Research ArticleONCOLOGY

Genomic agonism and phenotypic antagonism between estrogen and progesterone receptors in breast cancer

See allHide authors and affiliations

Science Advances  24 Jun 2016:
Vol. 2, no. 6, e1501924
DOI: 10.1126/sciadv.1501924
  • Fig. 1 Genomic agonism and phenotypic antagonism between ER and PR in breast cancer.

    (A to C) Estrogen and progestin independently regulate gene expression in the same direction for representative patients (A) P2, (B) P8, and (C) T47D cells. Axes denote log fold change of gene expression in response to estrogen (E) or progestin R5020 (P) treatment relative to vehicle (V). Green dots represent genes regulated in similar directions by estrogen or progestin. Red dots represent genes regulated in different directions by estrogen or progestin. (D) Box plot depicts the percentage of all ER- and PR-regulated genes in ex vivo cultured primary breast tumors (n = 8) for which progestin is an agonist or antagonist of estrogen-regulated gene expression. (E and F) Similarity matrices represent correlation between estrogen-, progestin-, and EP-regulated levels of transcriptomes in (E) a PR+ milieu and (F) a PR milieu. (G and H) Expression of estrogen and progestin-regulated genes in (G) a PR+ milieu (four ER+/PR+ ex vivo cultured human tumors and T47D, ZR75, and T47D PR-deficient cells with ectopic reexpression of PR) and (H) a PR-deficient milieu (four ER+/PR tumors and PR-deficient T47D and MCF7 cells). Tumors were treated ex vivo and cell lines in vitro with vehicle, estrogen, or progestin or concomitantly with both hormones (EP). All heat maps are row-normalized and include the union of ER- and PR-regulated genes. For any given gene, red (or blue) and green (or yellow) colors of a row-normalized heat map represent minimum and maximum magnitudes of normalized expression that are observed in response to various treatments. (I) Enrichment (P values) and Z scores of activation of functional processes by estrogen-, progestin-, and EP-regulated transcriptomes in five human tumor explants treated ex vivo for 24 hours. For cell models, RT-PCR assessments of RNA-seq were done as three independent experiments (three technical replicates per experiment) (fig. S3). RNA-seq was performed on one of the three biological replicates. For a subset of human tumors, the RT-PCR assessment of estrogen-mediated regulation was done for TFF1, GREB1, and PR genes (table S1). The lists of genes and their expression in response to various treatments are provided in tables S4 and S5.

  • Fig. 2 PR redirects ER binding to sites correlated with the binding of PR and ER/PR complexes.

    (A) Principal component analysis (PCA) plot displays 79% total variance between ER binding events in nine ER+/PR+ (green) and six ER+/PR (red) patient tumors [sequencing data were obtained from the study by Ross-Innes et al. (27)]. (B) Overlap of ER binding sites in PR+ and PR-deficient T47D cells treated with estrogen or EP. (C) Heat maps display intensity of sequencing obtained on anti-ER ChIP before and after remodeling by R5020 in PR+ T47D cells. The genomic window of the union of all ER binding sites observed before and after remodeling by PR is displayed. Overlap of at least 1 base pair (bp) was considered to categorize ER binding as lost, conserved, or gained. (D) PCA plot depicts 81% total variance between binding events for ER, PR, and ER/PR complexes observed upon treatment with estrogen, progestin, or EP. Binding events in PR+ and PR-deficient T47D cells are presented. All the binding sites and their annotations are provided in table S6. (E) Distributions of receptor binding around reprogrammed ER binding sites. Distributions for ER binding observed without progestin and binding for PR and ER/PR complexes on estrogen plus progestin treatment are plotted. (F) Distributions around transcription start sites for ER binding observed in six ER+PR and eight ER+PR+ human tumors. One outlier within the ER+/PR+ group was not included in the analyses. (G) Frequencies of binding events for ER, PR, and ER/PR complexes relative to their distance from transcription start sites. Hormone treatment is mentioned in parentheses. The numerical values for the total number of binding sites (gray) and the number of binding sites within the 3-kb promoter regions (white) are provided. ChIP-PCR assessments of ChIP-seq were done as three independent experiments and three technical replicates per experiment (fig. S4). ChIP-seq was performed on one of the three biological replicates.

  • Fig. 3 Progestin stimulation remodels nucleosomes to redirect ER binding to enhancers and binding sites enriched for BRCA1.

    (A) Anti-ER immunoprecipitation followed by immunoblotting for both ER and PR in T47D cells treated with different hormones. (B) Heat maps display intensity of sequencing obtained on reChIP-seq of anti-ER, followed by anti-PR or nonspecific immunoglobulin G (IgG) control. The genomic window of the binding of ER/PR complexes is displayed. (C) Capture of Associated Targets on Chromatin (CATCH) of estrogen response elements at PDZk1 and FHL2 loci pulls down distant progesterone response elements that interact with the pulled-down regions. The PCR enrichment of the pulled-down region, the interacting progesterone response elements, and random controls is shown. (D) Cumulative average of the percentage of H3K4me1+/H3K27ac1+ enhancers with the receptor binding within 5 kb. The enhancers are sorted from left to right in the increasing order of H3K4me1+ signal intensity. (E) BRCA1 and ELK4 binding motifs are highly enriched at binding sites for ER/PR complexes. The significance of the enriched motif is reported by P value. (F) Percentage overlap of ER binding sites with DNase-hypersensitive regions observed upon treatment of T47D or PR-depleted T47D cells with progestin [DNase-seq sequencing data were obtained from the study by Ballaré et al. (34)]. An overlap window of 20 kb is used for analyses. (G) Row-normalized heat maps depicting normalized expression of estrogen- and progestin-regulated genes in T47D cells after small interfering RNA (siRNA)–specific depletion of FOXA1, NF1C, or a nonspecific control. Heat maps for siFOXA1 represent genes that have FOXA1 binding within 100 kb of the gene’s promoter [FOXA1 binding data were obtained from the study by Hurtado et al. (26)].

