracellular Toll-like receptors (TLRs) specific for nucleic acids (1). TLR ligation on B cells enhances proliferation and production of autoantibodies and cytokines (2), and TLR ligation on plasmacytoid DCs induces them to secrete interferon-α (IFNα) (3). IFNα induces maturation of myeloid DCs that activate naive CD4+ T cells to provide help for B cells (4). Activated myeloid DCs also produce BAFF, a cytokine that enhances selection and survival of autoreactive B cells (5) and promotes isotype switching, giving rise to more ICs (6). BAFF-transgenic mice develop SLE independently of T cells, suggesting that T cells are dispensable for disease initiation if TLR-activating ICs are present.
In humans, IFNα can induce autoantibodies and clinical lupus (7). Furthermore, peripheral blood mononuclear cells (PBMCs) from patients with active lupus have up-regulated expression of a group of type I IFN–induced genes (8–10). IFNα is therefore deemed an important cytokine in SLE pathogenesis. In young lupus-prone (NZB × NZW)F1 (NZB/NZW) mice, administration of adenovirus expressing IFNα rapidly induces anti–double-stranded DNA (anti-dsDNA) antibodies, proteinuria, and glomerulonephritis (11), but this does not occur in BALB/c mice. Since some of the immunologic effects of IFNα are mediated independently of T cells, we wished to determine whether IFNα could bypass the need for T cells in the induction of SLE. Our data show that although IFNα induces T cell–independent class-switching and increases circulating interleukin-6 (IL-6) and BAFF, the generation of pathogenic autoantibodies still requires CD4+ T cells.
Twelve-week-old female NZB/NZW mice (The Jackson Laboratory) were treated with a single intravenous injection of 3.3 × 108 particles of IFNα adenovirus (AdIFNα; Qbiogene Morgan) that reproducibly induced proteinuria within 22–30 days. Controls received the same dose of β-galactosidase–expressing adenovirus (AdLacZ) or no treatment. Mice were bled and urine was tested weekly for proteinuria by dipstick (Multistick; Fisher Scientific). Groups of 5 AdIFNα-treated mice were killed at 13, 14, 15, 16, 17, 19, or 23 weeks of age, and controls were killed at 12 or 20 weeks of age. Groups of 5 mice received intraperitoneal injections of 1 mg anti-CD4 (GK1.5) antibody (BioXCell) weekly for 7 weeks starting on the day of adenovirus injection. All experiments using animals were carried out according to protocols approved by the Institutional Animal Care and Use Committees of Columbia University and the Feinstein Institute for Medical Research.
Serum immunoglobulin and anti-dsDNA antibody levels were measured as previously described (12). Standard curves were established using serial dilutions of murine IgM, IgG1, IgG2a, IgG2b, or IgG3 (Sigma-Aldrich). Enzyme-linked immunosorbent assay (ELISA) data on anti-dsDNA were normalized to findings in a high-titer serum assigned an arbitrary level of 512 units, and serial dilutions were run on each plate.
Serum levels of IL-6, IL-17, IL-21, BAFF, IFNγ, and tumor necrosis factor α (TNFα) were measured using a commercial multiplex assay (Assaygate). BAFF levels were also measured using an ELISA kit specific for murine BAFF (Axxora). Enzyme-linked immunospot (ELISpot) assays for total ASCs and for anti-dsDNA ASCs were performed as previously described (12). Spleen cells and PBMCs were analyzed for cell surface markers by fluorescence-activated cell sorting, as previously described (13), with follicular T helper cells gated as previously described (14).
Hematoxylin and eosin–stained sections were scored for renal damage as previously described (15). Cryosections (5 μm) of kidney and spleen were stained (13) using fluorescein isothiocyanate–conjugated anti-mouse IgG2a, IgG3 (Southern Biotechnology), or peanut agglutinin (PNA; Vector), or phycoerythrin (PE)–conjugated anti-mouse IgD, CD4, B220, CD11c (BD PharMingen), or F4/80 (Invitrogen). Images were captured using a Zeiss AxioCam digital camera connected to a Zeiss Axioplan 2 microscope. Glomerular Ig deposition was scored on a 1–4 scale by an observer who was blinded with regard to treatment group.
