Posttranslationally modified progesterone receptors direct ligand-specific expression of breast cancer stem cell-associated gene programs

Abstract

Background: Estrogen and progesterone are potent breast mitogens. In addition to steroid hormones, multiple signaling pathways input to estrogen receptor (ER) and progesterone receptor (PR) actions via posttranslational events. Protein kinases commonly activated in breast cancers phosphorylate steroid hormone receptors (SRs) and profoundly impact their activities.

Methods: To better understand the role of modified PRs in breast cancer, we measured total and phospho-Ser294 PRs in 209 human breast tumors represented on 2754 individual tissue spots within a tissue microarray and assayed the regulation of this site in human tumor explants cultured ex vivo. To complement this analysis, we assayed PR target gene regulation in T47D luminal breast cancer models following treatment with progestin (promegestone; R5020) and antiprogestins (mifepristone, onapristone, or aglepristone) in conditions under which the receptor is regulated by Lys388 SUMOylation (K388 intact) or is SUMO-deficient (via K388R mutation to mimic persistent Ser294 phosphorylation). Selected phospho-PR-driven target genes were validated by qRT-PCR and following RUNX2 shRNA knockdown in breast cancer cell lines. Primary and secondary mammosphere assays were performed to implicate phospho-Ser294 PRs, epidermal growth factor signaling, and RUNX2 in breast cancer stem cell biology.

Results: Phospho-Ser294 PR species were abundant in a majority (54%) of luminal breast tumors, and PR promoter selectivity was exquisitely sensitive to posttranslational modifications. Phospho-PR expression and target gene programs were significantly associated with invasive lobular carcinoma (ILC). Consistent with our finding that activated phospho-PRs undergo rapid ligand-dependent turnover, unique phospho-PR gene signatures were most prevalent in breast tumors clinically designated as PR-low to PR-null (luminal B) and included gene sets associated with cancer stem cell biology (HER2, PAX2, AHR, AR, RUNX). Validation studies demonstrated a requirement for RUNX2 in the regulation of selected phospho-PR target genes (SLC37A2). In vitro mammosphere formation assays support a role for phospho-Ser294-PRs via growth factor (EGF) signaling as well as RUNX2 as potent drivers of breast cancer stem cell fate.

Conclusions: We conclude that PR Ser294 phosphorylation is a common event in breast cancer progression that is required to maintain breast cancer stem cell fate, in part via cooperation with growth factor-initiated signaling pathways and key phospho-PR target genes including SLC37A2 and RUNX2. Clinical measurement of phosphorylated PRs should be considered a useful marker of breast tumor stem cell potential. Alternatively, unique phospho-PR target gene sets may provide useful tools with which to identify patients likely to respond to selective PR modulators that block PR Ser294 phosphorylation as part of rational combination (i.e., with antiestrogens) endocrine therapies designed to durably block breast cancer recurrence.

Keywords: Antiprogestin; Breast cancer; Cancer stem cells; ERK/MAP kinase (MAPK); Estrogen receptor (ER); Onapristone; Phosphorylation; Progesterone receptor (PR); RUNX2; SUMOylation.

Fig. 1

Total PR and phospho-Ser294 PR IHC staining in cell lines, tissue sections, and a breast cancer tissue microarray. a Cartoon depiction of PR ligand/kinase-dependent Ser294 phosphorylation, which blocks Lys388 SUMOylation and alters the recruitment of either co-activators or co-repressors, resulting in promoter-selective transcription. b PR Ser294 phosphorylation and total PR protein expression levels were measured in T47D breast cancer cell lines by western blotting, with and without progestin R5020 treatment. c PR levels were also measured in T47D cells on coverslips using IHC methods. d PR Ser294 and total PR protein levels were measured by IHC in a control tissue type (endometrial) to demonstrate effective PR Ser294 (and total PR) antibody specificity and sensitivity. e IHC in breast tumor sections (spots) of a tissue microarray. Six representative images demonstrate H-scoring classification: (column 1) 0% staining, (column 2) 20% positive cells, weak, (column 3) 100% positive cells, strong

