IgG hexamers initiate acute lung injury

Abstract

Antibodies can initiate lung injury in a variety of disease states such as autoimmunity, transfusion reactions, or after organ transplantation, but the key factors determining in vivo pathogenicity of injury-inducing antibodies are unclear. A previously overlooked step in complement activation by IgG antibodies has been elucidated involving interactions between IgG Fc domains that enable assembly of IgG hexamers, which can optimally activate the complement cascade. Here, we tested the in vivo relevance of IgG hexamers in a complement-dependent alloantibody model of acute lung injury. We used three approaches to block alloantibody hexamerization (antibody carbamylation, the K439E Fc mutation, or treatment with domain B from Staphylococcal protein A), all of which reduced acute lung injury. Conversely, Fc mutations promoting spontaneous hexamerization made a harmful alloantibody into a more potent inducer of acute lung injury and rendered an innocuous alloantibody pathogenic. Treatment with a recombinant Fc hexamer 'decoy' therapeutic protected mice from lung injury, including in a model with transgenic human FCGR2A expression that exacerbated pathology. These results indicate a direct in vivo role of IgG hexamerization in initiating acute lung injury and the potential for therapeutics that inhibit or mimic hexamerization to treat antibody-mediated diseases.

Fig. 1.. The 34-1-2S alloantibody binds to multiple MHC class I antigens to trigger acute lung injury.

A. Schematic showing approach for measuring affinity of the MHC class I alloantibody 34-1-2S for MHC class I monomers using surface plasmon resonance (SPR). B. SPR sensorgrams showing detection of binding of 34-1-2S to the K^b MHC class I antigen from H2^b mice and the K^d, D^d and L^d antigens from mice with the H2^d haplotype. C. Classical MHC class I antigens present in B6, B6.H2^d and B6.Con-K^d-on mice with summary of results from B. D. Lung vascular permeability and E. excess lung water measurements from LPS-primed B6.H2^d mice given intravenous (i.v.) doses of either 34-1-2S or isotype control, versus B6.Con-K^d-on mice given i.v. 34-1-2S. Depictions of IgG and MHC class I in A-C are based on protein data bank (PDB) entries 1HZH and 1RK1. B, D & E show means ± standard errors. Statistical tests used on D & F were ordinary one-way ANOVA with Dunnett’s test for differences relative to B6.H2^d + 34-1-2S i.v. group, with data log₁₀-transformed prior to analysis, *** = P<0.0001.

Fig. 2.. Classical complement activation on the pulmonary endothelium initiates 34-1-2S-induced acute lung injury.

A. Molecular model of C1 complex based on small angle scattering database entry SASDB38 (12). B. Lung vascular permeability and C. excess lung water measurements from LPS-primed B6.H2^d:C1qa^−/− mice and B6.H2^d:C1qa^+/+ littermates given i.v. 34-1-2S at 1 mg/kg. Horizontal gray lines are standard deviations of values from ‘no injury’ controls (B6.H2^d mice given LPS i.p. + mIgG2a isotype control i.v.) D. Survival of LPS-primed B6.H2^d:C1qa^−/− mice and B6.H2^d:C1qa^+/+ littermates given i.v. 34-1-2S at 4.5 mg/kg. E. Immunofluorescence staining for complement C3b/d, F. C1qa or G. C4b/d (red) as well as Acta2 (α-smooth muscle actin, cyan) and, in G., Scgb1a1 (club cell secretory protein, magenta) in lung sections from LPS-primed B6.H2^d:C1qa^−/− mice and B6.H2^d:C1qa^+/+ mice fixed 5 minutes after i.v. 34-1-2S at 1 mg/kg. Images in G. are maximum intensity projections sampling 240 μm from a cleared thick section. White arrowheads point to arterioles positive for complement components. B & C show means ± standard errors. P-values are from unpaired two-tailed t-tests on log₁₀-transformed data (B & C) or log-rank test (D), with group n=12.

Fig. 3.. Inhibiting IgG hexamer assembly reduces 34-1-2S-induced acute lung injury responses.

A. Molecular models of IgG hexamers based on PDB entry 1HZH, showing Fc:Fc and Fc:C1q interaction sites. B. Molecular model showing lysine residues in mouse IgG2a (mIgG2a), PDB entry 1IGT. C. Lung vascular permeability, D. excess lung water measurements and E. lung complement C3b/d and mIgG immunostains from LPS-primed BALB/c mice after i.v. injection of carbamylated or control 34-1-2S. F. Molecular model showing location of Fc domain lysine 439 (K439) in human IgG1 (hIgG1), PDB entry 1HZH. G. Lung vascular permeability, H. excess lung water measurements and I. lung complement C4b/d and Acta2 immunostains from LPS-primed B6.H2^d mice after i.v. injection with hIgG1-34-1-2S or hIgG1-34-1-2S with K439E mutation. J. Molecular model showing binding site of SpA-B to Fc domain of human IgG1 (hIgG1), PDB entries 1HZH and 5U4Y. K. Lung vascular permeability, L. excess lung water measurements and M. lung complement C4b/d and Acta2 immunostains from LPS-primed B6.H2^d mice after i.v. injection with hIgG1-34-1-2S either mixed with 75 μg SpA-B or vehicle control. Samples for injury measurements were collected at 2 hours after antibody injections and lungs were fixed for immunostaining at 5 minutes after antibody injections. Graphs show means ± standard errors. P-values are from unpaired two-tailed t-tests on log₁₀-transformed data, with group n=8 (C, D) or n=12 (G, H, K, L).