Seven days or 21 days after AdIFNα injection, mice were loaded intraperitoneally with 10 mg of BrdU (Sigma-Aldrich), followed by feeding for 10 days with water containing 1 mg/ml BrdU. Groups of 5 mice were killed upon completion of BrdU feeding or 2 weeks after BrdU was stopped. Spleen cells were stained with allophycocyanin-conjugated anti-B220, PE-conjugated anti-CD138 (Southern Biotechnology), or for intracellular IgG2a, and then with anti-BrdU according to instructions of the manufacturer (BD PharMingen).
Complementary DNA (cDNA) was synthesized from single GC B cells (CD19+IgM−IgD−PNA+) and PCs (IgD−B220intermediate CD138high) from spleens of 17-week-old IFNα-treated NZB/NZW mice, using Superscript (Invitrogen). PCR (40 cycles of 30 seconds at 95°C, 30 seconds at 58°C, and 1 minute at 72°C) was performed in 10-μl reaction mixtures containing 5 μl of FastStart PCR Master Mix (Roche), 0.5 μM of primers (IgG2a 5′-AAC-TAC-AAG-AAC-ACT-GAA-CCA-GTC–C, 3′-AAC-TGG-GTG-GAA-AGA-AAT-AGC-TAC-T; IgG3 5′-GGA-CAA-CAA-AGA-AGT-ACA-CAC–AGC, 3′-AAC-TTC-TTC-TCT-GAA-GCC-ATC-AGT), and 1 μl of cDNA. Cells producing a single band were then used to perform PCR of the VH region as previously described (16). PCR products were sequenced (Genewiz) and VH regions were compared with the germline sequences using Ig Blast (http://www.ncbi.nlm.nih.gov/igblast/). A polymerase error rate of <0.2% was calculated from 40 independent sequences (10,000 bp) of a germline-encoded heavy chain obtained from single-cell–sorted marginal zone (MZ) and follicular B cells from a transgenic NZB/NZW mouse.
Real-time PCR was performed in triplicate as previously described (17). Additional primers were as follows: TACI 5′-GAG-CTC-GGG-AGA-CCA-CAG, 3′-TGG-TCG-CTA-CTT-AGC-CTC–AAT; TLR-7 5′-TGA-TCC-TGG-CCT-ATC-TCT-GAC, 3′-CGT-GTC-CAC-ATC-GAA-AAC-AC; TLR-9 5′-GAA-TCC-TCC-ATC-TCC-CAA-CA, 3′-CCA-GAG-TCT-CAG-CCA-GCA-CT; CXCL12 5′-GAG-CCA-ACG-TCA-AGC-ATC-TG, 3′-TCT-TCA-GCC-GTG-CAA-CAA-TC; very late activation antigen 4 (VLA-4) 5′-CAA-ACC-AGA-CCT-GCG-AAC-A, 3′-GT-CTT-CCC-ACA-AGG-CTC-TC. The average of the raw data for each sample (Ct) was normalized to the internal control (housekeeping gene β-actin). Normalized expression data were log2-transformed and scaled to the expression value in a single naive mouse, set at an arbitrary value of 1 (0 by log scale). For display in the figures, the mean value in the naive controls was assigned an arbitrary value of 1.
Survival data were analyzed using Kaplan-Meier curves and a log rank test. Other comparisons were performed by Mann-Whitney U test. P values less than or equal to 0.05 were considered significant.
AdIFNα-treated NZB/NZW mice became proteinuric within 3–4 weeks (at age 15–16 weeks), followed by rapid death (Figure 1A). Lymphocytic infiltrates appeared in the renal pelvis of AdIFNα-treated mice at age 14 weeks and had enlarged by age 19 weeks (Figure 1C). Glomerular enlargement and damage with crescent formation (18, 19) occurred by 19–23 weeks of age (P < 0.001) (Figures 1B and C). By immunofluorescence staining, interstitial infiltrates of F4/80high mononuclear cells were visible after the onset of proteinuria and continued to increase until death (Figure 1D). In accordance with previous findings in this model (19), small infiltrates of CD4+ T cells and B cells appeared in the perivascular areas only in the late stages of disease (results not shown). Serum levels of BAFF increased starting 2 weeks after AdIFNα treatment (mean ± SD 8.1 ± 2.0 ng/ml versus 17.8 ± 1.4 ng/ml, 12-week-old naive mice versus 14-week-old AdIFNα-treated mice; P = 0.0025).
