Neonatal Fc Receptor Promotes Immune Complex–Mediated Glomerular Disease

The MHC class I–like neonatal Fc receptor (FcRn), a heterodimer comprising a heavy chain and β₂-microglobulin light chain, is the major regulator of IgG and albumin homeostasis.¹ Perinatally, FcRn mediates the transfer of IgG from mother to offspring, across the placenta in primates and trans-intestinally in suckling rodents. Throughout life, FcRn protects IgG and albumin from catabolism, explaining the unusually long t_1/2 and high serum levels of these proteins. IgG and albumin taken up by cells by pinocytosis bind strongly to FcRn at pH 6.0–6.5 in endosomes. FcRn-bound ligands are then recycled to the plasma membrane, where they dissociate at pH 7.4, whereas IgG and albumin not bound to FcRn are targeted to lysosomes for degradation. FcRn is thought to promote some autoimmune diseases because it protects pathogenic IgG from degradation. For instance, Fcrn^−/− mice are resistant to passive transfer of arthritis by K/BxN sera and autoimmune skin pathology induced by antibodies targeting autoantigens at the dermal–epidermal junction, although this protection can be overcome by excess autoantibodies.^2–4

In kidneys, FcRn is expressed in podocytes and proximal tubular epithelial cells.⁵ Overall, renal FcRn reclaims albumin but facilitates elimination of IgG.⁶ Tubular FcRn mediates IgG transcytosis.⁷ Podocytes use FcRn to clear IgG from the glomerular basement membrane (GBM).⁸ IgG accumulates in the glomeruli of aged Fcrn^−/− mice due to impaired clearance of IgG from the GBM, and saturating this clearance mechanism by excess ligand potentiates the pathogenicity of nephrotoxic sera in wild-type mice. Podocyte FcRn has been postulated to be involved in the clearance of immune complexes (ICs) present in pathologic conditions such as membranous nephropathy.⁵ Expression of FcRn in human podocytes is increased in various immune-mediated glomerular diseases.⁹ Given its role in IgG and albumin handling in the kidneys and systemically, FcRn can be expected to influence the development of immune-mediated kidney diseases at multiple levels. This conjecture awaits experimental verification.

To determine the role of FcRn in IgG-mediated glomerular disease, we asked how FcRn deficiency alters the course of disease in mice immunized with the NC1 domain of α3 type IV collagen (α3NC1). We chose this antigen because of its reported ability to induce disease in C57Bl/6 (B6) mice,¹⁰ corroborated in pilot studies (Supplemental Figure 1). Fcrn^−/− mice are hypoalbuminemic due to impaired albumin recycling,¹¹ and also exhibit reduced urinary albumin excretion.¹² As a control for this potential confounder, we used FCRN-humanized mice, which have normal serum albumin because human FcRn recycles mouse albumin but not mouse IgG.¹³

All mice immunized with α3NC1 developed circulating mouse IgG anti-α3NC1 antibodies, which reached the maximum titer about 6 weeks later and gradually declined thereafter. At all times, the levels of mouse IgG anti-α3NC1 antibodies in sera from Fcrn^−/− mice and FCRN-humanized mice were approximately 50%–70% lower than those in wild-type mouse sera (Figure 1A). The results were similar for mouse IgG1, IgG2b, and IgG2c anti-α3NC1 antibodies (Supplemental Figure 2). Wild-type B6 mice immunized with α3NC1 started developing progressive albuminuria 8–10 weeks later (Figure 1B). By week 14, the urinary albumin creatinine ratio increased approximately 100-fold, and hypoalbuminemia developed (Figure 1C). Urinary albumin excretion in Fcrn^−/− mice and FCRN-humanized mice immunized with α3NC1 was not significantly higher than in adjuvant-immunized control mice. No mice developed renal failure (Supplemental Figure 3).

At 14 weeks after α3NC1 immunization, kidneys examined by light microscopy showed mild glomerular pathology, with few crescents and relatively little inflammation (Figure 2A), similar to α3NC1-immunized DBA/1 mice with comparable albuminuria.^14,15 Electron microscopy showed extensive subepithelial IC deposits surrounded by an expanded GBM and effacement of podocyte foot processes in α3NC1-immunized B6 mice, whereas Fcrn^−/− mice had fewer subepithelial deposits (Figure 2B, Supplemental Figure 4). Immunofluorescence staining showed granular capillary loop deposition of mouse IgG, exogenous antigen, C3, and C5b-9, more intense in wild-type mice than in Fcrn^−/− mice and FCRN-humanized mice (Figure 2, Ca–Cp, Supplemental Figure 5). A loss of nephrin staining, indicative of podocyte injury, occurred in α3NC1-immunized B6 mice but not in Fcrn^−/− mice or FCRN-humanized mice (Figure 2, Cq–Ct).

Because B6 mice immunized with bovine GBM NC1 hexamers have normal kidney function and histology despite linear GBM deposition of IgG autoantibodies binding to mouse α345(IV) collagen (Supplemental Figure 1), the question arises as to what causes proteinuria in α3NC1-immunized mice. Because the clinical presentation, morphology, and effector mechanisms depend on where ICs are localized in the capillary wall, we compared IgG distribution in α3NC1-immunized mice and mice injected with anti-α3NC1 antibodies modeling anti-GBM autoantibodies. The distribution and relative abundance of mouse IgG, as imaged by immunoperoxidase immunoelectron microscopy and stochastic optical reconstruction microscopy (STORM), a method for super-resolution fluorescence microscopy, were concordant. In α3NC1-immunized mice, IgG deposition was abundant in the areas of expanded GBM and especially in regions corresponding to the subepithelial dense deposits seen by routine electron microscopy. By contrast, in mice injected with α3NC1-specific anti-GBM mAb, the IgG was confined to an ultrastructurally normal GBM that lacked subepithelial deposits (Figure 3).