Fig. 4.. Fc mutations promoting IgG hexamer assembly increase in vivo pathogenicity of alloantibodies.

A. Molecular model showing amino acids mutated in RGY-hIgG1 antibodies, based on PDB entry 1HZH. B. Negative stain electron micrographs showing single hIgG1-34-1-2S molecules and spontaneous solution-phase hexamers formed by RGY-hIgG1-34-1-2S (colored overlay highlights structures in expanded images). C. Lung vascular permeability and D. excess lung water measurements from LPS-primed B6.H2^d mice injected with control or RGY-mutated hIgG1-34-1-2S monoclonal antibodies (mAbs) at i.v. doses of either 0.3 or 1 mg/kg. E. Immunofluorescence staining showing pulmonary arterioles stained for complement C4b/d (red) and Acta2 (cyan) in lung sections from LPS-primed B6.H2^d mice given indicated treatments, with samples fixed 5 minutes after antibody injections. F. Lung vascular permeability and G. excess lung water measurements from LPS-primed B6.H2^d mice injected with control or RGY-mutated hIgG1-Kd1 (a novel mAb targeting only the H-2K^d MHC class I antigen) at 1 mg/kg i.v.. Graphs show means ± standard errors with horizontal line representing standard deviations from ‘no injury’ controls (LPS-primed B6.H2^d mice given hIgG1 isotype control i.v.). Log₁₀-transformed data were analyzed using an ordinary two-way ANOVA with Šídák's multiple comparisons test for effect of Fc mutation within dose level (C, D) or unpaired two-tailed t-test (F, G), with group n=12.

Fig. 5.. Treatment with recombinant Fc hexamer decoys prevents alloantibody-mediated acute lung injury.

A. Molecular representation of CSL777, an investigational recombinant Fc hexamer ‘decoy’ treatment which inhibits classical complement activation, based on PDB entry 7X13 (50). B. Molecular representation of SCIg, a pooled human immunoglobulin therapeutic with anti-inflammatory properties at high doses, based on PDB entry 1HZH. C. Lung vascular permeability and D. excess lung water measurements from LPS-primed BALB/c mice given i.p. vehicle or CSL777 at indicated doses 1 hour before i.v. injection with 34-1-2S or mIgG2a isotype control. E. Lung vascular permeability and F. excess lung water measurements from LPS-primed BALB/c mice given i.p. vehicle or 2000 mg/kg SCIg 1 hour before i.v. injection with 34-1-2S or mIgG2a isotype control. G. and H. Immunofluorescence showing pulmonary arterioles stained for complement C4b/d (red) and Acta2 (cyan) in lung sections from LPS-primed BALB/c mice given indicated treatments, with samples fixed 5 minutes after antibody injections. White arrowheads point to arterioles with endothelial positivity for C4b/d. Graphs show means ± standard errors with horizontal line representing standard deviations from ‘no injury’ controls (from vehicle + isotype control group). Log₁₀-transformed data were analyzed using either (C, D) an ordinary one-way ANOVA with P-values from Dunnett's multiple comparisons test for difference relative to vehicle + 34-1-2S group, or (E, F) two-tailed unpaired t-test, with group n=10.

Fig. 6.. Acute lung injury is complement-dependent in a model incorporating human FCGR2A-mediated pathology.

A. Lung vascular permeability, B. excess lung water and C. survival readouts from LPS-primed B6.H2^d:hFCGR2a^Tg/0 mice and B6.H2^d littermate controls given i.v. hIgG1-34-1-2S at 1 mg/kg. D. Immunofluorescence imaging of platelet sequestration (CD41, red, with Acta2 in cyan) in lungs from B6.H2^d:hFCGR2a^Tg/0 mice and littermates without hFCGR2A fixed at 20 minutes after hIgG1-34-1-2S injections, quantified in E. F. Lung vascular permeability, G. excess lung water and H. survival readouts from LPS-primed B6.H2^d:C1qa^+/+ and B6.H2^d:C1qa^−/− mice, as well as littermates of each genotype expressing hFCGR2A, given i.v. hIgG1-34-1-2S at 1 mg/kg. I. Lung vascular permeability, J. excess lung water and K. survival readouts from LPS-primed B6.H2^d:hFCGR2a^Tg/0 mice given either vehicle or 50 mg/kg CSL777 i.p. before i.v. hIgG1-34-1-2S at 1 mg/kg. A, B, E, F, G, I & J show means ± standard errors with horizontal gray lines showing means or standard deviations of values from ‘no injury’ controls (B6.H2^d mice given LPS i.p. + hIgG1 isotype control i.v.) and were log₁₀-transformed prior to analysis. P-values are from: (A, B, I, J) unpaired, two tailed t-tests; (F, G) two-way ANOVA with Šídák's multiple comparisons test; or (C, H, K) log-rank test, with group n=4 (E) or 12 (other graphs).