Journal list menu

Volume 289, Issue 6 p. 1552-1574
Original Article
Free Access

EGFR is a pivotal player of the E2/ERβ – mediated functional properties, aggressiveness, and stemness in triple-negative breast cancer cells

Konstantina Kyriakopoulou

Konstantina Kyriakopoulou

Biochemistry, Biochemical Analysis & Matrix Pathobiology Research Group, Laboratory of Biochemistry, Department of Chemistry, University of Patras, Greece

Search for more papers by this author
Elena Kefali

Elena Kefali

Biochemistry, Biochemical Analysis & Matrix Pathobiology Research Group, Laboratory of Biochemistry, Department of Chemistry, University of Patras, Greece

Search for more papers by this author
Zoi Piperigkou

Zoi Piperigkou

Biochemistry, Biochemical Analysis & Matrix Pathobiology Research Group, Laboratory of Biochemistry, Department of Chemistry, University of Patras, Greece

Foundation for Research and Technology-Hellas (FORTH)/Institute of Chemical Engineering Sciences (ICE-HT), Patras, Greece

Search for more papers by this author
Christoph Riethmüller

Christoph Riethmüller

Serendip GmbH, Center for Nanotechnology, Münster, Germany

Search for more papers by this author
Burkhard Greve

Burkhard Greve

Department of Radiotherapy–Radiooncology, University Hospital Münster, Germany

Search for more papers by this author
Marco Franchi

Marco Franchi

Department for Life Quality Study, University of Bologna, Rimini, Italy

Search for more papers by this author
Martin Götte

Martin Götte

Department of Gynecology and Obstetrics, University Hospital Münster, Germany

Search for more papers by this author
Nikos K. Karamanos

Corresponding Author

Nikos K. Karamanos

Biochemistry, Biochemical Analysis & Matrix Pathobiology Research Group, Laboratory of Biochemistry, Department of Chemistry, University of Patras, Greece

Correspondence

N. K. Karamanos, Biochemistry, Biochemical Analysis & Matrix Pathobiology Research Group, Laboratory of Biochemistry, Department of Chemistry, University of Patras, 26504 Patras, Greece

Tel: +30 2610997915

E-mail: [email protected]

Search for more papers by this author
First published: 19 October 2021
Citations: 6

Abstract

Triple-negative breast cancer (TNBC) is defined by aggressive behavior, limited response to chemotherapy and lower overall survival rates. The increased metastatic potential of TNBC is a combined result of extensive extracellular matrix (ECM) remodeling that leads to cytoskeleton rearrangement and activation of epithelial-to-mesenchymal transition (EMT). The overexpression of epidermal growth factor receptor (EGFR) in TNBC tumors has been linked to induced expression of EMT-related molecules. EMT activation has often been associated with increased metastasis and stemness. Recently, we described the crucial role of EGFR/estrogen receptor beta (ERβ) interplay in the regulation of invasion and cell–matrix interactions. In this study, we report on the EGFR-ERβ functional relationship in connection to the aggressiveness and cancer stem cell (CSC)-like characteristics of TNBC cells. ERβ-suppressed and MDA-MB-231 cells were subjected to downstream EGFR inhibition and/or estradiol stimulation to assess alterations in functional parameters as well as in morphological characteristics, studied by scanning electron, atomic force, and immunofluorescence microscopies. Moreover, the expression and localization of key EMT and CSC-related markers were also evaluated by real-time qPCR, immunofluorescence microscopy, and flow cytometry. EGFR inhibition resulted in an overall suppression of aggressive functional characteristics, which occurred in an ERβ-mediated manner. These changes could be attributed to a reduction, at the molecular level, of EMT and stemness-linked markers, most notably reduced expression of Notch signaling constituents and the cell surface proteoglycan, syndecan-1. Collectively, our study highlights the importance of EGFR signaling as a key effector of aggressiveness, EMT, and stemness in an ERβ-dependent way in TNBC.

Abbreviations

  • AFM
  • atomic force microscopy
  • CM
  • conditioned medium
  • CSC
  • cancer stem cells
  • E2
  • 17β-estradiol
  • ECM
  • extracellular matrix
  • EGFR
  • epidermal growth factor receptor
  • EMT
  • epithelial-to-mesenchymal transition
  • ER
  • estrogen receptor
  • GF
  • growth factor
  • HSPG
  • heparan sulfate proteoglycan
  • HUVECs
  • human umbilical vein endothelial cells
  • IF
  • immunofluorescence
  • MET
  • mesenchymal-to-epithelialtransition
  • MMPs
  • matrix metalloproteinases
  • PG
  • proteoglycan
  • RTK
  • receptor tyrosine kinase
  • SDC
  • syndecan
  • SEM
  • scanning electron microscopy
  • TF
  • transcription factor
  • TNBC
  • triple-negative breast cancer
  • Introduction

    Breast cancer is considered the most frequent form diagnosed in women and the second leading cause of cancer-related death in women after lung cancer [[1, 2]]. Given the complexity and heterogeneity of breast cancer tumors, in the last two decades there have been major treatment advances, such as targeted and immunotherapeutic interventions contributing to decreased mortality rates. However, the more aggressive types of breast carcinomas, for example, the triple negative (ERα-/PgR-/HER2-), have been associated with chemoresistance, distant-stage (metastatic) classification, and poor survival [[3]]. Metastasis involves intimate interactions of cancer cells with the extracellular matrix (ECM) of the surrounding stroma, which drive autonomous changes in cell motility affecting tumor progression [[4, 5]]. Modern approaches to understand the underlying ECM-driven mechanisms that are related to breast cancer aggressiveness, epithelial-to-mesenchymal transition (EMT), and stemness could help to a better diagnosis and treatment of triple-negative breast cancer (TNBC).

    ECM, an extensively studied network of macromolecules that surrounds and supports cells and tissues, has been established as a key regulator of cell processes and interactions [[6]]. Major matrix constituents, including heparan sulfate proteoglycans (HSPGs), participate in a vast array of cell functions such as migration, adhesion and invasion, to advance tumor progression [[7-9]]. In addition, the hyaluronan receptor CD44 and key enzymes, like matrix metalloproteinases (MMPs), modulate EMT-advancing signaling, expression of key cell adhesion molecules, and ECM remodeling in a feedback loop [[10-12]]. Moreover, ECM reorganization and EMT establishment are highly influenced by receptor tyrosine kinases (RTKs)-mediated signaling pathways that affect the metastatic potential of breast cancer cells [[13, 14]]. In particular, it is well established that the activation of epidermal growth factor receptor (EGFR) transmits extracellular mitogenic signals through the endocytosis of the receptor, activation of several downstream signaling cascades, and the subsequent induction of EMT, autophagy, and cell survival [[15]].

    EMT is a complex dynamic process that occurs normally during embryonic development but also drives cancer progression [[16]]. EMT is considered as one of the metastatic cascade’s driving forces as it bestows tumor cells the proper traits necessary for tumor dissemination and distant metastasis, namely, increased motility, invasion, intravasation, and extravasation [[17, 18]]. Still, for metastatic colonization to occur, cells mostly follow the reverse to EMT process, namely, mesenchymal-to-epithelial transition (MET), in order to proliferate and form the secondary tumor [[19]]. During EMT, epithelial cells that present apical-basal polarity and tight cell-cell junctions, shift toward the mesenchymal state resulting in a reorganized cytoskeleton, front-back polarity, and expression of distinct molecules and transcription factors (TFs) such as vimentin, fibronectin, ZEB 1/2, and SNAIL 1/2 [[20-22]].