Fig. 2

PR Ser294 phosphorylation and total PR H-scores are not correlated, and PR Ser294 phosphorylation H-scores were negatively associated with various tumor characteristics. a H-scores for total PR expression and phospho-Ser294 PR were compared among individual tumors spots from our TMA study. A Pearson correlation was calculated (r = 0.104, R ² = 0.0108). Tissue spots considered “positive” had an H-score of >20. Four quadrants were labeled (1−4) and discussed in the text. b Left: TMA spots were separated based on benign breast tissue (BBT) or tumor tissue (TT) pathological classification, and PR Ser294 phosphorylation H-scores were plotted (gray dots) with mean values (blue dots, ±95% CI). Right: H-score densities reveal wide distributions for both groups, but H-scores among TT samples are skewed toward zero. c Same analysis as Fig. 2b, except for total PR H-scores. d With IHC staining scores and patient metadata from our breast cancer TMA study, we used multiple regression to predict PR Ser294 H-scores from various factors. Significant variables (P < 0.05) have 95% CIs non-overlapping with the zero line

Fig. 3

Proliferation and biomarker expression in breast tumor explants in response to estrogen (E2), progesterone (P4), or combined P4 + U0126 treatment. a Post surgery, breast tumors were dissected and prepared for tissue explant experiments. Tumors were cut into small fragments and placed on sponges soaked in tissue culture medium. Sections were treated with vehicle (ethanol), E2 (1 or 10 nM) or P4 (1 or 10 nM) for 48 h. Tissue sections were then fixed, embedded, and processed for Ki-67 IHC staining. The percent of Ki-67-positive cells were plotted (mean ± SE). Comparing the groups via one-way ANOVA, followed by TukeyHSD posttest, indicated that only the P4 (10 nM) treatment was significantly different from vehicle (P = 0.0061, n = 6 explants per treatment condition). b Breast tumor explants were treated with vehicle, estradiol (E2, 10 nM), progesterone (P4, 10 nM), or a combination of P4 and MAPK inhibitor U0126 (1 μM) for 2 h. Explants were fixed, paraffin embedded, and stained for phospho-Ser294 PR expression, and H-scores were plotted. c–f Representative tumor explant IHC images after staining for total ER, total PR, pSer-294 PR, and phospho-ERK1/2 expression

Fig. 4

Select PR antiprogestins, mifepristone and aglepristone, induce PR Ser294 phosphorylation, but onapristone does not. a T47D cells expressing wild type (WT) PR or Ser294/SUMO-deficient PR (KR) were treated with vehicle (V), progesterone (P), mifepristone (M), aglepristone (A), onapristone (O), or a combination of progesterone and each antiprogestin. Cells were harvested for western blotting analysis and revealed that both mifepristone and aglepristone induce PR Ser294 phosphorylation, whereas onapristone does not. Co-treatment of progesterone and mifepristone or aglepristone also induce Ser294 phosphorylation, whereas onapristone blocks Ser294 phosphorylation even in the presence of progesterone. b Similar to western blotting analysis, T47D cells were treated as described above and analyzed for PR expression by immunofluorescence. Again, only onapristone effectively blocked PR Ser294 phosphorylation in both cells expressing WT or KR PR (highlighted by green box)

Fig. 5

Gene expression analysis in T47D cells treated with various ligand combinations demonstrates unique promoter selection. a Gene expression arrays were used to measure global changes in gene expression levels in T47D cells treated with different PR ligands: vehicle, progestin (P), mifepristone (M), aglepristone (A), onapristone (O), P + M, P + A, or P + O. Genes under high variance across these samples were isolated, and expression values were used for NMF clustering. Presented is the consensus matrix that indicates five major clusters are present in the samples. b Using the gene expression dataset (log2 fold change ≥2, BH P value ≤0.01), multiple sample comparisons (i.e., vehicle vs. P) were made and genes that were significantly regulated were isolated (rows). These genes were clustered via unsupervised hierarchal clustering methods, and two major branches were identified (clusters 1 and 2). In addition, sub-branches can also be seen, suggesting a total of five independent sample groups. c We identified 16 PR target genes that were specifically regulated in T47D breast cancer cells expressing Ser294 phosphorylation/SUMO-deficient PR (KR) and not regulated by WT PR (that is not phosphorylated and SUMOylated). In addition, we identified 101 genes that were specifically upregulated by WT PR (non-phosphorylated and SUMOylated PR) (not shown). d We compared the average expression of these 16 genes or 101 genes in the published TCGA breast cancer cohort of PR-negative tumors. Despite all of these tumors being PR-negative (by clinical IHC diagnosis), the activated PR target genes (KR) are expressed at significantly higher levels compared to genes upregulated by WT PR (P = 0.0003435)

