Categories
Prostate cancer

Aberrant androgen action in prostatic progenitor cells induces oncogenesis and tumor development through IGF1 and Wnt axes

Prostatic Osr1-lineage cells possess progenitor properties in expansion of prostatic epithelia through embryonic and pubescent development

Development of HGPIN and prostatic adenocarcinoma lesions in Osr1-Cre-driven hARtg transgenic mice but not in previous AR transgenic models regulated by the PB promoter implicates the critical role of prostatic Osr1-lineage cells in prostate tumorigenesis10. To assess the cellular properties of prostatic Osr1-lineage cells, we examined the molecular characterization of these cells using scRNA-seq analyses with urogenital sinus (UGS), and prepubescent and pubertal prostate tissues (Fig. 1a). Approximately 766, 5676, and 8253 cells from embryonic day E18.5 UGS tissues, and postnatal day P14 and P35 prostate tissues were obtained after filtering, respectively (Supplementary Fig. 1), and then merged and clustered after performing cell cycle regression. Both epithelial and stromal cell subsets were identified in the merged Uniform Manifold Approximation and Projection (UMAP) plots (Fig. 1b), aligning similar cell types based on their transcriptomic profiles using Seurat’s integrated method11,12 (Supplementary Fig. 2a–c). These cell subsets also appeared in individual UMAP plots, suggesting comparable cellular properties from UGS and prostate tissue samples (Fig. 1c, top panel). Osr1 expression appeared in both mesenchymal and epithelial cells at E18.5 UGS samples but significantly decreased in both total cells and total epithelial cells of P14 (p = 0.004 and p = 0.002, respectively) and P35 (p = 0.003 and p = 0.002, respectively) samples (Fig. 1c, bottom panel; Supplementary Fig. 2d, e). A significant positive correlation between expression of Osr1 with prostatic stem/progenitor cell markers including Itga6, Ly6a and Tacstd213, as well as Psca14, a prostate cancer stem cell marker, (Spearman r = 0.3, 0.31, 0.4, and 0.41, respectively), was identified in urogenital sinus epithelium (UGE) but not urogenital sinus mesenchyme (UGM) cells (Fig. 1d vs. Supplementary Fig. 2f), suggesting the progenitor properties of Osr1-lineage cells in prostatic epithelial development. Co-expression UMAP plots also showed overlaying Osr1 with the above cellular markers in epithelial cell clusters of E18.5 samples (arrows, Fig. 1e). To assess the differential fates of embryonic prostatic Osr1-expressing cells, we traced them through embryonic and postnatal prostate developmental stages using Osr1-Cre activated membrane-bound green fluorescent protein (mGFP) expression in Rosa26 mTmG-LoxP/+:Osr1Cre/+ (R26 mTmG/+:Osr1Cre/+) reporter mice (Fig. 1f). The expression of mGFP activated by Osr1-Cre mainly appeared in urogenital epithelium and prostatic buds at E15.5 and 18.5, respectively, and mGFP+ cells were expanded robustly in the epithelium of prostatic glands at P14, 35, and 56, through and after puberty (Fig. 1g). Co-immunofluorescence analyses (Co-IF) showed very limited Osr1+mGFP+ cells but many more mGFP+ cells in both E15.5 and E18.5 UGS samples (Fig. 1h, i). Robust mGFP+ cells also appear in postnatal prostate epithelia (Fig. 1g), demonstrating the ability of Osr1-expressing cells to expand prostatic epithelia through pubertal prostate development. Co-IF analysis of P56 prostate tissues showed the majority of mGFP+ cells co-stained with endogenous AR or CK8 (Fig. 1j; Supplementary Fig. 2g1–h4). Intriguingly, a portion of mGFP+ cells also showed positive staining for CK5 and p63, basal epithelial cell markers, in the above prostate sections (blue arrows, Fig. 1j; Supplementary Fig. 2i1–j4). No overlay between mGFP with vimentin (Vim) or smooth muscle actin (SMA) staining was observed (Fig. 1j; Supplementary Fig. 2k1–l4). These data consistently demonstrate the regulatory role of Osr1-lineage cells in prostatic epithelial development and growth. Accordingly, more mGFP+ epithelial cells were observed in prostate tissues, particularly in the anterior and dorsal lobes, isolated from R26 mTmG/+:Osr1Cre/+ mice at both P28 and P56 days than those from R26 mTmG/+:PBCre/+ counterparts (Supplementary Fig. 3af4). Moreover, TP63 + mGFP+ basal epithelial cells were detected in prostate tissues of R26 mTmG/+:Osr1Cre/+ mice but not in those of R26 mTmG/+:PB Cre/+ mice (yellow or blue arrows, Supplementary Fig. 3g1-1’ vs 3g2–2’). Taken together, these data demonstrate that prostatic Osr1-lineage cells possess basal progenitor properties and are able to expand prostatic epithelial cell populations during prostate development.

Fig. 1: Prostatic Osr1-expressing cells possess progenitor properties in expansion of prostatic epithelia through embryonic and pubescent development.
figure 1

