Neuro-immunology in a mouse model of anti-NMDAR encephalitis and assessment of treatment approaches

Abstract

Anti-N-methyl-D-aspartate receptor (NMDAR) encephalitis is a disorder mediated by autoantibodies against the GluN1 subunit of NMDAR. It occurs with severe neuropsychiatric symptoms that often improve with immunotherapy. Clinical studies and animal models based on patients antibody transfer or NMDAR immunization suggest that the autoantibodies play a major pathogenic role. Yet, there is an important need of models offering an all-inclusive neuro-immunobiology of the disease together with a clinical course long enough to facilitate the assessment of potential new treatments. Toward this end, eight-week-old female mice (C57BL/6J) were immunized (days 1 and 28) with GluN1356-385 peptide or saline with AddaVax adjuvant and pertussis toxin. After symptom development ([~]day 35), subsets of mice were treated with an anti-CD20 (day 35), a positive allosteric modulator (PAM) of NMDAR (NMDAR-PAM, SGE-301) from days 45 to 71, or both. GluN1-antibody synthesis, epitope spreading, effects of antibodies on density and function of NMDAR, brain immunological infiltrates, microglial activation and NMDAR phagocytosis, and antibody synthesis in cultured inguinal and deep cervical lymph nodes (DCLN) were assessed with techniques including immunohistochemistry, calcium imaging, confocal and super-resolution microscopy, electrophysiology, or flow cytometry. Changes of memory and behaviour were assessed with a panel of behavioural tests, and clinical/subclinical seizures with brain-implanted electrodes. Immunized mice, but not controls, developed serum and CSF NMDAR-antibodies (IgG1 predominant) against the immunizing peptide and other GluN1 regions (epitope spreading) resulting in a decrease of synaptic and extrasynaptic NMDAR clusters and reduction of hippocampal plasticity. These findings were associated with brain inflammatory infiltrates, mainly B- and plasma cells, microglial activation, colocalization of NMDAR-IgG complexes with microglia, and presence of these complexes within microglial endosomes. Cultures of DCLC showed GluN1-antibody production. These findings were associated with psychotic-like behaviour (predominant at disease onset), memory deficit, depressive-like behaviour, abnormal movements (15% of mice), and lower threshold for developing pentylenetetrazole-induced seizures (hypoactivity, myoclonic jerks, continuous tonic-clonic) which correlated with regional cFOS expression. Most symptoms and neurobiological alterations were reversed by the anti-CD20 and PAM, alone or combined. Initial repopulation of B cells, by the end of the study, was associated with re-emergence of clinical-neurobiological alterations, which were abrogated by PAM. Overall, this model offers an all-inclusive neuro-immunobiology of the disease, allowing testing novel treatments, supporting the potential therapeutic role of NMDAR-PAM, and suggesting an immunological paradigm of systemic antigen presentation and brain NMDAR epitope spreading, which along the DCLN might contribute to fine-tune the polyclonal immune response.