Apoptotic debris, autoantibody, and IgG–immune complexes (ICs) have long been implicated in the inflammation associated with systemic lupus erythematosus (SLE); however, it remains unclear whether they initiate immune-mediated events that promote disease. In this study, we show that PBMCs from SLE patients experiencing active disease, and hematopoietic cells from lupus-prone MRL/lpr and NZM2410 mice accumulate markedly elevated levels of surface-bound nuclear self-antigens. On dendritic cells (DCs) and macrophages (MFs), the self-antigens are part of IgG-ICs that promote FcγRI-mediated signal transduction. Accumulation of IgG-ICs is evident on ex vivo myeloid cells from MRL/lpr mice by 10 wk of age and steadily increases prior to lupus nephritis. IgG and FcγRI play a critical role in disease pathology. Passive transfer of pathogenic IgG into IgG-deficient MRL/lpr mice promotes the accumulation of IgG-ICs prior to significant B cell expansion, BAFF secretion, and lupus nephritis. In contrast, diminishing the burden IgG-ICs in MRL/lpr mice through deficiency in FcγRI markedly improves these lupus pathologies. Taken together, our findings reveal a previously unappreciated role for the cell surface accumulation of IgG-ICs in human and murine lupus.
Systemic lupus erythematosus (SLE) is a multisystemic autoimmune disease with genetic and environmental components that lead to autoimmunity and tissue-damaging inflammation (1, 2). There has long been an association between elevated levels of apoptotic debris and immune complexes (ICs) and their decreased clearance in SLE (3). Defects in opsonins such as mannose binding protein, complement components, and C-reactive protein reduce the clearance of apoptotic debris (4), and deficiency in DNase or RNase leads to poor lysosomal degradation (5). Although these defects heighten the burden of apoptotic debris and promote some of the phenotypes associated with lupus, ablation of opsonins or their receptors is insufficient to promote severe disease (6, 7). One consequence of heightened apoptotic debris is the exposure of the immune system to normally privileged nuclear self-antigens (8, 9). Cell-derived autoantigens exposed on apoptotic debris form ICs when bound by autoreactive IgG (henceforth referred to as IgG-ICs). Upon binding FcγRs or BCRs, they promote immune activation of B cells, macrophages (MFs), and dendritic cells (DCs) in part by delivering ligands to TLR7 and TLR9 (10, 11).
Activating FcγRs on human (FcγRI/IIa/IIc/IIIa) and murine (FcγRI/III/IV) phagocytic cells contain ITAMs that recruit Syk and activate the PI3K pathway (12). Activation of FcγRs (FcγRI, III, and IV in mouse) is regulated by coligation with ITIM-containing inhibitory (FcγRIIB) receptors. In mouse, FcγRIIB represses ITAM-containing FcγRs by recruiting SHIP to dephosphorylate PI(3,4,5)P3, thereby limiting downstream signal propagation (13, 14) and by Src homology region 2 domain–containing phosphatase 1 (SHP-1) through inhibitory signaling conditions called ITAMi that desensitizes receptor signal transduction (15).
Studies have identified FcγR polymorphisms as genetic factors influencing susceptibility to SLE and other autoimmune diseases (16, 17). Promoter polymorphisms that reduce FcγRIIB expression on germinal center and activated B cells are associated with murine and human SLE (18, 19). In addition, mice lacking FcγRIIB (20) develop lupus-like disease. Other functional polymorphisms in human FcγRIIa (R/H131) and FcγRIIIa (158V/F) decrease binding to IgG and are thought to diminish clearance of apoptotic debris, yet they are associated with lupus nephritis (21). Thus, polymorphisms in both activating and inhibitory FcγRs are associated with disease.
The pathogenic role of IgG-ICs in lupus has long been associated with their deposits in the kidneys and their ability to activate complement (C3) in lupus nephritis (22, 23). However, later studies showed that deposits of IgG and complement persist in the kidneys of lupus-prone mice when proteinuria and morbidity were diminished by blockade or genetic ablation of BAFF (24, 25). This indicates that IgG and complement deposits are not sufficient to induce lupus nephritis. Further studies using bone marrow chimeras showed that expression of FcγR on hematopoietic cells, rather than kidney mesangial cells, is required for lupus nephritis (26). This indicates that activation of the immune system through FcγRs on hematopoietic cells, rather than the deposits of IgG-IC in the kidney, is important in lupus nephritis. Studies also show that IgG-ICs promote autoantibody secretion in a TLR-dependent manner, and they contribute to immune responses associated with SLE in a TLR-independent manner (10, 11). However, it remains unclear how IgG-ICs plays a role in the pathogenic processes of SLE beyond internalizing TLR ligands to activate B cells.
In this paper, we show that IgG and apoptotic Ags (as IgG-ICs) accumulate on the surface of myeloid cells prior to the onset of SLE. In lupus-prone mice, nuclear self-antigens were displayed on hematopoietic cells, and in SLE patients, 67–75% with active disease accumulated nuclear self-antigens on peripheral blood B and T cells, and 10–40% displayed nuclear self-antigens on monocytes. On murine DCs and MFs, the Ags were contained within IgG-ICs bound to the activating FcRs, FcγRI and FcγRIV. In MRL/lpr mice, accumulation of IgG-ICs on FcγRI induced activation of the PI3K pathway and preceded lupus nephritis, whereas MRL/lpr mice lacking FcγRI (FcγRI−/−MRL/lpr) were protected. In contrast, inducing the accumulation of IgG-IC on the cell surface by passive transfer of anti-nucleosome IgG into AID−/−MRL/lpr mice promoted serological autoimmunity, BAFF secretion, and lupus nephritis. Taken together, these data identify that the accumulation of IgG-ICs on hematopoietic cells occurs during human and murine SLE and is associated with autoimmunity and disease pathogenesis.
