Dr. Davidson has received consulting fees, speaking fees, and/or honoraria from Biogen Idec (less than $10,000).
Systemic Lupus Erythematosus Basic Science Studies
Selective blockade of BAFF for the prevention and treatment of systemic lupus erythematosus nephritis in NZM2410 mice
Article first published online: 29 JAN 2010
Copyright © 2010 by the American College of Rheumatology
Arthritis & Rheumatism
Volume 62, Issue 5, pages 1457–1468, May 2010
How to Cite
Ramanujam, M., Bethunaickan, R., Huang, W., Tao, H., Madaio, M. P. and Davidson, A. (2010), Selective blockade of BAFF for the prevention and treatment of systemic lupus erythematosus nephritis in NZM2410 mice. Arthritis & Rheumatism, 62: 1457–1468. doi: 10.1002/art.27368
- Issue published online: 29 APR 2010
- Article first published online: 29 JAN 2010
- Accepted manuscript online: 29 JAN 2010 12:00AM EST
- Manuscript Accepted: 20 JAN 2010
- Manuscript Received: 11 SEP 2009
- New York SLE Foundation
- National Institute of Allergy and Infectious Diseases. Grant Number: R01-AR-049938-01
- National Institute of Arthritis and Musculoskeletal and Skin Diseases. Grant Number: R01-AI-059636-01A2
To determine whether BAFF or combined BAFF/APRIL blockade is effective in a mouse model of systemic lupus erythematosus (SLE) nephritis characterized by rapidly progressive glomerulosclerosis.
NZM2410 mice at early and late stages of SLE nephritis were treated with a short course of BAFF-R-Ig or TACI-Ig fusion protein. Proteinuria and serologic profile were evaluated every 2 weeks. Immunohistochemical, flow cytometric, and enzyme-linked immunospot analyses of the spleen, kidney, and bone marrow were performed after 8 weeks and after 33 weeks.
A short course of selective blockade of BAFF alone was sufficient to prevent and treat SLE nephritis in NZM2410 mice, despite the formation of pathogenic autoantibodies. Decreases in spleen size and B cell depletion persisted for more than 33 weeks after treatment and resulted in secondary decreases in CD4 memory T cell formation and activation of splenic and peripheral monocytes. Immune complex deposition in the kidneys was dissociated from renal damage and from activation of renal endothelial and resident dendritic cells.
Selective blockade of BAFF alone, which resulted in B cell depletion and splenic collapse, was sufficient to prevent and treat the disease in this model of noninflammatory SLE nephritis. This shows that the inflammatory microenvironment may be a determinant of the outcome of B cell modulation strategies.
Systemic lupus erythematosus (SLE) is an autoimmune disorder in which loss of tolerance to nucleic acids is associated with the development of pathogenic autoantibodies that damage target organs. Lupus nephritis develops in up to 60% of adult SLE patients and is even more common in children. Induction of remission of lupus nephritis requires the use of potent immunosuppressive agents, with significant adverse effects and frequent relapses (1).
B cells are therapeutic targets in SLE because they produce pathogenic autoantibodies and because they have multiple effector functions, including antigen presentation to T cells and cytokine production and migration to sites of inflammation (2). One way to modulate B cell function is by inhibiting the B cell survival molecule BAFF (B lymphocyte stimulator [BLyS; trademark of Human Genome Sciences, Rockville, MD]). Therapeutic antagonism of BAFF and its homolog APRIL (a proliferation ligand) is based on the discoveries that BAFF provides a crucial homeostatic signal for B cell survival and selection (3–6) and that soluble BAFF and APRIL are highly expressed in the serum of SLE patients (7) and in the target organs of SLE-prone mice (8, 9).
BAFF binds to 3 receptors, BAFF-R, TACI, and BCMA, which are differentially expressed during B cell ontogeny (10), whereas APRIL binds only to TACI and BCMA. Selective blockade of BAFF can be achieved with a soluble BAFF-R-Ig fusion protein or an antibody to BAFF, whereas blockade of both BAFF and APRIL is achieved with soluble TACI-Ig. Initial phase II and phase III studies of a selective antibody to soluble BAFF (belimumab) were recently completed (11), and studies of TACI-Ig (atacicept) are currently in progress. Questions remain about the mechanism of action of these reagents and about whether blocking both BAFF and APRIL will be more efficacious than blocking BAFF alone.