A previous study showed that AdIFNα treatment increases serum IgG levels in NZB/NZW mice (11). We found that this was due to an increase of IgG2 and IgG3, but not of IgG1. Serum levels of IgG2a, IgG2b, and IgG3 were higher in AdIFNα-treated mice than in naive or AdLacZ-treated controls (Figure 2A). Similarly, significant increases in serum IgG anti-dsDNA antibodies were detected in AdIFNα-treated mice at ages 15 weeks and 17 weeks (Figure 2B). In contrast, serum IgM levels decreased in AdIFNα-treated mice compared to 17-week-old controls (Figure 2A), and treatment did not affect the levels of circulating IgM anti-dsDNA antibodies (Figure 2B).
GCs appeared in the spleens 2 weeks after AdIFNα treatment and were sustained throughout the disease course (Figure 3B). Large numbers of IgG2a and IgG3 ASCs were found in extrafollicular areas and the splenic red pulp. In contrast, only a few small GCs and IgG ASCs were observed in the spleens of 20-week-old AdLacZ-treated controls. IgG2a and IgG3 deposits appeared in the glomeruli of treated mice at age 14 weeks. By age 19 weeks, heavy IgG deposition was found in the glomeruli of treated mice, whereas minimal IgG deposits were found in the kidneys of the control mice.
A 13.1-fold increase in the number of splenic IgG ASCs was observed at age 14–15 weeks in treated mice (P = 0.0007) versus pretreatment, and this increased further over time (Figure 3A). The number of splenic IgG anti-dsDNA ASCs was increased 10.4-fold and 17.9-fold at ages 16–17 weeks and 19 weeks, respectively, in treated mice compared to 20-week-old AdLacZ-treated controls (P = 0.0027 at age 16–17 weeks, P = 0.0159 at age 19 weeks). AdIFNα also induced a modest increase in IgM anti-dsDNA ASCs in the spleen (Figure 3A), which was not accompanied by an increase in circulating IgM anti-DNA antibodies (Figure 2B).
The vast increase in IgG ASCs in the spleen was not accompanied by a similar increase of these cells in the bone marrow of the treated mice. The frequency of total bone marrow ASCs in the treated mice was only 3.3- and 2.7-fold over that in naive controls at ages 16–17 weeks (P = 0.0046) and 19 weeks (P = 0.0242) (Figure 3A). Anti-dsDNA ASCs appeared at low levels in the bone marrow of treated mice after age 16 weeks, but this was not statistically significant. Furthermore, although the number of IgG ASCs in the spleens of 19-week-old AdIFNα-treated mice was essentially the same as that in untreated aged nephritic NZB/NZW mice, IgG ASCs were 3-fold less frequent in the bone marrow of the treated mice. Similarly, while the spleens of AdIFNα-treated mice exhibited approximately half as many IgG anti-dsDNA ASCs as those of untreated aged nephritic controls, the bone marrow of the AdIFNα-treated mice had a much lower frequency of IgG anti-dsDNA spots (Figures 3A and B). Taken together, these findings demonstrated that fewer long-lived PCs are present in AdIFNα-treated mice compared to spontaneously nephritic aged NZB/NZW controls.
To determine whether this was due to alterations in the bone marrow environment, we performed quantitative PCR for expression of CXCR4, CXCL12, vascular cell adhesion molecule 1 (VCAM-1), and VLA-4, the main molecules that attract and retain PCs in the bone marrow. There was a 2-fold increase in CXCL12 in the bone marrow of aged NZB/NZW mice compared to 14–20-week-old naive mice (P = 0.05). In contrast, CXCL12 expression was decreased in the bone marrow of 19-week-old AdIFNα-treated mice (P = 0.002) (Figure 4A). Similarly, VCAM-1 expression was increased in the bone marrow of aged NZB/NZW mice compared to naive mice (P < 0.001), but this increase did not occur in the bone marrow of 19-week-old AdIFNα-treated mice (Figure 4B). No differences in expression of VLA-4 or CXCR4 were detected between groups.