a Schematic of the single cell RNA-sequencing experiment performed. b Uniform Manifold Approximation and Projection (UMAP) plot showing the cell type cluster identities based on gene expression patterns. c Three individual UMAP plots of single cells at the indicated time points after separation from the original clustering. Gene expression UMAP plots displaying the expression pattern of the Osr1 gene at the indicated time points. Color intensity indicates the scaled expression level in each cell. d Heatmap of pairwise Spearman correlation between the indicated gene expression in epithelial cells from E18.5 male urogenital sinus (UGS). Colors reflect the level of correlation (positive correlation in red and negative correlation in blue); numbers show the correlation coefficient. The blue box highlights the correlation between Osr1 and other progenitor cell markers. e Blended expression UMAP plots displaying the expression of Osr1 and the indicated genes in the E18.5 male UGS. Blue arrows show the overlay (yellow color) of Osr1 and the indicated genes in the E18.5 male UGS. f Schematics of the R26mTmG/+ and Osr1Cre/+ alleles, shown in relation to the mating strategy for this experiment. g Representative fluorescence images of mouse UGS or prostate tissues at the indicated timepoints expressing mGFP reporter fluorescence (green) controlled by Cre recombination. Scale bars, 100 µm; 200 µm; 25 µm. h–j Representative fluorescence images of co-immunofluoresence (IF) staining for the indicated proteins using mouse UGS or prostate tissues at the indicated timepoints. Blue arrows indicate the overlay between mGFP and cytokeratin 5 (CK5) or p63, basal epithelial cell markers j. AR, androgen receptor; CK8, cytokeratin 8; Vim, vimentin; SMA, smooth muscle actin. Scale bars, 25 µm. Representative images from three independent experiments with similar results are displayed for each micrograph. See also Supplementary Figs. 13.

Aberrant expression of hARtg regulates transcriptome of prostatic Osr1-lineage cells and induces PIN and prostatic adenocarcinoma development

Development of HGPIN and PCa lesions in Osr1-Cre-driven hARtg mice suggests the promotional role of hARtg in Osr1-lineage cells in prostate tumorigenesis10. To better understand the oncogenic role of hARtg in Osr1-lineage cells, we generated R26 mTmG-LoxP/hAR-LoxP:Osr1Cre/+ (R26mTmG/hAR:Osr1Cre/+) mice, in which the expression of mGFP and transgenic AR co-occurs through Osr1Cre mediated activation (Fig. 2a). Both HGPIN and prostatic PCa lesions developed in R26mTmG/hAR:Osr1Cre/+ mice starting at 4 and 10 months, respectively (Supplementary Table 1) but not in control littermates, which is similar as observed in R26hAR/+:Osr1Cre/+ mice10. Prostate tissues isolated from 6- and 12-month-old R26mTmG/hAR:Osr1Cre/+ mice showed typical HGPIN and PCa lesions (Fig. 2b, c), respectively, which were used for preparing scRNA-seq analyses (see below). A uniform nuclear staining for hARtg appeared in both atypical and tumor cells within PIN and PCa lesions (Fig. 2d, e). Co-staining of hARtg and mGFP were also revealed in both atypical and tumor cells (Fig. 2f), demonstrating their origin deriving from Osr1-expressing cells. Interestingly, while the majority of atypical and tumor cells showed positive staining for both hARtg and CK8, a portion of atypical cells showed a clear overlay of hARtg and CK5 expression within PIN lesions (Fig. 2f). The expression of both hARtg and mGFP in atypical and tumor cells within HGPIN and PCa lesions demonstrate the critical role of transgenic AR in Osr1-lineage cells in inducing prostatic oncogenesis and promoting PIN and PCa development.

Fig. 2: Single cell transcriptomic analyses of prostatic intraepithelial neoplasia and adenocarcinoma tissues originating from Osr1-expressing cells with hARtg expression.
figure 2

a Schematic of the floxed human androgen receptor (hAR) transgene (R26hAR/+) and R26mTmG/+ alleles, as well as the corresponding recombined Osr1Cre/+ alleles, shown in relation to the mating strategy for this experiment. b–f Representative images of hematoxylin-eosin (H&E), immunohistochemistry (IHC), and co-IF staining using the indicated antibodies on adjacent prostate tissue sections from R26mTmG/hAR:Osr1Cre/+ mice. Scale bars, 100 µm; 25 µm. g Two individual UMAP plots of single cells isolated from prostatic intraepithelial neoplasia (PIN; gray) and prostate adenocarcinoma (PCa; dark blue) tissues, respectively. h UMAP visualization of the single cells following integration and clustering of the PIN (gray) and PCa (dark blue) cells. i Cell type cluster identities shown on the UMAP of the single cells based on their gene expression profiles. BE, basal epithelium; LE, luminal epithelium; Fib, fibroblasts; SM, smooth muscle cells; Endo, vascular endothelium; Leu, leukocytes. j UMAP plot from i is split by PIN or PCa cells colored by cell type. k UMAP plot of epithelial cells from PIN and PCa tissues. Epithelial cells were sub-clustered, re-clustered and colored by cell cluster. UrLE, urethral epithelium; OE, other epithelium. l UMAP plot from k is separated by PIN or PCa samples colored by cell type. m Dot plot of Ar, human AR transgene (hARtg), mGFP, as well as five cluster-specific genes for each epithelial cell cluster. Dot size indicates the percentage of cells in a cluster expressing each gene; color shows expression level. n Gene expression UMAP plots for the indicated genes after separating PIN (top) and PCa (bottom) samples. Color intensity indicates the scaled expression level. o Bar chart showing the cell counts within individual clusters from epithelial cells of PIN (gray) or PCa (black) tissues. Source data are provided as a Source Data file. p Table summarizing the characteristics for each cell cluster from PIN or PCa samples. Representative images with consistent results from three replicates are shown. See also Supplementary Fig. 4.