    EMT program activation has also been implicated in the acquisition of cancer stem cell (CSC) properties, even though the mechanistic link between EMT and CSC phenotype is not yet fully elucidated [[17, 23]]. CSCs have been identified in most carcinomas and usually represent a minor subpopulation of breast cancer cells with increased plasticity and tumor-initiating potential that can self-renew, differentiate, establish metastatic colonies, and, in general, promote relapse [[24, 25]]. Changes in the tumor microenvironment, that usually drive EMT, result in neighboring niches that, via secretion of paracrine factors (growth factors, cytokines, hormones, exosomes, etc.), modulate ECM and activate signaling pathways (most importantly Notch and Hedgehog) and autocrine loops (i.e., Wnt, TGFβ) implicated in stemness regulation [[26, 27]]. The identification of CSC-specific cell surface markers is a challenging process. In breast cancer, CSCs are mainly identified by the expression of surface markers CD44 and CD24, but also EMT-inducing proteins and TFs such as NOTCH1, OCT4, HEY1, and MUSASHI1 (MSI1) [[28, 29]].

    Estrogens and their receptors are of high importance for the progression of breast carcinoma. ERs can reside in the cytoplasm and nucleus, but also in the cell membrane upon binding of their main ligand, 17β-estradiol (E2) [[30]]. ERs can signal either through their genomic or nongenomic actions, in a ligand-dependent or ligand-independent way, respectively [[31]]. Notably, through their crosstalk with major ECM biomolecules, ERs contribute to ECM remodeling and EMT, and hence, they critically orchestrate various cellular processes, such as migration, adhesion to matrix substrates, and angiogenesis [[32-34]]. Furthermore, in the highly aggressive TNBC, EGFR has also been proven to be a significant prognostic factor, as its overexpression leads to aberrant EGF/EGFR signaling that induces EMT and promotes the invasive potential of cancer cells [[35]].

    Our group has shown that ERβ fine tunes the mRNA and protein expression of several matrix partners and EMT markers, to alter ECM composition, guide cell behavior, and cause changes in the morphological characteristics of the highly aggressive MDA-MB-231 cells [[33, 36]]. In addition, the EGFR/ERβ axis determines the morphological characteristics while mediating the adhesion and invasion potential of TNBC cells [[37]]. The significant role of ERβ in TNBC cells together with its established signature in ECM remodeling and interplay with EGFR, prompted us to take a step forward and assess the role of EGFR in the functional properties (i.e., migration, proliferation, angiogenesis) of TNBC cells with different ERβ status (ERβ+ and shERβ MDA-MB-231 cells) in the presence and absence of E2. In this study, we also examined the functional relationship between EGFR and E2/ERβ pathways in terms of EMT initiation and the increased stemness in TNBC cells.

    Results

    EGFR modulates major functional properties of TNBC cells

    ERs and EGFR are pivotal players of critical biological processes including proliferation and migration. Therefore, we evaluated the role of EGFR in the moderation of these functional cell properties in the presence and absence of E2. In the aggressive MDA-MB-231 cells, proliferation compared to control was significantly lower following EGFR inhibition (Fig. 1A), indicating the importance of EGFR function in sustaining cell growth. E2 upregulated cell growth, though only when EGFR was active and not in the presence of its inhibitor, AG1478, demonstrating the interdependence of their actions on cell proliferation. Besides, this underlines the importance of the active EGFR for the action of E2 in the ERβ+ MDA-MB-231 cells. Similarly, ERβ suppression in the MDA-MB-231 cells leads to 25% less cell growth, which was further reduced by 25% upon EGFR inhibition but notably it was not affected by E2 (Fig. 1A), hence supporting the findings that E2 action requires the ERβ-EGFR interplay. These results were consistent with data we obtained from flow cytometric cell cycle analysis (Table 1).

    Details are in the caption following the image
    EGFR promotes alterations in functional properties of triple-negative breast cancer cells. (A) Inhibition of EGFR attenuates cell proliferation in both MDA-MB-231 and shERβ MDA-MB-231 cells, while the effects of E2 depend on cells' ERβ status. (Β) Cell motility is downregulated following EGFR inhibition in both cell lines, whereas E2 addition induces migration in the ERβ + MDA-MB-231 cells. (C) Representative images of scratch wounds in 0 h and 24 h used to evaluate cell migration (scale bar = 20 μm). Each bar represents mean ± SD values from triplicate samples. Statistically significant differences are indicated accordingly: *P < 0.05, **P < 0.01 compared to MDA-MB-231 control cells; #P < 0.05 and ##P < 0.01 compared to shERβ MDA-MB-231 control cells.
    Table 1. Cell cycle analysis of MDA-MB-231 and shERβ MDA-MB-231 cells by flow cytometry. Cell cycle is impaired by both AG1478 and E2 treatments not only in ERβ+ but also in shERβ MDA-MB-231 cells.
    Condition Cell cycle phase (% of population)
    G1 S G2/M
    MDA-MB-231
    Control 80 11 9
    AG1478 87 6 7
    E2 76 15 9
    AG1478 + E2 87 6 7

    shERβ

    MDA-MB-231

    Control 63.1 21.7 15.2
    AG1478 63 16 21
    E2 64 21 15
    AG1478 + E2 66 18 16

    The effect of EGFR inhibition was prominent in the migration of breast cancer cells as well, since AG1478 treatment decreased the migration of the very motile MDA-MB-231 cells ca 30%. E2 addition promoted migration in a statistically significant manner. This effect was abolished upon EGFR inhibition, suggesting a synergy between active EGFR and E2/ERβ to promote the migratory potential of ERβ+ TNBC cells (Fig. 1B,C). In the shERβ MDA-MB-231 cells that already exhibited ca 60% less motility as compared to MDA-MB-231 cells, EGFR inhibition further reduced the migratory potential by 25%, whereas E2 addition and E2-AG1478 treatment had no significant effects (Fig. 1B,C).

    Besides proliferation and motility, tumor growth also relies on the formation of new blood vessels from existing ones, a procedure well known as angiogenesis. In vitro angiogenesis can be studied by various assays, the most common being tube formation assay [[38]]. To evaluate the effects of EGFR and E2/ERβ in tubulogenesis, human umbilical vein endothelial cells (HUVECs) were seeded on top of Matrigel and cultured in the presence of conditioned medium (CM). As CM, we used the supernatants of cell cultures following 24-h treatments, and after the incubation period ended, we imaged the samples and determined tube characteristics [[39]]. As depicted in Fig. 2Α, co-culturing HUVECs with CM derived from AG1478-treated MDA-MB-231 cells resulted in a significant reduction of total branching points/nodes (ca 55%), total number of formed tubes (ca 45%), and total tube length (ca 40%). Notably, no loops were formed among the tubes. CM derived from E2-treated cells caused a slight decrease but did not have a statistically significant effect on HUVECs’ tubulogenesis as compared to control cells (Fig. 2Α). However, when HUVECs were co-cultured with CM from cells treated with both AG1478 and E2, the suppressive effect of AG1478 was sustained, implying that active EGFR has a prominent role in the recorded action of CM even after the treatment with E2. Furthermore, incubation of HUVECs with CM derived from either EGFR-inhibited or E2-treated shERβ MDA-MB-231 cells significantly altered tube formation with lower number of branching points, number of tubes and total tube length, hence suggesting that both EGFR and E2 differentially fine-tune angiogenesis upon ERβ suppression in TNBC (Fig. 2Α). Besides, further examination of major angiogenesis-related markers’ expression via real-time qPCR confirmed the findings from the tube formation assay on the role of EGFR and E2 in tubulogenesis. Specifically, the expression of VEGFR, VEGF, and TGFβ is downregulated in a statistically significant manner upon EGFR inhibition in both cell lines (Fig. 2B), while E2 addition sustains the levels of VEGFR, and VEGF compared to the MDA-MB-231 control cells but reduces the expression of TGFβ in MDA-MB-231 cells and VEGF in shERβ cells.