We used BrdU incorporation to investigate whether the PCs in the spleens of AdIFNα-treated mice are long-lived. More than 90% of CD138+ PCs became BrdU positive after 10 days of BrdU feeding (Figures 4C and D). The vast majority of these cells, however, disappeared 2 weeks after BrdU withdrawal, leaving only 3% of CD138+ PCs positive for BrdU (Figure 4D). Similarly, the percentage of BrdU-labeled IgG2a PCs (all of which were positive both for intracellular IgG2a and for CD138) declined from ∼87% before BrdU withdrawal to ∼2% 2 weeks after BrdU withdrawal (Figure 4D). Similar results were obtained when BrdU feeding was delayed until day 21 after AdIFNα injection, at which time PCs were already present in large numbers in the spleen (Figures 3 and 4D). Taken together, these results show that almost no terminally differentiated long-lived PCs are present in the spleens during this time window.
We next investigated VH gene usage and somatic hypermutation of GC B cells and PCs of AdIFNα-treated mice, using single-cell PCR. IgG2a-producing PCs were approximately twice as frequent as IgG3-producing PCs (227 versus 90 of the 317 cells sequenced), whereas GC B cells produced IgG2a much more predominantly compared to IgG3 (101 versus 21 of the 122 cells sequenced; P < 0.02). The VH regions of 118 PCs (65 IgG2a and 53 IgG3) and 48 GCs (37 IgG2a and 11 IgG3) were analyzed. Of 51 unique genes identified, 27 were found only in PCs, 20 were shared between PCs and GCs, and 4 were found only in GCs. VH gene usage was distinct between IgG2a- and IgG3-secreting PCs (Table 1). Of the 47 genes found among PCs, 22 were found only in IgG2a PCs, 11 were found only in IgG3 PCs, and 14 were found in both IgG2a and IgG3 PCs. Seven overrepresented genes were examined for clonal expansion as evidenced by use of the same V–D–J junction in >2 sequences, and this was found in 2 of 7 (data not shown). There was a high rate of somatic mutation among GCs; this was somewhat lower in PCs, with no difference in the mutation rates between IgG2a and IgG3 PCs. A higher replacement:silent mutation ratio in the complementarity-determining regions compared with the framework regions (Table 1) suggests that both GCs and PCs were subject to post–somatic mutation selection.
|Mutations per sequence, mean ± SD*||No. of mutations per sequence, % of sequences||No. of VH genes (no. of sequences)†|
|CDR replacement||CDR silent||FR replacement||FR silent||Total||0–2||3–10||>11|
|GC B||3.2 ± 2.6||0.8 ± 1.2||2.9 ± 2.1||1.6 ± 1.3||8.5 ± 5.3||8.3||60.4||31.2||—|
|IgG2a PC||1.9 ± 2.0‡||0.5 ± 0.7||1.8 ± 1.7‡||1.4 ± 1.6||5.6 ± 4.3‡||21.5||69.2||9.2||22 (37/65)|
|IgG3 PC||2.5 ± 2.3||0.5 ± 0.8||1.9 ± 1.7‡||1.1 ± 1.3§||6.0 ± 4.2¶||26.4||64.2||9.4||11 (28/53)|
The first phenotypic change detected by flow cytometry was an increase in the number of class-switched B cells, seen 1 week after AdIFNα administration. Three weeks after virus administration, further changes included an increase in spleen size and in the absolute number of activated (CD69+) CD4+ T cells and B cells, and memory CD4+ T cells. By age 19 weeks, there was a marked increase in the absolute number of T cells and B cells of all subsets, except B1 B cells and naive T cells. Splenic and peripheral blood DCs also were increased in AdIFNα-treated mice (Table 2).
