Estrogen Receptor β-Mediated Inhibition of Actin-Based Cell Migration Suppresses Metastasis of Inflammatory Breast Cancer

Abstract

Inflammatory breast cancer (IBC) is a highly metastatic breast carcinoma with high frequency of estrogen receptor α (ERα) negativity. Here we explored the role of the second ER subtype, ERβ, and report expression in IBC tumors and its correlation with reduced metastasis. Ablation of ERβ in IBC cells promoted cell migration and activated gene networks that control actin reorganization, including G-protein-coupled receptors and downstream effectors that activate Rho GTPases. Analysis of preclinical mouse models of IBC revealed decreased metastasis of IBC tumors when ERβ was expressed or activated by chemical agonists. Our findings support a tumor-suppressive role of ERβ by demonstrating the ability of the receptor to inhibit dissemination of IBC cells and prevent metastasis. On the basis of these findings, we propose ERβ as a potentially novel biomarker and therapeutic target that can inhibit IBC metastasis and reduce its associated mortality. SIGNIFICANCE: These findings demonstrate the capacity of ERβ to elicit antimetastatic effects in highly aggressive inflammatory breast cancer and propose ERβ and the identified associated genes as potential therapeutic targets in this disease.

Figure 1.. ERβ expression is inversely associated with metastasis in IBC patients.

(A) Representative staining of ERβ in two cases of IBC tumors (scale bar, 100 μm). Images on the right display enlargement of cancer epithelium with ERβ-positive (black arrows) and negative cells (white arrows) (scale bar, 50 μm). (B) Kaplan-Meier estimates of metastasis-free survival (MFS) of patients with IBC (n=37). The plot compares groups of patients with high (score >8) and low (score ≤8) ERβ protein levels. Marks represent censored data. P-value refers to two-sided log-rank test. (C) Plot illustrates comparison of tumor mRNA levels of ERβ (ESR2) between IBC and non-IBC patients using two-tailed Student’s t-test. (D-E) Kaplan-Meier estimates of overall survival (OS) of patients with IBC (n=25) and non-IBC (n=58). Comparison was made between the group of patients with high (>2.68) and low (≤2.68) ESR2 expression. P-value was generated by two-sided log-rank test.

Figure 2.. ERβ is associated with less migratory morphology in IBC cells.

(A) ERβ levels in breast cancer cell lines. The expression was quantified and normalized to Actin-β. Relative band intensities were normalized to that of H1299 cells stably expressing ERβ that is set to 1. Recombinant short ERβ protein (477 amino acids (aa), 53.4 kDa vs. 530 aa of full length ERβ, 59 kDa) was loaded to demonstrate specificity of the ERβ antibody. (B) Morphology of IBC cell lines with varying ERβ expression. The more migratory morphology the cells have the less ERβ express (scale bar, 100 μm). (C) Protein levels of ERβ in control and different ERβ knockout clones. (D) ERβ levels in control, ERβ overexpressing, ERβKO knockout and ERβKO knockout after stable re-expression of ERβ KPL4 cells. (E) Morphology of KPL4 cells with different ERβ levels (scale bar, 100 μm). (F) ERβ levels and morphology of FC-IBC02 cells following knockdown of ERβ with siRNA pools (scale bar, 100 μm). Arrows indicate migratory cells.

Figure 3.. ERβ inhibits migration and decreases the invasiveness of IBC cells.

(A) Wound healing was assessed in KPL4 cells with different ERβ levels in IncuCyte live-cell imaging system. Wound healing assay was performed twice independently with similar results. The graph shows the wound area in different times following wound formation. (B) Invasion was assessed in Transwell chambers. Representative images of the cells that invaded to the lower compartment of the chamber are shown (scale bar, 200 μm). Cells were quantified in five independent fields and the graph indicates the mean (cell number per field) ± s.d. of three independent experiments, two-tailed Student’s t-test. (C) Invasion of control and ERβKO cells following treatment with vehicle (DMSO) or different concentrations of the ERβ agonist LY500307 for 16 hours. Graph indicates the mean of invaded cells per field ± s.d. of three independent experiments, two-tailed Student’s t-test. (D) FC-IBC02 cells were treated with control or ERβ-specific siRNA pools for 48 hours. Cells were then seeded in Transwell chambers and allowed to invade for 16 hours. Representative images of invaded cells are shown (scale bar, 100 μm). Graph illustrates the mean of cell number per field ± s.d. of three independent experiments, two-tailed Student’s t-test.

Figure 4.. ERβ inhibits metastasis of IBC tumors.

(A) 5X10⁵ control, ERβKO and ERβKO+ERβ KPL4 cells expressing firefly luciferase and GFP were delivered in NCG mice by tail-vein injection. Whole-body bioluminescence images are shown. Graph indicates total photon flux per second from lungs in vivo at the endpoint that was normalized to the signal at the day of cell injection ± s.d., two-tailed Student’s t-test. (B) Ex vivo bioluminescence images of lungs from NCG mice that were intravenously injected with KPL4 cells with different ERβ levels. (C) Representative H&E images of lungs of NCG mice that were injected through tail-vein with control and ERβKO KPL4 cells. Encircled tissue contains a metastatic cell aggregate (scale bar, 50 μm). Graph shows the mean of lung metastatic cell aggregates that are more than 5 cells large and counted in 15 randomly microscopic fields ± s.d., two-tailed Student’s t-test (n=3). (D) Firefly luciferase- and GFP-expressing (5X10⁵) control, ERβKO and ERβKO+ERβ KPL4 cells were injected into mammary fat pad of NCG mice. In vivo bioluminescence signal from lungs of indicated groups is shown and the graph presents the mean of total photon flux per second ± s.d., two-tailed Student’s t-test. (E) Ex vivo images of lung metastasis from mice with orthotopically injected cells using bioluminescence imaging. (F) Fluorescence images of bone sections. GFP-expressing tumor cells that were injected into the fat pad of NCG mice infiltrated the bone marrow (blue, nuclei of bone marrow cells) (scale bar, 50 μm). Graph indicates the mean of GFP-positive KPL4 cells per field ± s.d., two-tailed Student’s t-test. (G) Ex vivo lung metastasis in ovariectomized NCG mice 7 weeks following orthotopic injection with control KPL4 cells and pellets containing vehicle or LY500307. Graph depicts the mean of total photon flux per second ± s.d., two-tailed Student’s t-test.