Materials and Methods
C57BL/6 (B6) and MRL/MpJ-Tnfrs6lpr/J (MRL/lpr; JAX stock number 000485) colonies were maintained in an accredited animal facility at the University of North Carolina. FcγRI−/− (27, 28) and FcγRIII−/− mice (29) were obtained from Dr. A. Sperling (University of Chicago), FcγRIIB−/− (30) and FcγRIV−/− (31) from Dr. C. Jennette (University of North Carolina), and FcRγc−/− (32) from Dr. A. Szalai (University of Alabama at Birmingham). NZM2410 mice (33) were from Dr. G. Gilkeson (Medical University of South Carolina). MRL/MpJ mice were purchased from JAX (stock number 000486). AID−/−MRL/lpr mice were obtained from Dr. M. Diaz at National Institute on Environmental Health Sciences (34). FcγRI−/−B6 mice were crossed with MRL/lpr mice for 12 generations, followed by an intercross of FcγRI+/−MRL/lpr for 2 generations to produce FcγRI−/−MRL/lpr mice. Tail DNAs were analyzed by PCR (27).
Patients who showed SLEDAI scores ≥ 6, as defined by the Systemic Lupus Erythematosus Disease Activity Index (SLEDAI), were selected for enrollment after informed consent in accordance with the University of North Carolina institutional internal review board. Peripheral blood samples were collected with sodium heparin (BD Biosciences) during routine clinic visits.
Passive administration of anti-nucleosome Ab
AID−/−MRL/lpr, FcγRI−/−MRL/lpr, or B6 mice (18–21 wk old) were injected with 500 μg PL2-3 (anti-nucleosome, IgG2a) (35), F(ab′)2 of PL2-3, or Hy1.2 (isotype control IgG2a, anti-trinitrophenol [TNP]) i.v. per mouse every week for 2 or 5 wk. Five days after the fifth injection, mice were euthanized.
Splenocytes or total kidney cells were prepared into a single cell suspension. Splenic DCs and MFs were isolated using MicroBeads for CD11c+ cells (DCs) and CD11b+ cells (MFs) following the manufacturer’s instructions (Miltenyi Biotec).
For serum anti-nucleosome and anti-dsDNA IgM or IgG levels, plates were coated with dsDNA (10 μg/ml; Sigma-Aldrich) in the presence or absence of histone (40 μg/ml; Sigma-Aldrich). Total serum IgM or IgG levels were measured on a plate coated with anti-IgM or anti-IgG Abs. For serum BAFF, anti-mouse BAFF (5A8; Enzo) and biotinylated anti-mouse BAFF (1C9; Enzo) were used. Duplicates of serially diluted serum samples were tested.
To analyze the numbers of BAFF-secreting cells, isolated splenic DCs or MFs (1 × 106 and 0.5 × 106 cells/well) were incubated for 60 h at 37°C on ELISPOT plates (Millipore) coated with 5A8 and then detected using biotinylated 1C9.
Formalin-fixed (10%), paraffin-embedded kidneys were stained with H&E (8-μm sections). Glomerular changes and tubulointerstitial inflammation were assessed by a pathologist blinded to the experimental groups. The following criteria were used: glomerular lesions—0, no H&E changes; 1, minimal mesangial hypercellularity without visualized immune deposits; 2, focal immune deposits; 3, diffuse glomerulonephritis with widespread subendothelial immune deposits; and 4, global immune deposits with associated sclerosis. Interstitial inflammation was scored 0–3 based on the degree of tubulointerstitial involvement: 0, no infiltrate or inflammation; 1, <10%; 2, between 10 and 50%; and 3, >50% of the tubulointerstitium.
Urine protein was measured using Uristix strip following the manufacturer’s instruction (Siemens).
Snap-frozen kidney sections (8 μm) were stained with goat anti-mouse IgG-Fc conjugated with DyLight 488 (Jackson ImmunoResearch Laboratories) and PE-conjugated anti-mouse complement component 3 (C3) (Cedarlane Laboratories) and then visualized on LSM 710 confocal microscope (×20 magnification lens; Carl Zeiss).
Total splenocytes were stained with biotinylated 2.12.3 (mouse anti-mouse Smith [Sm], IgG2a) (36, 37) or goat anti-mouse IgG-Fc (Jackson Immunoresearch). Biotinylated anti-canine distemper virus (IgG2a) or goat anti-rabbit IgG were used as isotype control staining Abs (ISO). Biotinylated Abs were detected using streptavidin-Alexa Fluor 647 (Invitrogen). The mean fluorescence intensity (MFI) of surface-bound Ag or IgG staining was divided by the MFI values of ISO. For intracellular staining, fixed and permeabilized cells were stained with anti-mouse total Syk (Santa Cruz Biotechnology), pSyk (BD Biosciences), total Akt, pAkt-threonine (Thr308), total S6, or pS6 (Cell Signaling Technology). The expression levels of intracellular kinases were calculated as follows: (MFI of phosphorylated signaling molecule/MFI of ISO)/(MFI of total signaling molecule/MFI of ISO). The fold changes over average values of control mice from each group are graphed. For FcγRI staining, cells were blocked with rat serum and stained with biotinylated X54-5/7.1. Whole-blood cells from healthy donors or SLE patients were stained for surface DNA (anti-human DNA, 33H11; T. Winkler; University of Erlangen, Erlangen, Germany) (38), Sm (2.12.3), or IgG (anti-human IgG Fc-PE; Jackson ImmunoResearch Laboratories). 33H11 and 2.12.3 were conjugated to Alexa Fluor 647 following manufacturer’s instruction (Invitrogen). Samples were acquired on Cyan flow cytometer (Beckman Coulter). Cells were defined as follows: human and mouse B cells (CD19+) and T cells (CD3+), mouse DCs (CD11chiCD11b+), mouse MFs (CD11c−CD11bhi), and human monocytes (CD14+). Murine nonleukocytes are gated as CD45neg population.