The NZM2410 mouse is an inbred strain derived from NZB × NZW mice. NZM2410 mice manifest antibodies to nucleosomes and to double-stranded DNA (dsDNA), and they develop rapidly progressive glomerulosclerosis, with little lymphocytic infiltrate in the kidneys. These mice express high levels of interleukin-4 (IL-4), and they secrete large amounts of IgG1 antibodies (12). NZM2410 mice have a defect in the migration of plasma cells to the bone marrow and retain large numbers of plasma cells in their spleens (13). We therefore hypothesized that disease in these mice might be more responsive to TACI-Ig, which depletes splenic plasma cells (14), than to BAFF-R-Ig. Our findings show that BAFF-R-Ig and TACI-Ig were equally effective at preventing disease and that a short course of either agent induced sustained remission when used as a single therapeutic agent. This appears to be due to prolonged B cell depletion and a decrease in the inflammatory response to renal immune complex deposition.
MATERIALS AND METHODS
Treatment of NZM2410 mice.
NZM2410 mice were purchased from Taconic. Groups of 10 mice were treated at 14 weeks of age with 1 × 109 plaque-forming units of BAFF-R-Ig adenovirus (AdBAFF-R-Ig), TACI-Ig adenovirus (AdTACI-Ig), or β-galactosidase adenovirus (AdLacZ). Five mice received no treatment. These adenoviruses have previously been described in detail (14). We obtained blood and tested urine for proteinuria by dipstick (Multistix; Fisher) every 2 weeks. Mice were followed up until 55 weeks of age. Separate groups of 20 mice were treated at 22 weeks of age with the same adenoviruses. A total of 5–8 mice in each group were killed for mechanistic studies at 30 weeks, and the remaining mice were followed up until 55–62 weeks of age.
Ethical approval for animal experimentation.
Experiments using animals were performed under approved protocols of the Animal Institutes of Columbia University and the Feinstein Institute.
Measurement of serum levels of IgM, IgG, and anti-dsDNA antibodies.
Serum IgM and IgG levels were measured as previously described (15). To measure anti-dsDNA antibodies (16), Immulon 2 HB plates (Thermo Scientific) that had been precoated with 1 mg/ml of methylated BSA (Sigma-Aldrich) in phosphate buffered saline (PBS), were coated with 50 μg/ml of single-stranded DNA (ssDNA)/dsDNA for 30 minutes at 37°C, washed with PBS, and then blocked overnight in 0.1% gelatin containing 3% bovine serum albumin (BSA)/3 mM EDTA. Serum was diluted in 0.1% gelatin containing 2% BSA, 3 mM EDTA, and 0.05% Tween 20. Horseradish peroxidase–conjugated goat anti-mouse IgM or IgG (SouthernBiotech) was diluted in PBS containing 1% BSA and 0.05% Tween 20, and the plates were developed with ABTS. Enzyme-linked immunosorbent assay (ELISA) data obtained for each antigen were normalized to a high-titer mouse serum that was given an arbitrary level of 100 units and was run in serial dilutions on each plate.
Anti-dsDNA enzyme-linked immunospot (ELISpot) assay.
Spleens were harvested from 4–5 mice per group at ages 30 and 55 weeks. ELISpot assays for total Ig–secreting cells and for anti-dsDNA–secreting B cells were performed using plates coated as above and our previously described ELISpot protocol (15).
Measurement of serum levels of vascular cell adhesion molecule 1 (VCAM-1).
Frozen sera from 30-week-old mice were diluted to 1:2,000 and analyzed for serum VCAM-1 levels using an ELISA kit (R&D Systems) according to the manufacturer's instructions.
Flow cytometric analysis of spleens and kidneys.
Spleen cells from 5–10 mice per group were analyzed for lymphocyte markers as previously described (17). Cells isolated from PBS-perfused kidneys were analyzed for expression of type II major histocompatibility complex, Gr-1 (eBioscience), CD11b, F4/80, and CD11c (BD PharMingen) as previously described (9).
Bromodeoxyuridine (BrdU) labeling of renal CD11b+ cells.
