The estrogen signaling pathway reprograms prostate cancer cell metabolism and supports proliferation and disease progression

Abstract

Just like the androgen receptor (AR), the estrogen receptor α (ERα) is expressed in the prostate and is thought to influence prostate cancer (PCa) biology. Yet the incomplete understanding of ERα functions in PCa hinders our ability to fully comprehend its clinical relevance and restricts the repurposing of estrogen-targeted therapies for the treatment of this disease. Using 2 human PCa tissue microarray cohorts, we first demonstrate that nuclear ERα expression was heterogeneous among patients, being detected in only half of the tumors. Positive nuclear ERα levels were correlated with disease recurrence, progression to metastatic PCa, and patient survival. Using in vitro and in vivo models of the normal prostate and PCa, bulk and single-cell RNA-Seq analyses revealed that estrogens partially mimicked the androgen transcriptional response and activated specific biological pathways linked to proliferation and metabolism. Bioenergetic flux assays and metabolomics confirmed the regulation of cancer metabolism by estrogens, supporting proliferation. Using cancer cell lines and patient-derived organoids, selective estrogen receptor modulators, a pure anti-estrogen, and genetic approaches impaired cancer cell proliferation and growth in an ERα-dependent manner. Overall, our study revealed that, when expressed, ERα functionally reprogrammed PCa metabolism, was associated with disease progression, and could be targeted for therapeutic purposes.

Keywords: Endocrinology; Glucose metabolism; Oncology; Prostate cancer; Sex hormones.

Figure 1. ERα expression is heterogenous in PCa and, when nuclear (active), is associated with BCR.

(A) Kaplan-Meier of BCR-free survival following radical prostatectomy in patients from TCGA-PRAD cohort with high or low ERα protein expression levels (no distinction between nuclear and cytoplasmic localization). (B) Proportions of patients from TCGA cohort with high or low ERα protein expression levels, with and without BCR (**P < 0.0019, χ² test). (C–F) Analysis of the Belledant et al. (32) cohort. (C) Representative images of ERα IHC in 4 patients with PCa. Black and red arrows, respectively, highlight negative and positive staining. Scale bars: 50 μm. Original magnification, ×3.1 (enlarged insets in C and G). (D) Kaplan-Meier BCR-free survival following radical prostatectomy in patients with positive versus negative nuclear levels of ERα. (E) Proportions of patients from the TMA cohort with positive or negative nuclear levels of ERα, with and without BCR (***P < 0.001, χ² test). (F) Cox regression analyses of the effect of positive (Pos.) nuclear ERα levels on the risk of BCR (*P < 0.05, **P < 0.01, and ***P < 0.001). Boxes illustrate HRs with their respective 95% CIs. Results are shown without (left) and with (right) additional BCR risk factors. Reference groups for covariables: Gleason score of 6; T2c stage and below; presurgery PSA levels under 10 ng/mL; negative lymph node invasion and negative margins. (G–I) Analysis of an independent cohort of patients who received neoadjuvant hormonotherapy before surgery. (G) Representative IHC images of ERα expression in 4 patients with PCa. Black and red arrows, respectively, highlight negative and positive staining. Scale bars: 50 μm. (H and I) Kaplan-Meier survival analysis in patients with positive versus negative (Neg.) ERα nuclear levels in the development of metastasis (H) and overall survival (I). For Kaplan-Meier survival curves, the log-rank test P value is shown in the inset.

Figure 2. Estrogens modulate the normal prostate transcriptome in vivo, activating oncogenic pathways similar to those activated with androgen stimulation.

(A) Representative IHC images of ERα in normal mouse prostate lobes. Scale bars: 50 μm. Original magnification, ×1.68 (enlarged insets). (B) Quantification of ERα-positive staining and ERα staining intensity in normal mouse prostate lobes (n = ~2,700 cells/animal, n = 5 animals/lobe). (C–I) RNA-Seq analyses of the murine prostate transcriptome 24 hours after injections with vehicle (Ctl), testosterone (Testo), E₂, or both (T+E₂). Mice were castrated 3 days before injections to ensure hormonal deprivation. (C) Number of significantly differentially expressed genes (DEGs) following pairwise comparisons between conditions. The thresholds used were a fold change of 1.75 or more or –1.75 or less and a P value with a FDR of less than 5%. (D) GSEA normalized enrichment score (NES) following treatment with testosterone. (E and F) GSEA diagrams and heatmaps for the androgen response (E) and the OXPHOS (F) gene sets following testosterone treatment in vivo. (G) GSEA NES for enrichment following E₂ treatment in vivo. (H) GSEA diagram and heatmap for the cholesterol homeostasis gene set following E₂ treatment. For E, F and H, NESs, P values, and q values are indicated on each diagram, and only core genes of each pathway are shown. *q < 0.05, **q < 0.01, and ***q < 0.001 in GSEA (D and G). (I) Venn diagram of upregulated genes for each pairwise comparison. (J) Venn diagram of estrogen-responsive genes in breast cancer cells (MCF7), using the data set from (41), and in the mouse prostate. Circle and overlap sizes are not proportional to the number of genes. (K) qRT-PCR analysis of positive controls for androgenic (Pfkfb3 and Fkbp11) and estrogenic regulation (Pgr, Fkbp11, and Greb1). For B and K, results are shown as the average with SEM (n = 4 mice/treatment); ^#P < 0.10; **P < 0.01 and ***P < 0.001, by 1-way ANOVA.