    Details are in the caption following the image
    EGFR is an effector of angiogenesis in triple-negative breast cancer cells. (A) HUVEC cells cultured in the presence of CM derived from AG1478-treated cells present less tube formation ability, although the presence of E2 in the CM does not exert any added effect, even when derived from ERβ+ TNBC cells (scale bar = 20 μm). (B) Expression profiles of VEGFR, VEGF, and TGFβ showcase the effects of EGFR inhibition and E2 stimulation in the angiogenesis of ERβ-suppressed and MDA-MB-231 cells. Each bar represents mean ± SD values from triplicate samples. Statistically significant differences are indicated accordingly as in Fig. 1.

    EGFR influences cell morphology and cytoskeleton network and mediates the E2/ERβ effects

    Breast cancer cells' morphological characteristics are useful indicators of the interactions that take place between cells and tumor microenvironment. We have recently showed the importance of the EGFR/ERβ axis in the adhesion and invasion potential of breast cancer cells as well as the development of specific morphological structures (filopodia and extracellular vesicles) in artificial 3D cultures using various ECM substrates [[37]]. Herein, we assessed the impact of E2 and EGFR inhibition on cell morphology and cytoskeletal organization in standard 2D cultures by scanning electron microscopy (SEM), IF microscopy and AFM. As shown in Fig. 3A, MDA-MB-231 cells (control) exhibit three different phenotypes: (a) the flattened polygonal or cobblestone cells, which are morphologically comparable to epithelial cells, with no evident filopodia or lamellipodia; (b) the spherical or globular cells, which may be involved in ameboid movement and cancer invasion on the surrounding ECM; and (c) the elongated or fusiform cells, which resemble the shape of mesenchymal cells and have the potential to penetrate and invade the surrounding tissues by a proteolytic digestion of the ECM components (Fig. 3A–a and b). The potential invasion capability of globular and elongated cells is also confirmed by microvesicles distributed on their cytoplasmic surface (thick arrows in a). Most of EGFR-inhibited MDA-MB-231 cells (Fig. 3A–c and d) are flattened polygonal or cobblestone shaped cells that appear smoother, thinner, and aggregated with tighter cell-cell contacts (twisted arrows in d) compared to the untreated cells. A minor portion of other cells displays a globular phenotype (thick arrows in d). Importantly, the number of elongated or mesenchymal cells reduced upon EGFR inhibition, which also seems to partially inhibit the presence of microvesicles (thin arrows in c and d). On the other hand, E2-treated MDA-MB-231 cells exhibit all three different phenotypes found in control samples, yet they appear to be more isolated (Fig. 3A–e and f). Moreover, there is a significant increase in the proportion of the elongated or mesenchymal shaped cells as compared with control cells (arrows in e). Some of these cells develop thin and long filopodia, which display many microvesicles at their end and along their cytoplasmic surface (arrow in f). This observation suggests that E2 treatment evoked the growth of elongated mesenchymal cells that show a microvesicles release or uptake by the described filopodia. MDA-MB-231 cells treated with both AG1478 and E2 mainly include flattened polygonal shaped cells in tight contact that exhibit very smooth surface (Fig. 3A–g and h), along with some globular shaped cells and elongated mesenchymal ones (arrows in g). Interestingly, this pattern resembles that obtained following EGFR inhibition. Thus, the observed effects of EGFR and E2 in cell morphological characteristics come in agreement with our findings on their impact in the proliferation and migration of TNBC cells.

    Details are in the caption following the image
    EGFR inhibition and E2 induce alterations in morphological characteristics and surface nanotopography of TNBC cells. (A) MDA-MB-231 cells (a and b) exhibit three different relatively isolated phenotypes: (a) flattened polygonal or cobblestone cells, (b) globular and elongated mesenchymal ones (thin arrows), and (c) elongated cells with visible microvesicles on their cytoplasmic surface (thick arrows). MDA-MB-231 cells treated with AG1478 (c and d) are very flattened polygonal or cobblestone shaped cells with cell-cell contacts that appear melted to each other in aggregates (twisted arrows in d). Elongated or mesenchymal shaped cells are rare (thin arrows) and few globular cells with no microvesicles on their surface are present (thick arrows in d). E2-treated MDA-MB-231 cells (e and f) mainly present isolated, elongated or mesenchymal phenotype (arrows in e). Elongated mesenchymal cell develop thin long filopodium with many microvesicles at their end. Few microvesicles are visible also on the cytoplasmic surface of the filopodium (arrow in f). The pattern of the MDA-MB-231 cells treated with both AG1478 and E2 is composed mainly of very smooth, flattened polygonal cells in tight contact to each other and a few globular shaped and elongated mesenchymal ones (arrows in g). Only a few microvesicles are detectable in some cells (scale bars: a, c, e, g = 100 µm; b, d, f, h = 10 μm). (B) AFM micrographs resemble a 10 µm x 10 µm area. The height scale is color coded: bright is high (1 μm) and dark is low (0 μm). Treatment with AG1478 and/or E2 alters the cell surface architecture. Surface protrusions (green markings) represent microspikes or filopodial extensions, of < 300 nm, whose number and size are generally reduced upon EGFR inhibition.

    As EGFR and E2/ERβ axis seem to impact the observed cell morphology, we performed AFM to look closer at distinct morphological structures that arise from breast cancer cells. AFM is used for the detection of distinct morphologies on biological cells such as nanospikes, microvilli, filopodial extensions, or fibrous ECM networks. The AFM micrographs of MDA-MB-231 cells (Fig. 3B), demonstrated a reduction of surface-decorating filopodial structures (green markings) upon EGFR inhibition along with a more sheet-like topography. Moreover, E2 addition leads to broader and more dense filopodial-like protrusions, eventually extending to loop-like structures (green markings). These data further support the SEM findings regarding the effects of EGFR and E2 on the development of surface protrusions in MDA-MB-231 cells.

