Fig 2: Targeting FOSL1 and PIM1 in Club-like cells in vitro.(A, B) Dose-response of JQ-1 (A) and CX-6258 (B) on the number of organoids formed by sorted Ptenpc−/− LSCmed cells. Data are normalized to the DMSO condition (biological replicates, n = 5 (A) and n = 4 (B) independent experiments). ****p < 0.0001 versus DMSO (ANOVA analysis and Dunnett’s post hoc test). Exact p values are reported in Appendix Table S1. (C) Expression of FOSL1 and PIM1 in human HPV-10 cells determined by RT-qPCR. The results are normalized to the values obtained in LNCaP cells, represented by the horizontal dotted line (biological replicates, n = 3 independent experiments). *p < 0.05 versus LNCaP cells (unpaired t-test with Welch’s correction), p = 0.0157 (FOSL1), p = 0.0278 (PIM1). (D) Dose-response of JQ-1 and CX-6258 on the number of adherent HPV-10 cells. Data were normalized to the DMSO condition (each dot is the average of three biological replicates). ****p < 0.0001 versus DMSO (ordinary two-way ANOVA with Šídák multiple comparisons test). Exact p values are reported in Appendix Table S1. (E–H) Human PC-3 (E, F) and HPV-10 (G, H) cells were treated with T5224-PROTAC (‘PROTAC’) at 0, 8, and 24 h. The cells were collected at 48 h. The cell number (E, G) and cell viability (adherent + in suspension) (F, H) were determined by trypan blue staining (biological replicates, n = 4 independent experiments). The data were normalized to DMSO (control condition). *p < 0.05, **p < 0.01 (ANOVA analysis and Dunn’s post hoc test). (I) Human PC-3 cells were treated with siRNA (siScrambled or three different FOSL1 siRNA, as indicated) for 6 h and the cells were collected at 48 h. The expression of FOSL1 was measured by RT-qPCR (biological replicates, n = 2 independent experiments). The data were normalized to siScrambled (control condition). The three siRNA FOSL1 showed similar efficacy. ****p < 0.0001 (ANOVA analysis and Dunnett’s post hoc test). Exact p values are reported in Appendix Table S1. (J–L) Human PC-3 cells were treated with siScrambled or siFOSL1(1) for 6 h and the cells were collected at 72 h. Cell number (J) and cell viability (adherent + in suspension) (K) were determined by trypan blue staining. The expression of FOSL1 (L) was measured by RT-qPCR. The data are normalized to siScrambled (biological replicates, n = 3 independent experiments). **p < 0.01, ***p < 0.001, ****p < 0.0001 (two-tailed t-test). Exact p values are reported in Appendix Table S1. All error bars in this figure represent SD.

Fig 3: Time-course analysis of CD44 and FOSL1 expression after castration in Ptenpc−/− mouse prostates.(A, B) The expression of CD44, chosen as a selected marker of LSCmed-1 cells was assessed by immunohistochemistry in intact Ptenpc−/− mice (n = 6) and 5 days (n = 3), 21 days (n = 3), and 2 months (n = 7) post-castration. Both membrane and diffuse cytoplasmic staining was observed as previously reported in PCa (Omara-Opyene et al, 2004). (A) Quantification of CD44-positive cells was performed by QuPath. *p < 0.05, ***p < 0.001 (one-way ANOVA). Exact p values: p = 0.00231 (5 days), p = 0.00169 (21 days), p = 0.0005 (2 months). (B) Representative images are shown for the indicated time points. Scale bars = 200 µm and 50 µm in insets. (C, D) Immunofluorescence analysis of CD44, KRT4 and p63 showing (arrowheads) frequent co-immunostaining of CD44 and KRT4 in the luminal epithelium (C) and occasional co-immunostaining of CD44 and p63 in the basal layer (D). Representative images from castrated mice are shown. Scale bars = 50 and 10 µm in insets. (E–G) The expression of FOSL1 was assessed by immunohistochemistry in intact Ptenpc−/−mice (n = 2) and 5 days (n = 3), 21 days (n = 3), and 2 months (n = 3) post-castration. (E) Representative images from intact and 2-month castrated mice are shown. Scale bars = 50 µm. (F) Quantification of nuclear FOSL1-positive cells was performed by QuPath. *p < 0.05 (One-way ANOVA). Exact p values: p = 0.2383 (5 days), p = 0.1748 (21 days), p = 0.0132 (2 months). (G) For each timepoint, FOSL1-positive cells were stratified by QuPath according to low, medium and high nuclear staining intensity, as illustrated. *p < 0.05 (one-way ANOVA); exact p values are reported in Appendix Table S1. (H–J) Immunofluorescence analysis of FOSL1, KRT4, CD44, and p63 showing frequent co-immunostaining of FOSL1 and KRT4 (H) or CD44 (I) in the luminal epithelium, and occasional co-immunostaining of FOSL1 and p63 in the basal layer (arrowhead) (J). Representative images from castrated mice are shown. Scale bars = 50 µm. (K) Western blot showing the upregulation of FOSL1 expression in sorted LSCmed cells after castration (FOSL1/GAPDH ratio increased by 46%). Each lane represents two pooled animals.