Fig. 6

The phospho-Ser294 PR gene set is upregulated in ILC breast tumors. a Mean gene expression values for a phospho-Ser294 PR gene set (see Fig. 5c) were plotted (gray dots) for tumors classified as IDC, ILC, or mixed IDC/ILC by the TCGA project. The mean of all values within each tumor subset were plotted (blue dots, ±95% CI), and groups were statistically compared using ANOVA with TukeyHSD posttest. Adjusted P values are displayed. b A control analysis was repeated with a random set of 150 genes

Fig. 7

RUNX2 may facilitate SUMO-deficient PR target gene expression. a SLC37A2 genomic region contains multiple RUNX binding motifs and other regulatory regions (CpG islands and other transcription factor binding hot spots). The number (#) of RUNX binding motifs within three major regions are listed. b SLC37A2 expression in T47D cells was measured after treatment with progestin (R5020) by RT-qPCR. c SLC37A2 expression in MCF-7 cells was measured after treatment with progestin (R5020) and/or antiprogestin (mifepristone) by RT-qPCR. d SLC37A2 expression in BT474 cells was measured after treatment with progestin (R5020) and/or antiprogestin (onapristone) by RT-qPCR. e T47D cells expressing WT or KR PR were engineered to stably express shRNAs targeting RUNX2, resulting in approximately 50% reduction of RUNX2 mRNA levels. f In cells stably expressing shRNAs targeting RUNX2, expression of the KR PR target gene SLC37A2 was significantly reduced

Fig. 8

Mammosphere formation in T47D cells (empty vector, PR-B, PR-B K388R, and PR-B S294A). a Primary and b secondary mammospheres in T47D cells overexpressing empty vector, PR-B, PR-B K388R, or PR-B S294A and plotted as a percentage of mammosphere forming efficiency (MFE; see “Methods”). Cells were treated with vehicle (EtOH) or R5020 (10 nM). c Images of primary mammospheres (vehicle) from a. d Primary and e secondary mammospheres in T47D cells treated with vehicle (water) or EGF (20 ng/ml). Mammospheres were allowed to grow for 14 days prior to counting. f Primary mammospheres in T47D cells (PR-B or K388R) expressing shGFP or shRUNX2. g Images of primary mammospheres from f. Data is represented as the average ± SD of three readings. *p < 0.05, **p < 0.01, ***p < 0.001 compared to empty vector control (vehicle)

Fig. 9

Primary mammosphere formation in unmodified HER2+ BT474 breast cancer cells. a Primary mammospheres in HER2+ BT474 cells expressing endogenous estrogen receptor (ER) and progesterone receptor isoforms (PR-A and PR-B) plotted as a percentage of mammosphere forming efficiency (MFE; see “Methods”). Cells were treated with vehicle (EtOH) control or R5020 (10 nM) without or with increasing concentrations of the type II antiprogestin onapristone (0, 100, or 1000 nM). Data is represented as the average ± SD of three readings. *p < 0.05, **p < 0.01, ***p < 0.001 compared to vehicle control. Secondary mammospheres failed to form in onapristone-containing media (not shown). b, c Model depicting PR action in normal breast (b) vs. during neoplastic luminal tumor progression (c). Phosphorylation of PR Ser294 and p-PR target gene expression (HER2, RUNX2, AR, AHR, PAX2) in the cancer stem cell (CSC) or neighboring tumor cell compartments may occur during early luminal tumor progression of ER+/PR+ (luminal A type) breast cancers that progress toward ER+/PR-low (and HER2+) (luminal B type) tumors that are CSC-rich and thus more likely to become endocrine-resistant. Early addition of anti-progestins to anti-estrogen/ER-based therapies may prevent or delay the onset of endocrine therapy-resistant luminal breast cancer recurrence