    On the contrary, as seen in Fig. 4A, shERβ MDA-MB-231 cells (control) exhibit the following phenotypes: (a) the more numerous flattened polygonal or cobblestone cells, which appear aggregated and show tight contact with one another, (b) the globular ones, and (c) few elongated mesenchymal ones. No microvesicles or microvilli were detectable on the cytoplasmic surfaces (Fig. 4A–a and b). Most of the shERβ MDA-MB-231 cells treated with AG1478 present a very flattened polygonal or cobblestone cell phenotype and exhibit tight cell-cell contacts, while others, partially grouped cells, present a globular phenotype. Elongated or mesenchymal shaped cells are not easily detectable (Fig. 4A–c). The flattened cells seem to grow as aggregates and form colonies that are in contact one to each other. However, some of them exhibit short microvilli/filopodia on their surface, not visible in the untreated cells (Fig. 4A—d). Although in the E2-treated shERβ MDA-MB-231 cells the three phenotypes were still detectable with cell-cell contacts, E2 increases the number of the elongated or mesenchymal cells and globular ones compared to the untreated cells (Fig. 4A—e). Microvesicles are uniformly distributed on the cytoplasmic surface of the elongated and globular cells. These last also grow in groups of more than a dozen of cells, one up to each other (Fig. 4A–f). shERβ MDA-MB-231 cells treated with both AG1478 and E2 include many thin flattened and polygonal cells next to each other that show cell-cell contacts, partially grouped globular shaped cells and few elongated ones. Very few microvesicles or microvilli are present on the cell surface (Fig. 4A—g and h). Besides, contrary to the MDA-MB-231 cells, when we monitored the shERβ MDA-MB-231 cells with AFM (Fig. 4B) they generally exhibited a smoother surface throughout all different treatments. Moreover, the shERβ MDA-MB-231 cells displayed a clear tendency toward less protrusional volume under EGFR inhibition (green markings), hence reaffirming the importance of ERβ and EGFR for the formation of distinct cell morphological characteristics in TNBC.

    Details are in the caption following the image
    (A) Untreated shERβ MDA-MB-231 cells exhibit three different phenotypes: the more numerous flattened polygonal or cobblestone cells which appear aggregated in tight contact with each other, the globular ones in groups of few cells and few elongated mesenchymal ones (arrows). Few microvesicles or microvilli are detectable on the cytoplasmic surface. Most of the shERβ MDA-MB-231 cells treated with AG1478 appear as very flattened polygonal or cobblestone shaped cells, aggregating and showing tight cell-cell contacts, while others partially grouped display a globular phenotype. Few elongated or mesenchymal shaped cells were detectable (arrow). Flattened cells appear aggregated in colonies in contact one to each other. Some of them exhibit short microvilli/filopodia (needle shaped) on their surface, which are less observable in the untreated cells. When shERβ MDA-MB-231 cells were treated with E2 the three phenotypes are still detectable. E2 seems to slightly increase the number of the elongated or mesenchymal cells when compared to the untreated ones. All cell phenotypes are in contact one to each other, but microvesicles are clearly distributed on the cytoplasmic surface of the elongated cells and, in particular, in globular ones. In some areas, the globular cells also grow up one on another in large groups of cells (on the right). shERβ MDA-MB-231 cells treated with both AG1478 and E2 include many thin, flattened, and polygonal cells next to each other with cell-cell contacts, partially grouped, globular, and few elongated ones. Almost no microvesicles and few microvilli are present on cell surfaces (scale bars: a, c, e, g = 100 µm; b, d, f, h = 10 μm). (B) AFM micrographs resemble a 10 µm x 10 µm area. The height scale is color coded: Bright is high (1 μm) and dark is low (0 μm). The nanotopography of MDA-MB-231 cells is modified under ERβ-suppression and surface protrusions are significantly reduced, both in number and volume (green markings).

    The observed alterations on cell morphological characteristics and nanotopography upon EGFR inhibition and/or E2 treatment prompted us to further study the cytoskeleton organization by staining for F-actin. MDA-MB-231 cells display a very organized cytoskeleton network with actin stress fibers characteristic of the mesenchymal cells and limited cell-cell contacts (Fig. 5). On the other hand, AG1478-treated cells present a reduced and disorganized actin cytoskeleton with increased cell-cell contacts (Fig. 5). E2 addition resulted in more elongated cells with significantly reduced cell-cell contacts, whereas treatment with both AG1478 and E2 sustained a well-organized pattern of flattened cells in close contact along with a few elongated cells. Notably, in shERβ MDA-MB-231 cells due to their flattened morphology and increased spreading, F-actin filaments are very well outlined, even after EGFR inhibition that induces the actin reorganization (Fig. 5). These findings further support the notion that the synergy between EGFR and E2/ERβ is crucial for the cytoskeleton reorganization as well as the development of aggressive morphological features like filopodia and microvilli and the release of microvesicles.

    Details are in the caption following the image
    EGFR inflicts changes in the cytoskeleton network of TNBC cells. F-actin staining (red) shows reorganization of actin cytoskeleton in both cell lines after EGFR inhibition in the presence or absence of E2. Nuclei are shown in blue (DAPI) (scale bar = 50 µm).

    The role of EGFR in the expression of mesenchymal effectors and syndecan-1

    EMT program activation results in the suppression of epithelial characteristics and associated markers, while it induces the expression of mesenchymal ones to promote invasion and metastasis [[40]]. As shown in Fig. 6, EGFR inhibition and ERβ suppression led to decreased expression levels of important EMT mediators. Specifically, in MDA-MB-231 cells, treatment with AG1478 downregulates, in a statistically significant manner, the expression of vimentin (ca 40%), fibronectin (ca 20%), and ZEB2 (ca 25%) (Fig. 6). Exogenous addition of E2 downregulated the expression of vimentin (ca 45%), fibronectin (ca 20%), and ZEB2 (ca 15%) (Fig. 6). Interestingly, the combination treatment with AG1478 and E2 resulted in the suppression of all the examined markers' expression (Fig. 6). ERβ-suppression leads to a significant reduction in fibronectin mRNA levels (ca 40%) compared to MDA-MB-231 cells. In addition, EGFR inhibition resulted in lower gene expression levels compared to shERβ control cells: 20% for fibronectin, 20% for ZEB2, and 15% for vimentin. In shERβ MDA-MB-231 cells, E2 treatment does not have a great effect in the majority of EMT markers expression levels, suggesting that to exert its actions E2 needs ERβ and possibly the activation of EGFR pathway (Fig. 6).

    Details are in the caption following the image
    EGFR inhibition moderates key EMT markers’ and SDC-1 expression. AG1478 treatment significantly reduces the expression of vimentin, fibronectin, and ZEB2 even after E2 addition in both cell lines. SDC-1 levels are lower after EGFR inhibition in both cell lines but upregulated upon E2 addition only in the ERβ + MDA-MB-231 cells and when EGFR is active. Each bar represents mean ± SD values from triplicate samples. Statistically significant differences are indicated accordingly as in Fig. 1.

    Likewise, HSPGs have pivotal roles in breast cancer progression [[41]]. In particular, elevated syndecan-1 (SDC-1) levels are associated with tumor aggressiveness, grade and poor prognosis [[42]]. In addition, the expression levels of SDC-1 in MDA-MB-231 cells were downregulated following EGFR inhibition, while E2 induced the expression levels (Fig. 6). Besides the adverse effects that AG1478 and E2 had individually in the expression of SDC-1, their combination suppressed the mRNA levels of SDC-1. Upon ERβ suppression, treatment of MDA-MB-231 cells with AG1478 inhibited SDC-1 expression by ca 25% at the same levels as exogenous E2 addition (Fig. 6). Likewise, in the ERβ+ MDA-MB-231 cells, the immunoreactivity was lower after EGFR inhibition alone or in combination with E2 addition. Meanwhile, the distribution of SDC-1 was altered following both treatment with AG1478 and E2 (Fig. 7). Notably, ERβ suppression and E2 addition induced cytosolic diffusion of SDC-1, while after AG1478 treatment shERβ MDA-MB-231 cells showed lower immunoreactivity (Fig. 7).