NZM2410 mice ages 6–8 weeks were loaded intraperitoneally with 10 mg of BrdU (Sigma-Aldrich), followed by feeding with water containing 1 mg/ml of BrdU for up to 30 days. The water was protected from light and was changed daily. On days 0, 7, 15, and 30, groups of 3–5 mice were killed, and kidney cells were prepared and stained with anti-CD11b, anti-CD11c, and anti-F4/80 as above. BrdU was detected in CD11b+F4/80high and CD11b+CD11chigh cell populations using a BrdU flow kit (BD PharMingen) according to the manufacturer's protocol.
Immunohistochemistry and immunofluorescence staining.
Hematoxylin and eosin–stained renal sections were scored for glomerular and interstitial (inflammation and tubular atrophy) damage as previously described (18). Cryosections (5 μm) of spleen and/or PBS-perfused kidneys were stained for 1 hour at room temperature with phycoerythrin (PE)–conjugated anti-B220, fluorescein isothiocyanate (FITC)–conjugated anti-mouse IgM, PE-conjugated anti-mouse IgD, or PE-conjugated anti-mouse CD11b (all from BD PharMingen), as well as with FITC-conjugated anti-mouse IgG1 or FITC-conjugated anti-mouse IgG2a (both from SouthernBiotech) in 10% normal goat serum–PBS. Images were captured using a digital CCD camera (Diagnostic Instruments) connected to a Nikon microscope.
Real-time quantitative reverse transcription–polymerase chain reaction (RT-PCR).
Whole kidney messenger RNA was prepared from perfused kidneys obtained from 5–10 mice per group and real-time quantitative RT-PCR was performed with 61 primers for markers of inflammation as previously described (9), with the addition of hepatitis A virus cellular receptor 1 (HAVCR-1) (5′-CCAACATCAATCAGAGTCTCTACC-3′ [forward] and 5′-TGTCTCATGGGGACAAAATG-3′ [reverse]) and nephrin (5′-AACATCCAGCTCGTCAGCAT-3′ [forward] and 5′-AGGGCTCACGC TCACAAC-3′ [reverse]). Data were analyzed as previously described (9) and normalized to the mean values in 5 young NZM2410 control mice.
Proteinuria and survival data were analyzed using Kaplan-Meier curves and the log rank test. Between-group comparisons were examined with Wilcoxon's rank sum test.
Prevention of disease onset and induction of remission by treatment with TACI-Ig and with BAFF-R-Ig.
It has been shown that administration of AdTACI-Ig and AdBAFF-R-Ig results in detectable serum levels of the fusion proteins in normal and autoimmune mice for ∼6 weeks (14, 17). Strikingly, administration of either TACI-Ig or BAFF-R-Ig to NZM2410 mice at age 14 weeks resulted in almost complete prevention of SLE nephritis (Figure 1). After a single dose of AdBAFF-R-Ig, 90% of the mice remained free of proteinuria, and more than 40 weeks later, none of them had died (Figures 1A and B).
We next determined whether treatment at later stages of disease would induce remission. In these experiments, mice were treated at 22 weeks of age, when the first mice in this group were becoming proteinuric. Late treatment induced prolonged survival (Figures 1C and D), with ∼60% of the mice remaining proteinuria-free at ages 55–62 weeks. Notably, of the 7 AdTACI-Ig–treated and the 7 AdBAFF-R-Ig–treated mice that had >300 mg/dl of proteinuria at the time of treatment or developed proteinuria within the first 8 weeks after treatment, 4 mice and 5 mice, respectively, manifested complete remission, with disappearance of proteinuria and prolonged survival (P < 0.001 compared with controls). In contrast, all 10 control mice that developed proteinuria in the same time period died within 1 month. Thus, both TACI-Ig and BAFF-R-Ig prevent disease onset and induce remission of nephritis in NZM2410 mice.
Serum immunoglobulin levels and anti-dsDNA antibody production.
Mice treated with TACI-Ig had a profound decrease in serum IgM levels within 4 weeks that was maintained until the time the mice were killed. Serum levels of IgM were unaffected by BAFF-R-Ig (Figure 2A). In contrast, both TACI-Ig and BAFF-R-Ig induced a decrease in serum levels of total IgG (data not shown) and in serum levels of IgG1 and IgG2a (Figures 2B and C). Eight weeks after treatment, serum levels of anti-dsDNA antibodies were significantly lower in treated mice than in controls, but by 12 weeks after treatment, the levels were no longer significantly different from those in the control mice (Figure 2D). Similar results were observed in mice treated at 14 weeks of age (data not shown).