|12-week-old naive (n = 5)||13-week-old AdIFNα-treated (n = 5)||15-week-old AdIFNα-treated (n = 5)||16-week-old AdIFNα-treated (n = 5)||19-week-old AdIFNα-treated (n = 6)||19-week-old AdIFNα + anti-CD4− treated (n = 4)||20-week-old naive (n = 5)||20-week-old AdLacZ-treated (n = 5)|
|Total cell no., ×107||7.5 ± 2.6||7.8 ± 1.8||15.0 ± 4.2‡||12.0 ± 4.7||21.8 ± 7.7§||7.2 ± 1.7¶||5.0 ± 0.6||4.3 ± 1.0|
|CD19, ×107||2.7 ± 1.4||2.6 ± 0.5||4.7 ± 1.4||5.1 ± 2.3||10.8 ± 4.0§||3.6 ± 0.7¶||2.2 ± 0.5||1.9 ± 0.8|
|CD19/CD69, ×106||1.5 ± 1.4||1.7 ± 0.6||3.8 ± 1.5||6.1 ± 2.5‡||11.2 ± 4.7§||1.5 ± 0.2#||0.6 ± 0.7||0.6 ± 0.7|
|Follicular, ×107||1.5 ± 0.7||1.2 ± 0.3||2.0 ± 0.5||3.0 ± 1.6||5.2 ± 2.1§||1.4 ± 0.4¶||1.2 ± 0.3||1.1 ± 0.4|
|T1, ×106||1.8 ± 1.5||1.0 ± 0.8||1.8 ± 0.5||3.3 ± 2.3||7.6 ± 4.6‡||1.8 ± 0.2¶||1.3 ± 0.5||1.1 ± 0.5|
|MZ, ×106||2.5 ± 1.2||3.6 ± 0.4||5.2 ± 1.6**||5.9 ± 2.1**||9.3 ± 3.9§||8.8 ± 2.9||3.0 ± 0.5||2.2 ± 0.7|
|B1, ×106||5.0 ± 2.7||2.2 ± 0.6**||2.2 ± 0.6‡||2.3 ± 1.0||7.5 ± 2.1||ND||1.2 ± 0.4§||1.5 ± 0.4§|
|IgM–/IgD– (switched), ×106||0.4 ± 0.2||1.9 ± 0.4§||3.4 ± 2.7||4.3 ± 2.0§||11.8 ± 2.7§||1.4 ± 0.2¶||0.9 ± 0.2**||1.0 ± 0.3‡|
|CD4, ×107||2.2 ± 0.6||2.6 ± 0.7||4.4 ± 1.0||3.8 ± 1.0||5.9 ± 2.4§||0.0 ± 0.0¶||1.6 ± 0.3||1.1 ± 0.5|
|CD4/CD69, ×107||2.1 ± 0.9||2.2 ± 0.5||7.3 ± 2.5‡||8.0 ± 2.7‡||14.0 ± 6.1§||ND||0.9 ± 0.4||0.8 ± 0.5|
|CD4/CD44+CD62L (memory), ×107||0.4 ± 0.1||0.5 ± 0.1||1.3 ± 0.4§||1.2 ± 0.5§||3.1 ± 1.1§||ND||0.4 ± 0.1||0.3 ± 0.1|
|CD4/CD44–CD62L+ (naive), ×107||1.5 ± 0.5||1.9 ± 0.5||2.3 ± 0.5||1.9 ± 0.3||1.8 ± 0.8||ND||0.9 ± 0.2||0.6 ± 0.4|
|TFH, ×106||ND||ND||ND||ND||1.2 ± 0.3||ND||0.2 ± 0.1§||ND|
|CD8, ×107||1.1 ± 0.4||1.4 ± 0.2||1.9 ± 0.4||1.5 ± 0.4||1.8 ± 0.7||2.0 ± 0.3||0.8 ± 0.2||0.6 ± 0.3|
|CD11b/CD11c, ×106||0.7 ± 0.2||0.1 ± 0.1§||3.3 ± 1.1§||2.5 ± 1.2**||2.8 ± 2.0||0.7 ± 0.1¶||0.5 ± 0.3||0.5 ± 0.3|
|% of CD11b+CD11c cells in lymphocytes||ND||ND||ND||ND||3.1 ± 1.6§||1.1 ± 0.6††||0.7 ± 0.3||ND|
To evaluate the effects of AdIFNα treatment on the production of proinflammatory cytokines, we performed real-time PCR on spleen cells from AdIFNα-treated and age-matched naive NZB/NZW mice. Our data showed that the expression of IL4, IL6, IL10, IL21, and IFNγ was significantly elevated in the spleen cells from AdIFNα-treated mice compared to naive controls. No significant difference in the splenic expression of IL12, IL17, or TNFα was detected between AdIFNα-treated mice and naive controls (data available at www.feinsteininstitute.org/Feinstein/Autoimmune+Disease+Lab+Publications). We also measured serum levels of IL-6, IL-10, BAFF, and TNFα in naive and 19-week-old AdIFNα-treated mice, and all were increased in the AdIFNα-treated animals (data available at www.feinsteininstitute.org/Feinstein/Autoimmune+Disease+Lab+Publications).
