Integrating spatial and single-cell transcriptomics to characterize the molecular and cellular architecture of the ischemic mouse brain

10.1126/scitranslmed.adg1323

Abstract

Neuroinflammation is acknowledged as a pivotal pathological event after cerebral ischemia. However, there is limited knowledge of the molecular and spatial characteristics of nonneuronal cells, as well as of the interactions between cell types in the ischemic brain. Here, we used spatial transcriptomics to study the ischemic hemisphere in mice after stroke and sequenced the transcriptomes of 19,777 spots, allowing us to both visualize the transcriptional landscape within the tissue and identify gene expression profiles linked to specific histologic entities. Cell types identified by single-cell RNA sequencing confirmed and enriched the spatial annotation of ischemia-associated gene expression in the peri-infarct area of the ischemic hemisphere. Analysis of ligand-receptor interactions in cell communication revealed galectin-9 to cell-surface glycoprotein CD44 (LGALS9-CD44) as a critical signaling pathway after ischemic injury and identified microglia and macrophages as the main source of galectins after stroke. Extracellular vesicle-mediated Lgals9 delivery improved the long-term functional recovery in photothrombotic stroke mice. Knockdown of Cd44 partially reversed these therapeutic effects, inhibiting oligodendrocyte differentiation and remyelination. In summary, our study provides a detailed molecular and cellular characterization of the peri-infact area in a murine stroke model and revealed Lgals9 as potential treatment target that warrants further investigation

Fig. 3. Single-cell spatial architecture of the peri-infarct area.

(A) t-SNE of 12 major cell populations identified in the peri-infarct cortex in the PT group and the corresponding region in sham mice. Left: Integration of data from the sham and PT groups. Top middle: The sham group. Bottom middle: The PT group. Right: Stacked bar graphs of cell-type proportions in the sham and PT groups. (B) Cell-type annotation was performed on the basis of the expression of well-established marker genes. Left: Colored contours corresponding to the 12 cell types in (A). The expression of known markers is shown using the same layout (the size of the dots represents the proportion of cells expressing the gene, and colors ranging from green to yellow indicate that the expression increases gradually). Right: Bar graph depicting the mean number of UMIs per cell in each cell type. (C) Feature plots showing the expression of known markers. (D) Probabilistic cell map of the gene transcripts. The map shows pie charts of probabilities matching the scRNA-seq clusters. Image of an hematoxylin and eosin–stained tissue section from the brain of a mouse subjected to PT on the left. The outlines of brain regions are shown.

Fig. 5. Microglia and macrophage heterogeneity revealed by scRNA-seq and spatial transcriptomics.

(A) Representative RNAscope images to analyze the colocalization of Tmem119 (microglia marker) and Lgals9. Similar results were observed in at least three independent experiments. (B) t-SNE of five cell populations identified by microglia reclustering. (C) The percentage and cell numbers of various microglial subtypes. Left: Pie charts showing the percentage of the various microglial subtypes identified in (B) for both the sham and PT groups. Right: The number of microglial subtypes in the sham and PT groups. (D) Bar graphs showing the gene expression of three representative markers for each cluster. The height of each bar represents the log-normalized counts of cells. (E) Spatial feature plots of the activated microglial subtypes. (F) Violin plot of Lgals9 expression in each microglial subtype. One-way ANOVA implemented by R function aov was performed to identify the different expression of Lgals9 among the microglial subtypes. The Tukey post hoc test was used for conducting pairwise comparisons. (G) Dot plot of Lgals9 expression in microglia subtypes. (H) t-SNE of four cell populations identified by macrophage reclustering. (I) Dot plot showing the expression of the selected five genes in each macrophage subtype. (J) Spatial feature plots of the macrophage subtypes. (K) Violin plot of Lgals9 expression in each macrophage subtype. One-way ANOVA implemented by R function aov was performed to identify the different expression of Lgals9 among the macrophage subtypes. The Tukey post hoc test was used for conducting pairwise comparisons. (L) Dot plot of Lgals9 expression in macrophage subtypes.

Fig. 8. Lgals9 promoted the OPC proliferation and differentiation through Cd44 after ischemic stroke.

(A) t-SNE of 10 cell populations identified by OPC and oligodendrocyte reclustering. (B) Dot plot of Cd44 expression in OPC and oligodendrocyte subtypes. (C) Thermodynamic value of cell proliferation activity on the t-SNE plot. Blue dashed box, defining the developmental program leading to cell division in OPCs. (D) Schematic of LV microinjection, RVG-EV administration, flow cytometry analysis, and luxol fast blue (LFB) staining. (E to H) Flow cytometry analysis of the effects of LGALS9-CD44 on different stages of oligodendrocytes in the peri-infarct area of stroke mice, as measured OPCs (O1−O4−A2B5+) (E), pro-oligodendroblasts (pro-OLs; O1−O4+A2B5+) (F), immature oligodendrocytes (I-OLs; O1+O4−A2B5−) (G), and mature oligodendrocytes (M-OLs; O1+O4+A2B5−) (H). n = 4 samples (the tissues of peri-infarct area from four mice were pooled for one sample) per group. P < 0.05, *P < 0.01, and ***P < 0.001 using two-way ANOVA followed by Holm-Šidák post hoc multiple comparisons test. (I) LFB staining at days 3, 7, and 21 after PT. The plots on the right of the representative images were the analysis of myelinated area. ***P < 0.001 versus the sham group; ##P < 0.01 versus the PT + RVG-Vector-EV group; †P < 0.05 and ††P < 0.01 versus the PT + RVG-LGALS9-EV group using one-way ANOVA followed by the Holm-Šidák post hoc multiple comparisons test.