To gain in-depth mechanistic insight into the oncogenic role of hARtg, we performed scRNA-seq analyses using the above pathologically confirmed HGPIN and PCa tissues (see Methods). Both scRNA-seq samples prepared from PIN and PCa tissues underwent multiple steps of filtering following sequencing and alignment to the mm10 reference genome with addition of mGFP and hARtg sequences15. Post-filtering, 6332 or 9412 cells from PIN or PCa samples, with an average of 3751 or 3532 genes and 21,297 or 19,899 UMI counts per cell, respectively, were used in the study (Supplementary Fig. 4a, b). Both PIN and PCa samples were initially visualized individually using UMAP (Fig. 2g), and then merged and clustered, aligning similar cell subsets based on their transcriptomic profiles using Seurat’s integrated method11 (Fig. 2h). A total of 6 cell subsets were identified in both merged and individual PIN and PCa plots (Fig. 2i, j), demonstrating comparable cellular properties between these two samples. Specifically, the expression of hARtg and mGFP are comparable and restricted within the luminal and basal epithelial cell subsets in combined UMAP plots (Supplementary Fig. 4c). Those epithelial cell subsets were further validated with representative epithelial markers (Supplementary Fig. 4d). To gain higher resolution of hARtg + cells in PIN and PCa samples, epithelial cells were separated from other non-epithelial cells, and re-clustered following cell cycle regression (Supplementary Fig. 4e)16. Eleven cell clusters were yielded, including two basal, seven luminal, a urethral epithelial (UrLE)17, and another epithelial cluster, deemed other epithelia (OE) (Fig. 2k). Five highly expressed genes in each cell cluster were identified to represent their cellular properties (Fig. 2m). Prostatic basal epithelial cells showed high expression of Krt5, Krt14, and Trp63, while luminal cell clusters displayed elevated expression of Krt18, Krt19, and Pbsn, respectively18. The expression of hARtg and mGFP appears mainly in BE2 and LE1-3 clusters, and the expression of endogenous Ar is restricted within LE4-7 clusters. All cell clusters appeared in both individual PIN and PCa UMAP plots. However, luminal epithelial cell clusters, LE1 to 3, appeared predominantly in PCa samples, and the rest of the cell clusters showed abundantly in PIN samples (Fig. 2l). UMAP expression plots showed Krt5 and Krt8 expression selectively in basal and luminal epithelial cell clusters of both PIN and PCa samples while hARtg and mGFP expression was mainly revealed within the BE2 cluster in PIN, and LE1-3 clusters in PCa samples (Fig. 2n). Cell distribution analyses further confirmed luminal clusters (LE1-3) being mostly in PCa samples and other cell clusters being abundant in PIN samples (Fig. 2o). LE1-2 clusters showed relatively high expression of hARtg and mGFP but little or no expression of mouse Ar and Pbsn (Fig. 2p), suggesting their transformed and un-differentiated cellular properties. These data provide high-resolution insight into the cellular properties of prostatic hARtg+ atypical and tumor cells derived from Osr1-lineage, implicating the direct role of transgenic AR activation in PIN and prostate tumor development in R26mTmG/hAR:Osr1Cre/+ mice.

Transgenic AR expression elevates IGF1 signaling in atypical Osr1-lineage basal epithelial cells within PIN lesions

Identifying specific expression of hARtg in a subpopulation of atypical basal epithelial cells within BE2 cluster in R26mTmG/hAR:Osr1Cre/+ samples (Fig. 2m, n, p) suggests the important role of these hARtg+ basal epithelial cells in PIN initiation (Fig. 3a). Additionally, robust mGFP+TP63+ prostatic basal epithelial cells revealed in prostatic tissues of R26mTmG/+:Osr1Cre/+ mice but not in those of R26mTmG/+:PB Cre/+ counterparts (Supplementary Fig. 3g1, 2’). Moreover, only R26mTmG/hAR:Osr1Cre/+ mice developed HGPIN and PCa lesions but not in AR transgenic mice driven by PB promoters8,9,10,19. These data consistently demonstrate the significance of the hARtg+ atypical basal cells in initiating prostate oncogenesis. To gain direct insight into the regulatory role of hARtg in basal epithelial cells, we identified the differentially expressed genes (DEGs) between hARtg+ and hARtg−basal epithelial cells of PIN and PCa samples using a Wilcoxon Rank Sum test, which showed more than 5% of cells with adjusted p values < 0.05 by the Benjamini-Hochberg procedure and average log fold change >0.1 (Fig. 3a, b; Supplementary Data 1). Ingenuity pathway analysis (IPA) of the above DEGs identified significant enrichment in IGF1, Insulin secretion, Wnt/β-catenin, and JAK/STAT signaling pathways (Fig. 3c), which directly regulate prostate tumorigenesis20. Specifically, identifying enriched IGF1 and Insulin related signaling pathways in hARtg+ basal cells is intriguing. Multiple lines of evidence have shown that IGF1 signaling directly induces and promotes PIN development in both mouse prostate cancer models and human prostate cancers21,22,23. A significant increase in hARtg and Igf1r, as well as IGF1 signaling downstream genes, Jak2 and Mapk13 expression was further identified in hARtg+ compared to hARtg- basal epithelial cells using box plots (Fig. 3d). A correlation between hARtg with Igf1r or its downstream targets, Jak2 and Mapk13, was further identified using Spearman gene-gene correlation analysis (Fig. 3e). Quantitative reverse transcription-PCR (qRT-PCR) analyses showed the higher expression of IGF1 signaling effectors, including Igf1r, Jak2, and Mapk13, in PIN tissue samples of R26mTmG/hAR:Osr1Cre/+ mice than prostate tissues of R26mTmG/+:Osr1Cre/+ controls (Fig. 3f). Using triple-IF analyses, the regulatory role of transgenic AR in activating IGF1R axis was further assessed in PIN tissues of R26mTmG/hAR:Osr1Cre/+ mice. Whereas CK14 expression revealed in basal epithelial cells in both abnormal PIN prostatic glands and adjacent normal glands (Fig. 3g), co-expression of mGFP, representing hARtg-expressing cells, and IGF1R appeared mainly in atypical cells within PIN areas (Fig. 3g). Overlay of CK14 and mGFP, mGFP and IGF1R, or CK14 and IGF1R appeared selectively in atypical cells within the above PIN lesions (pink boxes and arrows, Fig. 3g) but no or very few cells with the above overlays were observed within normal glandular areas (blue boxes and arrows, Fig. 3g). Triple positive cells of CK14, mGFP, and IGF1R revealed only in atypical cells within abnormal prostatic glands (pink boxes pink arrows, Fig. 3g, right panels), but not in normal glandular cells (blue boxes and arrows, Fig. 3g, right panels). Quantified IGF1R expression in approximately a total of 500 to 800 CK14 + mGFP+ or CK14 + mGFP− cells from five different areas in each sample showed significantly more IGF1R expression in CK14 + mGFP+ cells than CK14 + mGFP− cells in three different experiments with three tissue samples prepared from three different R26mTmG/hAR:Osr1Cre/+ mice (p = 0.00001) (Fig. 3h).

