Androgen signaling activity negatively correlates with prostate cancer progression.

(A-B) mRNA expression in normal (green), primary (red) or castration-resistant (CRPC) prostate cancer samples. (A) AR expression. (B) AR target genes expression. (C) Principal component analysis of sample expression defines a trajectory of disease progression with an associated pseudotime score (extracted from Prostate Cancer Atlas). (D) Pseudotime correlation segregates normal, primary, CRPC and neuroendocrine (NEPC) samples. (E) Individual AR expression in the samples along the trajectory. (F) Relative change of expression of AR during progression. X-axis represents Pearson’s correlation coefficient between mRNAs and pseudotime as Y-axis displays the associated significance adjusted for false discovery rate (FDR). (G) Individual AR expression in the samples depending on pseudotime (X-axis) and sample type (color code). (H) CRPC samples are separated in four quartiles (1ST to 4TH QT) depending on their pseudotime. AR expression is then plotted for individual samples. (I-N) Individual expression and relative change of expression of three AR target genes (SPOP, KLK3, NKX3-1) during progression. * p<0.05; ** p<0.01; ****p<0.0001.

Canonical AR Signaling genes are downregulated in CRPC compared with primary cancer.

Based on 173 Normal samples, 708 Primary cancer samples and 484 CRPC samples. In yellow: six of the genes are significantly less expressed in CRPC than in normal samples.

Loss of correlation between AR expression and canonical AR signaling coincides with downregulated expression of SRD5A2.

(A) Individual AR expression (X-axis) versus ALDHIA3 expression (Y-axis) in the samples depending on sample type (color code). (B-D) Same expression separated by sample type. (E) SRD5A2 -coding for the DHT-producing enzyme-mRNA expression in normal (green), primary (red) or castration-resistant (CRPC) prostate cancer samples. (F-G) Individual expression and relative change of expression of SRD5A2 during progression. (H) Primary samples are separated in four quartiles (1ST to 4TH QT) depending on their pseudotime. SRD5A2 expression is then plotted for individual samples. ****p<0.0001.

Treatment-like effect induced by Ecdysone Receptor downregulation

(A-L and R-U) Confocal imaging of accessory glands (delimited by white dashes). White arrowheads indicate epithelial clones. Empty arrowheads indicate extra-epithelial tumors. Yellow arrowheads indicate a tumor cell population specific to EGFRλ EcR-KD genotype. (A-D) FISH detection of EcR mRNA in EGFRλ CTL and EGFRλ EcR-KD genotypes. (E-L) Shared phenotypes between the two genotypes. (M-O) Extra-epithelial tumor characteristics defined by 3D-reconstruction. (P) % of glands bearing extra-epithelial tumors 4, 5 or 6 days after clonal induction for the different genotypes. (Q) Comparison of the maximal extra-epithelial tumors percentage (at 6D) between the two genotypes. (R-U) Cleaved caspase 3 staining in the different tumor cell populations. (V) Quantification of the number of apoptotic extra-epithelial tumors at 5, 6 or 10 days after clonal induction for the different genotypes. Scale bars: 50μm. Number of experiments (N) and of individual flies (n) are given in Table 2. Statistical tests are explained in the Methods section; ns: non-significant; * p<0.05; ** p<0.01; *** p<0.001; ****p<0.0001.

Tumor escape induced by Ecdysone Receptor downregulation.

(A-T) Confocal imaging of accessory glands (delimited by white dashes in some images). White arrowheads indicate epithelial clones. Empty arrowheads indicate extra-epithelial tumors. Yellow arrowheads indicate the new tumor cell population which is specific to EGFRλ EcR-KD genotype. (A-D) Coracle staining indicates epithelial status of the cells. (E-H) Cleaved caspase 3 staining in the different tumor cell populations 10 days after clonal induction for the different genotypes. (I-L) Phospho-Histone 3 staining 6 days after clonal induction for the different genotypes. (M-P) Phospho-Src staining 6 days after clonal induction for the different genotypes. (Q-R) DAPI staining reveals nuclear shapes of cells of the new population (yellow empty arrowheads). (S-T) Multiple axes of growth (double-headed arrows) and F-actin reorganization (phalloidin staining, magenta) in the new tumor cell population. Scale bars: 50μm.

Basal extrusion is modified in case of tumor escape.

(A-T) Confocal imaging of accessory glands. White arrowheads indicate epithelial clones. Empty arrowheads indicate extra-epithelial tumors. Yellow arrowheads indicate the new tumor cell population which is specific to EGFRλ EcR-KD genotype. (A-B, E-F) XYZ imaging. Glands are delimited by white dashes in B and F. (C-D, G-H) XZY orthogonal reconstructions of the glands, showing the lumens and in-depth view of the epithelium. Epithelium pushed into the lumen is delimited by white dashes in G and H, as intrabasal tumor cells lay between this epithelium and the basement membrane (yellow). (I) % of glands bearing intrabasal tumors 6 or 7 days after clonal induction for the different genotypes. (J-L) Comparison of intrabasal vs other tumor characteristics defined by 3D-reconstruction (C: clones; T: extra-epithelial tumors; IB-T: intrabasal tumors). Scale bars: 50μm in XYZ views and 20μm in XZY reconstructions. ns: non-significant; ** p<0.01; *** p<0.001; ****p<0.0001.

Ecdysone-dependent molecular control of tumor escape and basal extrusion.