Isolated murine splenic DCs or MFs, or whole-blood cells from human samples, were stained for IgG, Sm (2.12.3), or DNA (33H11). Colocalization was quantified by calculating the Mander’s coefficient using ImageJ. For some experiments, cells were incubated with trypsin or DNase for 20 min (37°C) prior to the staining for surface-bound Ags.
Splenic MFs were isolated by incubation on a glass dishes for 2 h at 37°C. RNA was extracted using RNeasy Mini Kit (Qiagen). cDNA was generated using iScript cDNA synthesis kit (Bio-Rad). The following PCR primers were used: GAPDH, 5′-GGC-AAA-TTC-AAC-GGC-ACA-3′ (forward) and 5′-GTT-AGT-GGG-GTC-TCG-CTC-CTG-3′ (reverse); and FcγRI 5′-ACA-CAA-TGG-TTT-ATC-AAC-GGA-ACA-3′ (forward) and 5′-TGG-CCT-CTG-GGA-TGC-TAT-AAC-T-3′ (reverse). Quantitative real-time PCR was performed on ABI Prism 7500 Sequence Detection System (Applied Biosystems).
The Mann–Whitney U test was used to analyze human data. Kruskal–Wallis test (>3 groups) or Mann–Whitney U test (≤3 groups) was used to compare changes over control group. One-way ANOVA was used to compare changes between different conditions. Statistical analysis was performed using GraphPad Prism (GraphPad Software).
Nuclear self-antigens are displayed on hematopoietic cells
The clearance of apoptotic debris is critical in maintaining immune homeostasis; however in autoimmune-prone mice, continuous cell turnover can expose the immune system to self-antigens on apoptotic debris. To assess whether apoptotic debris was present on cells from nonautoimmune B6 mice, we stained splenocytes with Abs specific for Sm, a nuclear self-antigen evident on apoptotic debris (9). Although Sm was not detected on B6 T cells, the levels of Sm were heightened 7- to 10-fold on splenic DCs, MFs, and B cells when compared with isotype control Ab (Fig. 1A). Staining was not unique to anti-Sm as Abs specific for DNA (33H11), and nucleosomes (PL2-3) showed a similar staining pattern (Supplemental Fig. 1A), which by microscopy appeared punctate (Fig. 1B). To further characterize the Ags, we treated B6 splenocytes with trypsin (Fig. 1B) or DNase (Fig. 1C) prior to surface staining for DNA, nucleosome, or Sm. DNase treatment specifically removed DNA, whereas nucleosomes remained intact. As expected, all nuclear self-antigens were removed by trypsin in agreement with their display on the cell surface. These data indicate that nuclear self-antigens contained within apoptotic debris are displayed on the surface of hematopoietic cells.
Elevated levels of apoptotic debris have been associated with autoimmunity in murine and human SLE (39). To assess whether hematopoietic cells displayed elevated levels of nuclear self-antigens, we quantified the levels of Sm on cells from MRL/lpr mice aged 18–26 wk (Fig. 1D, 1F). Relative to isotype control, Sm levels increased 84- to 260-fold on DCs, MFs, and B cells. Surprisingly, T cells also displayed a 109-fold increase. The staining was specific because T cells from B6 mice did not display Sm (Fig. 1A). Furthermore, Sm was not evident on splenic nonleukocyte populations (CD45neg) from B6 or MRL/lpr mice (Fig. 1D), indicating that the accumulation of nuclear self-antigens is restricted to hematopoietic cells. No differences between genders were observed. To further assess whether nuclear self-antigens on splenic hematopoietic cells and PBMCs from B6 and MRL/lpr mice accumulated over time, we quantitated the levels of Sm over 30 wk (Fig. 1E). Surprisingly, by 3 wk of age, DCs, monocytes/MFs, B cells, and T cells from blood and spleen of MRL/lpr mice showed high levels in surface Sm (10-, 3-, 5-, and 30-fold on DCs, MFs, B cells, and T cells), compared with peripheral blood cells or splenocytes from B6 mice (single data points at weeks 3, 6, and 30). These levels declined during the next 6 wk. After week 9, the levels of Sm on splenic MFs and B cells steadily increased, reaching a maximal level at 30 wk when urine protein levels are high (score > 2) (Supplemental Fig. 1B). Similarly, the Sm levels on blood and splenic DCs and T cells, and blood monocytes showed steady increase after 9 wk, returning to levels found at week 3.