Fig. 3: Transgenic AR expression elevates IGF1 signaling in atypical Osr1-driven prostatic basal epithelial cells within PIN lesions.
figure 3

a UMAP plot indicating basal epithelial (BE) cells from PIN and PCa tissues of R26mTmG/hAR:Osr1Cre/+ mice (left), colored by hARtg expression (right). b Heatmap showing top 50 differentially expressed genes (DEGs) between hARtg + and hARtg– BE cells. c Ingenuity pathway analysis (IPA) pathway analysis of DEGs comparing hARtg + and hARtg– BE cells. Fisher’s exact test (two-sided). Ratio denotes the number of DEGs compared with the total number of genes associated with the canonical pathway. d Box-plots representing scaled expression for the indicated genes between hARtg + (n = 749) and hARtg– (n = 933) BE cells. Pink lines mark the median; top and bottom lines of boxes indicate the boundaries of the first and third quartiles, respectively; the top and bottom whiskers show the maximum and minimum values, respectively, excluding outliers. Wilcoxon rank-sum test (two-sided) followed by Benjamini-Hochberg correction. e Heatmap of pairwise Spearman correlation between indicated gene expression in BE cells. Numbers indicate correlation coefficient. f RT-qPCR analysis of the indicated genes shown as fold change in the indicated tissues. g Representative images of H&E and triple-IF staining for the indicated antibodies on adjacent prostate tissues from R26mTmG/hAR:Osr1Cre/+ mice. Pink and blue boxes indicate abnormal and normal glandular areas, respectively, and pink and blue arrows indicate atypical and normal cells, respectively. Nuclei were stained with DAPI. Scale bars, 200 µm; 50 µm; 12.5 µm. h Bar chart for the percentage of IGF1R + cells in mGFP− (black) or mGFP + (gray) CK14-expressing BE cells. i Diagram showing the mouse Igf1r locus containing two AR-binding sites. j hAR ChIP-qPCR analysis shown as percent input of Igf1r gene on its binding sites A and B in the indicated tissues. In f, h, and j, data are represented as mean ± SD of three biological replicates. Two-sided t-test, **p < 0.01. Representative images with consistent results from three replicates are shown. See also Supplementary Fig. 5 and Supplementary Data 1, 2. Source data and the exact p-values are provided in the Source Data file.

Identifying aberrant activation of IGF1 signaling pathways specifically in hARtg+ atypical basal cells within PIN lesions implicates a regulatory mechanism for hARtg in initiating prostatic oncogenesis through the activation of IGF1R axes. Given the AR functioning as a ligand-dependent transcriptional factor, we then performed chromatin immunoprecipitation sequencing (ChIPseq) analysis to examine if hARtg regulates IGF1R transcription. A specific enrichment of transgenic AR was identified within the Igf1r promoter region on mouse chromosome 7 in the human AR antibody immunoprecipitated samples prepared from PIN tissues of R26mTmG/hAR:Osr1Cre/+ mice (Supplementary Fig. 5; Supplementary Data 2). ChIP-quantitative polymerase chain reaction (ChIP-qPCR) analyses further showed the specific occupancy of the transgenic AR in both the promoter and enhancer regions of the Igf1r gene with the androgen response element sequences24, but not in the Untr4 locus, used as a negative control, in the same immunoprecipitated PIN samples (Fig. 3i, j). Using prostatic organoids derived from R26mTmG/hAR:Osr1Cre/+ mice, we also identified the higher expression of Igf1r transcripts in samples cultured with dihydrotestosterone (DHT) than those without DHT or treated with a combination of DHT and Enzalutamide (Enz), an antiandrogen (Supplementary Fig. 5c). Taken together, these lines of experimental evidence demonstrate a direct role of hARtg in regulating IGF1R expression, implicating an underlying mechanism by which aberrant AR induces PIN development via activating IGF1R signaling.

Activating Wnt signaling in hARtg positive prostatic tumor cells

Our finding of hARtg to activate IGF1R axes in atypical basal cells explores a mechanism for AR’s oncogenic role in prostate oncogenesis. Prostatic basal epithelial cells have been shown to possess the ability to initiate oncogenic transformation and to further transdifferentiate to luminal tumor cells in the presence of androgens25,26,27. To assess the underlying mechanisms for hARtg+ atypical cells to progress to prostate tumors, we analyzed the transcriptomic changes between LE1-2 tumor cell clusters possessing enriched hARtg+ tumor cells and normal LE5-7 clusters containing hARtg– cells (Fig. 4a, b). 1834 up-regulated and 3156 down-regulated DEGs were identified (Fig. 4b; Supplementary Data 3). A significant increase in the expression of Wnt/β-catenin target genes, such as Tcf4, Ccnd1, Axin2 and Lgr5, revealed in LE1-2 in comparison to LE5-7 clusters using either box plots (Fig. 4c) or UMAP expression plots (Fig. 4d). Strong correlations between hARtg and these Wnt downstream target genes were further confirmed in LE1-2 and LE5-7 cell clusters (Fig. 4e). GSEA using pre-ranked gene lists of LE1-2 versus LE5-7 further showed a significant enrichment in both Wnt and Myc target signaling pathways (Fig. 4f). Increased expression of Wnt/β-catenin downstream target genes, including Tcf4, c-Myc, Ccnd1, Axin2, and Lgr5, was also identified in RNA samples isolated from pathologically confirmed PCa samples in comparison to PIN and prostate tissues from age and sex-matched wild type (WT) mice using qRT-PCR (Fig. 4g). These data consistently demonstrate the activation of Wnt/β-catenin signaling pathways in hARtg+ prostate tumor cells of R26mTmG/hAR:Osr1Cre/+ mice.