Figure 5.. ERβ regulates genes that control cytoskeleton reorganization.

(A) Pie chart showing the number of ERβ-regulated genes in ERβKO KPL4 cells. (B) Graph of enriched GO terms across input gene lists with p-values. Of note, the third top-ranked GO term is associated with response to steroid hormone due to knockout of ERβ. (C) Heat map illustrating expression levels of up-regulated and down-regulated genes annotated to the enriched GO term that is associated with regulation of cell migration. (D) Control, ERβKO and ERβKO+ERβ KPL4 cells were stained with FITC-phalloidin to detect F-actin stress fibers (green) and DAPI for nuclei. Cell area was quantified based on F-actin staining. Data are shown as mean ± s.d., n=50 cells, Mann-Whitney t-test. (E) KPL4 cells with different ERβ levels were stained with FITC-phalloidin, anti-vinculin antibody for focal adhesions and DAPI. The number of focal adhesions per cell was determined based on vinculin staining. Data are shown as mean ± s.d., n=50 cells, Mann-Whitney t-test. (F) Active (GTP-loaded) RhoC that interacts with the GST-fused Rho binding domain of Rhotekin in KPL4 cells with different ERβ levels. (G) Active RhoC in KPL4 cells following treatment with the ERβ agonist LY500307 for 24 hours in estrogen-depleted media. (H) Following treatment with control or RhoC siRNA pools ERβKO cells were seeded in Transwell chambers and allowed to invade for 16 hours. Representative images of invaded cells are shown. Graph depicts the mean of invaded cells per field ± s.d. of three independent experiments, two-tailed Student’s t-test.

Figure 6.. GRP141 and ELMO1 are essential for the migratory phenotype of ERβKO cells.

(A) mRNA expression of GPR141 and GPR30 in control, two clones of ERβKO and ERβKO+ERβ KPL4 cells. Values are normalized to control cells that are set to 1 and represent the mean ± s.d. of at least three independent experiments, two-tailed Student’s t-test. (B) mRNA levels of ELMO1 in KPL4 cells with different ERβ levels. Graph depicts the mean ± s.d. of three experiments, two-tailed Student’s t-test. (C) mRNA of ELMO1, GPR141 and ERβ in FC-IBC02 cells following knockdown of ERβ with siRNA pools. Expression is shown as mean ± s.d., two-tailed Student’s t-test. (D) Protein levels of GPR141, ELMO1, ROCK1 and ERβ in KPL4 cells with different ERβ levels. The band intensities of GPR141 and ELMO1 were quantified and normalized to Actin-β. (E-G) Morphology and invasion of ERβKO KPL4 cells following treatment with siRNA against GPR141 or ELMO1 or 0.2 μg/ml pertussis toxin (PTX) in medium for 6 hours (scale bar, 100 μm). Graphs illustrate the % of migratory cells per field (mean from at least 3 different fields) (left) and the mean of invaded cells per field (right) ± s.d. of three independent experiments, two-tailed Student’s t-test. (H) Morphology and invasion of control cells stably transfected with empty vector or ELMO1 recombinant plasmid (scale bar, 100 μm). Levels of ELMO1 in control and ELMO1-overexpressing KPL4 cells are also indicated. Graphs show the % of migratory cells per field and the mean of invaded cells per field ± s.d., two-tailed Student’s t-test. (I) Levels of active RhoC in control and ELMO1-overexpressing KPL4 cells, and in ERβKO KPL4 cells after treatment with siRNA against GPR141, ELMO1, HER3 or 0.2 μg/ml PTX for 6 hours.

Figure 7.. ERβ directly regulates the expression of GPR141 and ELMO1 that are selectively expressed in IBC cells.

(A) Map illustrating the location and orientation of GPR141 and ELMO1 on chromosome 7 and their ERE-containing regulatory sites. (B) ERβ ChIP followed by qPCR at the GPR141 and ELMO1 regulatory elements in ERβKO and ERβKO+FLAG-tagged ERβ KPL4 cells following pull-down with an anti-FLAG antibody (n= 3 technical replicates). ChIP was performed twice independently with similar results. (C) ChIP-qPCR analysis showing interaction of ERβ with regulatory elements of GPR141 and ELMO1 as well as a previously validated ERβ-specific binding site that was served as positive control (POS) following treatment with E2 or LY500307 in estrogen-depleted media. (D) Luciferase assay showing expression of luciferase in control and ERβKO cells that is driven by GPR141 regulatory region containing wild-type, mutated or no ERβ-binding site (Mut, mutated; Del, deleted). Data are represented as mean ± s.d., two-tailed Student’s t-test. (E) ELMO1 and GPR141 protein levels in IBC and non-IBC breast cancer cell lines and normal mammary epithelial MCF-10A cells. (F) mRNA levels of GPR141 and ELMO1 in IBC and non-IBC breast cancer cell lines. Values are normalized to the expression of MCF-7 cells and represent the mean ± s.d., two-tailed Student’s t-test. (G) ELMO1 and RhoC protein levels in IBC cell lines. (H) Schematic illustration of the mechanism employed by ERβ to inhibit metastasis in IBC.