Estrogen Receptor Alpha Mutations in Breast Cancer Cells Cause Gene Expression Changes through Constant Activity and Secondary Effects

Abstract

While breast cancer patients with tumors that express estrogen receptor α (ER) generally respond well to hormone therapies that block ER activity, a significant number of patients relapse. Approximately 30% of these recurrences harbor activating mutations in the ligand binding domain (LBD) of ER, which have been shown to confer ligand-independent function. However, much is still unclear regarding the effect of mutant ER beyond its estrogen independence. To investigate the molecular effects of mutant ER, we developed multiple isogenic ER-mutant cell lines for the most common LBD mutations, Y537S and D538G. These mutations induced differential expression of thousands of genes, the majority of which were mutant allele specific and were not observed upon estrogen treatment of wild-type (WT) cells. These mutant-specific genes showed consistent differential expression across ER-mutant lines developed in other laboratories. WT cells with long-term estrogen exposure only exhibited some of these transcriptional changes, suggesting that mutant ER causes novel regulatory effects that are not simply due to constant activity. While ER mutations exhibited minor effects on ER genomic binding, with the exception of ligand independence, ER mutations conferred substantial differences in chromatin accessibility. Mutant ER was bound to approximately a quarter of mutant-enriched accessible regions that were enriched for other DNA binding factors, including FOXA1, CTCF, and OCT1. Overall, our findings indicate that mutant ER causes several consistent effects on gene expression, both indirectly and through constant activity. SIGNIFICANCE: This study demonstrates the multiple roles of mutant ER in breast cancer progression, including constant ER activity and secondary regulatory effects on gene expression and chromatin accessibility. GRAPHICAL ABSTRACT: http://cancerres.aacrjournals.org/content/canres/81/3/539/F1.large.jpg.See related commentary by Hermida-Prado and Jeselsohn, p. 537 See related article by Williams and colleagues, p. 732.

Figure 1.. ER mutations exhibit a mutation-specific expression profile.

(a) Principal component analysis of RNA-seq data displays the relationships between WT(blue), Y537S(yellow), and D538G(red) T-47D clones. (b) Heatmap shows expression levels for ligand-independent, mutant-specific shared, and allele-specific genes. (c) qPCR validation of ligand-independent (CISH) and mutant-specific (CASC5) expression. Error bars indicate average ± SEM for two clones for each genotype and each treatment. Student’s two sample t-test was used: **p<0.01, ***p<0.001, ****p<0.0001, and n.s. = not significant. (d) Enriched gene ontology terms for Y537S- and D538G-specific differentially regulated genes with Fisher’s exact test p-values are shown.

Figure 2.. Gene expression analysis from three independent studies reveals consistent ER mutant-specific patterns.

(a) Heatmap shows levels of consistent mutant-specific genes. (b) Examples of genes exhibiting consistent mutant-specific gene expression are shown. (c) Significantly enriched gene ontology terms are displayed for Y537S-specific and D538G-specific gene sets from the multi-study comparison.

Figure 3.. Constant activity of ER explains approximately half of T-47D mutant-specific genes.

(a) Principle component analysis of RNA-seq data captures the relationships between T-47D ER mutant clones and WT clones with short- or long-term E2 treatment. (b) Heatmap shows all mutant-specific genes separated by long-term E2-treatment expression (top: genes regulated by long-term E2 treatment; bottom: genes not regulated by long-term E2 treatment). (c) Percent of mutant-specific genes regulated by long-term ER activation in WT clones is shown. (d) Bar graphs show examples of genes that are regulated by long-term E2 treatment (TBCD) and that are differentially regulated by mutation only (COPS2). (e) Significantly enriched gene ontology terms are displayed for Y537S-specific and D538G-specific gene sets partitioned based on overlap with long-term E2 regulated genes.

Figure 4.. Limited changes in ER’s genomic binding are observed in ER mutant cells.

(a) Heatmap shows the binding signal of constant and mutant-specific ERBS in T-47D WT and mutant clones with addition of DMSO or E2. (b) Example locus of ligand-independent (constant) ER-binding, arrow indicates TSS of GREB1. All tracks are normalized to the same scale. GREB1 ligand-independent gene expression is shown on the right. (c) Heatmap displays the signal of mutant-enriched or depleted sites only. (d) Example locus of mutant-enriched ER-binding. All tracks are normalized to the same scale. ULK2 gene expression is shown on the right. (e) Distance from mutant-specific genes to mutant-enriched and -depleted ERBS is shown as a cumulative distribution.

Figure 5.. ER mutant cells exhibit large-scale alterations in chromatin accessibility.

(a) Heatmap displays ATAC-seq signal of T-47D mutant-enriched and -depleted chromatin accessibility sites. (b) Example locus of shared mutant-enriched accessible chromatin region is shown. Arrow indicates the TSS of CEP78. All tracks are normalized to the same scale. CEP78 gene expression is shown on the right. (c) Cumulative distribution graphs show distance from Y537S mutant-specific genes to Y537S mutant-enriched accessible chromatin. (d) Percent of mutant-enriched or -depleted accessible chromatin regions that also exhibit ER binding is shown.

Figure 6.. Transcription factors are observed at T-47D mutant-enriched accessible chromatin regions.

(a) Enriched transcription factor DNA binding motifs are found at T-47D mutant-enriched accessible chromatin. (b) Gene expression is shown for CTCF, FOX family genes, and POU family genes. Bars represent RNA-seq normalized counts ± SEM for two clones for each genotype and each treatment. (c) Heatmaps display the intensity (depth) of accessible chromatin (red) or DNA-binding of three factors: OCT1 (yellow), CTCF (green), and FOXA1 (blue) at mutant-enriched accessible chromatin regions. qPCR analysis of mutant-specific genes is shown after 72 hours of siRNA-mediated knockdown of OCT1 (d) or CTCF (e). Bars represent average expression ± SEM of two replicates for two clones for each genotype and each treatment. Student’s two sample t-test was used: *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001, and n.s. = not significant.