Microglial expression of CD83 governs cellular activation and restrains neuroinflammation in experimental autoimmune encephalomyelitis.

CD83 is expressed in a distinctive regional pattern associated with white matter.a Flow cytometric assessment of CD83 expression in CD83eGFP reporter animals under homeostatic conditions. Cells were pre-gated on single living cells, and the percentage of eGFP+ cells was assessed. Wild-type cells served as negative controls. b Summary of the cellular composition among CD45+eGFP+ cells. Data represent the mean of four different animals. c Histograms of eGFP fluorescence in different cell types isolated from the CNS. Wild-type cells served as negative fluorescence controls. Histograms are representative of four individual mice. d Assessment of eGFP signal in microglia from the cortex (Ctx), hippocampus (Hip), cerebellum (Cbm), brainstem (Stem), and spinal cord (SC). Median fluorescence was normalized to cortical microglia (n = 7, and each dot represents pooled tissue from two mice). Data are represented as mean ± SEM. Statistical significance was calculated with one-way ANOVA with Dunnett’s multiple comparison test. e Immunofluorescence of human brain tissue. Microglia are stained for Iba1 (red) and CD83 (green). Nuclear staining was performed with DAPI. Length of all scale bars: 100 µm.

Expression of CD83 in microglia increases with cellular activation but also during resolution of neuroinflammation.a CD83 expression on microglia upon acute isolation. Freshly isolated microglia were kept either at 4 °C or 37 °C for 6 h (blue and red line, respectively), and expression was assessed via flow cytometry. Cells from CD83-/- mice served as negative controls (dotted line) (n = 3 individual mice). b, c Analyses of CD83 expression in microglia from healthy (n = 4) or EAE mice (n = 6; peak of disease, day 16 post immunization, p.i.). Statistically significant differences were detected with two-tailed Mann–Whitney U-test. d Microglia were isolated from different brain regions of healthy or EAE CD83-eGFP mice. Cells were assessed for expression of CD11c and MHC-II; dot-plots contain color mapping to visualize eGFP fluorescence and are representative of three individual mice. e Representative histogram overlay of eGFP fluorescence shown in (d). f Comparison between eGFP signal and percentage of CD11c+ microglia in healthy mice, EAE mice at the peak of disease or during the recovery phase (i.e., day 28 p.i.). g Longitudinal expression analysis of Cd83 in RR-EAE mice (n = 4–8 individual mice, sample on the respective disease stage). Statistical significance was calculated with one-way ANOVA with Dunnett’s multiple comparison test. In all graphs, data are represented as mean ± SEM.

Deletion of CD83 in microglia exacerbates autoimmune neuroinflammation.a Schematic depiction of conditional knockout strategy. b Verification of successful deletion: microglia were isolated 7 weeks after tamoxifen injection via FACS. After RNA isolation, Cd83 gene expression was analyzed in CD83ΔMG cells in relation to controls (n = 12, pooled from three independent experiments). c, d Gene expression analysis of different gene transcripts in acutely isolated microglia. Expression levels were first normalized to Rpl4 and relative expression was calculated in relation to controls (n = 12, pooled from three independent experiments). e EAE course over 30 days, comparison between CD83ΔMG and CD83wtCre controls. Evaluation of the individual peak score of each mouse and the cumulative score, calculated from the area under the curve (AUC) of each mouse. (n = 15 for CD83wtCre and n = 17 for CD83ΔMG, data are pooled from three independent experiments). f Gating strategy of CNS-infiltrating and resident immune cells (pre-gated on CD45+ cells) isolated at the peak of disease (day 18 p.i.). g, h Relative percentage of microglia, monocytes, monocyte-derived cells (MDCs), and monocyte-derived dendritic cells (moDCs) among all CD45+ cells at the peak of disease (n = 18 for CD83wtCre, and n = 19 for CD83ΔMG, data are pooled from three independent experiments). i expression of Ccl2, Ccl4, and Ccl5 on on day 18 p.i. (n = 12 for CD83wtCre, and n = 13 for CD83ΔMG, data are pooled from two independent experiments). In all graphs, data are represented as mean ± SEM and two-tailed Mann–Whitney U-test was used to analyze the data.

Single-cell RNA-sequencing (scRNA-Seq) of microglia from healthy and EAE mice.a Experimental setup: naïve and mice at the peak of EAE from CD83wtCre and CD83ΔMG genotypes were used (n = 4 per condition and genotype; total number of animals: 16). Microglia were isolated and subjected to droplet-based scRNA-Seq. b UMAP of the filtered and integrated scRNA-seq dataset showing the heterogeneity of microglia in all four conditions: healthy control CD83wtCre, healthy control CD83ΔMG, EAE CD83wtCre, EAE CD83ΔMG; n = 3 per condition). c UMAP shown in (b) split between healthy and disease conditions. Clusters Gpr34+ and Ccl4+ are dominant in healthy animals. The remaining clusters are dominant in diseased animals. d Heat-map of the expression of the most relevant marker genes among each microglia cluster from the scRNA-seq dataset shown in (b). e Violin plots showing Cd83 expression for each cluster, split between CD83wtCre and CD83ΔMG animals.

