The C3aR promotes macrophage infiltration and regulates ANCA production but does not affect glomerular injury in experimental anti-myeloperoxidase glomerulonephritis

The anti-neutrophil cytoplasmic antibody (ANCA) associated vasculitides are autoimmune diseases associated with significant morbidity and mortality. They often affect the kidney causing rapidly progressive glomerulonephritis. While signalling by complement anaphylatoxin C5a though the C5a receptor is important in this disease, the role of the anaphylatoxin C3a signalling via the C3a receptor (C3aR) is not known. Using two different murine models of anti-myeloperoxidase (MPO) glomerulonephritis, one mediated by passive transfer of anti-MPO antibodies, the other by cell-mediated immunity, we found that the C3aR did not alter histological disease severity. However, it promoted macrophage recruitment to the inflamed glomerulus and inhibited the generation of MPO-ANCA whilst not influencing T cell autoimmunity. Thus, whilst the C3aR modulates some elements of disease pathogenesis, overall it is not critical in effector responses and glomerular injury caused by autoimmunity to MPO.


Introduction
The anti-neutrophil cytoplasmic antibody (ANCA) associated vasculitides (AAV) are diseases in which autoimmunity to the neutrophil granule proteins myeloperoxidase (MPO) or proteinase-3 (Pr3) can cause multi-organ injury, including rapidly progressive glomerulonephritis. The pathogenesis of AAV involves multiple steps. T and B cell tolerance to MPO or Pr3 is lost, resulting in the secretion of autoantibodies (ANCA). ANCA can bind to their cognate autoantigen on primed neutrophils, inducing them to activate [1] and lodge in the glomerulus. These intraglomerular neutrophils degranulate, producing reactive oxygen species and causing direct glomerular injury. Degranulation results in extensive glomerular deposits of non-leukocyte associated MPO in patients with AAV [2]. Murine models suggest that MPO-specific effector T cells recognize MPO deposited in glomeruli and contribute to glomerular injury [3,4]. PLOS

Induction of glomerulonephritis and assessment of autoimmunity
Anti-MPO IgG was generated by immunising Mpo -/mice with native murine MPO generated as previously described [23]. Briefly mice were immunised sub-cutaneously (s.c.) with 15μg MPO in FCA (Sigma-Aldrich, St Louis, MO) on day 0 followed by 10μg in Freund's incomplete adjuvant on days 7 and 14. On day 24 mice were humanely killed and blood was obtained. IgG fractions were obtained by Protein G affinity chromatography (GE Healthcare) and then extensively dialysed against PBS.
To induce glomerulonephritis age matched 6-10 week old C57B/L6 mice and C3ar -/mice received 100μg/g body weight anti-MPO IgG by tail vein injection. 1 hour later 0.5 μg/g bacterial lipopolysaccharide (LPS) (Sigma-Aldrich) was administered by a single intra-peritoneal injection. After 6 days, 24 hours before the end of experiments, mice were placed in metabolic cages for urine collection.
Autoimmune anti-MPO glomerulonephritis mice was induced as previously described [3]. Briefly, mice were immunised s.c. with 20μg MPO in FCA followed by 10μg MPO in FIA on day 7. On day 16 glomerulonephritis was triggered by injection of 0.12mg/g sheep anti-mouse GBM globulin in two divided doses. Mice were placed in metabolic cages on day 20 for assessment of proteinuria before being euthanased on day 21 by CO 2 inhalation.

Assessment of renal injury
Glomerular abnormalities were assessed on periodic acid-Schiff (PAS) stained, 4μm, formalin fixed, paraffin embedded sections using coded slides. Abnormalities scored included mesangial proliferation, capillary wall thickening, glomerular necrosis (defined as accumulation of PAS positive material combined with hypocellularity) and crescent formation (defined as two or more cells visible in Bowman's space). A minimum of 30 glomeruli per mouse were examined. To evaluate leukocyte infiltrate, kidneys were fixed in periodate lysine paraformaldehyde for four hours then washed with 20% sucrose. 5μm sections were stained with a three layer immunoperoxidase technique [24] using GK1.5 (anti-CD4), FA11 (anti-CD68) or Gr1 (anti Ly6g/c). The secondary antibody was rabbit anti-rat biotin (DAKO). A minimum of 30 glomeruli per section were scored. Albuminuria was measured by ELISA (Bethyl. Montgomery, TX). Creatinine was measured using standard methods at the biochemistry laboratory, Monash Medical Centre.