Fig. 4: Activating Wnt signaling pathways in hARtg positive prostatic tumor cells.
figure 4

a UMAP plot highlighting LE1-2 and LE5-7 clusters identified as hARtg + tumor cell clusters and hARtg-normal cell clusters, respectively. b Heatmap showing top 50 differentially expressed genes between LE1-2 (n = 5122) and LE5-7 (n = 2743) cell clusters. Yellow and purple indicate high and low expression, respectively. c Box plots representing scaled expression data for hARtg and Wnt downstream target genes applied to LE1-2 and LE5-7 cells. Pink lines mark the median; top and bottom lines of boxes indicate the boundaries of the first and third quartiles, respectively; the top and bottom whiskers show the maximum and minimum values, respectively, excluding outliers. P-values were computed using Wilcoxon rank-sum test (two-sided) and adjusted using Benjamini-Hochberg correction. d Gene expression UMAP plots displaying the expression patterns for the indicated Wnt downstream genes. Color intensity indicates the scaled expression level in each cell. e Heatmap of pairwise Spearman correlation between the indicated gene expression in LE1-2 and LE5-7 cells. Colors reflect the level of correlation (positive correlation in red and negative correlation in blue); numbers show the correlation coefficient. The blue box highlights the correlation between hARtg and other Wnt downstream genes. f GSEA enrichment plots highlighting the positive enrichment of Wnt signaling pathway with multiple gene sets comparing LE1-2 to LE5-7 cells. g RT-qPCR analysis of the indicated genes shown as fold change in normal prostate tissues from R26mTmG/+:Osr1Cre/+ control mice and PIN or PCa tissues from R26mTmG/hAR:Osr1Cre/+ mice. Data are represented as mean ± SD of three biological replicates. Two-sided t-test for PIN or PCa versus control, *p < 0.05, **p < 0.01. See also Supplementary Data 3. Source data and the exact p-values are provided in the Source Data file.

Activation of transgenic AR expression in Osr1-lineage cells upregulates Wnt signaling pathways in prostate tumor tissues

To gain more in-depth insight into transcriptomic changes in hARtg regulated prostatic oncogenesis, we performed bulk RNA-sequencing (RNA-seq) analyses using RNA samples isolated from microscopically confirmed PIN or PCa tissues with more than 80% atypical or tumor cells, respectively, from R26mTmG/hAR:Osr1Cre/+ mice, as well as prostate tissues from sex and age-matched WT mice. Using a median fold difference test (see the Methods), we identified 359, 2970 and 1887 DEGs with adjusted p value < 0.05 and fold change ≥ 2 by comparing RNA-seq samples between PIN or PCa versus WT controls, and PCa versus PIN samples, respectively (Fig. 5a; Supplementary Data 46). Crossing analyses from the above datasets identified common DEGs between each of the two or three groups (Fig. 5b). A group of common DEGs (n = 198) overlapping between PIN or PCa versus WT samples was identified (Fig. 5b; Supplementary Data 46). GSEA using pre-ranked DEG lists from PIN or PCa versus WT samples revealed four common significantly enriched signaling pathways, including the mitotic spindle, G2/M checkpoint, IL6-JAK-STAT3 signaling, and estrogen response pathways (red line, Fig. 5c). Much higher enrichment scores (ES) appeared in the samples of PCa versus WT than those of PIN versus WT, suggesting the promotional roles of these signaling pathways during disease progression. Interestingly, the significant enrichment of Wnt/β-catenin and Myc targets V1 signaling pathways only appeared in the gene list of PCa versus WT samples (Fig. 5c). GSEA with a pre-ranked DEG list by comparing PCa to WT samples further showed the significant enrichment (FDR < 0.25) in both Wnt/β-catenin signaling and Myc targets (Fig. 5d). IHC analyses using adjacent PCa tissue sections revealed positive nuclear staining for transgenic AR and β-catenin, as well as positive staining for phosphorylated IGF1R and for β-catenin downstream targets, including TCF4, c-Myc, Cyclin D1, AXIN2, and Lgr5 (Fig. 5f–m). In contrast, no specific staining was detected in normal prostate tissues isolated from age- and sex-matched R26mTmG/+:Osr1Cre/+ controls (Fig. 5o–v). These data consistently demonstrate aberrant activation of Wnt/β-catenin signaling pathways in prostate tumor cells of R26mTmG/hAR:Osr1Cre/+ mice, implicating the regulatory role of hARtg in activating Wnt/β-catenin to promote prostate tumor development and growth.

