EpCAM overexpression prolongs proliferative capacity of primary human breast epithelial cells and supports hyperplastic growth

Introduction The Epithelial Cell Adhesion Molecule (EpCAM) has been shown to be strongly expressed in human breast cancer and cancer stem cells and its overexpression has been supposed to support tumor progression and metastasis. However, effects of EpCAM overexpression on normal breast epithelial cells have never been studied before. Therefore, we analyzed effects of transient adenoviral overexpression of EpCAM on proliferation, migration and differentiation of primary human mammary epithelial cells (HMECs). Methods HMECs were transfected by an adenoviral system for transient overexpression of EpCAM. Thereafter, changes in cell proliferation and migration were studied using a real time measurement system. Target gene expression was evaluated by transcriptome analysis in proliferating and polarized HMEC cultures. A Chicken Chorioallantoic Membrane (CAM) xenograft model was used to study effects on in vivo growth of HMECs. Results EpCAM overexpression in HMECs did not significantly alter gene expression profile of proliferating or growth arrested cells. Proliferating HMECs displayed predominantly glycosylated EpCAM isoforms and were inhibited in cell proliferation and migration by upregulation of p27KIP1 and p53. HMECs with overexpression of EpCAM showed a down regulation of E-cadherin. Moreover, cells were more resistant to TGF-β1 induced growth arrest and maintained longer capacities to proliferate in vitro. EpCAM overexpressing HMECs xenografts in chicken embryos showed hyperplastic growth, lack of lumen formation and increased infiltrates of the chicken leukocytes. Conclusions EpCAM revealed oncogenic features in normal human breast cells by inducing resistance to TGF-β1-mediated growth arrest and supporting a cell phenotype with longer proliferative capacities in vitro. EpCAM overexpression resulted in hyperplastic growth in vivo. Thus, we suggest that EpCAM acts as a prosurvival factor counteracting terminal differentiation processes in normal mammary glands.


Introduction
EpCAM (also known as 17-1A, GA733-2, KSA, ESA, and EGP-40) is a homophilic, calcium-independent cell adhesion molecule of 39-42 kDa [1,2] expressed on most normal and cancerous epithelial tissues, cancer stem cells, embryonic stem cells and germ cells [3][4][5]. EpCAM is a type I transmembrane glycoprotein encoded by the TACSTD1 gene. The EpCAM protein contains an extracellular domain (EpEX) with a nidogen-like domain as well as thyroglobulinand epidermal growth factor-like repeats, a single transmembrane region, and a short intracellular domain (EpICD) consisting of 26 amino acids. EpCAM has been shown to be expressed on normal epithelial cells in situ at intercellular basolateral interfaces [1]. In regard to its function, it has been shown in the developing zebrafish, that EpCAMlacking mutants display defects both in epithelial morphogenesis and epithelial integrity [1,6]. Moreover, mutants show abnormal skin development with higher infection susceptibility and enhanced skin inflammation [1,6]. In regard to mammals, EpCAM −/− mice die in uterus at embryonic day 12, are developmentally delayed and display prominent placental abnormalities [7].
In tumor development and progression EpCAM has a controversial biological role [5]. As an adhesion molecule, EpCAM mediates homophilic cell-cell adhesion interactions thereby preventing metastasis [1,2]. In colorectal cancer EpCAM appears to act as molecule with protective function, since EpCAM deletions result in a higher risk to develop cancer [8] and overexpression of EpCAM in colorectal cancer cells has been shown to inhibit metastasis and invasion of tumor xenografts in mice [9].
On the other hand, it is known that EpCAM can abrogate E-cadherin mediated cell-cell adhesion thereby promoting metastasis [10]. Furthermore, it has been shown that EpCAM overexpression in cancer cells can support proliferation by enhancing Wnt signaling [11]. In breast carcinoma patients, high EpCAM expression was observed in less differentiated tumors [12] and was associated with larger tumors, nodal metastasis and worse survival of patients [13]. Moreover, high EpCAM expression correlated with poor prognosis in both node positive and node negative disease [14]. Due to its high expression in breast cancer tissue, EpCAM has emerged as an attractive target for treatment of breast cancer patients and recent studies with the humanized EpCAM antibody Adecatumumab showed already promising results in patients with EpCAM overexpression [15]. Moreover, the approval by the European Union in 2009 of the EpCAMspecific antibody Catumaxomab, adds a therapeutic option also in breast cancer patients with peritoneal carcinomatosis and malignant ascites [16].
Although it has been shown that EpCAM is expressed in normal epithelial cells [17] the role in normal breast tissue homeostasis is still unclear. In this study we analyzed effects of adenoviral overexpression of EpCAM on growth, migration and differentiation of normal breast epithelial cells. Moreover, we screened for genes altered by overexpression of EpCAM in normal epithelial cells of the breast and analyzed in vivo growth in a chicken xenograft model.

