Features of MOG required for recognition by patients with MOG antibody-associated disorders

Antibodies to myelin oligodendrocyte glycoprotein (MOG-Abs) define a distinct disease entity. Here we aimed to understand essential structural features of MOG required for recognition by autoantibodies from patients. We produced the N-terminal part of MOG in a conformationally correct form; this domain was insufficient to identify patients with MOG-Abs by ELISA even after site-directed binding. This was neither due to a lack of lipid embedding nor to a missing putative epitope at the C-terminus, which we confirmed to be an intracellular domain. When MOG was displayed on transfected cells, patients with MOG-Abs recognized full-length MOG much better than its N-terminal part with the first hydrophobic domain (P < 0.0001). Even antibodies affinity-purified with the extracellular part of MOG recognized full-length MOG better than the extracellular part of MOG after transfection. The second hydrophobic domain of MOG enhanced the recognition of the extracellular part of MOG by antibodies from patients as seen with truncated variants of MOG. We confirmed the pivotal role of the second hydrophobic domain by fusing the intracellular part of MOG from the evolutionary distant opossum to the human extracellular part; the chimeric construct restored the antibody binding completely. Further, we found that in contrast to 8-18C5, MOG-Abs from patients bound preferentially as F(ab')2 rather than Fab. It was previously found that bivalent binding of human IgG1, the prominent isotype of MOG-Abs, requires that its target antigen is displayed at a distance of 13-16 nm. We found that, upon transfection, molecules of MOG did not interact so closely to induce a Förster resonance energy transfer signal, indicating that they are more than 6 nm apart. We propose that the intracellular part of MOG holds the monomers apart at a suitable distance for bivalent binding; this could explain why a cell-based assay is needed to identify MOG-Abs. Our finding that MOG-Abs from most patients require bivalent binding has implications for understanding the pathogenesis of MOG-Ab associated disorders. Since bivalently bound antibodies have been reported to only poorly bind C1q, we speculate that the pathogenicity of MOG-Abs is mostly mediated by other mechanisms than complement activation. Therefore, therapeutic inhibition of complement activation should be less efficient in MOG-Ab associated disorders than in patients with antibodies to aquaporin-4 .