Fig. 5: Activation of transgenic AR expression in Osr1-expressing cells upregulates Wnt signaling pathways in prostate tumor tissues.
figure 5

a Heatmap showing the expression patterns of differentially expressed genes from the three comparisons of bulk RNAseq data from normal prostate tissues from R26mTmG/+:Osr1Cre/+ wild type (WT) mice and PIN or PCa tissues from R26mTmG/hAR:Osr1Cre/+ mice as indicated in the figure. Red and blue colors indicate up- and down-regulation, respectively. b Venn diagram depicting the number of differentially expressed genes from the different comparisons as labeled. c Radar chart of the enrichment score displaying differential enrichment of R26mTmG/hAR:Osr1Cre/+ (PIN or PCa) in comparison with wild type (WT) control samples. Enrichment score and false discovery rate (FDR) were calculated using a hypergeometric test from GSEA with DEGs (|Log2 fold-change| >1 and adjusted p value < 0.05). P values were adjusted for multiple testing with Benjamini-Hochberg correction. Red and blue lines indicate enriched hallmark gene sets in PIN and PCa tissues, respectively. d GSEA enrichment plots of pre-ranked gene list from differentially expressed genes comparing PCa to wild type samples, highlighting positive enrichment of Wnt signaling pathway with multiple gene sets. e-v Representative images of H&E and IHC staining using the indicated antibodies on adjacent PCa tissue sections from R26mTmG/hAR:Osr1Cre/+ mice and wild type mice. Scale bars, 100 µm; 25 µm. Images representative of consistent results from three independent experiments are shown. See also Supplementary Data 46.

Activation of human AR transgene expression initiates PIN formation and promotes tumor development through aberrant elevation of IGF1 and Wnt/β-catenin signaling pathways

It has been shown that aberrant activation of IGF1 signaling can either directly or via activating AKT and GSK3β axes stabilize cellular β-catenin to augment its activity for prostate tumor growth28,29,30,31. We then assessed the effect of IGF1 signaling in activating Wnt/β-catenin axes in hARtg + atypical basal cells. Co-IF analyses identified positive staining for IGF1R or phosphorylated IGF1R overlaid with transgenic AR staining in atypical HGPIN cells (Fig. 6a). Positive staining for phosphorylated forms of AKT, GSK3β, and ERK1/2 were also detected and overlaid with transgenic AR in those HGPIN cells on adjacent tissue sections (Fig. 6a). Positive staining for IGF1R, phosphorylated IGF1R, AKT, GSK3β, and ERK1/2 also overlaid with CK14 staining, demonstrating the basal cell properties of those double positive atypical cells (Fig. 6a). In contrast, there is no or very weak staining with transgenic AR, IGF1R, and phosphorylated IGF1R, AKT, GSK3β, and ERK1/2 in prostate tissues of age- and sex-matched R26mTmG/+:Osr1Cre/+ controls (Fig. 6b). These data suggest up-regulated IGF1R expression and activation of IGF1R and its downstream target, AKT, leading to the phosphorylation of GSK3β, and ERK1/2 in atypical basal cells of HGPIN lesions. Using co-IF approaches, we further assessed the activation of IGF1R and Wnt/β-catenin axes in hARtg+ tumor cells. Co-staining for phosphorylated IGF1R and GSK3β, as well as nuclear β-catenin and its downstream targets, TCF4 and Cyclin D1, with transgenic AR staining appeared specifically in prostate tumor cells (Fig. 6c) in comparison to normal prostatic epithelial cells of WT mouse prostate tissues (Fig. 6d), suggesting the regulatory loop of hARtg expression in elevating IGF1 signaling and activating Wnt/β-catenin axis during PIN and tumor development in R26mTmG/hAR:Osr1Cre/+ mice. These lines of scientific evidence are consistent with previous in vitro studies and provide more relevant data to demonstrate the co-regulation of androgen, IGF1, and Wnt/β-catenin axes in prostate tumor development and growth28,29,31.

Fig. 6: Activation of hAR transgene expression initiates PIN formation and tumor development through aberrant elevation of IGF1 and Wnt/ß-catenin signaling pathways.
figure 6

Representative images of H&E and co-IF staining for hAR or CK14, a basal cell marker, and IGF1 signaling downstream mediators, including IGF1R, phosphorylated-IGF1R (p-IGF1R), p-AKT, p-GSK3ß, and p-ERK1/2 in PIN tissue sections from R26mTmG/hAR:Osr1Cre/+ mice at 6 months age (a) and normal prostate tissue sections from age-matched R26mTmG/+:Osr1Cre/+ wild type mice (b). Scale bars, 100 µm; 50 µm; 25 µm. Representative images of H&E and co-IF staining for hAR and IGF1 downstream (p-IGF1R, p- GSK3ß) or Wnt downstream (ß-catenin, TCF4, and Cyclin D1) in PCa tissue sections from R26mTmG/hAR:Osr1Cre/+ mice at 12-month age (c) and normal prostate tissue sections from age-matched R26mTmG/+:Osr1Cre/+ wild type mice (d). Scale bars, 12.5 µm. Representative images with consistent results from three independent experiments are shown.

Aberrant transgenic AR induces Wnt/β-catenin signaling activation to promote prostate cancer development

