Additional file 1 of Unraveling the YAP1-TGFβ1 axis: a key driver of androgen receptor loss in prostate cancer-associated fibroblasts

Supplementary Materia 1: Supplemental Figure 1. Isolation and characterization of ex vivo culture of primary prostate fibroblasts. A) radical prostatectomy tissue wedges (i) containing regions of suspected malignancy or benign-adjacent tissue were sampled (ii) using 4 mm3 biopsy punchers. The top part of each biopsy core was removed (iii) for FFPE processing (iv) and subsequent histopathological evaluation. Remaining tissue was quartered (v) and similarly processed. B) Histopathological validation of the top section of biopsy cores from (Aiii-iv) via HE staining and p63 (brown) and AMACR (red) dual-immunohistochemistry. C) The remaining biopsy core was cut into small pieces and transferred to culture medium that supports fibroblast but not endothelial cell growth. Outgrowing fibroblasts were selectively enriched via trypsinization from any epithelial cells, which require additional collagenase treatment for detachment. D) Representative images of three primary fibroblast explant cultures isolated from different patients stained for mesenchymal markers vimentin (green) and CD90 (white) and the epithelial marker pan-cytokeratin (red). Nuclei were counterstained using Hoechst (blue). 22Rv1 PCa cells served as positive control for pan-cytokeratin (far right, upper panel). Negative control of a parallel stained fibroblast culture incubated without primary antibodies (far right, lower panel). Supplemental Figure 2. Ex vivo culture of primary prostate fibroblasts – pertaining to Supplemental Fig. 1. Single channel monochromatic images (from Supplemental Fig. 1D) of immunofluorescent validation of three primary fibroblast explant cultures isolated from different patients stained for mesenchymal markers vimentin (green) and CD90 (white) and the epithelial marker pan-cytokeratin (pan CK, red). Nuclei were counterstained using Hoechst (blue). 22Rv1 PCa cells served as a positive control (pos. ctrl) for pan-cytokeratin. Negative control (neg. ctrl) of a parallel stained fibroblast culture incubated without primary antibodies (far right, lower panel). Supplemental Figure 3. Transcriptomic analysis of primary prostate fibroblast cultures identifies two distinct CAF populations - pertaining to Figure 1 and 2. A) Heatmap depicting sample level expression of significantly upregulated genes (P.adj0.3) for each fibroblast cluster in the transcriptomic dataset (related to Fig. 1 C). B) clusterProfiler analysis of left: GO Biological Process for significantly upregulated genes of each fibroblast cluster, including two cycling CAF explant cultures (cycling), centre and right: the top 20 scoring Hallmark and REACTOME pathways for significantly upregulated genes of each fibroblast cluster (related to Fig. 1E-G). C-D) Expression of the indicated genes across each fibroblast cluster. Values denote C) average log2 signal intensity from quartile normalized bulk transcriptomic data and D) mean 2-ΔΔCt via qRT-PCR relative to the housekeeping gene TBP. Error bars denote S.E.M. Significance was calculated via a Kruskal-Wallis test with multi-comparison correction using the two-stage linear step-up method of Benjamini, Krieger and Yekutieli. E) Comparison of expression profiles for each fibroblast cluster with published scRNA-seq datasets using combined z-scores for each gene signature and represented as mean per fibroblast cluster (related to Fig. 1I-J). F) Expression of signature genes (combined z-score) for each primary fibroblast cluster in the TCGA prostate adenocarcinoma (PRAD) cohort (related to Fig. 2A-D). Boxplots were used to compare signature gene expression levels across Gleason scores, T and N stages and biochemical recurrence. The R package ggsignif was used to perform t-tests. G) Expression levels of substate-delineating markers used to perform survival analyses (Fig. 2 C) in different cell types of the Heidegger PCa scRNA-seq dataset. Source data for panels A-C, E-F are provided in the Source Data file. Supplemental Figure 4. Activation trajectory of fibroblast subpopulations- pertaining to Figure 1. A-C) Inferred pseudotime trajectory of CAF activation in cells annotated as CAF in the pan-cancer single cell CAF Atlas (1) available at https://gist-fgl.github.io/sc-caf-atlas/ . For orientation, A) the distribution of the original fibroblast clusters, pseudotime and cell activation state and B) expression of canonical markers for distinct stromal cell types are depicted. C) Distribution of C1-C3 fibroblast subpopulation markers on the inferred CAF activation trajectory. Supplemental Figure 5. Primary explant cultures depict distinct stable homogeneous fibroblast phenotypes. A) Sample level heatmap (clustered according to ward.D) for qRT-PCR-derived 2-ΔΔCT values of the indicated genes in an independent cohort of 29 primary human prostate explant cultures (qRT-PCR cohort) and fresh cultures from 3 donors used in the original bulk transcriptomic