Factors affecting IgG4-mediated complement activation

Abstract

Of the four human immunoglobulin G (IgG) subclasses, IgG4 is considered the least inflammatory, in part because it poorly activates the complement system. Regardless, in IgG4 related disease (IgG4-RD) and in autoimmune disorders with high levels of IgG4 autoantibodies, the presence of these antibodies has been linked to consumption and deposition of complement components. This apparent paradox suggests that conditions may exist, potentially reminiscent of in vivo deposits, that allow for complement activation by IgG4. Furthermore, it is currently unclear how variable glycosylation and Fab arm exchange may influence the ability of IgG4 to activate complement. Here, we used well-defined, glyco-engineered monoclonal preparations of IgG4 and determined their ability to activate complement in a controlled system. We show that IgG4 can activate complement only at high antigen and antibody concentrations, via the classical pathway. Moreover, elevated or reduced Fc galactosylation enhanced or diminished complement activation, respectively, with no apparent contribution from the lectin pathway. Fab glycans slightly reduced complement activation. Lastly, we show that bispecific, monovalent IgG4 resulting from Fab arm exchange is a less potent activator of complement than monospecific IgG4. Taken together, these results imply that involvement of IgG4-mediated complement activation in pathology is possible but unlikely.

Keywords: IgG4-related disease; antibodies; complement activation; fab arm exchange; glycoengineering; primary membranous nephropathy.

Figure 1

Schematic overview of the human IgG antibody structure and antibody-mediated complement activation. (A) Generalized overview of the complement system, which is activated through three pathways: the classical, lectin and alternative pathway. The classical pathway is activated by binding of C1q to antigen-bound IgM and IgG, which then forms the C1 complex with its proteases C1r/s. The lectin pathway is activated by the binding of lectins, such as mannose-binding lectin (MBL), to carbohydrates on cell surfaces. C3b deposition induced by these pathways or spontaneous hydrolysis of C3 can activate the alternative pathway, mediated by Factor B and D, which results in an amplification loop of C3b deposition. C3b can opsonize the target cell for phagocytosis and can lead to the assembly of the membrane attack complex with factors C5b to C9 through the terminal pathway and subsequent cell lysis. (B) A structural representation of the hinge variation and half molecule exchange of IgG4 via switch of inter-chain to intra-chain disulfide bonds at amino acid positions 228 and 409. (C) General structure of an IgG monomer and schematic representation of N-linked glycan composition of human IgG Abs. The glycans are attached to asparagine (N) at position 297 in the C_H2 domain and have a biantennary heptasaccharide core (solid line) and variable extensions (dash line), such as fucose, galactose and/or sialic acid. Novel glycosylation sites can be introduced within the Fab region of matured antibodies.

Figure 2

IgG4 activates complement at high antigen densities and antibody concentrations. (A) C3b deposition induced by anti-biotin IgG1 and IgG4 antibodies on biotinylated human serum albumin (HSA; 30, 60, 120 and 240 μM biotin; corresponding to approximately 2, 4, 8 and 50 biotins per HSA molecule respectively), in the presence of 2.5% human serum. All antibodies were tested in a two-fold serial dilution starting from 10 µg/mL for IgG4 and 2 µg/mL for IgG1 antibodies in four independent assays. (B) The capacity of biotin-specific IgG1 and IgG4 antibodies to induce complement-mediated lysis (as percentage) of biotinylated human red blood cells at various antigen densities (0.25, 0.5, 1, and 2.5 mM biotin) in the presence of 10% human serum. All antibodies were tested in a two-fold serial dilution starting from 20 µg/mL for IgG4 and 10 µg/mL for IgG1 antibodies in triplicates. The data was normalized to a 100% lysis control (RBCs with saponin).