Assessment of immunity
Immunity was assessed on day 10 after immunisation with 20μg MPO in FCA and at day 21, at the end of the autoimmune anti-MPO glomerulonephritis model. The spleen and draining lymph nodes were harvested and a single cell suspension was obtained. IFN-γ and IL-17A ELI-SPOT was performed according to the manufacturer's instructions (Ebioscience, San Diego, CA) with 5x10 5 cells per well. Cells were incubated for 18 hours at 37˚C with 10μg/ml heat inactivated recombinant murine MPO (rMPO). Total anti-MPO IgG was measured by ELISA on MPO coated plates, with IgG detected with sheep-anti mouse IgG-HRP (Sigma-Aldrich). Antibody subclasses were measured using subclass specific goat anti-mouse IgG-HRP (Southern Biotech, Birmingham, AL). Serum BAFF was measured by ELISA (RnD Systems, Minneapolis, MN). For assessment of B cell development, bone marrow was flushed from tibiae and femora of naïve mice and analysed by flow cytometry.

Statistics
Data are presented with each dot representing a mouse and a bar representing median value. Prism 6 (Graphpad, San Diego, CA) software was used with analysis by Mann-Whitney U test. Ã p<0.05, ÃÃ p< 0.01, ÃÃÃ p<0.001.

Endogenous C3a does not exacerbate glomerular injury induced by passive transfer of anti-MPO IgG but promotes glomerular macrophage recruitment
Passive transfer of anti-MPO IgG into mice induces a neutrophil mediated necrotising glomerulonephritis [25]. Glomerular injury is exacerbated by the co-administration of lipopolysaccharide (LPS) [26]. We used this enhanced model to investigate the role of the endogenous C3a in the effector phase of anti-MPO IgG induced glomerulonephritis. Wild type (WT) and C3ar -/mice received anti-MPO IgG and LPS. Renal injury was studied after 7 days. Both groups of mice developed glomerulonephritis with a similar degree of histological glomerular injury (Fig 1A-1C). However, fewer glomerular macrophages were observed in C3ar -/mice with glomerulonephritis (Fig 1D-1F). Numbers of glomerular neutrophils were not different between groups and, consistent with histological findings, albuminuria was similar between groups ( Fig 1G and 1H). Therefore, in this model, the C3aR is not required for the development of anti-MPO IgG induced glomerular injury, but does promote macrophage infiltration to the inflamed glomerulus.

The C3aR does not promote injury in experimental autoimmune anti-MPO glomerulonephritis
We then examined the role of C3aR in an autoimmune model of anti-MPO glomerulonephritis that is mediated by the T cell effector response to glomerular MPO [3]. In mice, autoimmunity to MPO induced by immunisation does not result in ANCA of sufficient pathogenicity to cause disease. However, injection of a sub-nephritogenic dose of anti-glomerular basement membrane (GBM) globulin induces glomerular neutrophil recruitment and deposition of MPO. Glomerular MPO is recognised by antigen specific effector T cells with resulting necrotising glomerulonephritis [3,4]. Several strands of evidence confirm that this disease is due to MPO-specific T cell effectors. These include lack of injury in Mpo -/mice or OVA immunised mice, a similar degree of injury in B cell deficient mice and the induction of glomerulonephritis by transfer of MPO-specific, but not OVA-specific CD4+ T cell clones [3,4,27]. As this model is dependent on glomerular neutrophil recruitment by anti-GBM globulin, we first confirmed that this parameter was not affected by the absence of the C3aR (150 minutes after anti-GBM IgG, WT 1.17±0.17 vs. C3ar -/-1.13±0.16; mean ± SEM, neutrophils/glomerular cross section). Having excluded this potential confounder, we induced autoimmune anti-MPO glomerulonephritis in WT and C3ar -/mice. Both groups had similar severity of histological injury and albuminuria, with no differences in the numbers of neutrophils, macrophages or CD4 + T cells between WT and C3ar -/mice (Fig 2A-2G).

C3a suppresses humoral autoimmunity to MPO
As ANCA production is critical in the pathogenesis of AAV, we measured the development of anti-MPO humoral autoimmunity in this disease model. MPO-ANCA IgG titres were increased in C3ar /mice ( Fig 3A and S2 Fig). Levels of anti-MPO IgG1 and IgG2b isotypes, but not IgG2c and IgG3 were significantly higher in C3ar -/than wild-type mice ( Fig 3B).
To further investigate the differences in humoral autoimmunity, we analysed B cell development in the bone marrow and spleen. In naïve mice, the proportion of bone marrow B cells that were proB-preB (B220 + IgM low IgD low ), transitional (B220 + IgM + IgD low ) or mature (B220 + IgM int IgD + ) was not different between WT and C3ar -/mice. In mice immunised with MPO in FCA, there was no difference in the total number of splenic B220 + B cells, Follicular B cells (B220 + CD21 + CD23 + ), marginal zone B cells (B220 + CD21 + CD23 low ) or plasma cells (B220 low/int CD138 + ) between groups (Fig 4). B cell activating factor (BAFF), expressed by myeloid and bone marrow stromal cells is elevated in AAV[28] and is important in B cell development and differentiation. Levels of serum BAFF in immunised WT and C3ar -/mice were not different between groups.

