Hypoxia and ERα Transcriptional Crosstalk Is Associated with Endocrine Resistance in Breast Cancer

Abstract

Estrogen receptor-alpha (ERα) is the driving transcription factor in 70% of breast cancers and its activity is associated with hormone dependent tumor cell proliferation and survival. Given the recurrence of hormone resistant relapses, understanding the etiological factors fueling resistance is of major clinical interest. Hypoxia, a frequent feature of the solid tumor microenvironment, has been described to promote endocrine resistance by triggering ERα down-regulation in both in vitro and in vivo models. Yet, the consequences of hypoxia on ERα genomic activity remain largely elusive. In the present study, transcriptomic analysis shows that hypoxia regulates a fraction of ERα target genes, underlying an important regulatory overlap between hypoxic and estrogenic signaling. This gene expression reprogramming is associated with a massive reorganization of ERα cistrome, highlighted by a massive loss of ERα binding sites. Profiling of enhancer acetylation revealed a hormone independent enhancer activation at the vicinity of genes harboring hypoxia inducible factor (HIFα) binding sites, the major transcription factors governing hypoxic adaptation. This activation counterbalances the loss of ERα and sustains hormone-independent gene expression. We describe hypoxia in luminal ERα (+) breast cancer as a key factor interfering with endocrine therapies, associated with poor clinical prognosis in breast cancer patients.

Keywords: breast cancer; endocrine resistance; estrogen receptor alpha; hypoxia.

Figure 1

Effect of hypoxia on ERα expression, proliferation and survival of MCF7 cells. (A) Western blot showing ERα protein abundance after 1-month exposure to hypoxia inducer CoCl2, in the presence or absence of 10 nM of E2 for 24 h. For proteasome inhibition, cells were treated for 8 h with 2.5 µM of MG-132. ERK1/2 was used as the normalizing control. pAKT Serine 473 and total AKT abundance were also measured. (B) Proliferation count of MCF7 cells treated for 6 days with EtOH, 10 nM E2 or 1 µM 4OHT, in presence of 2% of FBS steroid-free. Values are expressed in fold change compared to the number of seeded cells at the beginning of the treatment. (C) Cell cycle analysis using FACS using propidium iodide. MCF7 cells were treated for 48 h with EtOH, 10 nM E2 or 1 µM 4-OHT. (D) TUNEL assay showing percentage of cell death between control and CoCl2-treated cells, in presence or absence of 10 nM of E2 for 72 h. * p-value < 0.05 and *** p-value < 0.001 with a Mann–Whitney test for comparisons against the control. ### p-value < 0.05 with a Mann–Whitney test for comparisons against CoCl2 treatment.

Figure 2

Transcriptomic analysis of the E2 response under hypoxic stress. (A) Venn diagram representing E2-regulated genes both in control and CoCl2-treated MCF7 cells (FC > 1.8–FDR < 0.05). (B) Scatter plot showing the Log2FC upon E2 treatment for each gene both in control and CoCl2-treated MCF7 cells. (C) Heatmap showing supervised clustering (k-means method) of total E2-regulated genes both in control and CoCl2-treated MCF7 cells (FC > 1.8–FDR < 0.05). Control and CoCl2-treated MCF7 cells were treated with EtOH or 10 nM of E2. Transcriptomes from four independent biological samples were analyzed by microarray. Graphs on the left represent the average expression value (Log2) of all genes for each cluster. The number of genes in each cluster are indicated. (D) RT-qPCR experiments of representative genes validating the expression profiles found in the six different clusters, * p-value < 0.05 and ** p-value < 0.01 with a Mann–Whitney test for comparisons against the control. Diagram representing E2-regulated genes both in control and CoCl2-treated MCF7 cells (FC 1.8–FDR 0.05).

