Abstract
Conventional rodent neuromyelitis optica spectrum disorder (NMOSD) models using patient-derived immunoglobulin G (IgG) are potentially affected by the differences between the human and rodent aquaporin-4 (AQP4) extracellular domains (ECDs). We hypothesized that the humanization of AQP4 ECDs would make the rodent model lesions closer to human NMOSD pathology. Humanized-AQP4-expressing (hAQP4) rats were generated using genome-editing technology, and the human AQP4-specific monoclonal antibody (mAb) or six patient-derived IgGs were introduced intraperitoneally into hAQP4 rats and wild-type Lewis (WT) rats after immunization with myelin basic protein and complete Freund's adjuvant. Human AQP4-specific mAb induced astrocyte loss lesions specifically in hAQP4 rats. The patient-derived IgGs also induced NMOSD-like tissue-destructive lesions with AQP4 loss, demyelination, axonal swelling, complement deposition, and marked neutrophil and macrophage/microglia infiltration in hAQP4 rats; however, the difference in AQP4 loss lesion size and infiltrating cells was not significant between hAQP4 and WT rats. The patient-derived IgGs bound to both human and rat AQP4 M23, suggesting their binding to the shared region of human and rat AQP4 ECDs. Anti-AQP4 titers positively correlated with AQP4 loss lesion size and neutrophil and macrophage/microglia infiltration. Considering that patient-derived IgGs vary in binding sites and affinities and some of them may not bind to rodent AQP4, our hAQP4 rat is expected to reproduce NMOSD-like pathology more accurately than WT rats.