Fig 1. Endogenous C3a does not promote anti-MPO glomerulonephritis but supports macrophage recruitment to the inflamed glomerulus.
Anti-MPO glomerulonephritis was induced in WT and C3ar -/mice (n = 14/group, data from two independent experiments) by injection of 100μg/g anti-MPO IgG and 0.5μg/g LPS. On day 7 histological glomerular injury was assessed. Both groups of mice developed glomerulonephritis including segmental necrosis. There was no difference in degree of injury (A) between WT (B) and C3ar -/-(C) mice. Glomerular leucocyte influx was assessed by immunohistochemistry (n = 7/group). Compared to WT mice (D-E), the number of glomerular macrophages was reduced in C3ar -/mice. (F) There was no difference in glomerular neutrophils (G), Albuminuria, measured by urinary albumin: creatinine ratio was not different between groups (H). ÃÃ p<0.01 by Mann-Whitney U Test. C3a does not promote cellular autoimmunity to myeloperoxidase CD4 + T cells participate in AAV in several ways. They provide help to B cells for the generation of humoral immunity and also effect glomerular injury in crescentic glomerulonephritis via Th1 and Th17 responses [29][30][31]. In vitro, C3aR ligation on dendritic cells results in increased activation and functional capacity to activate T cells [32,33]. T cell intrinsic C3aR has also been reported to be important in Th1 responses and in inhibiting the generation of Foxp3 + T regulatory cells (Tregs) [34][35][36]. We therefore analysed cellular immune responses in mice at two time points: early (10 days) and later (21 days) after immunisation. Ten days after MPO immunisation there were no differences in MPO-stimulated Th1 or Th17 responses measured by IFN-γ and IL-17A ELISPOT (Fig 5A and 5B). Furthermore, proportions of CD4 + cells that were CD44 hi (activated T cells) (Fig 5C), or CD25 + Foxp3 + Tregs (Fig 5D and 5E) were similar between groups. T follicular helper cells (TFH) are important for the formation of germinal centres and subsequent antibody response, the proportion of CD4 cells with CXCR5 hi PD-1 hi TFH phenotype were also similar between groups (Fig 5F and 5G).
Similar to findings at day 10, at day 21 there was no difference in MPO-stimulated IFN-γ and IL-17A producing cells measured by ELISPOT (Fig 6A and 6B). Additionally at this time point, the proportion of CD4 T cells that were CD44 + activated or CD25 + Foxp3 + Tregs were similar, both in the draining lymph nodes and spleen (Fig 6C-6E).

