MRI guided spatial transcriptomics of meningeal-based inflammation in SJL EAE.

(A) Schematic describing the experimental paradigm. SJL mice underwent brain MRI 6-, 8-, and 10-weeks post immunization with PLP139-151. Brain slices from regions with meningeal inflammation were collected and processed for spatial transcriptomics on the 10x Genomics platform. (B–C) Behavior scores (B) and mouse weights (C) of the EAE cohort. Red arrows indicate MRI time points, black arrow indicates time of tissue harvesting (N = 4). (D) Representative post-contrast MRI brain images, white arrows indicate areas of meningeal-based inflammation. (E) Representative images of H&E-stained tissue sections mounted on spatial transcriptomics slides (left, naïve; right, EAE). (F) Spatial feature plots from naïve (top row) and EAE (bottom row) representative samples demonstrate altered expression of genes related to complement (i), immune infiltration (ii), antigen presentation (iii), dendritic spines, and astrocyte activation (iv).

PROGENy analysis reveals spatially restricted pathway activity differences between naïve and EAE.

(A–B) Heat map displaying averaged PROGENy pathway analysis results. (C) Representative spatial plot showing activity of the JAK-STAT, NFkB, and TNFa signaling pathways. (D) Comparison of JAK-STAT, NFkB, and TNFa pathway activities between groups. (E) Representative spatial plot showing activity of the Trail, PI3K, and TGFb signaling pathways. (F) Comparison of Trail, PI3K, and TGFb pathway activities between groups. (Naïve mouse N = 4, sample N = 5; EAE mouse N = 4, sample N = 6; multiple T tests corrected for multiple comparisons with the Benjamini, Krieger, and Yekutieli method; * p < 0.05, ** p < 0.01, *** p < 0.001).

Unbiased clustering reveals a group of spots enriched in inflammatory genes.

(A–B) UMAP dimensionality reduction plots colored by (A) cluster or (B) group. (C) Bar plot showing the proportion of spots in each cluster by sample. (D) Number of spots in cluster 11 by group (N = 11; Student’s two-tailed T test). (E) Representative spatial feature plots of naïve and EAE samples showing the spatial distribution of each cluster. (F) MA plot comparing differences in gene expression between cluster 11 and all other clusters averaged across samples. Red and blue spots represent genes in cluster 11 that are significantly increased or decreased, respectively (adjusted p-value < 0.05, log 2 fold change > 1). (G) Bar plot of top 15 genes enriched in cluster 11 compared to other clusters. (H) Tree plot displaying gene set enrichment results using the gene ontology (GO) database. Spots in cluster 11 were compared to other spots and gene set sizes ranging from 10-500 were included (adjusted p-value < 0.05).

Subclustering of spots adjacent to meningeal immune follicles reveals a subset of active immune patterns.

(A) UMAP dimensionality reduction plots showing subclustering of cluster 1 colored by (left) group or (right) cluster. (B) UMAP dimensionality reduction plots showing subclustering of cluster 2 colored by (left) group or (right) cluster. (C) Representative spatial feature plot showing the locations of cluster 1 and 2 subclusters. (D) Distance from the center of indicated subclusters to the nearest point of cluster 11 (N = 11; Student’s two-tailed T-Test). (E–G) Tree plot displaying gene set enrichment results using the gene ontology (GO) database for subcluster 1_3 (E), 1_4 (F), and 2_6 (G) compared to other spots in their respective clusters. (H) Venn diagram shows overlap of significantly enriched GO gene sets between cluster 11 and subclusters 1_3, 1_4, and 2_6, with (I) 31 gene sets elevated in all. GO gene set of size ranging from 10-500 were included (adjusted p-value < 0.05).

Trajectory analysis reveals gradients of gene expression originating from meningeal lymphoid follicles.

(A) Trajectories were drawn based on spatial cluster plot (i) from C11 to C2 (ii). (B) Representative plot of B2m relative expression along the trajectory length. Green line: B2m expression; black line: ideal model fit, “logarithmic descending” (top) or “gradient descending” (bottom); red line: residual area under the curve (AUC) representing the difference between B2m expression and the ideal model. (C) Barplot showing residual AUC of B2m relative expression along the trajectory direction compared to “logarithmic descending” or “gradient descending” (Student’s two-tailed T-test). (D) Representative plot of C3 relative expression along the trajectory length. Green line: C3 expression; black line: ideal model fit, “logarithmic descending” (top) or “gradient descending” (bottom); red line: residual area under the curve (auc) representing the difference between C3 expression and the ideal model. (C) Barplot showing residual AUC of C3 relative expression along the trajectory direction compared to “logarithmic descending” or “gradient descending” (Student’s T-test). (F) Genesets that were previously identified as significantly enriched in C11 were selected for trajectory analysis. Residual AUCs were calculated for “logarithmic descending”, “gradient descending”, “logarithmic ascending”, and “gradient ascending” ideal fits and displayed on (i) a heatmap sorted by “gradient descending”. (ii) Representative feature plot demonstrating deeper penetration of upper genesets (related to antigen presentation and processing, microglial activation, IL-6 production, interferon gamma) response relative to other gene sets (B cell activation, T cell activation, TNF production, complement, humoral immune response). (G, I) Representative images of RNAscope labeling for (G) Fcgr3, B2m, Cd74, and (I) Gfap, C3 in SJL mice 11 weeks after EAE induction. Yellow dashed lines indicate the areas of leptomeningeal inflammation, scale bars represent 100μM. (H, J) Barplots representing the percent of marker-positive cells present at distances from leptomeningeal inflammation. Lines represent best fit curves from exponential regression. N = 3 animals per group; bars represent mean, error bars represent standard error.