cohort. Sample numbers are indicated beneath the heatmap. Font color denotes the subpopulation annotated using the different methodologies (C1, blue; C2, yellow; C3, red; n.d., not determined). B) Expression level of the indicated genes in patient-matched C1 and C2 cultures and a non-matched C3 culture maintained longitudinally for the indicated number of passages. Data are representative of longitudinal cultures from a total of six different donors. C) Immunofluorescence using the antibodies stated against the indicated markers in three different primary human prostate explant cultures that clustered in the independent explant cohort (A) into C1/C2/C3 subpopulations. Font color denotes pseudo-coloring in the displayed merged images. Nuclei were counterstained using Hoechst 33342. Images are representative of three experiments using explant cultures from seven different patients. Supplemental Figure 6. Flow cytometry-based single cell analysis of primary human prostate fibroblast cultures. A-H) Flow cytometry analysis of thirty-six biobank cultures selected randomly or on the basis of their substate annotation per the original transcriptomic dataset (Supplemental Table S2) and/or qRT-PCR of substate-delineating markers (Supplemental Fig. S4 A) and stained for the cell surface stromal markers FAP, PDPN, CD90/THY1, CD140b/PDGFRB, CD140a/PDGFRA, MCAM and CD105/ENG. A-B) tSNE dimensional reduction analysis of flow cytometry (FC) data from twenty-three primary prostate fibroblast cultures (training set) based on their expression of the aforementioned stromal makers. Sample counts are stratified to visualize A) the original 13 clusters identified via XShift or B) sample number. C) Heatmap of flow cytometry data displaying percentage of counts in the positive marker gate and percentage of counts in samples annotated as C1, C2 or C3 subpopulations as defined and marked * in panel H). D) Gating strategy developed using the most distinguishable stromal surface markers identified from C), which yielded eight gated clusters from the original 13 clusters. E) tSNE plot overlaid with the 8 gated clusters identified by applying the gating strategy in D). F) Heatmap clustered based on the percentage of gated clusters for 36 fibroblast cultures (extended set), including the 23 explant culture samples from the training set. FC sample numbering of the extended set corresponds to that of the training set in the tSNE plot in A). Annotation of donor-related cultures that were also used in the qRT-PCR cohort (Supplemental Fig. S4A) and original transcriptomic cohort (Fig. 1, Supplemental Table 2) are depicted underneath the corresponding FC sample number. Data are scaled by column to highlight the most prominent cluster in each sample. G) tSNE plot overlaid with sample counts from cultures annotated as C1, C2 or C3 subpopulations as defined and marked * in panel H). Grey shading denotes ungated cells of samples that did not display a single enriched cluster or were not determined (n.d.). H) Stacked bar plot displaying the percentages of gated clusters within each primary prostate fibroblast culture sample (n = 36), grouped according to the predominant gated cluster composition of each sample. Sample numbering beneath each bar corresponds to the sample numbering in F). Cultures enriched for a single cluster (≥55%) were annotated as C1, C2 or C3 (marked *). Cultures enriched for two of either clusters 6 and 10 or clusters 10 and 1 were annotated as C1/C2 or C2/C3, respectively. All other cultures were designated n.d. (not determined). I) qRT-PCR of the indicated gene in primary prostate fibroblast cultures sorted via FACS for cluster 1, 6+10 combined (lower panel) or cluster 6 or 10 separately (upper panel). Genes associated with general fibroblast activation/contractile markers are indicated in black, while C1, C2, and C3 subpopulation-delineating markers (from Fig. 1) are shown in blue, yellow, and red, respectively. Values represent mean 2-ΔΔCt via qRT-PCR, normalized to the housekeeping gene TBP. Data are representative of two independent sorting experiments using five fibroblast cultures isolated from different donors. F-H) Font color denotes the subpopulation annotated using the different complementary methodologies (C1, blue; C2, yellow; C3, red; black font, not determined; black font with yellow-red box denotes cultures considered to comprise a mix of C2/C3 substates). Supplemental Figure 7. Expression of fibroblast subpopulation-delineating markers across PCa disease stages - pertaining to Figure 2E-G. UMAP visualization of stromal cell A) cluster distribution and B) feature plots displaying expression of the indicated genes across different disease stages in the single cell Prostate Cancer Cell Atlas [27] available at https://pccat.net . For clarity, only fibroblast clusters are displayed. Supplemental Figure 8. Expression of fibroblast substate-delineating marker genes during PCa progression - pertaining to Figure 2G. Bar plots depicting expression scores of A) the indicated fibrobla