Microglia nodule formation is a common feature in inflammatory brain diseases mediated by T lymphocytes such as viral and paraneoplastic encephalitis, multiple sclerosis, and Rasmussen encephalitis (RE). However, its role has not been fully understood yet. We hypothesized that, in RE, microglial nodules provide an environment for the initiation of the later dominating T-cell cytotoxicity. In RE stage 0, small primary microglia nodules could be identified in the absence of T cells. These primary nodules showed inflammasome activation and endosomal Toll-like receptor upregulation. In stage 1, T cells migrate into the parenchyma and intermingle with microglial cells, thereby forming secondary nodules in which neurons are destroyed. Whole-genome transcriptome analysis at this point showed upregulation of several inflammatory pathways including interferon signaling and major histocompatibility complex-I signaling. Inflammatory profiles, like the ones observed in RE, could be induced upon TLR3 stimulation in neonatal microglial cell cultures. Taken together, our results point towards activation of endosomal TLRs, resulting in increased interferon signaling, inflammasome activation, and chemokine upregulation as early steps in RE pathogenesis. This activity sets the scene for subsequent infiltration of T cells and destruction of neurons. Similar to RE, this microglial microenvironment might be a crucial step in other T-cell-mediated inflammatory brain diseases.Electronic supplementary materialThe online version of this article (10.1007/s00401-019-01958-5) contains supplementary material, which is available to authorized users.

Human leucine-rich glioma-inactivated protein 1 encephalitis (LGI1) is an autoimmune limbic encephalitis in which serum and cerebrospinal fluid contain antibodies targeting LGI1, a protein of the voltage gated potassium channel (VGKC) complex. Recently, we showed that a feline model of limbic encephalitis with LGI1 antibodies, called feline complex partial seizures with orofacial involvement (FEPSO), is highly comparable to human LGI1 encephalitis. In human LGI1 encephalitis, neuropathological investigations are difficult because very little material is available. Taking advantage of this natural animal model to study pathological mechanisms will, therefore, contribute to a better understanding of its human counterpart. Here, we present a brain-wide histopathological analysis of FEPSO. We discovered that blood–brain barrier (BBB) leakage was present not only in all regions of the hippocampus but also in other limbic structures such as the subiculum, amygdale, and piriform lobe. However, in other regions, such as the cerebellum, no leakage was observed. In addition, this brain-region-specific immunoglobulin leakage was associated with the breakdown of endothelial tight junctions. Brain areas affected by BBB dysfunction also revealed immunoglobulin and complement deposition as well as neuronal cell death. These neuropathological findings were supported by magnetic resonance imaging showing signal and volume increase in the amygdala and the piriform lobe. Importantly, we could show that BBB disturbance in LGI1 encephalitis does not depend on T cell infiltrates, which were present brain-wide. This finding points toward another, so far unknown, mechanism of opening the BBB. The limbic predilection sites of immunoglobulin antibody leakage into the brain may explain why most patients with LGI1 antibodies have a limbic phenotype even though LGI1, the target protein, is ubiquitously distributed across the central nervous system.