These data were confirmed by the findings of ELISpot analyses (Figure 3). Eight weeks after treatment, the treated mice had a significantly lower frequency and total number of IgG and IgG anti-DNA–secreting cells in the spleen than did the controls. Only TACI-Ig–treated mice had significantly lower numbers of total IgM–secreting cells (Figures 3A and B). When the mice whose SLE nephritis remained in remission were killed at 55 weeks of age and these parameters were compared with those in old controls, the differences were no longer significant (Figures 3C and D). Mice treated with TACI-Ig had a significantly lower frequency of IgG-secreting cells in the bone marrow 8 weeks after treatment than did either BAFF-R-Ig–treated mice or controls; anti-dsDNA–secreting B cells were not detected in either of the treated groups. By 55 weeks of age, there was no longer any difference between the 3 groups, although there was still a trend toward lower numbers in the TACI-Ig–treated mice (Figure 3E).
To further confirm the effects of treatment on antibody-producing cells, we performed immunohistochemical analysis of spleens from 30-week-old mice. Consistent with the ELISpot data, the spleens of treated mice had fewer plasma cells in the red pulp than did those of the controls (Figures 4A–C). To determine whether germinal centers were present, we stained the spleens with peanut agglutinin. Despite the marked B cell depletion and thin follicles in the spleens, germinal centers were still present in the treated mice (Figures 4F–H).
Spleen cell phenotype, as determined by flow cytometry.
We further examined by flow cytometry the effect of BAFF/APRIL blockade on lymphocytes (Table 1). Both TACI-Ig treatment and BAFF-R-Ig treatment depleted spleen B cells by 50–75%; this depletion persisted for many months. B cell subset analysis showed a preferential depletion of transitional type 2 (T2), marginal zone (MZ), and follicular cells, with sparing of the transitional type 1 (T1) subset. It has been shown that NZM2410 mice have large numbers of B-1a cells in the spleen (19). Interestingly, treatment with TACI-Ig, but not BAFF-R-Ig, depleted B-1a cells early in the treatment course, which suggests that APRIL, with or without BAFF, supports B-1a cell survival, a result similar to that observed in studies of TACI-Ig–transgenic mice (20). Untreated mice manifested a 2–3-fold expansion of B cells of all subsets with age, but this was prevented by BAFF blockade. Even at age 55 weeks, the treated mice had significantly fewer T2, MZ, and follicular B cells than did the 8-week-old controls.
|Numbers and percentages of cells per spleen||Young untreated control mice, 8 weeks old (n = 12)||AdBAFF-R-Ig–treated mice||AdTACI-Ig–treated mice||AdLacZ-treated mice, 30 weeks old (n = 8)|
|30 weeks old (n = 8)||55 weeks old (n = 9)||30 weeks old (n = 8)||55 weeks old (n = 8)|
|Cell count, ×107||4.0 ± 0.7||5.8 ± 3.0†||6.9 ± 3.7‡||2.9 ± 1.0§||4.1 ± 2.2†||11.5 ± 4.2§|
|CD4 T cell count, ×106||13.8 ± 3.1||24.4 ± 11.7||20.5 ± 12.4||13.6 ± 5.4‡||16.7 ± 8.9||40.9 ± 23.9†|
|No. of activated T cells, ×106||0.7 ± 0.32||6.2 ± 3.04¶||12.6 ± 3.97¶||2.4 ± 2.32#||10.8 ± 0.55¶||9.0 ± 4.84§|
|% activated CD4 cells||6.6 ± 1.6||23.7 ± 7.4¶||30.5 ± 10.0¶||17.6 ± 19.8||46.8 ± 10.7¶||19.0 ± 8.5§|