To assess whether accumulation of nuclear self-antigens was unique to the MRL/lpr background, we examined other models. In NZM2410 mice, Sm levels were heightened on the surface of all hematopoietic cells, reaching levels that were 35- to 78-fold higher than isotype control (Fig. 1F). Similarly, Sm levels were elevated on cells from MRL/MpJ mice, but not on cells from B6/lpr or B6/Merkd (Fig. 1F), suggesting that accumulation of nuclear self-antigens is associated with the lupus-prone background.
SLE patients with active disease exhibit heightened levels of surface-bound nuclear Ags
The accumulation of nuclear self-antigens in two spontaneous models of murine SLE suggests an inherent defect in immune cells that could be present in human SLE. To assess this, we analyzed circulating mononuclear cells from healthy controls and SLE patients for surface Sm and DNA (Fig. 2). We chose SLE patients experiencing active disease (SLEDAI > 6) because active disease in humans might be most like disease in MRL/lpr mice (Supplemental Table I). Similar to nonautoimmune mice (Fig. 1A), healthy controls showed low levels of Sm and DNA on B cells and monocyte but not T cells (Fig. 2A, 2B). In contrast, 67–75% of SLE patients showed more than a 2-fold increase of Sm and DNA on B and T cells compared with healthy controls. Among the patients with active disease, there was significant variation in the levels of Sm and DNA on the B cells. In the case of DNA, two SLE patients showed 30- to 40-fold increases, four showed 2- to 7-fold increases, and the rest showed less than a 2-fold increase. On blood monocytes, 10–40% of SLE patients showed a 2- to 3-fold increase of Sm and DNA (Fig. 2A). The findings that blood monocytes from SLE patients accumulate less nuclear self-antigens compared with healthy controls is consistent with the murine data where blood monocytes showed similar levels of surface Sm as B6 blood monocytes (Fig. 1E). In the patients whose monocytes and B and T cells showed an accumulation of nuclear self-antigens (Fig. 2A), microscopy showed punctate staining of DNA (Fig. 2C; T cell example) similar to that seen in lupus-prone mice, suggesting receptor aggregation. Thus, the data show that nuclear self-antigens accumulate on hematopoietic cells in human and murine lupus.
Nuclear self-antigens are bound to FcγRs as IgG-ICs
Multiple cell surface receptors clear apoptotic debris, including FcγRs. One possibility is that the nuclear self-antigens displayed on myeloid cells represent ICs bound to FcγRs. To test this idea, we quantified the levels of Sm on immune cells from B6 mice that were deficient in individual FcγRs (Fig. 3A). We found that loss of activating receptor FcγRI or FcγRIV reduced Sm levels 40 or 45% on B6 DCs and 50% on B6 MFs. The receptor(s) displaying self-antigen on B and T cells remains unclear; however, consistent with the limited expression of FcγRs on B and T cells, loss of any FcγR did not reduce the levels of Sm by >10% (data not shown).
We reasoned that if accumulated nuclear self-antigens were part of ICs bound to FcγRI/FcγRIV, surface IgG levels would be elevated on myeloid cells. In B6 mice, DCs and MFs consistently showed low levels of IgG (Fig. 3B) and Sm (Fig. 1A, 1E) over 30 wk, and they failed to develop lupus nephritis (Supplemental Fig. 1B). In MRL/lpr mice, the levels of IgG remained low (at levels of B6 mice) until 10 wk of age and then steadily increased until 20 wk, rising sharply between weeks 20 and 30 (14-fold on DCs, 5-fold on MFs compared with B6). This was consistent with the changes in the levels of Sm between 10 and 30 wk of age (Fig. 1E). On DCs and MFs, 70% of the surface IgG was IgG2a and IgG2b (Supplemental Fig. 1C) that appeared prior to high levels of proteinuria (score > 2; Supplemental Fig. 1B). In addition, SLE patients that accumulated Sm and DNA on blood monocytes also expressed surface IgG (Supplemental Fig. 1D). Display of nuclear self-antigens on myeloid cells did not involve FcμR because IgM was not found on DCs or MFs from B6 or MRL/lpr mice (Supplemental Fig. 1E).
To further define the role of ICs in the surface display of self-antigen, we assessed whether Sm and IgG colocalized and showed punctate staining. We found a 6- and 4-fold increase in the colocalization of Sm and IgG on MRL/lpr DCs and MFs that exhibited punctate staining consistent with receptor aggregation (Fig. 3C, 3D). To assess whether cell signaling was occurring, we measured the activation state of kinases coupled to FcγRs. We found that the levels of pSyk were increased 2- to 3-fold, pAkt-Thr308 1.4-fold, and pS6 2-fold in both MRL/lpr DCs and MFs, suggesting chronic activation of the PI3K pathway (Fig. 3E–G). Thus, myeloid cells from MRL/lpr mice accumulate IgG-ICs containing nuclear Ags that are bound in part by FcγRI and FcγRIV.
One possibility was that accumulation of IgG-ICs on MRL/lpr DCs and MFs might reflect increased expression of FcγR. We found that DCs and MFs from MRL/lpr mice showed 3- and 2-fold increases in FcγRI compared with B6 mice (Fig. 3H). Because the elevated levels of FcγRI could reflect an increased rate of transcription, we quantitated FcγR message levels by quantitative PCR. Surprisingly, we found that the levels of FcγRI mRNA from MRL/lpr mice were comparable to those in B6 mice (Fig. 3I). This indicates that accumulation of IgG-ICs on the cell surface is associated with increased surface expression of FcγRI; however, these increased levels do not reflect increased transcription, suggesting that the FcγRs might be recycled.