Subepithelial ICs, a hallmark of human membranous nephropathy (MN), form when IgG antibodies bind to podocyte antigens, such as phospholipase A2 receptor (PLA2R) and neutral endopeptidase (NEP), or to planted antigens, such as cationic BSA.^16–18 Subsequent expansion of the GBM, complement activation, and podocyte injury by C5b-9 cause proteinuria. Although it is unexpected, formation of subepithelial ICs in α3NC1-immunized mice may be explained by exogenous α3NC1 deposited in glomeruli acting as a planted antigen.¹⁹ Alternatively, anti-α3NC1 antibodies in complex with α3NC1 antigen may act as surrogate antipodocyte antibodies, because α3NC1-containing ICs bind to podocytes.²⁰ After four immunizations with α3NC1 monomers, B6 mice and DBA/1 mice eventually develop crescentic GN by 26 and 10 weeks, respectively.^10,14 The combination of subepithelial ICs and crescentic anti-GBM antibody GN was most recently described in a series of eight patients with circulating anti-α3NC1 autoantibodies but undetectable anti-PLA2R autoantibodies.²¹

In contrast to wild-type B6 mice, congenic Fcrn^−/− mice and FCRN-humanized mice did not develop albuminuria after α3NC1 immunization. Their resistance to proteinuria was associated with lower serum titers of anti-α3NC1 IgG antibodies and reduced glomerular deposition of IgG, antigen, C3, and C5b-9. Because C5b-9 is an essential mediator of podocyte damage and proteinuria by subepithelial ICs,^22,23 reduced complement activation potentially explains the attenuated glomerular pathology in FcRn-deficient mice. The resistance of FCRN-humanized mice indicates that FcRn promotes IC-mediated glomerular disease due to its interaction with IgG rather than albumin. We propose that FcRn promotes the development of subepithelial ICs and subsequent glomerular injury primarily by maintaining higher serum levels of pathogenic IgG (Supplemental Figure 6). However, we cannot formally exclude a possible pathogenic role of podocyte FcRn, whose stimulation by ICs may induce maladaptive signaling.⁹ Future studies in mice with podocyte-specific ablation of FcRn would address this possibility.

Our findings identify FcRn as a potential target for therapeutic intervention in IC-mediated glomerular diseases, typically treated with nonspecific immunosuppressants that are toxic and sometimes ineffective. More specific therapies include ablation of B cells by rituximab. In patients with idiopathic MN who respond to rituximab therapy, serum levels of anti-PLA2R IgG autoantibodies decline over a period of many months, and their disappearance is followed by resolution of proteinuria.²⁴ The slow decline in proteinuria is problematic for patients already suffering from complications of nephrotic syndrome, who would benefit from ancillary therapies that reduce pathogenic IgG antibodies more rapidly. This may be achieved by inhibiting FcRn.

One implementation of this concept is therapy with high-dose intravenous Ig (HD-IVIG). HD-IVIG accelerates the degradation of IgG by saturating FcRn,²⁵ one of the mechanisms that explain the beneficial effects of HD-IVIG therapy in some autoimmune diseases.³ In pregnant women with circulating anti-NEP alloantibodies mediating antenatal MN, treatment with HD-IVIG reduces the titers of IgG alloantibodies by approximately 30% within 2–3 weeks.²⁶ However, HD-IVIG is inefficient, because large amounts of IgG (1–2 g/kg) cause relatively modest reductions in pathogenic IgG titers. Specific FcRn inhibitors recapitulate this activity of HD-IVIG more effectively at lower doses. By reducing pathogenic IgG levels, function-blocking anti-FcRn mAbs ameliorate experimental myasthenia gravis in rats,²⁷ and engineered IgG “Abdegs” that bind with high affinity to FcRn ameliorate arthritis transferred by K/BxN serum.²⁸

To assess the translational potential of our findings, we asked whether pharmacologic blockade of human FcRn can reproduce the effects of genetic FcRn deficiency. To this end, FCRN-humanized and Fcrn^−/− mice were passively immunized with human IgG containing anti-α3NC1 (Goodpasture) autoantibodies. To inhibit human FcRn, we used a lysine analog of SYN1436 (Figure 4A),²⁹ a peptide that binds with subnanomolar affinity to human FcRn, thus preventing IgG binding.³⁰ In vivo, SYN1436 reduces IgG levels in cynomolgus monkeys by 80%.³⁰ Serum anti-α3NC1 autoantibodies in FCRN-humanized mice treated with anti-FcRn peptide, but not with control peptide, sharply decreased to the same levels as in Fcrn^−/− mice (Figure 4B), and were no longer detected after 4 days. In mice, human IgG elicits murine anti-human IgG antibodies, forming ICs that can deposit in glomeruli, as shown in active serum sickness models. Glomerular deposition of ICs containing human IgG was abolished in mice treated with anti-FcRn peptide, but not with control peptide (Figure 4C). Linear GBM deposition of human anti-GBM IgG was not observed, because the epitopes recognized by Goodpasture autoantibodies are completely inaccessible in the mouse GBM.³¹ These results provide proof of concept that therapies targeting human FcRn effectively lower serum levels of pathogenic human IgG autoantibodies, which could be beneficial in patients with IgG-mediated kidney diseases. Because FcRn also mediates the trans-placental transfer of IgG from mother to the fetus, FcRn inhibition may be particularly attractive for preventing antenatal MN caused by maternal anti-NEP alloantibodies.