|No. of memory T cells, ×106||2.2 ± 0.6||14.4 ± 7.2¶||14.0 ± 7.1¶||7.3 ± 4.6¶#||13.0 ± 7.8¶||26.2 ± 14.8§|
|% memory CD4 cells||16.5 ± 6.8||63.2 ± 21.2¶||71.0 ± 11.5¶||50.8 ± 21.0¶||76.1 ± 11.6¶||67.6 ± 13.1§|
|CD8 T cell count, ×106||3.1 ± 1.22||4.6 ± 2.64||7.6 ± 3.24||3.3 ± 1.36#||8.5 ± 0.52||7.7 ± 3.68§|
|B cell count, ×106||16.3 ± 5.7||8.9 ± 3.9¶#||9.3 ± 7.7#||3.1 ± 1.5¶#||10.6 ± 10.1†||42.6 ± 16.1#|
|% B cells||46.3 ± 2.4||16.0 ± 4.4¶#||17.1 ± 8.7†¶||10.3 ± 2.0¶#||21.8 ± 13.2||35.4 ± 6.7|
|No. of activated B cells, ×106||0.2 ± 0.15||0.5 ± 0.34¶#||0.8 ± 0.51†||0.4 ± .032§||1.9 ± 0.18||3.7 ± 2.13§|
|% activated CD19 cells||1.0 ± 0.9||5.4 ± 3.8¶||5.1 ± 1.5¶||8.8 ± 4.7¶||9.2 ± 1.3¶||9.5 ± 5.7§|
|No. of T1 cells, ×105||8.9 ± 3.8||7.7 ± 3.5#||6.0 ± 5.2#||1.8 ± 0.9¶#||4.3 ± 5.2#||37.5 ± 24.3†|
|No. of T2 cells, ×105||13.7 ± 7.4||0.7 ± 0.6¶#||1.2 ± 1.3¶#||0.6 ± 0.7¶#||1.7 ± 1.9¶#||15.9 ± 17.6|
|No. of MZ cells, ×105||12.8 ± 10.3||0.9 ± 0.5¶#||1.9 ± 1.6¶#||0.5 ± 0.3¶#||1.9 ± 1.8¶#||31.5 ± 38.5|
|No. of follicular cells, ×106||10.8 ± 3.8||2.9 ± 1.3§¶||4.3 ± 3.6§¶||1.3 ± 0.6¶#||6.2 ± 5.9†¶||24.3 ± 9.2†|
|No. of B-1 cells, ×106||3.3 ± 1.6||5.1 ± 2.7||4.1 ± 2.4||1.5 ± 0.5¶#||3.6 ± 3.2‡||9.8 ± 6.6†|
|% B-1 cells||19.1 ± 3.8||51.9 ± 20.7†¶||40.7 ± 12.9¶#||41.3 ± 8.5¶#||34.8 ± 4.2†¶||23.4 ± 10.2|
|CD11b cell count, ×106||1.2 ± 0.9||1.9 ± 1.4#||2.9 ± 1.8†||0.7 ± 0.5#||2.9 ± 2.9‡||7.8 ± 4.4§|
BAFF blockade did not prevent the activation of CD4 T cells that occurs as disease progresses, but the reduction in spleen size caused a significant decrease in the number of total and memory CD4 T cells in treated mice as compared with nephritic controls; the decrease in memory CD4 T cells persisted up to age 55 weeks. BAFF blockade also attenuated the expansion of the CD11b macrophage/dendritic cell (DC) population in the spleen, and this persisted up to age 55 weeks. Peripheral blood monocytes expanded from a mean ± SD of 4.1 ± 0.5% to 22.8 ± 10.8% of peripheral blood lymphocytes by 30 weeks; this was not prevented by either BAFF-R-Ig or TACI-Ig treatment (data not shown). However, peripheral blood monocytes from treated mice expressed lower levels of CD11c at 30 weeks (mean fluorescence intensity [MFI] 479.5 versus 650.77 in untreated controls; P < 0.01), and this effect persisted at 55 weeks (data not shown).
Findings of renal damage assessments.
To determine whether the autoantibodies produced in the treated mice were pathogenic, we performed immunofluorescence staining of kidneys from four 40-week-old AdBAFF-R-Ig–treated mice without proteinuria and from five 25–30-week-old proteinuric controls. All 4 of the treated mice had extensive immune complex deposition in their renal glomeruli and were no different from the untreated controls (Figures 4D and E). Nevertheless, the treated mice had significantly less renal damage than did the control mice (Figures 4I and J and Figure 5A).
Kidneys from nephritic NZM2410 mice had significantly less expression of a set of markers of inflammation previously detected in the kidneys of (NZB × NZW)F1 mice with proliferative SLE nephritis (9) (Figure 5D). Immunohistochemical analysis confirmed the lack of inflammatory infiltrates previously reported (12), with no difference in the number of renal parenchymal B cells, T cells, or CD11c+ cells between young mice and controls with new-onset proteinuria (results not shown).