To gain dynamic and deep insight into the oncogenic role of hARtg in prostate tumorigenesis, we conducted single-cell trajectory analyses to unveil transcriptomic changes that were specifically induced by hARtg expression and governed cell fate decisions through PIN initiation and progression to prostatic adenocarcinomas32. Using Monocle232, we observed that the pseudotime trajectory plots of hARtg + epithelial cells from merged PIN and PCa samples displayed a starting point composed mainly of BE2, a PIN cell cluster, that differentiated to the cell trend with luminal cell properties, and then further progressed to two main luminal cell branches mainly possessing LE1-2 tumor cells (Fig. 7a). In contrast, hARtg– epithelial cells showed a different trajectory fate, which started with BE1 cells and differentiated into a single luminal cell branch containing normal LE5-7 cells (Fig. 7b). The cellular properties of different cell branches were further assessed using trajectory expression plots with different cellular markers. The expression of hARtg in mGFP+ Osr1-lineage cells was only observed in hARtg+ cell samples (Fig. 7c). Whereas the expression of Krt14 mainly appeared in the start of the branches, composing BE2 or BE1 cells, in hARtg+ or hARtg− cell plots, respectively, Krt8 expression showed at later cell branches in pseudotime in both hARtg+ and hARtg−samples, representing their luminal cell properties (Fig. 7c, d). Intense expression of endogenous Ar and Pbsn appeared at the later trend of the luminal cell branch in hARtg– cells (Fig. 7d). In contrast, in hARtg+ cell plots, both endogenous Ar and Pbsn expression only appeared earlier in pseudotime and decreased as cells progressed toward the two luminal tumor cell branches, indicating their undifferentiated and transformed cellular properties (Fig. 7c). Importantly, the expression of Wnt/β-catenin target genes, including Tcf4, Ccnd1, Axin2, and Lgr5, was observed in later luminal cell branches of hARtg+ cells through cell trajectory plots in comparison to those of hARtg– cells (Fig. 7c vs d), implicating the activation of Wnt/β-catenin signaling during the course of tumor development. Using pseudotemporal kinetic analysis, we further demonstrated increased expression of Tcf4, Ccnd1, Axin2, and Lgr5 in the later trend of tumor cells (fate 2) in comparison to the middle trend of cells (fate 1) in hARtg+ cell trajectory plots (Fig. 7e), showing the activation of Wnt/β-catenin signaling correlating with the course of tumor development. Taken together, the above analyses provide a dynamic and high-resolution image for transgenic AR induced Wnt signaling activation through prostate cancer development.

Fig. 7: Aberrant IGF1 and Wnt/β-catenin signaling pathways regulate prostate cancer development.
figure 7

Pseudotime trajectory plots displaying a predicted directional path of hARtg + (a) or hARtg– epithelial cells (b) colored by cluster identity (left) and by pseudotime (right). Expression of the indicated genes projected onto the pseudotime trajectory in hARtg + (c) and hARtg– epithelial cells (d). Red and purple color intensities indicate the scaled expression levels of cellular markers, including hARtg, mGFP, Krt14, Krt8, endogenous Ar (mAr), Pbsn, and Wnt downstream target genes, including Tcf4, Axin2, Ccnd1, and Lgr5, respectively, in each cell. e Linear pseudotime expression plots for indicated Wnt downstream target genes in hARtg + epithelial cells. Lines on each plot correspond to the path of tumor progression moving from the start point (left) to indicated branch tips (right) as in a (right). f Oncoprint outlining genetic alterations and expression of the indicated genes in 488 human primary prostate cancer samples and 444 metastatic castration-resistant prostate cancer samples. Datasets of Prostate Adenocarcinoma (TCGA, PanCancer Atlas) and Metastatic Prostate Adenocarcinoma (SU2C/PCF Dream Team) for primary and advanced prostate cancer, respectively, were extracted from cBioPortal. Colors show genetic alteration as indicated in legend. See also Methods section. g Mutual exclusivity panel analysis depicting the co-occurrence of alterations of the indicated genes in human primary PCa and metastatic castration-resistant prostate cancer. P values were computed using Fisher’s exact test (one-sided) and adjusted using Benjamini-Hochberg correction. h Scatter plots showing the Spearman correlation of co-expression between AR and IGF1R, CTNNB1, or MYC. r = spearman’s correlation coefficient, p = p value. Two-sided t-test. See also Supplementary Fig. 8.

Correlative activation of androgen, IGF1, and β-catenin signaling occurs in human prostate cancer samples

Next, we assessed aberrant AR, IGF1, and Wnt/β-catenin signaling pathways in both human primary PCa samples from the TCGA Pan-Cancer Atlas and advanced PCa samples in cBioPortal33,34. Aberrant activation of AR signaling through AR gene amplification or increasing its expression was detected in 5% of 488 primary PCa samples. Within these samples, about 42% and 55% of them also co-existed with aberrant alterations in IGF1R and Wnt/β-catenin downstream target genes, MYC and CCND1, respectively (Fig. 7f, top panel). Intriguingly, similar alterations in AR amplification and increasing expression were identified in 61% of 444 total advanced PCa samples. Increased co-occurrence of altered IGF1R and gain-of-function alterations and mutations in Wnt/β-catenin pathways also appeared (Fig. 7f, bottom panel). Odds ratio testing showed significant co-occurrences of alterations of AR and IGF1R, MYC, or CCND1 genes and their transcripts in both primary and advanced PCa samples (Fig. 7g). Specifically, significantly positive correlations were observed between the transcripts of AR with IGF1R, CTNNB1, or MYC in primary PCa samples (Fig. 7h). Enrichment of AR on the IGF1R gene locus was also observed in AR ChIP-seq datasets of human PCa samples in comparison to controls with normal prostate tissues (Supplementary Fig. 6), further supporting our mouse ChIP-seq analyses (Supplementary Fig. 5). These lines of scientific evidence provide the importance and clinical relevance of aberrant AR activation in altering IGF1 and Wnt/β-catenin signaling pathways in human prostate tumorigenesis.