Tissue samples
A Human Breast Cancer Tissue Array, with matched metastatic carcinoma tissue (BR10010-2-BX), including TNM and pathology grade (50 cases, 100 cores) was purchased from Biocat and was composed of primary breast carcinoma (n = 50) with corresponding lymph node metastasis (n = 50). Samples from normal breast tissue (n = 5) were obtained in form of paraffin-embedded tissue block slides with normal breast tissue (Breast T2234086-BC). Detailed information about all tumor samples can be found on the supplier's web site (http:// www.biocat.com)

Generation of polarized cultures of HMECs
HMECs of 3 independent donors were seeded on a transwell 0.4 μm polyester membrane (Costar) coated with growth factor reduced matrigel (BD Biosciences, 50 μg/cm 2 ). Cell culture medium was exchanged daily. Until day 4-5 cells formed confluent monolayer and until day 12 they polarized. The polarization status of the culture was confirmed by transepithelial resistance measurement using the STX2 electrode and the EVOM epithelial voltohmmeter (World Precision Instruments, WPI). For transepithelial resistance calculations we used the following formule: (measured valueblank value) × 18.1.

Immunofluorescence
Cells were seeded on a Matrigel coated eight-well culture slides (Falcon BD Labware). Polarized 3D cultures cells were fixed, permeabilized and stained directly on Matrigel coated transwells. After being fixed in 4% paraformaldehyde and permeabilized with 0.2% Triton-X-100 cells were blocked with PBS containing 3% BSA for 45 min at room temperature (RT). All antibodies of the immunohistochemistry section and additional antibodiesanti human ZO-1 (clone 1/ZO-1, BD Biosciences Pharmingen), anti human E-cadherin (clone 32A8, Cell Signaling Technology), anti-β-catenin (clone 14/ Beta-Catenin, BD Biosciences Pharmingen) were applied in a 1:100 dilution at RT for two hours. After washing in PBS cells were incubated with secondary fluorochrome-labeled antibodies (FITC-labeled/PE-labeled rabbit anti mouse, Dako Cytomation) and nuclei were counterstained with TO-PRO-3 Iodide or DAPI (both Molecular Probes). Cells were embedded in fluorescent mounting medium (Dako Cytomation) and viewed by a Fluorescence Microscope (Zeiss Axiovert 200, Carl Zeiss, Axiovision Software).

PNGaseF treatment
Enzymatic deglycosylation of total protein (20 μg) was performed with PNGaseF enzyme (New England Biolabs) according to manufacturer's protocol. Thereafter, protein extracts were analyzed by Western Blot.

Adenoviral overexpression of EpCAM
Replication-defective adenoviruses were generated with the Ad-Easy Adenoviral vector system (Stratagene) according to the manufacturer's instructions and as described elsewhere [19]. In brief, the EpCAM cDNA (NM_002354, Openbiosystems) was subcloned into the pShuttle CMV GFP vector and sequenced. Recombinant adenoviral DNA was generated in BJ5183 bacteria cells using a double-recombination event between cotransfected adenoviral backbone plasmid vector, pAdEasy-1, and a shuttle vector carrying the gene of interest. For generation of replicationdefective adenovirus recombinant DNA was transfected into HEK293 cells using Lipofectamin 2000 (Invitrogen). All viral titers were determined by qPCR for the gene coding for the encapsulation signal (for:5-cgacggatgtggcaaaagt, rev: 5-cctaaaaccgcgcgaaaa) and the respective viral plasmid DNA standards. HMECs were transfected with a multiplicity of infection (MOI) of 100 viruses/cell and tested for gene and protein expression 24 to 118 hours after transfection. All cell proliferation, migration and in vivo assays were performed at least 24 hours after adenoviral transfection to allow efficient EpCAM overexpression.

Flow cytometry
For FACS analysis of membranous EpCAM staining, cells were washed in PBS, resuspended and incubated with the first anti-EpCAM antibody (sc-25308, Santa Cruz Biotechnology, 1 μg/mL) for 30 min and then with the second, PE-labeled antibody (PE-labeled rabbit anti mouse, Dako Cytomation). Thereafter, cells were washed with PBS, resuspended and stainings were evaluated by a FACSCalibur (Becton-Dickinson, Heidelberg, Germany).

Microarray data set generation and analysis
Gene expression profiling analysis was performed at the Expression Profiling Unit of the Medical University Innsbruck. RNA quantity was determined by optical density measurements and RNA integrity using the 2100 Bioanalyzer (Agilent Technologies, Palo Alto, CA). Fifty ng high quality RNA were processed using the WT Expression Kit (Ambion) and the WT Terminal Labeling Kit (Affymetrix). The resulting biotinylated targets were hybridized to Affymetrix Human Gene ST 1.0 v microarrays. Microarrays were washed and stained in an Affymetrix fluidic station 450, fluorescence signals were recorded by an Affymetrix scanner 3000 and image analysis was performed with the GCOS software (Affymetrix). Raw and preprocessed microarray data have been deposited at the Gene Expression Omnibus accession number GSE37172 and GSE39071. http://www.ncbi.nlm.nih. gov/geo/query/acc.cgi?token=xhyxvqggwgewqts&acc=GSE37 172 http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?token=h xspfguwciquuhg&acc=GSE39071.