To confirm the lack of renal damage in treated mice, we performed real-time RT-PCR of 2 informative biomarkers of renal damage (lipocalin 2 and HAVCR-1), a biomarker of endothelial activation (VCAM-1), and a biomarker of podocyte loss (nephrin). Control mice had elevated expression of lipocalin 2, HAVCR-1, and VCAM-1 in the kidneys and decreased expression of nephrin. Treated mice had normal levels of these markers, even at 55 weeks of age (Figure 5C). Treated mice also had lower serum levels of VCAM-1 at 30 weeks than did the controls (mean ± SD 1.8 ± 0.2 μg/ml in 8-week-old NZM2410 mice, 4.1 ± 0.4 μg/ml in 30-week-old NZM2410 mice, 2.7 ± 0.2 μg/ml in 30-week-old AdTACI-Ig–treated mice, and 2.6 ± 0.6 μg/ml in AdBAFF-R-Ig–treated mice; P < 0.005 for treated mice versus age-matched controls).
Resident populations of CD11b+ mononuclear phagocytes in normal murine kidneys are heterogeneous. The major population is F4/80highMHChighCD11cintermediateGr-1intermediateCX3CR1high; these cells are currently referred to as resident renal DCs. A second common population is F4/80lowCX3CR1lowGr-1highLy6Chigh; this is the resident macrophage population (21, 22). One feature of renal damage in NZB × NZW mice with SLE is activation of renal DCs, with up-regulation of CD11b expression. In addition, a small population of F4/80lowCD11chigh myeloid dendritic cells becomes markedly expanded in nephritic NZB × NZW mouse kidneys (9).
Analysis of the kidneys of young NZM2410 mice revealed that their major CD11b+ population manifested the renal DC phenotype (Figure 5F). Control NZM2410 mice with proteinuria had marked up-regulation of CD11b on renal DCs, as evidenced by immunohistochemistry (Figures 4K and L) as well as by flow cytometry. However, renal DCs from treated 55-week-old mice without proteinuria did not have increased expression of CD11b (Figure 4M and Figure 5). Even kidneys from mice that had proteinuria at the time of treatment and then entered prolonged remission after BAFF blockade had normal CD11b expression on renal DCs, comparable to that of young controls. This was confirmed by flow cytometry of isolated renal cells (mean ± SD MFI for CD11b 23,560 ± 4,234 in treated mice versus 37,713 ± 10,248 in controls; P < 0.001) (Figure 5E).
Using BrdU feeding, we determined that the half-life of renal CD11b+F4/80lowCD11chigh cells in young NZM2410 mice was 10 days, whereas that of the major CD11b+F4/80highCD11cintermediate renal DC population was 13 days. Since the number of these cells in the kidney remained constant during the course of this experiment, the rate of BrdU incorporation reflects cell turnover (Figure 5C). In contrast to the findings in the nephritic NZB × NZW mice, the CD11b+F4/80lowCD11chigh myeloid DC population did not expand in the nephritic kidneys of NZM2410 mice.
Animal models of lupus nephritis are very useful in elucidating the pathogenesis of renal inflammation and for testing new therapies. Disease in humans is heterogeneous, however, and it is increasingly recognized that multiple animal models are needed to elucidate disease mechanisms and to evaluate new therapeutic strategies. The promise of BAFF blockade as an effective therapy for SLE warrants an analysis of its mechanism of action in diverse SLE models.
We have performed extensive studies of the mechanism of action of BAFF blockade in 2 different models of murine lupus (for review, see ref.23). In (NZB × NZW)F1 mice, which develop proliferative glomerulonephritis similar to class IV nephritis in humans, BAFF blockade prevents disease onset, but it only induces remission when administered together with the T cell–costimulatory antagonist CTLA-4Ig (14, 17). In male NZW × BXSB mice, which overexpress Toll-like receptor 7 and have severe proliferative glomerulonephritis, BAFF blockade attenuates disease, but there is no added benefit of CTLA-4Ig (24).