Figure 3. Estrogens activate oncogenic pathways in a PCa mouse model.

(A) Representative of H&E staining and staining for AR and ERα in prostates from 24-week-old WT and PCa-developing mice. Black and red arrows, respectively, highlight negative and positive staining. Scale bars: 50 μm. Original magnification, ×3.1 (enlarged insets). (B) Western blot of prostate samples from WT and PCa-developing mice. Phosphorylated S6 (p-S6) shows activation of the mTOR signaling following prostate-specific deletion of Pten in tumors. S6 was used as the loading control. exp., exposure. (C–I) RNA-Seq analyses of mouse PCa tumors following a 24-hour treatment in vivo with vehicle, testosterone, E₂, or both. Mice were castrated 3 days before injections to ensure steroid deprivation. (C) Number of DEGs following pairwise comparisons. (D and F) NES of GSEA following treatment with testosterone (D) or E₂ (F). ^#q < 0.05, ^##q < 0.01, and ^###q < 0.001. (E, G, and I) GSEA diagrams and heatmaps for the androgen response following testosterone treatment (E), the mTORC1 gene set following E₂ treatment (G), and the OXPHOS gene set following testosterone plus E₂ treatment (I). Only core genes are shown. (H) Venn diagram of upregulated genes for each pairwise comparison. (J and K) qRT-PCR analysis of positive controls (J) and metabolic genes (K) following treatments. Results are shown as the mean ± SEM (3–4 mice/condition). (L–O) Single-cell RNA-Seq analyses from tumoral murine prostates, with and without treatment with E₂ (n = 2 mice/condition). (L) Esr1 expression in Pbsn-positive epithelial cells (in log scale of [counts/10K (CP10K) + 1]). (M) Greb1 expression in mesenchymal and epithelial Pbsn–positive clusters. (N and O) NES of GSEA analysis enriched following E₂ treatment in Pbsn-positive epithelial cells (O), with the GSEA diagram for the OXPHOS gene set (N). *P < 0.05, **P < 0.01, and ***P < 0.001, by 1-way ANOVA (J and K) or 2-tailed Student’s t test (M).

Figure 4. The ERα transcriptional program promotes PCa cell metabolism and proliferation.

(A) Western blot of AR and ERα expression in in vitro models: 1 ERα-positive breast cancer cell line (MCF7), 1 ERα-negative mammary gland cell line (MCF10A), and 6 human PCa cell lines (α-tubulin was used as a loading control). exp., exposure. (B–F) RNA-Seq analyses of VCaP cells following 24 hours of treatment with vehicle, the synthetic androgen R1881, E₂, or both (R + E₂). (B) GSEA NES following treatment with R1881. (C) GSEA diagrams and heatmap for the androgen response gene set following treatment with R1881 and qRT-PCR analysis of KLK3 expression (encodes PSA). Values are shown as the average with the SEM of 4 independent experiments performed in triplicate. (D) GSEA NESs showing enrichment following treatment with E₂. ^#q < 0.05, ^##q < 0.01, and ^###q < 0.001 (B and D). GSEA diagrams and heatmaps for the OXPHOS (E) and androgen response (F) gene sets following treatment with E₂ in VCaP cells. For C, E, and F, the NES, P values, and q values are indicated on each diagram, and only core genes for each pathway are shown. (G) VCaP proliferation assay following treatment with either R1881, E₂, or both. One representative experiment of 4 independent experiments is shown. Results are shown as the mean ± SEM (n = 6–8/treatment group). (H) VCaP OCR profiles following 72 hours of treatment with either R1881, E₂, or both. Complete mitochondrial stress test results with basal and maximal OCR capacities are shown. Oligo, oligomycin; Rot.+A.A., rotenone + antimycin A. One representative independent experiment of 3 is shown. Data show the mean of normalized data to cell numbers ± SEM (n = 10–12/treatment). *P < 0.05, **P < 0.01, and ***P < 0.001, by 1-way ANOVA (C, G, and H).

Figure 5. ERα activation induces cancer cell metabolism, notably by promoting glucose consumption and usage.

