Single-cell spatial transcriptomic profiling defines a pathogenic inflammatory niche in chronic active multiple sclerosis lesions - snRNA-seq

Compartmentalized inflammation is considered a critical factor in driving the progression of multiple sclerosis (MS). Yet, the mechanisms sustaining its persistence remain poorly understood. A hallmark of this persistent and slowly evolving inflammatory process are chronic active MS lesions. In this study, we created a high-resolution, single-cell molecular and spatial atlas of chronic inflammation in MS. To accomplish this, we combined single-nucleus RNA sequencing (snRNA-seq) with single-cell spatial transcriptomics using multiplexed error-robust fluorescence in situ hybridization (MERFISH) to examine MS lesions, specifically focusing on those exhibiting chronic active immune pathology characterized by lymphocyte presence. Our integrative profiling uncovered the molecular landscape of glial and immune cells, their disease-associated states, and the surrounding microenvironments. Within the lesion rim, we identified CD8+ T cell niches with inflamed, foamy microglia, characterized by an interferon response and dysfunctional lipid metabolism. To investigate the function of these microglia, we deleted ATP-binding cassette transporters A1 and G1 (ABCA1/G1) in microglia of mice and induced experimental autoimmune encephalomyelitis (EAE). Our findings revealed that inhibiting cholesterol efflux increased the formation of lipid-storing phagocytes, which actively drove inflammatory processes in EAE. Moreover, we found that pharmacologically targeting sterol metabolism presents a promising therapeutic strategy for mitigating inflammation in EAE. Thus, we have created a high-resolution map of immune niches in chronic active MS lesions, leading to the discovery of lipid-associated microglia as key drivers of neuroinflammation. Overall design: Frozen autopsy brain tissue from fourteen multiple-sclerosis donors and three neurologically healthy controls was sectioned at –20 °C into 80–100 µm slices, and an 8 mm punch (~25 mg) of white matter or matched control tissue was collected from each region of interest . Nuclei were liberated with the 10x Genomics Chromium Nuclei Isolation Kit, using an extended lysis–debris-removal protocol tailored to lipid-rich white matter; the final suspension was stained with propidium iodide and counted before loading. Single-nucleus libraries were prepared with Chromium 3'-v3.1 reagents and sequenced on an Illumina NovaSeq 6000 to ~30 000 read pairs per nucleus, generating ~20 000 quality-filtered nuclei per sample and 306 731 nuclei in total . Reads were aligned to the GRCh38 reference with CellRanger 7.1.0 ( --include-introns ), and nuclei were retained if they contained 700–25 000 UMIs, 500–7 500 genes and < 5 % mitochondrial transcripts. Per-sample objects were log-normalised in Seurat v5, highly variable genes selected, and batch effects removed with the HarmonyIntegration workflow; clustering was performed in PCA–UMAP space and the cleaned layers were merged with JoinLayers to yield a unified atlas. Finally, CytoSPACE was used to project the transcriptomic clusters onto matched MERFISH sections, assigning each nucleus to demyelinated, perilesional, or healthy white-matter domains and thereby linking cell states to their spatial lesion context. *rds files: slimmed Seurat object with normalized expression values, essential cell-level metadata, and the final dimensional reductions used in the associated publication.