In both mouse models, consistent with the findings in normal mice (25, 26), BAFF blockade depletes transitional type 2, follicular, and marginal zone B cells, with sparing of germinal center responses, allowing the generation of high-affinity pathogenic autoantibodies by somatic mutation. In both models, prolonged B cell depletion has significant effects on spleen size. This results in decreased numbers of activated DCs and T cells in treated mice versus untreated controls, with less overall inflammatory burden and therefore less tissue damage. In both models, BAFF-R-Ig and TACI-Ig treatment are equally effective. However, plasma cells, which express BCMA and TACI but not BAFF-R, are depleted only by TACI-Ig. Short-lived IgM-producing plasma cells are more susceptible to depletion by TACI-Ig than are IgG-producing plasma cells, and TACI-Ig does not deplete long-lived bone marrow plasma cells in either of the lupus models we have studied. This is different from the reported findings in nonautoimmune mice (27) and may reflect the presence of other cytokines that support plasma cell survival in an inflammatory environment. TACI-Ig has also been reported to deplete short-lived IgM- and IgG-secreting plasma cells from the spleens of lupus prone MRL/lpr mice, resulting in the disappearance of autoantibodies from the serum (28).
NZM2410 mice are different from other SLE models for several reasons. First, they produce high levels of serum IL-4 and IgG1 autoantibodies (12). Second, they have a defect that delays the migration of long-lived plasma cells to the bone marrow and accumulates large numbers of these cells in the spleen (13, 14). Third, these mice have abnormalities in their B cell subsets, with loss of MZ B cells and an expansion of the B-1a subset (19). Fourth, NZM2410 mice develop a noninflammatory glomerulosclerosis, with few mediators of inflammation and infiltrating cells in the kidneys (12). Serum levels of BAFF, as in other SLE strains, increase with age in NZM2410 mice (data not shown).
We show here that a short course of a selective BAFF inhibitor completely prevents disease onset and is also highly effective at inducing complete and long-lasting remission after the onset of proteinuria in NZM2410 mice. The immediate effects of BAFF blockade were depletion of T2, MZ, and follicular B cells and a delay in the generation of the class-switched anti-dsDNA response. Because plasma cells are located in the spleens of NZM2410 mice, the contraction of spleen size induced by BAFF blockade resulted in a loss of plasma cells and a decrease in total serum IgG levels. The loss of plasma cell niches appeared to be the major reason for this decrease, since there was no difference between serum IgG levels in BAFF-R-Ig–treated and TACI-Ig–treated mice despite the more profound plasma cell depletion induced by blockade of both BAFF and APRIL.
Homeostatic mechanisms that prolong IgG half-life when serum levels of IgG are low (29) could help explain why the effects on serum levels of total IgG and autoantibodies were not as profound as would be predicted by the decrease in the actual number of antibody-forming cells. Even after the inhibitors were no longer present, the immune effects persisted for many months, including a contraction of the spleen size, attenuation of the expansion of splenic CD11b+ cells and CD4 memory T cells, and prolonged B-2 cell depletion. Nevertheless, germinal centers continued to generate autoreactive plasma cells that competed effectively for new plasma cell niches in the recovering spleens and produced autoantibodies capable of depositing in renal glomeruli.
It has been postulated that the expanded B-1a cell population in NZM2410 mice may be pathogenic because of enhanced antigen presentation function (19); B-1a cells were the major splenic B cell subset in the treated mice but were clearly not sufficient to mediate disease. Conversely, splenic B-1a cells also secrete IL-10 (30), a feature associated with regulatory B cell function (31). The phenotype of regulatory B cells is not fully defined, but there was no change in the percentage of CD5+CD1dhigh B cells (32) in the spleens of treated mice (data not shown). It remains to be determined whether the increased ratio of B-1a cells (or a regulatory subset of these cells) to B-2 cells exerted protective effects in the treated NZM2410 mice.
BAFF blockade also prevented the expansion of splenic CD11b+ cells and prevented the up-regulation of CD11b on resident renal CD11b+ cells. Although there was no alteration in the expansion of peripheral blood monocytes that normally occurs in NZM2410 mice with age, cells from treated mice expressed lower levels of CD11c. Since monocytes and dendritic cells turn over rapidly in the peripheral blood and spleen (33) and since most renal DCs turn over in <30 days in NZM2410 mice, these effects are most likely due to an alteration of the inflammatory environment, rather than a direct effect of BAFF inhibition on DCs themselves (34). It has been reported that activated monocytes express BAFF receptors (35); however, we were not able to detect by flow cytometry any differences in either TACI or BAFF-R expression on isolated renal mononuclear cells from treated and nephritic mice (data not shown).