Using real-time PCR, we detected a 1.8-fold increase in TLR-7 expression in sorted splenic CD19+ B cells from AdIFNα-treated mice compared with naive controls (P = 0.0073) (Figure 4E). However, there was no difference in the expression of TLR-9 or TACI (data not shown).
To investigate the role of T cell help in this model, we depleted CD4+ T cells with anti-CD4 antibody. Ninety-nine percent of CD4+ cells were depleted from peripheral blood within the first week of anti-CD4 antibody treatment (data not shown), and splenic CD4+ T cells were still completely depleted after 7 weeks of weekly treatment (Table 2). The spleens of these mice were smaller and contained fewer B cells than those of age-matched AdIFNα-treated controls (Table 2). This was due to a significant decrease of follicular B cells, with no effect on the IFNα-driven expansion of MZ B cells. Furthermore, there was no increase in CD69 expression on B cells in anti-CD4–treated mice, and class-switching was partially inhibited by the antibody (P = 0.0159 versus 12-week-old or 20-week-old naive controls). Expansion of splenic DCs was also inhibited by anti-CD4. Nevertheless, isolated B cells from anti-CD4–treated mice expressed increased levels of TLR-7, similar to those from AdIFNα-treated mice (Figure 4E).
We next investigated whether T cell help is required for immunoglobulin production in AdIFNα-treated mice. Serum levels of IgM were higher in anti-CD4 antibody–treated mice than in age-matched AdIFNα-treated controls, with no difference in serum levels of IgM anti-dsDNA (Figure 2). Surprisingly, anti-CD4 did not prevent the increase in serum levels of total IgG2a and IgG3 detected at week 17 (Figure 2), despite partial inhibition of class-switching and almost complete blockade of PC formation in the spleens (Table 2 and Figure 3). This was not due to an increase in the frequency of either bone marrow or peritoneal PCs (data not shown). It is possible that the increase in IgG observed in AdIFNα-treated mice was derived from class-switched B cells that were producing low amounts of Ig per cell as a direct result of AdIFNα stimulation (20). In addition, loss of Ig in the urine may have resulted in an underestimate of total Ig production in the control IFNα-treated mice.
Despite the overall increase in serum IgG levels, serum levels of IgG2a and IgG3 anti-dsDNA antibodies in the anti-CD4–treated mice were significantly lower than in AdIFNα-treated controls (P = 0.0057 for IgG2a; P = 0.016 for IgG3). ELISpot analysis revealed significantly lower numbers of both total and anti-dsDNA IgG ASCs in the spleens of anti-CD4–treated mice compared with AdIFNα-treated controls (Figure 3A). Immunofluorescence staining showed that GCs failed to form in the spleens of anti-CD4–treated mice, and only a few IgG2a and IgG3 ASCs were detected (Figure 3B). Renal deposition of IgG2a and IgG3 was greatly reduced by anti-CD4 antibody treatment (Figure 3B) (mean ± SD immunofluorescence score 1.0 ± 0.7 in anti-CD4–treated mice versus 3.6 ± 0.5 in AdIFNα-treated controls; P < 0.01), and the degree of renal damage was significantly less than in AdIFNα-treated controls (P < 0.0001, glomerular and interstitial damage) (Figure 1B).