(A-B) % of glands bearing extra-epithelial (A) or intrabasal tumors (B) 6 days after clonal induction for different genotypes. (C-D’) Confocal imaging of accessory glands. Yellow arrowheads indicate intrabasal tumors. (C, D) XYZ imaging. (C’, D’) Corresponding XZY orthogonal reconstructions of the glands, showing the lumens and in-depth view of the epithelium. Epithelium pushed into the lumen is delimited by white dashes. (E-H) β3 Tubulin staining 6 days after clonal induction for different genotypes. Glands are delimited by white dashes, white arrowheads indicate tumors clones. (I-I’) Confocal imaging of accessory glands. (I) XYZ imaging. (I’) Corresponding XZY orthogonal reconstruction of the glands. (J) % of glands bearing extra-epithelial (left) or intrabasal tumors (right) 6 days after clonal induction for different genotypes. (K) Comparison of proliferation rates defined by 3D-reconstruction (E: epithelial clones; IB-T: intrabasal tumors; IB: intrabasal cells in EGFRλ Tub60D-KD genotype). Scale bars: 50μm in XYZ views and 20μm in XZY reconstructions. ** p<0.01; *** p<0.001; ****p<0.0001.

Numbers of independent experiments (N) and individuals (n) used in the different experiments.

EGFRλ clonal expression induces tumorigenesis in the drosophila accessory gland

White arrowheads: mildly hyperproliferative, hypertrophic cells are well integrated into the epithelium with undisturbed cell-cell interaction (A-F, Coracle staining). Thick empty arrowheads: some of these epithelial tumor cells express higher levels of GFP (A-B), reflecting modified gene expression, and higher levels of epithelial markers. However, these markers do not properly localize at the membrane (C), indicating a loss of cohesion with neighbor cells. Empty arrowheads: some tumor cells are able to undergo epithelial basal extrusion to form extra-epithelial tumors. These cells are devoid of epithelial markers, indicating a change in their fate (F). Scale bars: 50 micrometers.

Schematic representation of the late steps of ecdysone synthesis and canonical Ecdysone Signaling.

Six different genes (underlined) were independently targeted by RNAi to block tumor cell production of ecdysone (Phm, Sad), metabolization of active 20-hydroxyecdysone (Shd), or EcR Signaling (EcR, Hr3, Hr4). Results are shown in Figure 6.

Ecdsyone deprivation does not promote tumorigenesis by itself.

(a-c) In the absence of expression of an oncogene (CTL), clonal induction of NLS-GFP (a) EcR RNAi (b) or Sad RNAi (c) brings no evident phenotype to the clonal cells. (d) It does not promote tumorigenesis. (e-g) In the absence of average clone volume (e) or difference in proliferation rate compared to control (g), the only noticeable phenotype is a slightly decreased cell size in clones expressing EcR or Sad RNAi (f). Chi2 test; **P < 0.01; ***P < 0.001. Representative images in (a-c) from three or more experiments. Scale bars: 20 μm.

Downregulation of canonical Ecdysone Signaling induces intrabasal tumorigenesis.

(a-d) Co-expression of Hr3 RNAi suppresses only partially extra-glandular tumors (empty arrowheads in (a-b)) but leads to the appearance of highly proliferative, intrabasal tumor cells into the glands (yellow arrowheads in (c-d)). (e-h) This new population displays a strong staining for classical aggressiveness markers (yellow arrowheads, pH3 staining in (e-f) and pSrc staining in (g-h)). Representative images from three or more experiments. Scale bars: 50 μM.

Loss of autocrine Ecdysone Signaling induces the same paradoxical response as loss of other members of the same pathway.

(a-d) Co-expression of Phm RNAi suppresses only partially the formation of extra-epithelial tumors (empty arrowheads in (a-b)) but leads to the appearance of highly proliferative intrabasal tumors (yellow arrowheads in (c-d)). (e-l) Co-expression of Sad (RNAi (e-h) or Phm RNAi (i-l) induces early apoptosis of extra-epithelial tumors (at 6D, cleaved-caspase 3 staining), mimicking initial effect of steroid deprivation on human tumors (empty arrowheads). (m) Quantification of this early apoptosis phenomenon appearing at 6D for the coexpression of Sad RNAi. (n) qPCR quantification of whole glands expression of Sad for the given genotypes. “WT”: glands baring the EGFRλ genetic background, but unable to produce clones, i.e. “wild type” glands. Compared to these, oncogene induction (column 2) induces a significant overexpression of Sad mRNA; co-expression of EcR RNAi (column 3) still significantly increases Sad mRNA expression. Compared to this condition, co-expression of Sad RNAi (column 4) significantly decreases Sad mRNA expression, as expected. Representative images in (a-l) from three or more experiments. Scale bars: 50 μM.

Efficient downregulation of Tub60D has little effect on epithelial tumor cells.

(a-d) Co-expression of Tub60D RNAi does not affect the phenotype of extra-epithelial tumors. (e-l) Downregulation of Tub60D efficiently suppresses the accumulation of β3 Tubulin in clonal cells (e-h) and tumor cells (i-l). Remaining staining can be seen in other cell types such as tracheal cells (blue arrowheads in (e-i). (m-r) Characteristics of the different types of tumor cells co-expressing Tub60D RNAi. (m-q) Extraepithelial tumors co-expressing Tub60D RNAi display the same size (m), cell number (n) and proliferation rate (o) as EGFRλ CTL tumors. (pq) Intra-epithelial clones co-expressing Tub60D RNAi are composed of cells of same size (p) and display the same proliferation rate (q) as as EGFRλ CTL intra-epithelial clones. Chi2 test; ns non significant. Representative images in (a-l) from three or more experiments. Scale bars: 50 μM.