Discussion
A significant body of evidence implicates complement in the pathogenesis of AAV. To date, the C5aR has been identified as a key mediator of neutrophil activation and glomerular injury. Here, we assessed whether the signalling though the C3aR mediated autoimmunity or glomerular injury, and thus had potential as a therapeutic target. C3aR has previously been described to play both pathogenic and protective roles in experimental inflammatory diseases. Animal models in which the C3aR is pathogenic include antibody-induced arthritis [37], adriamycin induced nephropathy [38], complement mediated tubulointerstitial injury [21] and renal ischemia-reperfusion injury [39]. However, the C3aR is protective in lupus-like disease in the MRL/ lpr mouse [40], endotoxic shock [22] and intestinal ischemia-reperfusion injury [18]. In the current studies, although we found that the C3aR had biological effects, there were no net effects on glomerulonephritis and renal injury in the two complementary disease models studied.
Passive transfer of anti-MPO IgG with LPS models pathological neutrophil activation with glomerular injury being caused by neutrophil degranulation, and the consequent release of inflammatory mediators and reactive oxygen species [25]. Whilst absence of the C3aR did not influence overall injury it was associated with an attenuation of glomerular macrophage accumulation, suggesting a role in this context for local C3a. In vitro and in vivo evidence supports a role for the C3aR in renal macrophage recruitment. In vitro, C3a acts as a chemotaxin in a murine macrophage cell line [41]. In vivo evidence derives from a model of complement  induced renal injury in which Crry -/-C3 -/kidneys are transplanted into syngeneic complement sufficient recipients, resulting in unrestricted renal complement activation with a inflammatory cell influx. When Crry -/-C3 -/kidneys were transplanted into C3ar -/recipients, there was a marked attenuation of the monocyte-macrophage influx, suggesting a role for the C3aR in monocyte trafficking to areas of complement activation [21]. splenocytes that were B220 + B cells, (F) the proportion of B220 + cells that were CD21 + CD23 + follicular B cells, or (G) CD21 + CD23 low marginal zone B cells was similar between groups. (H) The proportion of splenocytes that were B220 low/int CD138 + plasma cells also did not differ between groups. (I) Representative flow cytometry plot of splenocytes showing gating of plasma cells. (J) Serum B cell activating factor (BAFF) was similar between groups.
https://doi.org/10.1371/journal.pone.0190655.g004 Whilst AAV is classically described to be "pauci-immune" with minimal glomerular complement deposition, detailed analysis of kidney biopsies from humans with AAV reveals evidence of local complement activation with detectable C3c, C3d and C5b-9 deposition [14,42,43], additional evidence for renal complement activation in AAV is the elevated urinary C3a observed in patients with active disease [42]. This complement activation, which is likely to be initiated by neutrophil secreted factors such as properdin [44], MPO [45] and neutrophil extracellular traps [46] may be one of the early signals contributing to macrophage recruitment to the inflamed glomerulus. Depletion studies have implicated macrophages as important injurious mediators in murine models of glomerulonephritis including disease induced by anti-MPO IgG [47,48]. In the current experiments, the observed reduction in glomerular macrophages in C3ar -/mice was not paralleled by reduced glomerular injury. This may be because when compared to studies using pharmacological or genetic depletion, the magnitude of the reduction in infiltrating macrophages was insufficient affect a change in injury.
While passive transfer of anti-MPO IgG allows study of the role of effector pathways responding to autoantibodies, it cannot be used to examine the contribution of the adaptive immune system to glomerular injury. AAV is an autoimmune disease in which both B and T cells play essential roles in the pathogenesis. We used the model of experimental autoimmune C3aR in anti-MPO glomerulonephritis anti-MPO glomerulonephritis to model these processes and found that absence of C3aR did not affect albuminuria, histological injury or glomerular leucocyte influx. The discrepancy between results for glomerular macrophage infiltration between passive transfer of anti-MPO IgG and autoimmune anti-MPO glomerulonephritis is likely to be because the dominant signals driving glomerular macrophage recruitment in the model of autoimmune anti-MPO GN are T helper cell derived costimulatory molecules, such as CD154-CD40 interactions [49] and cytokines, such as IL-17A [29] and IFN-γ [50].
C3ar -/mice had an increased humoral immune response to MPO with higher MPO-ANCA titres. Elevated autoantibody titres associated with accelerated renal injury have also described in C3ar -/mice when backcrossed to the MRL lupus-prone strain [40]. In human cells, C3a has been reported to directly suppress polyclonal antibody response from isolated B lymphocytes, [20,51] whilst both activation and suppression of T lymphocytes have shown [34,51] C3ar -/mice have also been reported to have an enhanced Th2 response and higher anti-OVA IgG titres after epicutanteous sensitisation [52]. In contrast, models of infection have found that absence of the C3aR results in unaltered [53] or attenuated [54] humoral immunity. The enhanced humoral immunity in C3ar -/mice is not likely to be due to elevated C3 levels (and enhanced generation of C3d which is a potent B cell adjuvant), as serum C3 in this strain has previously be described to be similar to that of WT mice [37,40]. Because of the observed differences in anti-MPO IgG titres we investigated whether C3ar -/mice had any numerical or developmental alteration in the B lymphocyte compartment. However, B cell number and development in the spleen and bone marrow appeared similar to WT mice.
Given the previous descriptions of the important role that C3aR plays in influencing adaptive immunity we investigated the effect of absence of this receptor on the generation of T cell mediated immunity. We found no difference in the generation of Th1 or Th17 effector responses, or in the proportion of T cells that had a CD25 + Foxp3 + regulatory phenotype. This is in contrast to previous reports of elevated T regulatory cells in C3ar -/mice [35,55]. The model of autoimmune anti-MPO glomerulonephritis relies on immunisation with MPO in Freund's complete adjuvant to break immune tolerance and generate autoimmunity. Other investigators who have reported increased T regulatory cells in C3ar -/mice used animal models that either require no adjuvant or use incomplete Freund's adjuvant. It is possible that differences in the additional immune signals provided by the adjuvant may account for this discrepancy. Additionally, whilst C3aR expression in the mouse has previously been reported to be extensive in both myeloid and lymphoid lineages, this has been called into questions by recent findings using a C3aR reporter mouse in which expression of C3aR was evident on macrophages and some dendritic cell subsets but not bone marrow neutrophils, B or T cells [56,57]. In contrast, neutrophil, macrophage and T cell expression have been reported in humans [34,[58][59][60][61]. Potential species differences in C3aR distribution should therefore be considered when interpreting studies using murine models.
In summary, these data suggest a role for C3a in driving glomerular macrophage recruitment and suppressing humoral immunity in these pre-clinical models of AAV. However, the lack of attenuation of immune mediated glomerular injury in either model does not support the C3aR as a putative therapeutic target in this disease.