Microglial CD83-deficiency causes an over-activated phenotype supportive of pro-inflammatory pathways.a Expression of CD11c and MHC-II on microglia from the CNS on day 18 p.i. Representative dot-plots show gating for CD11c+/MHC-II+ cells, which were quantified in the attached bar chart (n = 18 for CD83wtCre and n = 19 for CD83ΔMG, data are pooled from three independent experiments). b Expression of Tmem119 on FACS-sorted microglia at the peak of disease (day 18 p.i.; n = 12 for CD83wtCre and n = 13 for CD83ΔMG, data are pooled from two independent experiments). c Clustering of EAE-enriched microglia from CD83wtCre and CD83ΔMG animals using Vertex frequency clustering (VFC). Cells of clusters Gpr34+ and Ccl4+ (enriched in healthy controls) were excluded from VFC. d Mean relative likelihood for each of the MHCII+ VFC clusters to be enriched in the EAE condition in CD83ΔMG animals compared with CD83wtCre animals. e Heat-map of the top 10 differentially expressed genes between VFC clusters MHCII+ 0 and 2 in EAE. f Volcano-plot showing differentially expressed genes between microglia from CD83wtCre and CD83ΔMG animals in EAE condition. Log2 fold change threshold was set to 0.25 and adjusted p-value threshold to 0.05. g Heat-map of the top 10 Gene Ontology Biological Process pathways between CD83wtCre and CD83ΔMG animals in EAE condition based on the differentially expressed genes. h Expression of Lpl and Trem2 on microglia on day 18 p.i. (n = 12 for CD83wtCre and n = 13 for CD83ΔMG, data are pooled from at least three independent experiments). i Expression of Ccl2 in microglial cultures 6 h after treatment with 10 µg/ml myelin debris. Expression is depicted as fold-change over untreated cells (n = 9 for CD83wtCre and n = 11 for CD83ΔMG, pooled from four independent experiments). In all graphs, data are represented as mean ± SEM and two-tailed Mann–Whitney U-test was used to analyze the data.

The inflamed CNS of CD83ΔMG exhibit disturbed T cell balance and a pro-inflammatory, disease-fostering milieu.a, b Flow cytometric analyses of regulatory T cells (Tregs) in the inflamed CNS. Representative dot-plots (a) and quantitative assessment of the percentage of CD25+FoxP3+ among CD4+ T cells (b, n = 10 for CD83wtCre and n = 11 for CD83ΔMG; pooled from two independent experiments) are shown. c, d Flow cytometric analyses of IFN-γ producing T cells in the inflamed CNS. Single cells from mice were isolated on day 18 p.i., re-stimulated with PMA/ionomycin for 5 h in the presence of Golgi transport inhibitors and intracellularly stained for IFN-γ. Representative dot-plots (c) and quantitative assessment of the percentage of IFN-γ producing CD4+ T cells (d, n = 10 for CD83wtCre and n = 11 for CD83ΔMG; pooled from two independent experiments) are shown. e Expression of Th1-related transcripts in the spinal cords of EAE animals on day 18 p.i. RNA was extracted from lumbar parts of spinal cords, reversely transcribed and qPCR was performed for Ifng, Tbx21 (T-bet) and Il12a (n = 10 for CD83wtCre and n = 11 for CD83ΔMG, pooled from two independent experiments). f TNF expression levels in spinal cords and sorted microglia or MDCs derived from EAE mice on day 18 p.i. (n = 10 for CD83wtCre and n = 11 for CD83ΔMG; pooled from two independent experiments). g Mmp9 expression levels in spinal cords of EAE mice on day 18 p.i. (n = 10 for CD83wtCre and n = 11 for CD83ΔMG; pooled from two independent experiments). Gene expression was normalized to Rpl4 and relative expression was calculated as ratio between CD83ΔMG and CD83wtCre. In all graphs, data are represented as mean ± SEM and Mann–Whitney U-test was used to analyze the data.

Graphic summary.During neuroinflammation, CD83ΔMG microglia get over-activated due to incapability to cope with myelin debris, which leads to enhanced production of chemokines (e.g., CCL2). Consequently, the CNS of CD83ΔMG mice experiences an elevated influx of pathogenic monocytes and T cells, the latter of which rather differentiate to Th1 cells than to regulatory T cells. The elevated amount of IFN-γ induces TNF-α production from microglia as well as monocyte-derived cells, which causes elevated expression of MMP9 and subsequent damage to the blood-brain barrier. On the other hand, CD83-deficient microglia fail to upregulate sufficient amounts of Lpl and Trem2 to clear accumulating myelin debris, thus perpetuating the pro-inflammatory milieu instead of leading to the resolution of inflammation.