Aberrant activation of Wnt/β-catenin signaling pathways enhances AR-mediated prostate tumor growth

Multiple lines of evidence have shown a promotional role of Wnt/β-catenin in PCa growth and progression35,36. Specifically, an interaction between the AR and β-catenin has been identified in PCa cells, directly augmenting tumor cell growth37,38,39. Using organoid cultures derived from prostatic tumor cells of R26mTmG/hAR:Osr1Cre/+ mice, we directly assessed the role of Wnt/β-catenin in hARtg+ tumor cell growth. The developed prostatic organoids were treated with the anti-androgen, enzalutamide (Enz), and Wnt inhibitors, ICG-00140 and iCRT341 alone or in combination (Fig. 8a). It has been shown that iCRT3 can disrupt the interaction between AR and/β-catenin and inhibit AR-mediated transcription and cell growth in PCa cells41. Measuring average sizes of individual organoids and the organoid forming efficiency showed significant reduction in samples treated with Enz, ICG-001, and iCRT3 alone or in combination in comparison with vehicle-treated controls (Fig. 8b, c). Accordingly, more and larger organoids developed in vehicle-treated samples than in those treated with different inhibitors in the brightfield images (Fig. 8d). Specifically, far fewer and smaller organoids revealed in samples treated with Enz combined with Wnt inhibitors, ICG-001 or iCRT3. Histological analyses recapitulated similar prostatic adenocarcinoma lesions in vehicle-treated organoids as observed in tumor tissues of R26mTmG/hAR:Osr1Cre/+ mice (Fig. 8d). Abnormal glandular structures resembling PIN lesions represented in samples treated with Enz, or Wnt inhibitors alone. Appearances of minor pathological changes to normal glandular structures were observed in samples treated with both Enz and Wnt inhibitors. IHC showed CK5 and CK8 staining in organoid cells, indicting their epithelial properties (Supplementary Fig. 7a1–b6). Positive nuclear staining for transgenic AR appeared in organoids treated with vehicle, ICG-001, and iCRT, whereas samples treated with Enz alone and in combination with Wnt inhibitors showed both nuclear and cytoplasmic staining for transgenic AR (Fig. 8d). Positive staining for c-Myc and Cyclin D1, the downstream targets of β-catenin, appeared in vehicle-treated samples and slightly in Enz-treated samples, but not in those treated with Wnt inhibitors alone or in combination with Enz (Fig. 8d). Measuring Ki67 positive cells in the organoid samples revealed the inhibitory effects with Enz and Wnt inhibitors. Among them, the most robust inhibition appeared in samples treated with Enz and iCRT3 (Fig. 8e). Taken together, these data demonstrate the inhibitory effects of both antiandrogen and Wnt inhibitors on the growth of hARtg+ tumor organoids.

Fig. 8: Co-inhibition of androgen and Wnt/β-catenin signaling pathways in prostate tumor growth.
figure 8

a Schematic representation of the experimental design for the ex vivo organoid culture performed. Organoids derived from PCa cells of R26mTmG/hAR:Osr1Cre/+ mice were treated with vehicle, antiandrogen (10 µM Enzalutamide; Enz), Wnt inhibitor (10 µM ICG-001 or 10 µM iCRT3), or combination of Enz and ICG-001 or iCRT3 two times for 6 days. See also Methods section. b Quantification of individual organoid size. Organoids per treatment group (n = 50) examined over three independent experiments. The center blue bar indicates the median value in each group. c Quantification of organoid forming efficiency showing the percentage of organoids above 50 μm diameter per total cells seeded at day 0 in a well. The center line represents the median value; the box borders represent the lower and upper quartiles (25% and 75% percentiles, respectively); the ends of the bottom and top whiskers represent the minimum and maximum values, respectively. d Representative images of brightfield, H&E and IHC staining for the indicated antibodies of the organoids with the indicated treatments. Scale bars, 400 µm; 50 µm; 25 µm. e Quantification of Ki67 positive cells per total cells in the organoids with the indicated treatments. f Weights of xenografts (n = 6) from groups treated as indicated. Data are represented as mean ± SD (n = 6 replicates per data point). Student’s t-test, **p < 0.01. g Schematic representation of the experimental design using in vivo kidney capsule transplantation. See also Methods section. h Histological and IHC analysis of graft tissues with the indicated treatment. Representative images of H&E and IHC images for indicated antibodies on adjacent sections from graft tissues with the indicated treatments. Scale bars, 100 µm; 25 µm. In c, e, and f data are represented as mean ± SD of six independent samples over three biological replicates. Two-sided t-test, *p < 0.05, **p < 0.01. Representative images with consistent results from three independent experiments are shown. Source data and the exact p-values are provided in the Source Data file.

Using in vivo tissue grafting assays, we further examined co-inhibition of AR and Wnt signaling in hARtg+ tumor growth. Specifically, we tested iCRT341, a Wnt inhibitor that also showed inhibition of AR signaling in PCa cells, alone or with Enz. Prostate tumor cells isolated from R26mTmG/hAR:Osr1Cre/+ mice were implanted under the kidney capsule of SCID mice (Fig. 8g). Four weeks after implantation, SCID mice were administered Enz, iCRT3, or both, as well as vehicle, and analyzed four weeks post treatment. Although grafts treated with Enz or iCRT3 alone appeared significantly smaller in size and less in weight than vehicle-treated group, samples treated with both Enz and iCRT3 showed the lowest weights and smallest size among the groups (Fig. 8f). Histologically, vehicle-treated samples retained abnormal pathological characteristics of prostate adenocarcinomas as observed in prostate tumor tissues of R26mTmG/hAR:Osr1Cre/+ mice whereas Enz or iCRT3-treated samples displayed minor pathological changes. Samples treated with both Enz and iCRT3 showed much less pathological changes than other controls (Fig. 8h). Positive staining for CK5 and CK8 was observed in all grafted samples (Supplementary Fig. 7g1–i4). While robust nuclear staining for hAR was observed in vehicle treated samples, much less staining appeared in the other treated samples (Fig. 8h). Positive nuclear staining for β-catenin revealed in the vehicle and Enz-treated grafts. In contrast, only cellular membrane staining of β-catenin appeared in samples treated with iCRT alone and in combination with Enz (Fig. 8h). Positive staining for c-Myc and Cyclin D1, the downstream targets of β-catenin, also only revealed in vehicle treated samples (Fig. 8h). These data reaffirm that the co-inhibition of AR and Wnt signaling pathways represses the growth of hARtg+ tumor cells.

Leave a Reply