Figure 3

IgG4 glycovariants can induce C3b deposition to different extent. (A) C3b deposition induced by anti-biotin IgG1 and IgG4 glycovariants (high mannose (HM), high galactose (HG), normal galactose (NG) and low galactose (LG)), and IgG1 PG-LALA in the presence of biotinylated human serum albumin (240 μM biotin) and 2.5% human serum. Relative C3b deposition of the IgG4 glycovariants at concentrations 4 µg/mL and 8 µg/mL was normalized to IgG4 HG, which was set to 1 (dashed line), is shown as bar graphs. (B) Percentage of complement-mediated lysis of biotinylated human red blood cells (2.5 mM biotin) by the anti-biotin IgG1 and IgG4 glycovariants. The percentage of lysis was calculated based on the 100% lysis control (red blood cells with saponin). The percentage of complement-mediated lysis of the IgG4 glycovariants at concentrations 3.75 µg/mL and 7 µg/mL normalized to IgG4 HG is shown as bar graphs. Kruskal-Wallis test with Dunn’s multiple comparisons test of the IgG4 glycovariants was performed. (C) To test unspecific complement-dependent cytotoxicity (CDC), anti-biotin IgG1, IgG4 and IgG1 PG-LALA (eliminated Fc-mediated effector functions) were tested in the presence of biotinylated RBCs and 10% normal or heat-inactivated human serum. (D) C3b deposition induced by anti-trinitrophenyl (TNP) IgG4 glycovariants variants (HG and LG), which were produced as wild type IgG4, and a variant with D86N substitution in the light chain to introduce an additional glycosylation site in the Fab region. All experiments were performed at least three times, shown are representative figures for the C3b deposition and the mean ± standard deviation for the lysis experiments.

Figure 4

IgG4 activates complement via the classical pathway. (A) C3b deposition by anti-biotin IgG1, IgG4 and IgG1 PG-LALA (eliminated Fc-mediated effector functions) in presence or absence of either C1q, mannose-binding lectin (MBL), or complement factor B (FB) inhibitors, biotinylated human serum albumin (HSA; 240 μM biotin) and 2.5% human serum. Antibodies with different glycan profiles were tested: high mannose (HM), high galactose (HG), normal galactose (NH) and low galactose (LG). The molar ratios of target to inhibitor are 1:4.5 for C1q, 1:8.5 for MBL, and 1:3.5 for FB. (B) The percentage of complement-mediated lysis of biotinylated human red blood cells (2.5 mM biotin) by IgG1, IgG4 and IgG1 PG-LALA in presence or absence of C1q, MBL, or factor D (FD) inhibitors and 10% human serum. The molar ratios for target to inhibitor were 1:3 for C1q, 1:3 for MBL, and 1:2 for FD. The percentage of lysis was calculated based on the 100% lysis control (red blood cells with saponin). (C) The relative C3b deposition and (D) percentage of complement-mediated lysis of biotinylated human red blood cells induced by IgG4 glycovariants are shown as bar graphs with an antibody concentration of either (C) 8 µg/mL, or (D) 15 µg/mL. (E) To test binding of purified C1q (600 μg/mL) to the mAbs, IgG1, IgG4 and IgG1 PG-LALA were tested in the presence of biotinylated human serum albumin (HSA; 240 μM biotin). Kruskal-Wallis test with Dunn’s multiple comparisons test was performed for statistical comparison of the IgG4 glycovariants per inhibition condition: *p < 0.05; **p < 0.01. All antibodies were tested in a two-fold serial dilution starting from 8 µg/mL for C3b deposition and 15 µg/mL for CDC in at least three independent experiments.

Figure 5

Fab arm exchange reduces antibody-mediated C3b deposition. Bispecific IgG1 and IgG4 variants were generated by half molecule exchanging anti-biotin and anti-trinitrophenyl (TNP) clones (both normal galactosylation) with IgG1 K409R adalimumab and IgG4 natalizumab respectively under mild reduction conditions. C3b deposition by (A) anti-biotin and (B) anti-TNP monospecific (full dots) and bispecific (empty dots) IgG1 and IgG4 in the presence of either TNPlated (1 mM 2,4,6-trinitrobenzenesulfonic acid) or biotinylated (240 μM biotin) human serum albumin and 2.5% human serum was tested. Binding to either biotin or TNP was measured for all variants (right panel). All experiments were performed at least three times, shown are representative figures.