  • Fig. 4 Presence and activity of PR contribute to the prognostic value of ER.

    (A) Frequency of hypermethylation of PR locus in ER+ TCGA tumors categorized on the basis of PR status. (B) Hypermethylation of PR gene locus correlates with loss of PR expression in ER+ TCGA tumors, measured using reverse-phase protein arrays. The horizontal axis displays four PR-specific methylation probes from the Human Methylation 450k array. (C) Frequency of copy number variation of PR gene locus in TCGA and METABRIC cohorts categorized on the basis of prediction analysis of microarray 50 (PAM50) breast tumor subtypes (22). (D) Overall survival in the TCGA cohort classified by positive or negative correlation to estrogen-regulated signature scores. Curves are presented for before (red) and after (green) progestin-mediated reprogramming of estrogen signaling. (E) Overall survival as determined by the differential tumor staining for ER in PR-negative (red) and PR+ (green) patient cohorts from METABRIC. Summaries of the patient cohorts are provided in table S7.

  • Fig. 5 Cytotoxic tumor regression on combination therapy with tamoxifen and PR antagonist CDB4124.

    (A) T47D xenografts were grown in ovariectomized nude mice containing estrogen silastic implants and were treated with placebo, tamoxifen, CDB-4124, or tamoxifen plus CDB4124. The average tumor volume at the start of therapies was 125 mm3, and percentage change in tumor volume is shown (n = at least 7). P values are calculated using mixed linear modeling. Control group is plotted until day 49 because a significant number of mice in the control group died after day 49. (B) Genomic agonism: In isolation and in combination, activated ER and PR regulate the expression of most of the genes in similar directions. The magnitude of gene expression on joint estrogen plus progestin treatments correlates with those observed with progestin alone, but not estrogen alone. Phenotypic antagonism: Individually, estrogen and progestin activate most of the oncogenic pathways in similar directions, but progestin lacks the degree of activation induced by estrogen. When both ER and PR are active, PR opposes ER-regulated phenotypes, suggesting phenotypic antagonism between these hormones. (C) Model for ER/PR crosstalk. PR remodels chromatin and redirects ER binding to antagonize estrogen signaling and to potentiate response to antiestrogens. Genomic agonism and the phenotypic antagonism between ER and PR highlight the prognostic and therapeutic value of PR in ER+/PR+ breast cancers. **<0.005.

Supplementary Materials

  • Supplementary material for this article is available at http://advances.sciencemag.org/cgi/content/full/2/6/e1501924/DC1

    table S1. Clinical information of tumors.

    table S2. PCR primers for ChIP-PCR and reChIP-PCR.

    table S3. PCR primers for CATCH chromosome capture.

    table S4. Gene expression changes observed in eight ER+/PR+ patient tumors and three ER+/PR+ cell models in response to various combinations of estrogen and progestin treatments.

    table S5. Gene expression changes observed in four ER+/PR patient tumors and two ER+/PR-deficient cell models in response to various combinations of estrogen and progestin treatments.

    table S6. Binding sites for ER, PR, and ER/PR complexes in ER+/PR+ T47D and ER+/PR-deficient T47D cells.

    table S7. Summaries of patient cohorts.

    fig. S1. Progestin is a genomic agonist of estrogen-regulated gene expression.

    fig. S2. Progestin is a phenotypic antagonist of estrogen-induced cell proliferation, invasion, and migration.

    fig. S3. Progestin modulates estrogen-regulated gene expression.

    fig. S4. PR redirects ER to sites enriched for motifs of PR and PR-associated cofactors.

    fig. S5. Noncompetitive interactions between ER and PR.

    fig. S6. Depletion of FOXA1 or NF1C insignificantly impacts the effects of PR on ER-regulated gene expression.

    fig. S7. PR-regulated genes are enriched for breast cancer signatures, and PR contributes to the prognostic value of ER.

  • Supplementary Materials

    This PDF file includes:

    • table S1. Clinical information of tumors.
    • table S2. PCR primers for ChIP-PCR and reChIP-PCR.
    • table S3. PCR primers for CATCH chromosome capture.
    • Legends for tables S4 to S6
    • table S7. Summaries of patient cohorts.
    • fig. S1. Progestin is a genomic agonist of estrogen-regulated gene expression.
    • fig. S2. Progestin is a phenotypic antagonist of estrogen-induced cell proliferation, invasion, and migration.
    • fig. S3. Progestin modulates estrogen-regulated gene expression.
    • fig. S4. PR redirects ER to sites enriched for motifs of PR and PR-associated cofactors.
    • fig. S5. Noncompetitive interactions between ER and PR.
    • fig. S6. Depletion of FOXA1 or NF1C insignificantly impacts the effects of PR on ER-regulated gene expression.
    • fig. S7. PR-regulated genes are enriched for breast cancer signatures, and PR contributes to the prognostic value of ER.

    Download PDF

    Other Supplementary Material for this manuscript includes the following:

    • table S4 (Microsoft Excel format). Gene expression changes observed in eight ER+/PR+ patient tumors and three ER+/PR+ cell models in response to various combinations of estrogen and progestin treatments.
    • table S5 (Microsoft Excel format). Gene expression changes observed in four ER+/PR? patient tumors and two ER+/PR-deficient cell models in response to various combinations of estrogen and progestin treatments.
    • table S6 (Microsoft Excel format). Binding sites for ER, PR, and ER/PR complexes in ER+/PR+ T47D and ER+/PR-deficient T47D cells.

    Files in this Data Supplement:

Navigate This Article