FcγRI−/−MRL/lpr mice show reduced levels of IgG-ICs and diminished autoantibody levels
Our findings indicate that IgG-ICs accumulate on hematopoietic cells in human and murine SLE and induce signal transduction by activating FcγRs. To define whether FcγRI contributes to disease, we used MRL/lpr mice deficient in FcγRI (FcγRI−/−MRL/lpr). Previous studies have found that MRL/lpr mice have elevated splenic and lymph node cellularity because of enhanced lymphoproliferation (40). Loss of FcγRI in MRL/lpr mice reduced the numbers of splenic DCs and MFs to levels found in B6 mice and partially reduced the numbers of T and B cells (Fig. 4A). This suggests that FcγRI plays a significant role in the expansion of myeloid cells in MRL/lpr mice. Although B and T cells typically do not express FcγRI, their numbers were reduced in FcγRI−/−MRL/lpr mice, suggesting that loss of FcγRI may indirectly impact lymphocytes.
To define whether FcγRI contributes to the accumulation of ICs in vivo, we quantitated the levels of surface Sm and IgG on DCs and MFs from FcγRI−/−MRL/lpr mice. Compared with MRL/lpr mice, the levels of Sm were reduced 30–40% on DCs and MFs (Fig. 4B), and the levels of IgG were decreased to levels found on B6 cells (Fig. 4C). This reduction was similar to the contribution of FcγRI in the low level of IgG-ICs displayed on DCs and MFs from B6 mice (Fig. 3A). The levels of Sm on B and T cells were not altered in the absence of FcγRI, consistent with the lack of FcγRI expression on these cells (Fig. 4B) and the idea that other receptors are involved in the display of self-antigens on lymphocytes.
To establish whether the accumulation of IgG-ICs contributes to the autoimmune phenotype of MRL/lpr mice, we quantitated the levels of intracellular pSyk, pAkt, pS6, serum BAFF, and autoantibody levels in FcγRI−/−MRL/lpr mice. In these mice, the levels of pSyk and pS6 in MFs and DCs, and the levels of pAkt-Thr308 in DCs were diminished to the levels found in B6 mice, whereas the levels of pAkt-Thr308 in MFs were only reduced 15% (Fig. 4D–F). Furthermore, FcγRI−/−MRL/lpr mice showed a 200-fold decrease in serum anti-nucleosome Ab (Fig. 4G), and a 95-fold decrease in serum anti-dsDNA Abs (Fig. 4H) when compared with MRL/lpr (FcγRI+/+) mice. Similarly, FcγRI−/−MRL/lpr mice showed a 4-fold decrease in serum BAFF (Fig. 4I). Despite these significant improvements in the serological phenotype associated with autoimmunity, the levels of autoantibody and BAFF remained elevated compared with B6 mice. This is consistent with the idea that other receptors display self-antigens on lymphocytes, and the findings that loss of FcγRI alone does not ablate the accumulation of IgG-ICs on myeloid cells. Despite the multiple ways IgG-ICs accumulate on myeloid cells, the data show that FcγRI plays a significant role in the disease of MRL/lpr mice coincident with defects that lead to accumulation of IgG-ICs and chronic FcγR activation.
FcγRI−/−MRL/lpr mice fail to develop nephritis
To assess whether FcγRI is important in lupus nephritis, we quantitated the numbers of renal hematopoietic cells and assessed renal pathology in FcγRI−/−MRL/lpr mice. We found that compared with MRL/lpr mice, the number of T cells were reduced 6-fold, DCs 2-fold, and MFs 4-fold. Interestingly, the number of B cells that infiltrated the kidney was less than B6 controls making the reduction 89-fold (Fig. 4J). This indicates that migration of DC, MF, T, and B cells to the kidney is dependent on FcγRI. In addition, the kidneys showed significantly reduced glomerular inflammation (Fig. 4K) and urine protein levels (Fig. 4L), although levels remained higher than B6 controls, consistent with a partial role of FcγRI in kidney disease.
FcγRI−/−MRL/lpr mice failed to develop lupus nephritis coincident with significantly reduced serological phenotype. This could reflect reduced autoantibody production that diminishes the formation of IgG-ICs, hence, the accumulation on the cell surface and heightened signal transduction. Alternatively, lupus nephritis might develop independent of FcγRI. To sort out these possibilities, we passively transferred pathogenic anti-nucleosome IgG2a (PL2-3) into FcγRI−/−MRL/lpr mice. We found that 5 wk of PL2-3 injection did not change the levels of Sm and IgG on the surface of DCs and MFs or induce glomerular inflammation or proteinuria in FcγRI−/−MRL/lpr mice compared with PBS-treated control mice (Supplemental Fig. 2A–E). However, deposits of IgG and C3 remain evident in the kidney of MRL/lpr mice regardless of disease pathology or the presence of FcγRI (Supplemental Fig. 2F). Collectively, the data show that FcγRI plays a major role in many of the serological and cellular phenotypes associated with autoimmunity (Fig. 4A–I). It also partially contributes to kidney disease (Fig. 4J–L). This suggests that FcγRIV may also play a role in kidney pathology (41). We are currently backcrossing FcγRIV−/− mice to the MRL/lpr background to assess this possibility.
