Fig 1: cisTOPIC identifies 16 clusters of cells spanning 30 topics from 14,251 cells.a Single-cells from 18 prostate samples were clustered using cisTOPIC based on their chromatin accessibility profiles. Cells are coloured according to their sample IDs on the UMAP. b Single-cells from 18 prostate samples formed 16 clusters based on their shared chromatin accessibility profiles. Cells are coloured according to their cluster IDs on the UMAP. c Heatmap showing the topic distribution across 16 clusters. d Epithelial and stromal cell types are identified in localized prostate tumours. Lymphoid, myeloid and fibroblasts (grey tones) are removed from downstream analyses. Outer clusters show high-MYC accessibility (blue tones). The middle cluster shows higher accessibility to genes associated with inflammatory response (red tones). e Gene scores are shown for epithelial cell type markers luminal (KRT8), basal (KRT5 and KRT14) and neuroendocrine (Chromogranin A); common prostate epithelial cell markers AR, TMPRSS2 NKX3.1; prostate cancer cell markers AMACR, EPCAM, CDH1, C-MYC and PTEN.
Fig 2: Basal-like MMTV-PYMT tumor cells arise from the luminal lineage. A, B Strategy to trace basal and luminal lineage tumor cells in MMTV-PyMT mice. Lineage-specific and tamoxifen-inducible CreERT2 was used to specifically label basal (A) or luminal (B) epithelial cells with GFP. C Representative flow cytometry plots of luminal and basal tumor cells inheriting the basal lineage GFP label. D, E Percentage of basal and luminal cells expressing basal lineage GFP in Krt5-CreERT/Rosa26-mTmG/MMTV-PyMT tumors (n = 7, p = 0.0513, unpaired t test) (D) and the normal Krt5-CreERT/Rosa26-mTmG mouse mammary gland (n = 4, p = 0.028, unpaired t test) (E). F Representative flow cytometry plots showing luminal and basal tumor cells inheriting the luminal lineage GFP label. G, H Percentage of basal and luminal cells expressing luminal lineage GFP in Krt8-CreERT/Rosa26-mTmG/MMTV-PyMT tumors (n = 8, p = 0.4938, unpaired t test) (G) and the normal Krt8-CreERT/Rosa26-mTmG mouse mammary gland (n = 4, p = 0.0005, unpaired t test) (H). I Representative images of TSA staining of Krt5-CreERT/Rosa26-mTmG/MMTV-PyMT and Krt8-CreERT/Rosa26-mTmG/MMTV-PyMT tumors. Samples were stained with Krt8 (purple), Krt5 (red), Krt14 (white), and GFP (green). Yellow arrows point to Krt5+ /GFP+ tumor cells, and green arrows point to Krt14+ /GFP+ tumor cells. J Quantification of lumino-basal tumor cells expressing GFP and strictly basal tumor cells expressing GFP from TSA-stained images of Krt8-CreERT/Rosa26-mTmG/MMTV-PyMT tumors. K Representative images of TSA staining of lung metastases from Krt5-CreERT/Rosa26-mTmG/MMTV-PyMT and Krt8-CreERT/Rosa26-mTmG/MMTV-PyMT mice. Samples were stained with Krt8 (purple), Krt14 (white), GFP (green), and DAPI (blue). White arrows point to Krt14-expressing cells
Fig 3: Sox10 is a potential driver of luminal-to-basal plasticity. A UMAP of Krt8-CreERT/Rosa26-mTmG/MMTV-PyMT tumor cells sorted for GFP. Unsupervised clustering divided the cells into 18 different clusters. B Heatmap showing the basal, mature luminal, and luminal progenitor gene signature activity in each cluster. Cluster 13 (purple box) has high expression of mature luminal genes, while clusters 15 and 16 (orange box) have high expression of basal genes. Cluster 6 (red box) has high expression of both mature luminal and basal genes, and is identified as the lumino-basal cluster. C Ternary plots showing the distribution of relative activity of luminal progenitor, mature luminal, and basal gene signatures across all tumor cells profiled by scRNA-seq. In the left plot, cells assigned to cluster 6 are highlighted in red, while cells assigned to cluster 13 are highlighted in the right plot. D Volcano plot showing genes upregulated and downregulated in cluster 6 as compared to cluster 13. Light blue points are basal marker genes, dark blue points are luminal marker genes, yellow points are mesenchymal genes, orange points are epithelial genes, red points are potential plasticity driver genes, and green points are SOX10 target genes. E Boxplots showing the differences in distribution of POSTN, SOX10, and SPP1 gene expression between the ERα low and ERα high groups of tumors obtained from TCGA. F Representative images of SOX10 IHC staining in luminal B and low-ER TMAs. G Quantified SOX10 expression in luminal B and low-ER TMAs
Fig 4: Low-ER tumors are distinct from luminal B tumors. A Details of the luminal B tumors included in TMAs as a comparison to Low-ER tumors. B, C H&E staining showing typical features of a luminal B tumor (high-grade invasive ductal carcinoma ER+ (> 90%) PR+ (30%) HER2- high Ki67 tumor). B Lower power image showing a poorly differentiated carcinoma with irregular, infiltrative borders (1X H&E). C Tumor cells infiltrating as small solid nests with scattered tubule formation within a desmoplastic stroma. Note the lack of necrosis, solid growth, and a tumor associated lymphoplasmacytic infiltrate typically seen in breast carcinomas with basal-like characteristics (20X H&E). D, E ER expression (D) and intensity (E) levels of the luminal B tumors. F Representative images of Krt5 IHC staining in low-ER TMAs. G Quantified Krt5 expression in luminal B and low-ER TMAs. H Representative images of TSA staining containing heterogeneous populations in low-ER TMAs. Samples were stained with Krt8 (purple), Krt5 (cyan), Krt14 (white), ER (green), and DAPI (blue). White arrows point to basal tumor cells, yellow arrows point to luminal tumor cells, and green arrows point to lumino-basal tumor cells. I Quantification of different cell phenotypes found within 13 of the low-ER TMA cores which were found to harbor heterogeneous populations (TMA cores with homogenous populations were excluded)
Supplier Page from Abcam for Anti-Cytokeratin 8 antibody [EP1628Y] (Alexa Fluor® 647)