(A) Schematic overview of glucose metabolism through glycolysis to allow pyruvate synthesis, which can then fuel the mitochondrial TCA cycle for respiration. Note that not all enzymatic reactions are shown (dashed lines symbolize intermediate steps). αKG, α-ketoglutarate; Succ., succinate; Fum., fumarate; Oxalo., oxaloacetate. (B and C) Quantification of lactate (B), alanine (B), and TCA cycle intermediates (C) in VCaP cells following 72 hours of treatment with E₂ or the synthetic androgen R1881 by gas chromatography–mass spectrometry (GC-MS). (D and E) Quantification of ¹³C incorporation from ¹³C-glucose in lactate and alanine (D) and TCA cycle intermediates (E) in VCaP cells following 72 hours of treatment with E₂ or R1881. ¹³C-glucose allowed the enrichment of a heavier isotopomer with a mass of plus 3 (m+3) for lactate and alanine and a mass of plus 2 (m+2) for citrate, succinate, and malate if it feeds the TCA cycle. (F) Changes in VCaP cell numbers following 168 hours of treatment with either E₂, the inhibitor of mitochondrial respiration metformin (Met), or both (Met + E₂). The changes in cell numbers were normalized in percentages according to the control treatment. Results are shown as the mean ± SEM of 2 independent experiments (n = 16/treatment group). (G) Quantification of amino acids connected to energy synthesis pathways in VCaP cells following 72 hours of treatment with E₂ or R1881 by GC-MS. For B–E and G, results are shown as the mean ± SEM of 1 representative experiment (n = 5/conditions) of 3 independent experiments. (H) Western blot of the mTOR signaling pathway with phosphorylation of downstream targets (S6 and S6K) following hormone treatment. α-Tubulin was used as a loading control. *P < 0.05, **P < 0.01, and ***P < 0.001, by 1-way ANOVA, respective to control conditions or as indicated. For D and E, P values are only shown for metabolites with ¹³C labeling. ^#P < 0.10 (G).

Figure 6. SERMs and fulvestrant inhibit E₂-dependent stimulation of mitochondrial respiration, proliferation, and growth of PCa cells.

(A) VCaP OCR profiles following a 72-hour treatment with E₂, tamoxifen (Tamox), raloxifene (Ralox), toremifene (Torem), and fulvestrant (Fulv). Results from a complete mitochondrial stress test of 1 experiment are presented, with basal and maximal OCR capacities shown as the average of 2 of 3 independent experiments. Data indicate the mean ± SEM (n = 8–12/treatments per experiment). Changes in VCaP cell number following 168 hours of treatment with anti-estrogens cotreated with E₂ (B), or with hormone cotreatment with fulvestrant or enzalutamide (C), normalized to control. One representative experiment of 3 independent experiments is shown. Data indicate the mean ± SEM (n = 6–8/condition). (D) Kaplan-Meier of survival and tumor growth of castrated mice with VCaP xenografts under either a placebo or E₂ pellet treatment and injected weekly with vehicle or fulvestrant (n = 5–10 mice/condition). The log-rank test P value is shown. Changes in tumor growth were quantified on the basis of tumor volume at castration adjusted at 0%. Tumor growth is shown up to 90 days, at which point most E₂-treated tumors were harvested. Colored arrows indicate mice reaching ethical limit points. (E and F) Bright-field images (E) and changes in organoid growth (F) of 3 PDO lines after 14–15 days of treatment with vehicle, E₂, fulvestrant, or both. (G) qRT-PCR analysis of ESR1 expression in the PDO lines shown in E. Results are shown as a fold change compared with PDO 3. (H and I) Bright-field images (H) and changes in organoid growth (I) in PDO 1 after 15 days of treatment with vehicle and E₂, with and without ESR1 knockdown. Scale bars: 300 μm (E and H). Results in F and I are shown as the mean ± SEM (n = 4 replicates/condition). NS, nonsignificant; *P < 0.05, **P < 0.01, and ***P < 0.001, by 1-way ANOVA.

Figure 7. ESR1 expression is increased following ADT, and its transcriptional signature is associated with PCa progression.

(A) Heatmap of the ERα-score in patients from TCGA-PRAD data set (30, 31). The ERα-score is the predicted transcriptional activity of ERα. The legend shows DEGs with increased (red) or decreased (blue) expression following E₂ treatment in VCaP cells. (B and C) Kaplan-Meier of BCR-free survival following surgery for patients from TCGA-PRAD (B) and the Taylor et al. (C) data sets, separated by high and low ERα-scores. The log-rank P values are shown. (D) ESR1 (encodes ERα) expression in PCa tumors before and after ADT in the Eur Uro 2017 data set (52). adj, adjusted. (E) ESR1, ESR2, and PGR gene expression in PCa tumors before and after ADT in the Eur Uro 2014 data set (53) (n = 7 paired samples). (F) ESR1, ESR2, and PGR gene expression in PCa tumors before and after ADT plus docetaxel in the BioMed Central (BMC) cancer data set (54) (n = 4 paired samples). (G) ESR1, ESR2, and PGR gene expression in PCa tumors before and after ADT in the GSE183100 data set (n = 73 samples). (H and I) Bright-field images (scale bars: 300 μm) (H) and changes in organoid growth (I) of the PDO 1 line after treatment with vehicle and the anti-androgen enzalutamide (Enza) cotreated or not with E₂. (J) ERα-score in the SU2C data set (55), separated by tumor localization in the prostate (n = 5) and metastases in either adrenal glands (n = 2), bone (n = 82), lymph nodes (LN) (n = 79), liver (n = 26), and other sites (n = 14). NS, nonsignificant; *P < 0.05, **P < 0.01, and ***P < 0.001, by 1-way ANOVA (I and J) or 2-tailed Student’s t test, as appropriate (E–G).