Finally, we examined whether CD4+ T cells are the major source of inflammatory cytokines in AdIFNα-treated mice. IL4, IL10, IL21, and IFNγ gene expression levels were not significantly different between anti-CD4–treated mice and naive controls and were significantly lower in anti-CD4–treated mice than in AdIFNα-treated controls. In contrast, anti-CD4 had only a partial effect on IL6 gene expression in the spleens of AdIFNα-treated mice, with levels intermediate between those in naive and IFNα-treated controls (data available at www.feinsteininstitute.org/Feinstein/Autoimmune+Disease+Lab+Publications). Anti-CD4 did not prevent the increase in serum levels of IL-6 or BAFF, but it did prevent the increase in serum levels of TNFα that was likely derived from target organs (data available at www.feinsteininstitute.org/Feinstein/Autoimmune+Disease+Lab+Publications).
This study explored the immunologic events that underlie IFNα-induced acceleration of lupus in NZB/NZW mice. The earliest event we were able to detect was class-switching of splenic B cells, which occurs before formation of GCs or an increase in circulating BAFF, and is observed even when T cells are depleted. Two weeks after AdIFNα injection, large GCs and IgG ASCs appear in the spleens, and a week later B and T cell activation becomes evident, accompanied by the appearance of serum autoantibodies, renal IgG deposition, glomerular damage, onset of proteinuria, and increasing renal infiltration with mononuclear phagocytes.
IFNα enhances GC formation and antigen-specific antibody responses in several experimental models (21, 22), especially in the context of weak immunogens (23). It also induces differentiation of CD40-activated B cells into plasmablasts; terminal differentiation to Ig-secreting PCs is then mediated by IL-6 secreted in response to CD40 ligation on plasmacytoid DCs (24). However, expression of IL-6 by splenic cells in AdIFNα-treated mice is only partly dampened by CD4+ T cell depletion, suggesting that it is induced by other mechanisms in our model, such as enhanced TLR signaling or direct stimulation of B cells (20). IFNα-enhanced TLR signaling also inhibits shedding of inducible costimulator (ICOS) ligand from B cells, resulting in increased T cell help through ICOS within GCs (25). Finally, IFNα signaling on DCs is required for the development of follicular T helper cells (26). In accordance with these mechanisms, we noted the rapid appearance of enlarged GCs, increased numbers of follicular T helper cells, elevated expression of IL-21, and accumulation of large numbers of IgG-secreting PCs in the spleens of AdIFNα-treated mice.
IFNα also mediates T cell–independent class-switching by several mechanisms. It induces up-regulation of TLRs on DCs and B cells (6, 27) in a type I interferon receptor–dependent manner (20, 28), and we have shown herein that it up-regulates expression of TLR-7, but not TLR-9, in the splenic B cells of mice, similar to findings in human B cells (20). Activation of murine B cells through TLR-7 or TLR-9, together with CD40- or cytokine-mediated costimulation, promotes class-switching to IgG2a, IgG2b, and IgG3 and inhibits the generation of IgG1 (29, 30). Second, IFNα amplifies the response of B cells to TLR-mediated signals (31, 32), enhances B cell production of IL-6, and induces antibody secretion from both naive and memory B cells. Furthermore, TLR-activating ICs direct the formation of extrafollicular foci in which somatic mutation can occur (33). Third, IFNα induces DCs to produce BAFF and APRIL, which enhance T cell–independent class-switching and Ig secretion (5, 11, 34, 35). BAFF amplifies TLR expression on B cells, and TLR ligation in turn up-regulates expression of the BAFF receptor TACI (6, 36). Overexpression of BAFF induces class-switching in a myeloid differentiation factor 88–dependent manner; in BAFF-transgenic mice that express a supraphysiologic (50–100-fold) increase in serum BAFF, this is sufficient to induce autoimmunity even in the absence of T cells (6).
The above findings suggest that a T cell–independent mechanism involving ICs, TLRs, and BAFF might initiate or propagate disease, and raise questions regarding the importance of T cells in the break of tolerance of autoreactive B cells and the value of T cell–directed therapies in SLE. Our study in NZB/NZW mice showed that IFNα up-regulates BAFF and IL-6 production, enhances B cell TLR-7 expression, promotes T cell–independent class-switching to IgG, and facilitates expansion of the MZ B cell subset that is the source of pathogenic IgG autoantibodies in BAFF-transgenic mice (6). Nevertheless, these events were not sufficient to initiate autoimmunity in the absence of CD4+ T cells in our model.