Fig 4: FOSL1/AP-1 complex is a major transcriptional hub for Ptenpc−/− LSCmed cells emerging post-castration.(A–D) Identification of the main upstream regulators of the three LSCmed cell subpopulations by SCENIC analysis. The top-20 (fold change) activated (red) and inactivated (blue) transcription factors in LSCmed-0 (A), LSCmed-1 (B) et LSCmed-2 (D) cells are represented as waterfall plots. The red arrow in B identifies Fosl1. The transcription factor network of the three most pertinent regulons of LSCmed-1 cells is represented in (C) (see Fig. EV3 for LSCmed-0 and LSCmed-2 cell transcription factor networks). (E, F) Fosl1 expression (E) and AUCell Score of FOSL1 regulon (F) in the three LSCmed cell subpopulations. ns, not significant; *p < 0.05; ****p < 0.0001 (Wilcoxon test, Holm-method adjusted). Exact p values: Fosl1: p = 3.7e-21 (0 vs 1), 1.3e-06 (0 vs 2), p = 0.11 (1 vs 2); Fosl1 regulon: p = 4.8e-26 (0 vs 1), 1e-07 (0 vs 2) and p = 0.031 (1 vs 2), n = 140 (cluster 0), n = 130 (cluster 1), n = 28 (cluster 2). (G) Correlation of Fosl1 expression and FOSL1 regulon in the three LSCmed cell subpopulations. (H) Projection of Fosl1 expression during pseudotime (pval = 5.01e-21, qval = 6.37e-19). (I, J) Expression of FOSL1 (I) and FOSL1 regulon (J) in SU2C patient cohort (Data Ref: Abida et al, 2019a), according to the tumor molecular subtype (n = 125 ARPC, n = 40 NEPC, n = 76 SCL, n = 25 WNT). Boxplots display median (line), 25th–75th percentiles (box), whiskers at 1.5×IQR, and individual outliers. ns, not significant; **p < 0.01; ***p < 0.001; ****p < 0.0001 (Kruskal–Wallis with Holm-corrected Wilcoxon post hoc tests). See Appendix Table S1 for exact p values.

Fig 5: Induction of FOSL1, FOSL1 regulon, PIM1 and LSCmed cluster 1 and 2 expression features upon enzalutamide treatment in human prostate cancer (DARANA study).In the DARANA (Dynamics of Androgen Receptor Genomics and Transcriptomics After Neoadjuvant Androgen Ablation) study (ClinicalTrials.gov NCT03297385), high-risk prostate cancer patients were treated with neoadjuvant enzalutamide for 3 months before prostatectomy, with tumor biopsies taken before and after treatment. RNA sequencing data of naive (“pre”, n = 42) and neoadjuvantly-treated (“post”, n = 52) samples (Data Ref: Linder et al, 2022b) was interrogated for LSCmed cluster signatures (A), FOSL1 (B), FOSL1 regulon (C), and PIM1 (D) expression. Enzalutamide treatment induces FOSL1, FOSL1-regulon and LSCmed-1 and LSCmed -2 gene expression and decreases LSCmed-0 gene expression. *p < 0.05; **p < 0.01, ****p < 0.0001 (Wilcoxon test). Exact p values: p = 0.0092 (LSCmed-0), p = 1.02e-40 (LSCmed-1), p = 0.0146 (LSCmed-2), p = 8.82e-18 (FOSL1), p = 5.46e-58 (FOSL1 regulon), p = 0.0378 (PIM1)