    Details are in the caption following the image
    SDC-1 (green) distribution and immunoreactivity. Nuclei are shown in blue (DAPI) (scale bar = 50 μm). The distribution as well as the intensity, calculated as corrected total cell fluorescence (CTCF), of SDC-1 is influenced upon treatment with AG1478 in both cell lines. Statistically significant differences are indicated accordingly as in Fig. 1.

    EGFR controls the expression of major CSC markers and stemness of breast cancer cells

    EMT activation bestows several aggressive properties in breast cancer cells while altering several signaling cascades, which are the main regulators of CSC initiation [[43, 44]]. Therefore, we moved to evaluate whether the various treatments influence stemness in TNBC. First, we employed flow cytometry to characterize the CD44/CD24 phenotype of our cells. Since all our samples were CD44high/CD24−/low, we focused on the changes in the CD44 expression. As seen in Fig. 8, the CD44high population is decreased upon treatment with AG1478 and E2. Specifically, after EGFR inhibition CD44high levels were lower ca 20% in MDA-MB-231 cells and ca 30% in shERβ MDA-MB-231 ones, while after E2 addition expression was downregulated ca 25% and 32%, respectively.

    Details are in the caption following the image
    Protein levels of the stem cell specific cell surface marker, CD44, were detected via flow cytometry. EGFR inhibition and E2 stimulation have a statistically significant impact in CD44high levels in both ERβ+ and ERβ- MDA-MB-231 cells (FACS histograms). Statistically significant differences are indicated accordingly as in Fig. 1.

    Furthermore, analysis of the expression patterns of key molecules connected with conferring stem cell-like abilities in cancer cells, namely, NOTCH1, HEY1, OCT4, and MSI1, was performed in order study these markers’ profiles following treatments with AG1478 and/or E2 (Fig. 9). In MDA-MB-231 cells, EGFR inhibition downregulates the expression levels of NOTCH1 (30%), HEY1 (40%), OCT4 (35%), and MSI1 (40%). E2 showed no effect on the expression of OCT4 and MSI1 even though it downregulated the expression of NOTCH1 and HEY1 (Fig. 9). Based on these results, it is plausible to suggest that E2 exerts its actions only when EGFR is not inhibited. Furthermore, when we suppress ERβ in the MDA-MB-231 cells, MSI1 expression is not detectable and NOTCH1 along with HEY1 are also expressed at lower levels (Fig. 9). Treatment with AG1478 further decreases the levels of NOTCH1 (11%), HEY1 (30%), and OCT4 (30%), while addition of E2 does not affect their levels in a statistically significant manner.

    Details are in the caption following the image
    Cancer stem cell markers gene expression in TNBC cells. Expression profiles of molecules linked with stemness (HEY1, NOTCH1, OCT4, MSI1) are altered upon EGFR inhibition; each bar represents mean ± SD values from triplicate samples. Statistically significant differences are indicated accordingly as in Fig. 1.

    Various ways for assessing CSC properties have been proposed with the most used being sphere formation and colony formation assays. Hanging drop is a simple cost-effective method used for assessing spontaneous sphere formation efficiency of CSCs [[45, 46]]. Both cell lines were able to form compact circular spheres with variations in spheroid areas, depending on the treatment. As shown in Fig. 10, in MDA-MB-231 cells EGFR inhibition and E2 addition both caused reduction of the spheroid size of MDA-MB-231 cells by 1000 μm2 (ca 50%) and 600 μm2 (ca 40%), though E2 exerted its actions only when EGFR was not inhibited, as in combination treatment E2 caused no further size reduction. shERβ MDA-MB-231 cells formed ca 25% smaller spheroids, a size further reduced upon AG1478 treatment, whereas E2 addition had no significant effect compared to the shERβ control cells. On the other hand, colony formation has been described as the ability of CSCs to form colonies when seeded on cell culture dishes in low concentrations after limiting dilutions [[47]]. EGFR inhibition significantly limited the ability of both MDA-MB-231 and shERβ cells to form colonies, whereas E2 had an effect only in the ERβ positive MDA-MB-231 cells where its addition prompted the formation of less colonies compared to the untreated control cells. (Fig. 10).

    Details are in the caption following the image
    EGFR fine tunes cancer stem cell-like characteristics of TNBC cells. Typical CSC properties, colony, and sphere formation are modulated by EGFR in both ERβ+ and shERβ MDA-MB-231 cells; each bar represents mean ± SD values from triplicate samples. Statistically significant differences are indicated accordingly as in Fig. 1.

    The colonies formed were also stained for CD44 and imaged by IF microscopy. As is evident in Fig. 11, the MDA-MB-231 formed colonies are cell aggregates with CD44 localization mainly in the plasma membrane; EGFR inhibition results in more compact colonies with lower CD44 intensity, while with E2 addition CD44 can be seen not only in the membrane but also in the Golgi apparatus. ERβ-suppression clearly promoted the formation of more compact and tightly connected colonies with a shift in CD44 localization toward the Golgi apparatus, especially in the EGFR-inhibited shERβ MDA-MB-231 cells. Overall, in both cell lines treatment of cells with AG1478 and E2, alone or in combination, not only changes the distribution of CD44 but also reduces the fluorescent intensity, thus substantiating the crucial role of EGFR signaling in the regulation of CSC-like characteristics in TNBC.

    Details are in the caption following the image
    CD44 (green) distribution and fluorescence within the colonies is affected by EGFR and ERβ. Nuclei are shown in blue (DAPI) (scale bar = 50 μm). Statistically significant differences are indicated accordingly as in Fig. 1.

    Discussion

    TNBC accounts for almost 15% of total breast cancer cases every year worldwide [[1]]. Although distinct aggressive features including higher-grade, higher size, and higher probability of metastasis characterize TNBC tumors, treatment options are limited and new therapeutic strategies have been proven a challenge [[48]]. Despite the estrogen receptor (ER)-negative status of TNBC, the G protein-coupled estrogen receptor (GPER) and ERβ are still present and active. Indeed, GPER expression is prevalent in TNBC and some of the rapid effects of E2 could be attributed to its binding to and activation of the receptor [[49]]. Additionally, E2-mediated GPER activation promotes, among others, EGFR signaling in an SRC tyrosine kinase-integrin-mediated way [[50]]. Regardless, the epidemiological and laboratory reports on GPER’s actions have been conflicting and inconsistent [[51]]. Consequently, deeper investigation of ERβ could provide much needed insight since its actions in TNBC are not fully elucidated yet. Furthermore, EGFR is significantly overexpressed in TNBC, and despite the fact that various clinical trials have targeted the receptor’s activation, the results until now have not been promising [[52, 53]]. EGFR and ERβ are known to interact and together modulate ECM remodeling of TNBC cells [[37]]. To that end, in the current study we chose the triple-negative, ERβ+ MDA-MB-231 cells, as well as the stably transfected, shERβ MDA-MB-231 cell line generated in our laboratories, to showcase the synergistic effects of EGFR and ERβ in the aggressiveness and CSC-like capabilities of TNBC cells.