Passive transfer of IgG heightens surface ICs coincident with early autoantibody secretion
Our results show that loss of FcγRI markedly reduces the accumulation of IgG-ICs on myeloid cells and the pathologies associated with SLE (Fig. 4). To begin to understand their role in disease, we defined the temporal order of events leading to lupus-related pathologies. For this, we developed an in vivo model where we passively transferred pathogenic anti-nucleosome IgG2a into IgG-deficient MRL/lpr mice. Our selection of anti-nucleosome IgG2a (PL2-3) was based on in vitro experiments showing that coculturing PL2-3, but not anti-TNP (Hy1.2; IgG2a), during the derivation of MRL/lpr bone marrow–derived DCs promoted the accumulation of IgG and secretion of BAFF (Supplemental Fig. 1F, 1G). Bone marrow–derived DCs from B6 mice treated with PL2-3 failed to secrete BAFF or increase the level of surface IgG supporting the findings that accumulation is unique to mice genetically prone to lupus (Fig. 3B–D). In addition, anti-nucleosome (PL2-3) is pathogenic, and the IgG2a isotype binds to FcγRI and FcγRIV and is a highly displayed isotype on the surface of myeloid cells from MRL/lpr mice (Supplemental Fig. 1C).
Our model uses AID−/−MRL/lpr mice since deficiency in activation-induced cytidine deaminase (AID) prevents class switch. These mice also fail to develop serological phenotypes of autoimmunity or lupus nephritis (34). Thus, passive transfer of anti-nucleosome IgG2a (PL2-3) would induce disease pathology and allow us to order the events that occur during the onset of autoimmune disease. AID−/−MRL/lpr mice were intravenously treated with PL2-3 weekly for 2 or 5 wk (Fig. 5A). Separate cohorts were treated with PBS, isotype control Ab, or F(ab′)2 of PL2-3. We found that after 2 wk of PL2-3 treatment, the levels of surface IgG increased 2-fold on DCs and MFs and showed a punctate staining similar to that found on MRL/lpr mice (Figs. 5B, 3C). By 5 wk, the levels of IgG on DCs increase to 3-fold, whereas on MFs, the levels remained comparable to those at 2 wk. Surface accumulation of IgG was not evident when PL2-3 was injected into B6 mice (Supplemental Fig. 2B) or when AID−/−MRL/lpr mice were treated with F(ab′)2 of PL2-3 or an isotype control Ab (anti-TNP, Hy1.2, IgG2a) (Fig. 5B). To assess whether the treatment of PL2-3 prolonged or enhanced signaling from FcγRs, we quantitated the levels of pSyk in ex vivo DCs and MFs (Fig. 5C). After 5 wk of PL2-3 treatment, pSyk levels in DCs were increased 1.8-fold, and in MFs, the levels were increased 1.5-fold compared with PBS-treated mice. This suggests that PL2-3 treatment promotes the surface accumulation of IgG-ICs and activates FcγRs on myeloid cells.
To define whether the accumulation of IgG-ICs was associated with autoantibody titers in PL2-3–treated AID−/−MRL/lpr mice, we enumerated B cells and measured serum autoantibody levels after 2 and 5 wk of PL2-3 treatment. We found that by 2 wk, there was a 1.5-fold increase in the number of splenic B cells that reached 3.6-fold after 5 wk of injection (Fig. 5D). The numbers of DCs, MFs, and T cells were not different at 2 or 5 wk postinjection (data not shown). Initially, the expanded B cells secreted low levels of anti-dsDNA IgM (2 wk); however, by 5 wk of treatment, the levels of anti-nucleosome and anti-dsDNA IgM were increased 6- and 7-fold, respectively (Fig. 5E, F). Because the levels of total IgM were not affected (Fig. 5G), the data suggest that the increase in autoantibody titers and the number of B cells was due to activation and expansion of autoreactive B cells. Thus, accumulation of IgG-ICs on the surface of myeloid cells is concurrent with B cell expansion and the initial production of autoantibody.
Heightened levels of BAFF allow autoreactive B cells to survive during the transitional stage of development (42). To determine whether accumulation of IgG-ICs promotes BAFF secretion in vivo, we enumerated BAFF-secreting MFs and DCs in AID−/−MRL/lpr mice after 2 and 5 wk of PL2-3 injection (Fig. 5H). We found that production of BAFF in MRL/lpr MFs reached a maximal 11-fold increase at 2 wk, the time point correlated with the maximal accumulation of IgG-ICs on MFs (Fig. 5B), then declined to 4-fold over PBS or F(ab′)2 controls. In contrast, BAFF secretion in DCs increased 4-fold at 2 wk and an additional 10-fold by 5 wk post–PL2-3 injection. Thus, surface IgG-ICs accumulate rapidly on MFs concurrent with early BAFF secretion but preceding the significant increases in the number of B cells and the levels of autoantibodies. In contrast, the accumulation of IgG-ICs on DCs and their secretion of BAFF are delayed, occurring concomitantly with B cell expansion and heightened autoantibody.
