To further investigate the effect of CD83-deletion on microglial activation during EAE, we re-analyzed the scRNA-Seq data with the aim to identify possible subclusters within the EAE-derived samples. We calculated the sample-associated relative likelihood with MELD38 and performed vertex frequency clustering on the merged MHCII+, Stmn1+, Mki67+ and Cxcl10+ clusters, and each major cluster was divided into subclusters based on the visual distribution of the EAE-wildtype associated likelihood. This process resulted in six subclusters, three of which contained cells of both original MHC-II+ clusters (Fig. 5c). While CD83cKO and WT cells were equally likely distributed in VFC cluster 1, which largely covers the original Cxcl10+ cluster, WT cells were rather associated with VFC2, and CD83cKO cell with VFC3 (Supplementary Fig. 5e). Interestingly, we detected a striking diversification of WT and cKO microglia when comparing the MHC-II+ subclusters: WT cells were more likely to be represented in MHC-II+ sub-cluster 0, whereas CD83-deficient cells had a higher likelihood to be present in MHC-II+ sub-cluster 2 (Fig. 5d). These subclusters also differed significantly in their gene expression pattern: MHC-II+ sub-cluster 0 retained more
