Molecular characterization of breast cancer cell pools with normal or reduced ability to respond to progesterone: a study based on RNA-seq

Abstract

Background: About one-third of patients with estrogen receptor alpha (ERα)-positive breast cancer have tumors which are progesterone receptor (PR) negative. PR is an important prognostic factor in breast cancer. Patients with ERα-positive/PR-negative tumors have shorter disease-free and overall survival than patients with ERα-positive/PR-positive tumors. New evidence has shown that progesterone (P4) has an anti-proliferative effect in ERα-positive breast cancer cells. However, the role of PR in breast cancer is only poorly understood.

Methods: We disrupted the PR gene (PGR) in ERα-positive/PR-positive T-47D cells using the CRISPR/Cas9 system. This resulted in cell pools we termed PR-low as P4 mediated effects were inhibited or blocked compared to control T-47D cells. We analyzed the gene expression profiles of PR-low and control T-47D cells in the absence of hormone and upon treatment with P4 alone or P4 together with estradiol (E2). Differentially expressed (DE) genes between experimental groups were characterized based on RNA-seq and Gene Ontology (GO) enrichment analyses.

Results: The overall gene expression pattern was very similar between untreated PR-low and untreated control T-47D cells. More than 6000 genes were DE in control T-47D cells upon stimulation with P4 or P4 plus E2. When PR-low pools were subjected to the same hormonal treatment, up- or downregulation was either blocked/absent or consistently lower. We identified more than 3000 genes that were DE between hormone-treated PR-low and control T-47D cells. GO analysis revealed seven significantly enriched biological processes affected by PR and associated with G protein-coupled receptor (GPCR) pathways which have been described to support growth, invasiveness, and metastasis in breast cancer cells.

Conclusions: The present study provides new insights into the complex role of PR in ERα-positive/PR-positive breast cancer cells. Many of the genes affected by PR are part of central biological processes of tumorigenesis.

Keywords: Breast cancer; CRISPR/Cas9; Estrogen receptor alpha; Gene expression; Progesterone receptor; RNA sequencing.

Fig. 1

Characterization of T-47D control cells and PR-low. a PR function was assessed in the three PR-low (red) and three control (green) pools with a PRE-driven luciferase construct (2 × PRE-TK-luc). Each treatment was done in triplicate. Error bars represent the SEM. b Endogenous progesterone (P4)-responsive genes were measured in T-47D control cells (green) and PR-low pools by RT-qPCR. Samples were measured in triplicates. P4 and P4/estradiol (E2) treatment was for 8 h. Fold change was calculated relative to untreated as described in Material and methods. Error bars represent the SEM

Fig. 2

Differential expression analyses of RNA-seq samples. a Differentially expressed (DE) genes (BH-adjusted P < 0.05) between T-47D control cells (green) and PR-low pools (red) and among treatments. The data for control and PR-low are mean values of three independent biological replicates. The number of DE genes is shown for each comparison. Venn diagrams representing overlaps of DE genes between P4-responsive and P4/E2-responsive genes in control (b), P4-responsive and P4/E2-responsive genes in PR-low (c), P4-responsive genes in control and P4-responsive genes in PR-low (d), P4/E2-responsive genes in control and P4-responsive genes in PR-low (e), and DE genes between control and PR-low after vehicle, P4 and P4/E2 treatments (f). P4, progesterone; E2, estradiol; PR, progesterone receptor

Fig. 3

Differentially expressed (DE) genes (BH-adjusted P < 0.05) in T-47D control cells and PR-low pools after progesterone (P4) treatment. Genes are aligned along the x-axis according to their log2 fold change (P4/vehicle) in T-47D control cells from highest to lowest (green if significant and black if not significant). The same arrangement of the genes on the x-axis was used for the corresponding PR-low pools (red if significant and grey if not significant). Shown are mean values of three independent biological replicates. About 95% of genes in PR-low pools were less responsive to P4 treatment than control cells resulting in nonsignificant regulation (4411 genes in gray). Induction/repression of PR-responsive genes was still significant in 2104 genes (red) as compared to controls (6515 genes in green). The number of differentially expressed genes in control and PR-low is depicted in the table (inset). ↑Higher, ↓lower, ns stands for not significant, *significant in control and PR-low, **significant in control and/or PR-low. PR, progesterone receptor

Fig. 4

Gene Ontology (GO) term enrichment analysis. Biological processes enriched for differentially expressed (DE) genes between P4/E2-treated T-47D control cells and P4/E2-treated PR-low pools. a The 30 most significantly deregulated biological processes (P-value < 0.01). For each GO annotation, the % of significantly DE genes from the total number of annotated genes is depicted. REVIGO [34] was used to assess the functional redundancy among the biological processes; redundant GO annotations are shown in the same color. + Involved in breast cancer. *Involved in cancer. b Interactive graph created by REVIGO, redundant GO annotations are removed. The sizes of the circles indicate the relative frequencies of GO terms in the underlying GO annotation database (UniProt) (frequency = percentage of annotated human proteins in UniProt with a GO term). Darker shading indicates lower p-values; similar GO terms are connected by lines, with thicker lines indicating closer similarity; the length of the lines is arbitrary. + Involved in breast cancer. *Involved in cancer. BC, breast cancer