Fig. 4: Applying SPACECAT to a KP autochthonous lung tumors uncovers spatial heterogeneity of immune infiltrate.
a Experimental schematic and representative stitched images of KP lung tumors slices in FITC fluorescence channel before photoactivation (left) and after photoactivation of regions within the white dashed boundary (right). Scale bar = 500 μm. Created using BioRender.com (2021). b UMAP embedding and Louvain clusters of 4343 single cells, colored by annotated cell type and state. From Tumor 1, 1634 were from the non-photoactivated, total population and 686 were from the Healthy/Tumor Border. From Tumor 2, 1489 were from the non-photoactivated population and 534 from the Healthy/Tumor Border. c Hierarchical clustering of cell types and dot plot of top differentially expressed genes between each cluster and all other cells. All genes significant by Bonferroni-adjusted p value <0.001. Dot color represents average expression per cluster; dot size represents percentage of expression per cluster. d UMAP embedding with points corresponding to cells only in Tumor 1. Points colored by sort condition; yellow, green, and gray represent triplicate Seq-Well arrays from the Whole Tumor. Red points represent single cells photoactivated and sorted from the Healthy/Tumor Border. e UMAP embedding with points corresponding to cells only in Tumor 2. Points colored by sort condition; blue, yellow, and red represent triplicate Seq-Well arrays from the Whole Tumor. Black points represent single cells photoactivated and sorted from the Healthy/Tumor Border. f Relative cell-type compositions colored by region and tumor. Individual dots represent proportions in the triplicate technical replicates of three Seq-Well arrays from the Whole Tumor for each tumor (blue for Tumor 1, red for Tumor 2) or single Seq-Well array from the Healthy Tumor Border (black). X-axis presented in log scale, and length of bar indicates mean proportion across triplicate technical replicates. g Expression of individual genes among Monocyte or Macrophage cells by tumor region in Tumor 1 and 2 (Healthy/Tumor Border vs. Whole Tumor). Two-sided likelihood ratio test for single-cell expression with Bonferroni-adjusted p value ***<0.001, **<0.01, *<0.05 (n genes = 27,923); p values for Tumor 1 and Tumor 2, respectively for: Tnf: 5.103442 × 10−59 and 1.230068 × 10−07, Cstb: 4.442659 × 10−06 and 1.851645 × 10−05, Lsp1: 3.122840 × 10−03 and 3.862113 × 10−02, Il1a: 1.376249 × 10−30 and 5.629442 × 10−03, Il1rn: 2.620537 × 10−19 and 9.675296 × 10−07, Actb: 5.392148 × 10−12 and 1.491589 × 10−02. h, i Volcano plots of differentially expressed genes between Monocyte and Macrophage cells by tumor region; dashed lines correspond to Bonferroni-adjusted p value = 0.05. h Healthy/Tumor 1 Border (n = 480 cells) vs. Whole Tumor 1 (n = 674 cells); blue: higher relative expression in Whole Tumor 1, black: higher relative expression in Healthy/Tumor 1 Border; n genes = 27,923. i Healthy/Tumor 2 Border (n = 104 cells) vs. Whole Tumor 2 (n = 527 cells); red: higher relative expression in Whole Tumor 2, black: higher relative expression in Healthy/Tumor 2 Border; n genes = 27,923. j Heatmap of 321 genes negatively and 360 genes positively correlated with Healthy/Tumor Border Score among Tumor 1 Monocyte/Macrophage cells (blue: lower relative expression, red: higher relative expression). Cells are ordered by expression of Healthy/Tumor Border Score (color bar represents original location of each cell: blue, Whole Tumor; black, Healthy/Tumor Border). All genes significant by p value < 0.05. k p Values of significantly enriched gene ontologies among genes positively correlated with Healthy/Tumor Border Score (black bars) and genes negatively correlated with Healthy/Tumor Border Score (blue bars). Dashed lines correspond to p value = 0.05. Source data are available as a Source data file.