Figure 3

Genome-wide reprogramming of ERα cistrome under hypoxic stress. (A,B) Venn diagrams showing the overlaps of genomic regions bound by Erα detected by ChIP-seq, both in control or CoCl2-treated MCF7 cells, following a 50 min treatment with E2 or EtOH. The Erα ChIP-seq signal was aligned and averaged within a −2/+2 kbp window centered on ERBS belonging to the 3 categories in the upper Venn diagrams. (C) The stacked histogram illustrates the percentage of conserved, lost and gained ERBSs according to each cluster. (D) Graphs representing the mean of ERα ChIP-seq signal obtained at the center of the conserved ERBSs located at the vicinity of the TSSs of the genes clustered in C4, C5 and C6 obtained from transcriptomic analysis. (E) Heatmap of ERα targeted CHIP-qPCR experiments on conserved ERBSs at the vicinity of genes belonging to the clusters C4, C5 and C6. Data shown are the mean of relative enrichment of five independent experiments normalized to an internal control. (F) ERα and H3K27Ac CHIP experiments on CoCl2-specific ERBSs. Data shown are the mean of relative enrichment of five independent experiments normalized to an internal control. ** p-value < 0.01 with a Mann–Whitney test for comparisons against the control. # p-value < 0.05 with a Mann–Whitney test for comparisons against CoCl2 treatment.

Figure 4

HIF1α regulates a subset of ERα target genes and activates ERα-bound enhancers. (A) Heatmap of H3K27Ac targeted CHIP-qPCR experiments on enhancers located at the vicinity of genes belonging the clusters C4, C5 and C6. Data shown are the mean of relative enrichment of five independent experiments normalized to an internal control. (B) Analysis of transcription factor binding enrichment using ChEA algorithm on genes belonging to the clusters 3 and 6. (C,D) Venn diagram showing overlap between the 584 hormone-induced differentially expressed genes and the respective HIF1α and HIF2α bound genes in MCF7 as determined from (Mole et al., [33], using GREAT online tool and their respective distribution across the six different gene clusters. (E) Bar graphs showing mRNA expression of ERα representative target genes belonging to clusters 4 and 6, both in control or CoCl2-treated MCF7 cells, treated with control or HIF1α/ HIF2α targeting siRNAs. qPCRs for HIF1α and HIF2α mRNA silencing are shown as the control. Values represent the mean +/− SEM of three experiments and are expressed as fold induction compared to untreated cells (Cont, EtOH) transfected with control Si RNA (* p < 0.05, student′s t-test).

Figure 5

Hypoxia prevents a tamoxifen inhibitory effect on a subset of ERα target genes associated with poor survival prognostic of patients with luminal breast cancers. (A) RT-qPCR shows CXCL12, PGR, GREB1 and AREG expression level after 24 h exposure to CoCl2, in presence or absence of 10 nM of E2 and/or 1 μM 4-hydroxytamoxifen (4-OHT) for 24 h. Values represent the mean +/− SD of triplicate and are expressed as a fold induction as compared to untreated cells. (** p < 0.05, student’s t-test). (B) Histopathological features and ERα status measured by immunohistochemistry of the METABRIC cohort (n = 1980 patients) extracted from cBioportal. (C) Prognostic values of the different gene signatures (Cluster C1, C3, C4, C6) corresponding to E2 and hypoxia regulated genes identified in Figure 2, in the METABRIC cohort (mRNA expression z-scores relative to diploid samples z-score = 2, n = 1980). (D) HIF1α mRNA expression prediction value regarding overall survival (OS) and relapse free survival (RFS) in the METABRIC cohort (mRNA expression z-scores relative to diploid samples, z-score = 1.5, n = 1980). (E) HIF2α mRNA expression prediction value regarding overall survival (OS) and relapse free survival (RFS) in the METABRIC cohort (mRNA expression z-scores relative to diploid samples, z-score = 1.5, n = 1980). (F) HIF1α and HIF2α mRNA expression prediction value regarding overall survival (OS) in ERα (+) patients from the Breast Cancer Kaplan–Meier Plotter resource (Auto-select cutoff, n = 754). (G) HIF1α and HIF2α mRNA expression prediction value regarding overall survival (OS) in ERα (−) patients from the Breast Cancer Kaplan–Meier Plotter resource (Auto-select cutoff, n = 520).