    Since EGFR and ERβ are differentially involved in a plethora of cell functions in breast cancer, we examined their role in the regulation of cell proliferation, cell cycle, migration, and angiogenesis. E2 addition induces both cell proliferation and migration through ERβ activation in MDA-MB-231 cells, whereas it inhibits migration in shERβ MDA-MB-231 cells. This could be attributed to the fact that E2, apart from ERs, can also mediate signaling cascades through other cell membrane receptors, such as G protein-coupled receptors [[31]]. Meanwhile, EGFR inhibition leads to lower growth rates and motility, an effect more prominent in the ERβ-suppressed cells. In general, these results reveal an important synergy of the EGFR and ERβ pathways for the induction of proliferation and motility in TNBC. Angiogenesis, the formation of new capillary blood vessels from existing vasculatures, is a required process for tumor progression and metastasis and depends on a delicate balance that exists between stimulators and inhibitors [[54, 55]]. EGFR inhibition and E2 addition significantly reduced angiogenesis in vitro in both cell lines and altered the expression profiles of associated markers, hence highlighting the importance of EGFR and E2-mediated signaling in the activation of the angiogenic switch.

    Additionally, we demonstrated significant alterations in the morphological characteristics and cytoskeletal elements of MDA-MB-231 cells when treated either with the EGFR inhibitor, AG1478, E2, or their combination (AG1478+E2). SEM images show a less aggressive phenotype in the MDA-MB-231 cells after EGFR inhibition with more cell-cell contacts, smoother surface, and less microvesicles, which is mainly maintained even in the presence of E2. These morphological changes were coupled with reorganization in the actin network of the cytoskeleton. In addition, the nano-topographic images obtained by AFM clearly show substantial differences in the cell surface architecture, with alterations in the morphological structures (loops, sheets, ruffles) manifested after each treatment. These ultrastructural observations confirm that EGFR inhibition and ERβ suppression promote the shift toward a less aggressive phenotype of TNBC cells.

    Likewise, EGFR inhibition in MDA-MB-231 cells substantially reduces the expression of EMT mediators, including vimentin, fibronectin and ZEB2, whereas E2 needs ERβ and the activation of EGFR to exert its actions. When ERβ was suppressed, EGFR inhibition further downregulated the mRNA levels of the mesenchymal markers. Moreover, our results show that EGFR inhibition causes a downregulation of SDC-1 expression in both cell lines, while it also alters its localization and distribution. SDC-1, an HSPG that facilitates cell anchorage, has also emerged as a molecular marker of aggressiveness and CSC phenotype in breast cancer [[56-58]]. Different SDCs are known to regulate cell processes like motility and angiogenesis by binding to different growth factors and their receptors [[59]]. Interestingly, a recent publication has revealed that altered expression of SDC-1 impacts the recycling of RTKs in breast cancer cells [[60]], suggesting interdependence in the regulation of various pathways. Hence, our data reaffirm the crucial role of EGFR while revealing a mechanistic link between the EGFR and ERβ pathways, which promotes EMT and alters cell behavior of TNBC cells.

    EMT progression is correlated with the generation of tumor-initiating cells and induction of stemness [[24]]. Thus, we assessed the cell surface markers CD44 and CD24, since a CD44high/CD24−/low profile is considered typical of mesenchymal, CSC-like subpopulations [[61]]; the more enriched this population is, the more tumor-seeding abilities it has [[62]]. Both ERβ+ and shERβ MDA-MB-231 cells are characterized as CD24−/low; hence, we focused our study in the levels of CD44 measured as mean fluorescence intensity. Treatment with EGFR or E2 resulted in significantly lower CD44 levels and consequently in a lower CSC phenotype in both cell lines. Along with the downregulation of SDC-1, which resulted in a similar phenotype in MDA-MB-231 cells [[57]], these data suggest that EGFR and E2 modulate the CSC phenotype in TNBC cells.

    The expression profiles of the established markers of stemness NOTCH1, HEY1, MSI1, and OCT4 showed a downregulation following EGFR inhibition, with additional reductions of NOTCH1 and OCT4 expression levels in shERβ MDA-MB-231 cells. Notably, Musashi-1, a protein with a main role in NOTCH-mediated CSC proliferation and therapeutic resistance in TNBC cells [[57]], was undetectable in shERβ MDA-MB-231 cells. Since one of CSCs abilities is the increased metastatic colony-seeding potential, we employed colony and hanging drop sphere formation assays to explore CSCs' functional properties. Our analysis shows that EGFR modulates the colony and sphere formation properties displayed by the CSC-subpopulation in both cell lines and, along with ERβ, modifies the expression and distribution of CD44 within the colonies. Collectively, these results suggest that EGFR pathway promotes the NOTCH-mediated induction of CSC formation.

    Conclusions

    Even though TNBC has been extensively studied various underlying mechanisms that drive the disease are still unknown. In this study, we used the highly aggressive, ERβ+ MDA-MB-231 cell line, as a widely accessible and established model, to study the EGFR-ERβ interplay. In the MDA-MB-231 cells, EGFR signaling and E2-activation of ERβ-pathway appear to advance growth, motility, and angiogenesis of breast cancer cells, while in the ERβ suppressed cells E2 addition inhibits migration seemingly by advancing alternative signaling cascades. Moreover, the obtained results depict EGFR/ERβ axis as effector of alterations in the morphological characteristics and cytoskeleton organization. Furthermore, the acquired data reveal that EMT stimulation and matrix composition are tightly regulated by EGFR in an E2/ERβ-dependent manner and that, in turn, affects the expression of CSC-related markers and CSC properties possibly via NOTCH-signaling activation.

    In conclusion, herein we highlighted the importance of EGFR signaling in TNBC as a key effector of aggressiveness and stemness in an ERβ-dependent way. Further investigation, in other in vitro and in vivo models, of the mechanisms by which the ERβ-mediated EGFR signaling modulates the aggressive behavior and CSC-like characteristics of TNBC cells will benefit breast cancer management by offering high quality prevention, early detection, and personalized treatment services.

    Materials and methods

    Chemicals and reagents

    Dulbecco's modified Eagle's medium (DMEM), fetal bovine serum (FBS), PBS, trypsin-Na2EDTA, L-glutamine, penicillin, and streptomycin were all obtained from Biosera LTD (Courtaboeuf Cedex, France). Endothelial cell growth medium MV (C-22020) was purchased from PromoCell GmbH (Heidelberg, Germany). The cytostatic agent cytarabine was supplied by Sigma Chemical Co. (St. Louis, MO, USA). The allosteric inhibitor AG1478 (EGFR) and 17-β estradiol (E2) were supplied by Sigma Chemical Co. (St. Louis, MO, USA). All other chemicals used were of the best commercially available grade.