Despite autoantibody formation and renal immune complex deposition, renal damage was prevented or reversed by a short course of BAFF blockade in NZM2410 mice. We have previously shown that in NZB × NZW mice, the major molecular renal biomarkers of proteinuria onset and of renal remission reflect activation of renal mononuclear phagocytes and endothelial cells (9). These were both attenuated in the treated NZM2410 mice. These findings show that deposition of immune complexes in the kidneys does not inevitably result in tissue damage but that activation of renal effector mechanisms is also required. For example, both the absence of activating Fc receptors on circulating monocytes and caspase inhibition can completely prevent renal inflammation in SLE mice despite immune complex deposition (36, 37). We have shown that remission of nephritis in NZB × NZW mice can be induced by a combination of cyclophosphamide and costimulatory blockade despite continued renal immune complex deposition (38).
The data presented here similarly demonstrate that immune complex deposition is not sufficient to cause glomerulosclerosis and proteinuria in NZM2410 mice. Podocyte loss, as assessed by the expression of the surrogate marker nephrin, was prevented in treated mice. Thus, the glomerular filtration barrier, consisting of podocytes, glomerular endothelial cells, and the basement membrane in between (39), was preserved in mice treated with BAFF inhibition. This tripartite structure has a set of highly integrated functions. Podocytes synthesize the glomerular basement membrane, and podocyte-derived vascular endothelial growth factor protects against thrombotic microangiopathy. During inflammation, endothelin released from damaged endothelium injures podocytes, and this injury is further compounded by local hypoxia. Podocyte loss then amplifies the endothelial damage (40). While immune complexes contribute to this process, additional inflammatory signals are likely required. The decrease in the numbers of B cells, which themselves have proinflammatory functions, and of activated T cells and DCs may all contribute to the sustained remission in NZM2410 mice treated with BAFF blockade. Our data in sum suggest that strategies for the prevention and treatment of SLE nephritis do not necessarily require autoantibody depletion but should be directed at decreasing systemic inflammation and protecting the local renal cells whose activation is associated with proteinuria onset.
It is clear that BAFF blockade consistently demonstrates the same effects on spleen cell populations and autoantibodies in different murine models of SLE, the only exception being the MRL/lpr model, in which all the plasma cells derive from an extrafollicular source and are depleted by TACI-Ig (28, 41). We have recently demonstrated similar effects of selective BAFF blockade in humans, with depletion of naive B cells, sparing of memory cells and plasma cells, and only a modest effect on circulating autoantibodies (42). Despite these similarities, the clinical effect of BAFF blockade on SLE nephritis varies in the different murine models. While prevention of SLE nephritis is achieved by selective blockade of BAFF in all models, treatment of active disease is not always effective. For example, in NZB × NZW mice with spontaneous disease onset, remission of nephritis is only achieved if BAFF blockade is combined with administration of CTLA-4Ig (14). In the same model, when disease is accelerated by type I interferon, BAFF blockade is beneficial only during the short initiation phase and does not prevent disease onset once autoantibodies are present, even when given before the onset of proteinuria (Liu Z, Davidson A: unpublished observations).
One reason for the differences observed in different models is that NZB × NZW mice with proteinuria have much higher expression of multiple mediators of inflammation in the kidneys than do NZM2410 mice with proteinuria. Thus, the clinical outcome of BAFF blockade in nephritis may depend on the presence of other mediators that collaborate with autoantibodies in the effector phase of renal inflammation or with BAFF in the reactivation of memory B cells or in supporting plasma cell survival. This observation has 2 implications of relevance to the treatment of human SLE. First, BAFF blockade may be most effective as a preventive therapy in patients with quiescent disease after induction of remission. Second, the presence of mediators of inflammation, for example, circulating mediators of inflammation (43) or the interferon signature, may need to be taken into account when examining the outcomes of clinical trials of BAFF inhibition.
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. Ramanujam, Bethunaickan, Davidson.
Acquisition of data. Ramanujam, Bethunaickan, Huang, Tao, Davidson.
Analysis and interpretation of data. Ramanujam, Bethunaickan, Madaio, Davidson.
- 1Clinical and laboratory features of SLE nephritis. In: Wallace DJ, Hahn BH, editors. Dubois' lupus erythematosus. 7th ed. Philadelphia: Lippincott Williams & Wilkins; 2007. p. 1112–30..