Passive transfer of anti-nucleosome IgG promotes renal disease
Renal failure is one of the leading causes of mortality in human and murine SLE. To understand whether the accumulation of IgG-ICs on the surface of cells precedes lupus nephritis, we assessed renal pathology after 2 and 5 wk of PL2-3 treatment. At 2 wk, H&E-stained kidney sections from PL2-3–treated mice did not show morphologic changes despite the accumulation of IgG on DCs and MFs (data not shown; Fig. 5B). After 5 wk of PL2-3 treatment, extensive tubular and glomerular inflammation, including the formation of fibrocellular crescents was evident (Fig. 5I). These changes were much like those found in MRL/lpr mice. In accordance, scores for glomerular and tubulointerstitial inflammation, and proteinuria were increased in MRL/lpr mice and in PL2-3–treated AID−/−MRL/lpr mice, but not in B6-, PBS-, or F(ab′)2-treated AID−/−MRL/lpr mice (Fig. 5J–L). The ability of PL2-3 to induce kidney pathology was specific to lupus-prone mice because B6 mice treated with PL2-3 did not develop renal pathology (Supplemental Fig. 2C–F). Our data in the passive Ab transfer model show that the accumulation of IgG-ICs on the cell surface precedes glomerulonephritis. This is much like the timeline of the accumulation of surface IgG in MRL/lpr mice where IgG-ICs increase at 10 wk of age prior to high levels of proteinuria (score > 2) (Fig. 3B, Supplemental Fig. 1B). Collectively, our data show that passive transfer of anti-nucleosome IgG into AID−/−MRL/lpr mice induces the surface accumulation of IgG-ICs on myeloid cells as the early Ab response begins, promoting chronic FcγRI signaling and BAFF secretion that leads to extensive B cell expansion and lupus nephritis.
Our findings identify a previously unidentified defect that promotes the accumulation of nuclear Ags (Sm, DNA, and nucleosomes) on the surface of hematopoietic cells. This defect was evident in two genetically unrelated strains of lupus-prone mice and on PBMCs from SLE patients. On myeloid cells, the self-antigens were part of IgG-ICs bound by the activating receptors FcγRI and FcγRIV. B and T cells also accumulated self-antigens; however, the receptors involved remain unknown. We used several mouse models to define whether accumulation of IgG-ICs on the surface of myeloid cells contributes to the pathogenesis of SLE. In the AID−/−MRL/lpr model treated with anti-nucleosome, we showed that IgG-ICs accumulated concomitantly with the activation of Syk and secretion of BAFF but prior to the significant expansion of autoreactive B cells and lupus nephritis. F(ab′)2 of anti-nucleosome IgG did not elicit changes in the serological phenotype or renal disease, indicating that the autoimmune phenotype required IgG-ICs/FcγR interactions. In SLE patients experiencing active disease (SLEDAI > 6), nuclear self-antigens were displayed on the surface of peripheral blood B cells and to a lesser extent on T cells and monocytes. In addition, loss of FcγRI (FcγRI−/−MRL/lpr) reduced surface IgG, decreased signal transduction (pSyk, pAkt, and pS6), diminished serum BAFF, and reduced kidney disease. Taken together, the data show that after the early autoantibody response, chronic interaction of IgG-ICs and activating FcγRs amplifies the disease process. It also provides insight into how FcγRs on circulating myeloid cells, rather than kidney mesangial cells, might contribute to renal pathology in SLE (26, 43).
Autoreactive B cells are normally maintained in an unresponsive state as a result of chronic engagement of the BCR by self-antigens (44). The findings that membrane-bound apoptotic self-antigens are present at low levels on the surface of DCs, MFs, and B cells from B6 mice (Fig. 1A) raises the possibility that they deliver tolerogenic signals to B cells (45). This is supported by previous studies in the hen egg lysosome (HEL) model of tolerance, showing that membrane-bound HEL induces a stronger BCR signal than soluble HEL (45), and in the Sm model of tolerance, showing that soluble Sm or snRNPs fail to maintain the unresponsive state associated with anergy (46). Thus, in nonautoimmune mice, the low levels of nuclear self-antigens on DCs, MFs, and B cells may tolerize low-affinity autoreactive B cells. However, when a high burden of nuclear Ags accumulate on hematopoietic cells in MLR/lpr mice, they could promote a break in tolerance (44) by providing a source of TLR agonists that chronically stimulate TLR7 and TLR9 (47) or by providing a source of membrane-bound nuclear Ags that activate autoreactive BCRs “in trans.” Similarly, accumulated IgG-ICs on myeloid cells constitutively cross-link activating FcγRs and heighten cytokine secretion (Figs. 3A–G, 4B–F, 5B, 5C, 5H, Supplemental Fig. 1F, 1G).
Our data show that hematopoietic cells from both B6 and MRL/lpr mice display nuclear self-antigens; however, only MRL/lpr mice develop disease. The low levels of self-antigens displayed on B6 hematopoietic cells and their lack of autoimmune disease suggest the need for a threshold level of activation to promote the autoimmune pathology. Another possibility is that a protective signal is conferred by opsonins other than IgG that coat apoptotic debris (48). We found that B6 cells displayed low levels of Sm; however, IgG was barely detectable (Figs. 1A, 1E, 1F, 3B–D). Similarly, disease-free MRL/MpJ or 3- to 9-wk-old MRL/lpr mice showed high levels of surface-bound nuclear Ags despite low levels of surface IgG (Figs. 1E, 1F, 3B). This might reflect the binding of C-reactive protein (CRP), an acute phase protein that opsonizes apoptotic debris and binds to activating FcγRI and inhibitory FcγRIIB (49). Whether or how the exchange of CRP for IgG could promote disease remains unclear; however, CRP might confer unique downstream signals that are protective (50) because CRPTg(NZB/NZW)F1 mice and CRP-treated MRL/lpr mice show delayed disease (51, 52). Although CRP-bound apoptotic debris may be protective in a non-autoimmune-prone environment, in MRL/lpr mice, the high burden of apoptotic debris could activate intracellular TLRs, triggering a break in tolerance and the production of autoreactive Abs at early ages. This would increase the production of IgG-ICs, thus amplifying the downstream effects of FcγRI. It is also supported by the data in Fig. 4B and 4C showing that, in FcγRI−/−MRL/lpr mice, the levels of Sm were moderately decreased (30–40%), whereas the decrease in IgG were more significant (50–60%) and coincident with the lack of disease pathology. The data are consistent with the idea that binding of pathogenic IgG to FcγRI gives rise to lupus pathology.