    Cell cultures

    MDA-MB-231 (high metastatic potential, ERα-negative, ERβ-positive) breast cancer cell line was obtained from the American Type Culture Collection (ATCC, Baltimore, MD, USA) and routinely cultured as monolayers at 37 °C in a humidified atmosphere of 5% (v/v) CO2 and 95% air. shERβ ΜDA-MB-231 (ERβ-suppressed cells) were previously described [[33]]. Breast cancer cells were cultured in complete medium [DMEM supplemented with 1.0 mm sodium pyruvate, 2 mm l-glutamine, 100 IU·mL−1 penicillin, 100 μg·mL−1 streptomycin, 10 μg·mL−1 gentamicin sulfate, 2.5 μg·mL−1 amphotericin B, and 10% fetal bovine serum (FBS)]. Puromycin dihydrochloride (Sigma-Aldrich, Steinheim, Germany; 0.8 μg·mL−1) was included in the cultures of shERβ ΜDA-MB-231 cells. Cells were harvested by trypsinization with 0.05% (w/v) trypsin in PBS containing 0.02% (w/v) Na2EDTA. All experiments were conducted in serum-free conditions. Cell cultures were performed as described above prior to AG1478 (2 μm) and/or E2 (10 nm) treatment for 24 h. When the two treatments were combined, AG1478 (2 μm) treatment was conducted 30 min prior to E2 (10 nm) addition.

    HUVECs were purchased from PromoCell (Heidelberg, Germany) and cultured in endothelial cell growth medium MV (EGM) (PromoCell GmbH, cat. no. C-22020, Heidelberg, Germany) supplemented with 15% FBS. The cultures were maintained in an atmosphere of humidified 95% air, 5% CO2 (v/v) at 37 °C.

    In vitro cell proliferation assay

    Breast cancer cells were seeded in 96-well plates and incubated in complete medium until 70%–80% confluency, followed by 16-h starvation in serum-free medium. Afterward, cells were treated with vehicle (control), AG1478 (2 μm), and/or E2 (10 nm) in serum-free medium for 24 h. To assess cell proliferation, we used the Premix WST-1 (water-soluble tetrazolium salt) Cell Proliferation Assay System (Takara Bio Inc., Göteborg, Sweden) where Premix WST-1 was added at a ratio 1 : 10, and the absorbance at 450 nm was measured (reference wavelength at 650 nm), according to manufacturer’s protocol.

    In vitro wound healing assay

    25 x 104 breast cancer cells were seeded in 12-well plates and incubated in complete medium. Cells were then starved with serum-free medium for 16 h. The confluent cell layers were then scratched using a 100-μL pipette tip. Detached cells were removed by washing twice with PBS, and cells were incubated for 40 min at 37 °C with serum-free media containing 10 μΜ of the cytostatic agent cytarabine (Sigma-Aldrich), to minimize a possible contribution of proliferation, and then, they were photographed at 10× magnification [OLYMPUS CKX41 microscope with a color digital camera CMOS (SC30)]. Afterward, the cells were treated accordingly and incubated for 24 h under serum-free conditions. Images were captured, and wound surface was quantified using ImageJ software (National Institute of Mental Health, Betesda, Maryland, USA).

    Angiogenesis assay

    In vitro angiogenesis was evaluated by using tube formation assay, as previously described [[38, 63]]. Briefly, a 96-well plate was carefully coated on ice with Matrigel® (#354230, Corning GmbH, Kaiserslautern, Germany) to avoid air bubbles and placed in an incubator at 37 °C for 30 mins to allow Matrigel® to form a gel. HUVEC cells were harvested by trypsinization and counted. 18 000 HUVECs in 50 μL media (with supplements) were plated per well and supplemented with 100 μL of conditioned media (CM). CM was collected from untreated (control) and treated (AG1478 and/or E2) MDA-MB-231 and shERβ MDA-MB-231 breast cancer cell cultures. The plate was then placed in the cell culture incubator at 37 °C, 5% CO2 for a period of 6–8 h. Tube formation was examined by an inverted microscope at 10× magnification. Quantification of total tube length and related parameters was generated by image processing via the Wimtube online image analysis tool developed by Wimasis [[39]].

    SEM imaging

    Breast cancer cells were cultured in polystyrene flasks as described, treated for 24 h accordingly, and then fixed in a Karnovsky’s solution for 30 min at 4 °C. Small central portions of the flasks with adhering cells were obtained by a cutter, rinsed three times with 0.1% cacodylate buffer, dehydrated with increasing concentrations of ethanol and finally dehydrated with hexamethyldisilazane (Sigma-Aldrich, Inc.) for 15 min. All specimens were mounted on appropriate stubs, coated with a 5-nm palladium gold film (Emitech 550 sputter-coater) to be observed under a SEM (Philips 515, Eindhoven, The Netherlands) operating in secondary-electron mode.

    Surface nano-texture analysis (nAnostic)

    Contact mode atomic force microscopy (AFM) on cultivated cells was performed as described before [[64]]. In this study, cells were chemically stabilized by glutaraldehyde fixation (1% final concentration). Briefly, AFM measurements were carried out in PBS-buffered solution (pH 7.4) using a Multimode AFM equipped with Nanoscope III controller and software version 5.30 sr3 (Digital Instruments, Santa Barbara, CA, USA). Silicon-nitride tips on V-shaped gold-coated cantilevers were used (0.01 N·m−1, MLCT, VEECO, Mannheim, Germany). Imaging was performed at ambient temperature with forces < 1 nN at 1–3 scan lines per second (1–3 Hz) with 256 × 256 pixels resolution. Scan areas of 10 μm2 were chosen.

    Immunofluorescence microscopy

    For immunofluorescence microscopy, 4 × 104 breast cancer cells were seeded on glass coverslips in 24-well plates and incubated for 24 h in complete medium followed by 16-h serum starvation. The medium was replaced with serum-free medium containing the appropriate agents, and cells were incubated for 24 h. Cells were first washed with phosphate-buffered saline (PBS), fixed in 4% paraformaldehyde (PFA) in PBS, washed three times with PBS, permeabilized with freshly made 0.1% Triton X-100 in PBS, washed three times with PBS-Tween 0.01%, and blocked with 5% BSA/PBS-Tween 0.01% for 1 h at room temperature (RT). Then, primary antibodies in 1% BSA/PBS-Tween 0.01% were added and samples were incubated at 4 °C overnight. On the next day, the appropriate secondary antibodies in 1% BSA/PBS-Tween 0.01% were added and incubated for 1 h in the dark. Finally, cells were washed, stained, and mounted with DAPI. Between each step, cells were washed 3 times with PBS-Tween 0.01%. Visualization was performed using a fluorescent phase contrast microscope (OLYMPUS CKX41, QImaging Micro Publisher 3.3RTV, Teledyne Photometrics, Tucson, Arizona, USA) at 60x. Slides were stained with the following antibodies and reagents: Phalloidin CruzFluor™ 594 Conjugate (dilution 1 : 1000, sc-363795, Santa Cruz Biotechnology, Inc., Heidelberg, Germany), syndecan-1 (dilution 1:50, ab714, Abcam, Cambridge, UK), anti-CD44 (Hermes-3) (1 μg·mL−1, generously provided by Dr. S. Jalkanen), Alexa Fluor 488 goat anti-rabbit (Biotium, #PSF006), and Alexa Fluor 488 goat anti-mouse (Biotium, #PSF006). The level of cellular fluorescence from fluorescence microscopy images was determined with ImageJ software based on an established protocol from Queensland Brain Institute Advanced Microscopy Facility. The corrected total cell fluorescence (CTCF) was calculated with the following formula: CTCF = Integrated Density – (Area of selected cell x Mean fluorescence of background readings).

