Estrogen receptor silencing induces epithelial to mesenchymal transition in human breast cancer cells

Abstract

We propose the hypothesis that loss of estrogen receptor function which leads to endocrine resistance in breast cancer, also results in trans-differentiation from an epithelial to a mesenchymal phenotype that is responsible for increased aggressiveness and metastatic propensity. siRNA mediated silencing of the estrogen receptor in MCF7 breast cancer cells resulted in estrogen/tamoxifen resistant cells (pII) with altered morphology, increased motility with rearrangement and switch from a keratin/actin to a vimentin based cytoskeleton, and ability to invade simulated components of the extracellular matrix. Phenotypic profiling using an Affymetrix Human Genome U133 plus 2.0 GeneChip indicated geometric fold changes ≥ 3 in approximately 2500 identifiable unique sequences, with about 1270 of these being up-regulated in pII cells. Changes were associated with genes whose products are involved in cell motility, loss of cellular adhesion and interaction with the extracellular matrix. Selective analysis of the data also showed a shift from luminal to basal cell markers and increased expression of a wide spectrum of genes normally associated with mesenchymal characteristics, with consequent loss of epithelial specific markers. Over-expression of several peptide growth factors and their receptors are indicative of an increased contribution to the higher proliferative rates of pII cells as well as aiding their potential for metastatic activity. Signalling molecules that have been identified as key transcriptional drivers of epithelial to mesenchymal transition were also found to be elevated in pII cells. These data support our hypothesis that induced loss of estrogen receptor in previously estrogen/antiestrogen sensitive cells is a trigger for the concomitant loss of endocrine dependence and onset of a series of possibly parallel events that changes the cell from an epithelial to a mesenchymal type. Inhibition of this transition through targeting of specific mediators may offer a useful supplementary strategy to circumvent the effects of loss of endocrine sensitivity.

Figure 1. Morphology and growth rates of MCF7 and pII cells.

Panels A and B: Phase-contrast optical photomicrographs of single colonies of MCF7 and pII respectively. Scale bar represents 100 µm. Panel C: Cell shape analysis was performed using an ImageJ freeware package; histograms show mean cell parameter ± SEM of 4 independent determinations quantifying morphological differences between MCF7 and pII cells (p < 0.04). Panel D: Equivalent numbers of MCF7 and pII cells (10⁴) were seeded into 12-well plates and grown over 6 days. Cells were harvested from triplicate wells on the days indicated and growth determined by cell counting. Differences at days 4, 5 and 6 were significant with p≤0.0001.

Figure 2. Motility assay of MCF7, pII and MDA231.

Cells were plated into individual wells of a 6 well microwell dish and allowed to grow to ∼80% confluency. A scratch was made across the centre of the cell monolayer with a yellow eppendorf pipette tip (as shown in cartoon) creating a parallel space. Damaged cells were removed by carefully aspirating off the medium and rinsing with PBS before addition of fresh medium containing 2% serum. Upper panels show comparative microscopic view of MCF7 (a–b) and pII (c–d) cells in the scratched area at 0 time (a, c) and after 24 h [b, d] for a typical experiment. Chart shows the combined quantitated results from 3 independent experiments for the cell lines indicated. Histobars represent the mean [± standard deviation] width of the original scratch space at 0, 24 and 48 h. Wound closure indicates movement of cells into the scratch. *Significantly different from MCF7 cells at respective times with p≤0.006.

Figure 3. Confocal laser scanning microscope analysis of individual MCF7 (A) and pII (B) cells showing differences in both intensity and arrangement of F actin filaments visualised by phalloidin staining as described in Methods.

Microspikes and lamellipodia-like structures are visible in the pII cells. Both photographs were enhanced equally by auto-contrast in Adobe Photoshop. Scale bar indicates 10 µm.

Figure 4. Invasion assay of MCF7, E2, pII and MDA231 cells through simulated ECM protein components.

Cells were plated into the upper chambers of the cell invasion plate and incubated for 48 h prior to measurement of the fluorescence intensity of the invading cells in the bottom chambers. MCF7 and E2 cells were considered non-invasive (<20 FU/m), whereas pII and MDA231 cells progressively invaded the BME (>100 FU/m). Histobars represent the mean ± SEM of at least 3 independent determinations of the fluorescent intensity of the invading cells in the bottom chamber. This experiment was performed three times with similar outcome. pII and MDA231 significantly different from MCF7 with p<0.0001.

Figure 5. Agarose spot assay.

Panel A is a cartoon to illustrate agarose spot inside well with cells added some of which settle around the periphery. MCF7 (B) and pII (C, D) cells were plated into individual wells of a 12 well dish containing an agarose spot, and photographed after 24 h of incubation. MCF7 cannot penetrate even at high density whereas pII do so even at very low density (D). Arrows delineate the periphery of the agarose spot beneath. Scale bar represents 400 µm. This experiment was repeated at least ten times with identical outcome. The numbers of pII cells entering the agarose varied with density of cells settling at the periphery, and time.

Figure 6. Biological processes affected by ER silencing.

GO terms most significantly affected or disrupted secondary to ER silencing. Each GO term is boxed in a rectangle and shaded according to the p value as indicated. Enrichment (N, B, n, b), where indicated is defined as (b/n)/(B/N), where N - is the total number of genes, B - is total number of genes associated with a specific GO term (‘target’ set and ‘background set’), n - is number of genes in the ‘target set’, b - is number of genes in the ‘target set’ associated with a specific GO term.

Figure 7. Cellular components affected by ER silencing. See legend to Figure 6.

Figure 8. Realtime PCR analysis of epithelial and mesenchymal markers in MCF7, YS1.2 and YS2.5 cells.

Extracted RNA was converted to cDNA and amplified using the Taqman procedure as described in Methods. Histograms represent means of duplicate determinations. Similar results were obtained in two separate experiments. Expression was normalized to the level found in MCF7 cells (arbitrarily fixed as 100), with β actin used as an internal control. Photomicrograph shows the morphology of the two original adjacent G418 resistant colonies from which YS1.2 and YS2.5 were cloned. Scale bar represents 100 µm.