The IgG PCs induced by IFNα in NZB/NZW mice may derive from GCs or extrafollicular foci. As shown by single-cell PCR, GC B cells predominantly gave rise to IgG2a ASCs whereas the PC compartment contained a higher percentage of IgG3 ASCs, suggesting that some of the IgG3 PCs may derive from the extrafollicular pathway. In addition, a significant proportion of PCs producing either IgG2a or IgG3 used VH genes that are not present in GC B cells, indicating that they may have arisen from outside the GCs. Furthermore, IgG2a and IgG3 PCs have different repertoires, suggesting different origins. Although 24% of the PCs had <2 somatic mutations, compared with only 9% of the GC cells, most were extensively mutated, suggesting that they had received help from T cells either inside or outside the GCs. The absence of autoreactive antibodies despite evidence of active class-switching in T cell–depleted mice provides strong evidence that clonal expansion and/or somatic mutation is required for the initiation of pathogenic autoreactivity in NZB/NZW mice.
Our data strongly suggest that the IgG PCs induced by IFNα in NZB/NZW mice are not long-lived PCs. Impaired homing of PCs to the bone marrow is characteristic of NZW and NZB/NZW mice (15, 37). The mechanism for the absence of these cells in the bone marrow of AdIFNα-treated mice is yet to be determined. One explanation is that IFNα affects the expression of critical chemoattractant molecules or the responsiveness of PCs to these molecules or to extrinsic signals provided in survival niches of bone marrow (38, 39). We observed a decrease of CXCL12 expression in the bone marrow and a failure to up-regulate VCAM-1 in the bone marrow of IFNα-treated mice compared with mice with spontaneous disease, suggesting that in AdIFNα-treated mice, the bone marrow does not provide optimal support for PC survival. The down-regulation of CXCL12 might be due to the observed increase in serum TNFα levels (40). Alternatively, sphingosine 1-phosphate (S1P) receptor is required for PC egress from the spleen (41), and IFNα inhibits S1P-mediated egress of CD69+ lymphocytes from the spleen by modulating expression of the S1P1 receptor (42). Whether IFNα affects the expression of S1P1 receptor on PCs remains to be determined.
There are also survival niches for PCs in the red pulp of spleen (43), and 40% of the PCs in the spleens of aged NZB/NZW mice are long-lived (44). It is possible that survival niches and soluble survival factors are limited in the spleen and therefore cannot accommodate the large numbers of PCs rapidly generated in IFNα-treated NZB/NZW mice.
IFNα activates immature myeloid DCs and monocytes, causing them to up-regulate the expression of costimulatory molecules and produce inflammatory cytokines (27, 45, 46). Mature DCs activate naive Th1 CD4+ T cells that secrete IFNγ and provide help to B cells (47). Consistent with this mechanism, we observed CD11c+ DCs in the spleens and peripheral blood of AdIFNα-treated NZB/NZW mice. However, the expansion of the DCs occurs downstream of T cell activation, as they failed to accumulate in the spleens of AdIFNα-treated mice in the absence of T cells.
The NZB/NZW mouse is an excellent model for human lupus nephritis. However, studies of therapeutic interventions in this mouse strain are hampered by the stochastic and delayed spontaneous disease onset. Because of its high reproducibility and the synchronized onset of disease, the model of IFNα-induced disease allows investigation of the effects of therapies on defined stages of the disorder including remission induction studies, which pose a higher bar for therapeutic intervention. Such studies may help predict appropriate interventions for patients with an IFN signature.
All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Davidson had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study conception and design. Liu, Bethunaickan, Davidson.
Acquisition of data. Liu, Bethunaickan, Huang, Lodhi, Solano.
Analysis and interpretation of data. Liu, Bethunaickan, Lodhi, Madaio, Davidson.
We would like to thank Anton Kuratnik and Elif Alpoge for technical assistance.