    RNA isolation, reverse transcription, and real-time qPCR analysis

    Breast cancer cells were grown in serum-containing medium up to 70%–80% confluence. Cells were serum starved for 16 h. Afterward, AG1478 (2 μm) or/and E2 (10 nm) were added for 24 h according to the experimental plan in serum-free culture medium. When both agents were added, AG1478 treatment for 30 min was conducted prior to E2 addition. Total RNA was isolated from cells, using NucleoSpin® RNA II Kit (Macherey-Nagel, Duren, Germany). Quantification of the isolated RNA was done by measuring its absorbance at 260 nm. RNA purity was ensured by evaluating the 260/280 nm and 260/230 nm ratios of all RNA extracts. Total RNA was reverse-transcribed, using thePrimeScript™1st strand cDNA synthesis kit perfect real time (Takara Bio Inc., Göteborg, Sweden) and KAPA Taq ReadyMix DNA Polymerase (KAPA BIOSYSTEMS, Wilmington, MA, USA). Real-time PCR analysis was conducted according to manufacturer’s instructions. For the amplification, we utilized the Rotor Gene Q (Qiagen, Germantown, MD, USA). All reactions were performed in triplicates, and a standard curve was always included for each pair of primers for assay validation. For quantification purposes, the point of product accumulation in the early logarithmic phase of the amplification plot was defined by assigning a fluorescence threshold above the background, defined as the threshold cycle (Ct) number. Relative expression of different gene transcripts was calculated by the ΔΔCt method. Τhe Ct of any gene of interest was normalized to the Ct of the normalizer (ACTB). Fold changes (arbitrary units) were determined as 2−ΔΔCt. Genes of interest and utilized primers are listed in Table 2.

    Table 2. qPCR primer sequences
    Gene Primer Sequence (5′-3′)
    SYBR Green probes
    ACTB
    F TCAAGATCATTGCTCCTCCTGAG
    R ACATCTGCTGGAAGGTGGACA
    FN1
    F CCAAGCATCACCCTGGGAGT
    R CGAAGCAGAACAGGCAATGTG
    VIM
    F TCAGCATCACGATGACCTTGAA
    R CTGCAGAAAGGCACTTGAAAGC
    ZEB2
    F TGGGCTAGTAGGCTGTGTCCA
    R TCATCTTCAACCCTGAAACAGAGG
    SDC1
    F AGGACGAAGGCAGCTACTCCT
    R TTTGGTGGGCTTCTGGTAGG
    SDC2
    F CTCTGGACTCTCTACATCCTCATCAG
    R GAGTGTATCCTATTGATGACGATGACTAC
    MSI1
    F AGCTTCCCTCTCCCTCATTCC
    R GGAATTCCAGGGTCCTGAGC
    TaqMan probes
    HEY1 hs 00232618 m1a
    OCT4 hs 00742896 s1a
    NOTCH1 hs 00413187 m1a

    Flow cytometry

    For the detection of stem cell specific cell surface markers, CD44 and CD24, breast cancer cells were harvested by incubation with 2 mm EDTA in Ca2+/Mg2+-free PBS buffer for 10 min at 37 °C with gentle agitation. 1 x 106 cells in 100 μL PBS/0.1% BSA (PBS/BSA) were incubated with 10 μL of anti-CD44-APC, anti-CD24-PE, and the APC and PE isotype controls (BD-Bioscience, Heidelberg, Germany) for 30 min at room temperature in the dark. Subsequently, cells were centrifuged (450 g, 5 min) and washed twice with 300 μL PBS/BSA. Cells were resuspended in 1 mL PBS/BSA before analysis in a flow cytometer. Flow cytometric analysis took place on a CyFlow Space (Sysmex/Partec, Görlitz, Germany) equipped with a 25 mW 638 nm red laser diode and a 20 mW 488 nm blue argon laser, fluorescence emission was detected at 675 nm (BP675/20 nm) and 590 nm (BP 590/50 nm). Data analysis was done by using FloMax software (Quantum Analysis, Münster, Germany). Isotype controls were set in the first quadrant, and gates were administered. CD44(−) CD24(−) resides in the Q3, CD44(+) CD24(−) in Q4, CD44(+) CD24(+) in Q2, and CD44(−) CD24(+) in Q1. CD44 and CD24 expression was also measured as mean x (mean fluorescence index = MFI) over the whole population by setting a region-gate (RN1) in FL2. Histograms from flow cytometry analysis of CD44 levels are displayed in Fig. 8.

    Colony formation assay

    Untreated and treated breast cancer cells were seeded in 35-mm culture dishes (1000 cells·dish−1) (Nunc, Langenselbold, Germany) and incubated with DMEM supplemented with 20% (v/v) FBS for 7–10 days. The media was renewed every 2 days. Colonies with more than 50 cells were counted using a microscope.

    Hanging drop assay

    To assess the spheroid formation ability and compare the spheroid size of the cells, we first prepared a solution of 106 cells·mL−1 in complete medium and placed several 20 μL drops into the lid of a Petri dish, after which we added 7 mL sterile PBS to the bottom of the dish and left it in the cell incubator for 1 week. Pictures of the spheroids inside the drop were taken using a Zeiss Axiophot camera, and their comparative size was obtained measuring the area occupied by the spheres using the software NIH ImageJ (NIH, Bethesda, MD, USA).

    Statistical analysis

    For each assay, three individual experiments were conducted. Data in diagrams are expressed as mean ± standard deviation (SD). Statistically significant differences were evaluated using one-way ANOVA followed by Tukey's post hoc test. Statistical analyses and graphs were made using GraphPad prism 9 (GraphPad Software, San Diego, CA, USA). Statistically significant differences are indicated by asterisks: * (P ≤ 0.05), ** (P < 0.01), *** (P < 0.001) compared to MDA-MB-231 control cells; # (P < 0.05) and ## (P < 0.01) compared to shERβ MDA-MB-231 control cells. Nonstatistically significant comparisons (P > 0.05) are not displayed.

    Acknowledgements

    We acknowledge funding by EU-H2020-MSCA-RISE grant GLYCANC #645756 (to MG, NK, CR, and ZP). KK was co-financed by Greece and the European Union (European Social Fund-ESF) through the Operational Programme ‘Human Resources Development, Education and Lifelong Learning’, in the context of the project ‘Strengthening Human Resources Research Potential via Doctorate Research’ (MIS-5000432), implemented by the State Scholarships Foundation (ΙΚΥ). ZP acknowledges funding by the Action for the Strategic Development on the Research and Technological Sector (MIS5048492), funded by the Operational Programme ‘Competitiveness, Entrepreneurship and Innovation’ (NSRF 2014-2020), and co-financed by Greece and the European Union (European Regional Development Fund).

      Conflicts of interest

      The authors declare no conflict of interest.

      Author contributions

      NKK performed conceptualization; CR, BG, MF, and MG involved in methodology; KK, CR, B., and MF validated the document; KK and BG performed formal analysis; KK, EK, CR, BG, and MF investigated the document; CR, BG, MF, MG, and NKK involved in resources; KK and ZP wrote original draft preparation; KK, EK, ZP, CR, BG, MF, MG, and NKK reviewed and edited the document; MG and NKK visualized and supervised the study; MG and NKK involved in project administration; KK, ZP, CR, MG, and NKK involved in funding acquisition. All authors have read and agreed to the published version of the manuscript.

      Peer Review

      The peer review history for this article is available at https://publons.com/publon/10.1111/febs.16240.