It was striking that AID−/−MRL/lpr mice passively administered pathogenic IgG2a developed fulminant lupus nephritis, whereas AID−/−MRL/lpr mice treated with F(ab′)2 of IgG2a or FcγRI−/−MRL/lpr mice treated or untreated with intact IgG2a were void of severe renal disease. The lack of severe disease in the latter mice was observed despite the presence of renal IgG and C3 deposits (Supplemental Fig. 2F). Taken together, these findings indicate that the interaction of IgG-ICs and FcγRI plays crucial roles in the pathogenesis of lupus nephritis, but deposits of IgG/C3 in the kidney appear not to be sufficient to induce fulminant disease. It is possible that passive transfer of anti-nucleosome IgG2a into AID−/−MRL/lpr mice induces deposits in the kidney independent of FcγRs by forming ICs with nuclear Ags that deposit on the glomerular basement membrane (53). However, because resident renal cells such as mesangial cells, podocytes, and renal endothelial cells do not express FcγRI (43), our data indicate that lupus nephritis is dependent on constant binding of IgG-ICs to FcγRI expressed on hematopoietic cells. Whether disease depends on the secretion of FcγR-dependent cytokines or signals that promote the migration of cells to the kidney is under investigation. It is noteworthy that although a previous study showed lupus nephritis in MRL/lpr mice lacking the Fc common γ-chain (FcRγc, a subunit of murine FcγRI, III, and IV) (54), another study had found that FcRγc-deficient mice maintain a partially functional FcγRI (28). Coupled with our data, it raises the possibility that FcγRI contributed to disease in the MRL/lpr model.
An interesting observation in autoimmune MRL/lpr mice is that the burden of IgG-ICs varies depending on the source of cells. All hematopoietic splenocytes from MRL/lpr mice accumulate ICs (Fig. 1D–F); however, the levels of surface IgG-ICs on blood monocytes (week 30) were comparable to those from B6 mice (Fig. 1E). This was also evident in human SLE where the levels of surface-bound nuclear Ags on circulating monocytes were comparable to those from healthy donors (Fig. 2A), despite reports that human blood monocytes express elevated levels of FcγRI (55). Circulating monocytes activate and differentiate into MFs upon migration to tissues. Thus, it is possible that accumulation of IgG-ICs on blood monocytes promotes their migration to the tissues leaving only blood monocytes that display low levels of nuclear self-antigens. Given that activating FcγRs account for 50–60% of the display of nuclear self-antigens on myeloid cells (Figs. 3A, 4C), another possibility is that MFs use receptors that are not highly expressed on blood monocytes for the display of IgG-opsonized apoptotic debris.
What leads to the accumulation of IgG-ICs on the surface of cells is currently under investigation. Possible defects include failure to internalize IgG-ICs bound by FcγRI, disrupted trafficking to the lysosome, diminished degradation in the lysosome, or aberrant recycling of IgG-ICs (56–58). Our study provides a new insight into how apoptotic debris and IgG-ICs contribute to heightened BAFF secretion and lupus nephritis. Recent advances in treating lupus show that 43–58% of patients treated with anti-BAFF reduce SLEDAI scores more than 4 points compared with 36–46% in control subjects (59). Therefore, understanding the molecular events that lead to the accumulation of nuclear self-antigen might prove fruitful in providing a means to reduce high BAFF and simultaneously other activating FcγR-mediated events that promote autoimmune pathologies.
We thank Drs. Charles Jennette (University of North Carolina) for FcγRIV−/−B6 mice and FcγRIIB−/− mice; Alex Szalai (University of Alabama at Birmingham, Birmingham, AL) for FcRγc−/− mice, Anne Sperling (University of Chicago) for FcγRI −/− and FcγRIII−/− mice; Gary Gilkeson (Medical University of South Carolina) for NZM2410 mice; George Tsokos (Beth Israel Deaconess Medical Center) for B6/lpr mice, Mark Hogarth (Burnet Institute) for anti-FcγRI Ab (X54-5/7.1); and Thomas Winkler and Joachim Kalden (University Erlangen, Nuernberg, Germany) for anti-DNA Ab (33H11). We also thank Diane Carnathan and Chris Hilliard for technical assistance in the early study establishing the accumulation of IgG-ICs on hematopoietic cells. We thank the Lineberger Comprehensive Cancer Center Biostatistics Core Facility for support.
This work was supported by National Institutes of Health (NIH) Grants R01AI070984 and R21AI105613, Alliance for Lupus Research, the National Center for Advancing Translational Sciences (NIH) through Grant 1UL1TR001111, the Flow Cytometry Core (NIH/National Cancer Institute Grant P30CA016086), and the Microscopy Services Laboratory (NIH Grant CA 16086-26). A.J.M. was supported by NIH Grant 5T32AI07273-27.
The online version of this article contains supplemental material.
Abbreviations used in this article:
activation-induced cytidine deaminase
hen egg lysosome
isotype control staining Ab
mean fluorescence intensity
systemic lupus erythematosus
Systemic Lupus Erythematosus Disease Activity Index
The authors have no financial conflicts of interest.