Fig 2: Significantly enriched GSEA pathways for key genes. (A) Apoe. (B) Abca1. (C) Hexb. The upper part of the abscissa indicates high expression, and the lower part indicates low expression.

Fig 3: Immune infiltration in thalamic haemorrhage. (A) Relative percentages of 25 immune cell subsets. (B) Pearson correlation between 25 kinds of immune cells. Blue indicates a positive correlation, and red indicates a positive correlation. (C) The difference in immune cell content between control and disease samples, with control samples in blue and disease samples in red. (D–F) Pearson correlation between key genes Abca1, Apoe and Hexb and immune cells, positive correlation on the right and negative correlation on the left.

Fig 4: GW3965 upregulates LXR downstream-gene expression. (A) Representative images of confocal tile scan showing whole day 7 lesion labelled with DAPI (blue), GFAP (Cyan), ABCA1 (green) and Iba1 (red). Scale bar, 200 µm. (B) Representative image of a three-dimensional (3D) reconstruction of a perihaematomal region in the treatment group using Imaris. Scale bar, 3 µm. (C) Representative confocal images of perihaematomal region labelled with DAPI (blue), GFAP (cyan), ABCA1 (green) and Iba1 (red). The left corner within dotted lines depicts lesion centre. Scale bar, 100 µm. (D–F), Bar graphs comparing the ABCA1+ percentage (D), Iba1+ABCA1+ over Iba1+ proportions (E) and GFAP+ABCA1+ over GFAP+ proportions (F) between GW3965 treatment and DMSO control within the lesion region at day 7; the sample size was n=10 per group. (G) Representative images of confocal tile scan showing whole day 7 lesion labelled with DAPI (blue), Iba1 (cyan), GFAP (green) and ApoE (red). Scale bar, 200 µm. Scale bars are not equal. (H), Representative image of a 3D reconstruction of a perihaematomal region in the treatment group using Imaris. Scale bar, 7 µm. (I), Representative confocal images of perihaematomal region labelled with DAPI (blue), Iba1 (cyan), GFAP (green) and ApoE (red). The left corner within dotted lines is the lesion centre. Scale bar, 100 µm. (J–L), Bar graphs comparing the ApoE+ percentage (J), GFAP+ApoE+ over GFAP+ proportions (K) and Iba1+ApoE+ over Iba1+ proportions (L) between GW3965 treatment and DMSO control within the lesion region at day 7; the sample size was n=10 per group. Significance is indicated as *p<0.05, **p<0.01, ****p<0.0001; two-tailed, unpaired t-test (D, F, K); or two-tailed, unpaired t-test with Welch’s correction (E); or Mann-Whitney U test (J, L). Bar graphs show individual data points and represent mean±SD. ApoE, apolipoprotein E; LXR, liver X receptor.

Fig 5: Single-cell RNA sequencing of annotated macrophages from human IPF. Macrophages from a published IPF dataset (GSE136831) were re-mined. (A) Split identity UMAP analysis of clusters annotated at 0.1 resolution using the Seurat analysis package. (B) Cluster-based and disease-based top-5 non-redundant genes from annotated clusters. (C) Split identity UMAP shows expression distribution for TREM2 in control and IPF lungs. (D) Split identity bubble plot of selected genes associated with macrophage maturation (TREM2, MERTK, ATP6V0D2, ITGAX), recruitment and activation (ITGAM, CD163, HLA-DRB6, CTSK, COL3A1, COL4A2, MMP7, MMP9, FN1, SPP1, TGFBI), and lipid homeostasis (APOE, CD36, LRP1, LPL, LIPA, ABCA1, ABCG1, FABP5, PPARG). (E) Split identity UMAP showing expression distribution for MERTK in control and IPF lungs. (F) Split identity UMAP showing expression distribution for FN1 in control and IPF lungs. (G) Split identity UMAP showing expression distribution for SPP1 in control and IPF lungs. (H) Enrichr-based Reactome pathway analysis for the top-15 predicted pathways between controls and IPF samples. Single gradient color heatmap shows logarithmic adjusted p-values. Values above the arbitrary threshold (-log of adjusted p-value) of 6 were color-coated in black. (I) Enrichr-based Reactome pathway analysis comparing significant pathways in mononuclear myeloid clusters from the human IPF dataset, SP-CI73T spatial transcriptomics, and SP-CI73T single-cell sequencing datasets. Single gradient color heatmap shows logarithmic adjusted p-values. Values above the arbitrary threshold (-log of adjusted p-value) of 10 were color-coated in black.