Quantitative RT-PCR analysis
Total RNA was isolated from HMECs using the TriReagent (Sigma Biochemicals), according to manufacturer's instructions. For microarrays, RNA was purified by cell lysis and nucleic acid extraction using the RNeasy Kit (Qiagen). Thereafter, viral and genomic DNA in the RNA samples was digested with the RQ1 DNAse (Promega). The cDNA was amplified from 1 μg total RNA by the use of the SuperScript II Reverse Transcriptase Kit (Invitrogen Life Technologies). For validation, real time RT-PCR was performed using a SensiMix SYBR No-ROX Kit (Bioline) and a Rotor-Gene 6000 detection system (Corbett Research). Primers were designed to amplify specific GAPDH (for: 5-CTGACCTGCCGTCTAGAAAA; rev: 5-GAGCTTGACAAA GTGGTCGT), TATA Box Binding Protein (for: 5-GGAGCCAAGAGTGAAGAACA; rev: 5-AGCACAAGGCCTTCTAACCT) and EpCAM (for: 5-GCTGGTGTGTGAACACTGCT; rev: 5-ACGCGTTGT GATCTCCTTCT).

Real time cell proliferation and migration assay (xCelligence system)
Real time cell proliferation and migration experiments were performed using the RTCA DP instrument (Roche Diagnostics GmbH), which was placed in a humidified incubator maintained at a 5% CO 2 at 37°C. For proliferation assay cells were seeded in complete medium in 16-well plates (E-plate 16, Roche Diagnostics GmbH) at density of 5,000 cells/well. The plate containing gold microelectrodes on its bottom was monitored every 10 minutes for 4 hours (adhesion process), then once every 30 min, until the end of experiment, which was in total 72 hours (cell proliferation). Cell migration was performed using special 16-well plates with 8 μm pores (CIM-plate 16, Roche Diagnostics GmbH). These plates, resembling conventional transwells, have microelectrodes placed on the underside of the membrane. Cells were seeded into the upper chamber at a density of 20,000 cells/well in a serum free medium and the lower chamber was filled with complete medium. The plate was monitored every 15 minutes for 12 hours. Data analysis was performed using RTCA software 1.2 supplied with the instrument.

Chicken chorioallantoic membrane (CAM) xenograft model
On embryo development day 0 fertilized chicken eggs (Gallus domesticus, Charles River) were placed in a 75-80% humidified 37°C incubator (Grumbach) to allow normal embryo development. On day 3 eggs were opened, egg shells removed and embryos were placed in a sterile Petri dish in an egg incubator to induce CAM development. On day 8, when chorioallantoic membrane (CAM) and its vasculature were well developed, all experiments were performed. HMECs were transfected one day before the experiment either by EpCAM adenoviruses or GFP control adenoviruses (both MOI = 100). 3.0 × 10 5 cells were resuspended in a 30 μL drop of ice-cold growth-factor reduced Matrigel (Becton Dickinson) containing TGF-β1 in a concentration of 1.7 ng/mL and the mixture solidified for 30 min at 37°C. Subsequently, 4 onplants per chicken were grafted on the CAM. Growth of HMECs onplants was inspected on a daily basis using a stereo fluorescence microscope (Olympus SZW 10). On day 6 post-grafting chicken embryos were sacrificed with hypothermia, xenografts cut out and stored either in 4% paraformaldehyde for immunohistochemical studies or in TRI-reagent (Sigma Biochemicals) for RNA isolation.

Statistical analyses
Statistical analyses were performed with the GraphPad Prism™ 5.0 (GraphPad Software, Inc.) software for Windows. All tests of statistical significance were twosided. Student's T test, two-way ANOVA and Mann-Whitney U Tests were used to study differences between two groups. Statistical analyses of quantitative PCR data were performed according to the delta Ct method described by Pfaffl et al. [20] and p values were calculated with the Student's T Test. Data analysis of microarrays was performed in R (http://www.r-project.org) using packages from the Bioconductor project. The custom Ensembl transcript based CDF package (v13) from the brainarray group was used for probe set definitions. GeneChip raw expression values were preprocessed using the RMA method. After preprocessing a representative transcript probe, the set was selected for each gene as described previously [21]. In brief, a combination of average and variation of expression of a probe set across all samples was used to select the most informative transcript probe set for a gene. The moderated t-test was employed to assess significance of differential expression of a probe set between EpCAM overexpressing and control samples. The resulting raw p-values were adjusted for multiple hypotheses testing with Benjamini and Hochberg's method for a strong control of the false discovery rate. Raw and preprocessed microarray data have been deposited at the Gene Expression Omnibus (accession number GSE37172, GSE39071).

Expression of EpCAM in normal breast tissue and primary human mammary epithelial cells
In the mammary gland all epithelial cells express EpCAM with the exception of myoepithelial cells. There were no significant changes in immunoreactivity between luminal and basal cells ( Figure 1A). In clear contrast to all tumor samples analyzed, normal polarized epithelia had a strict localization of EpCAM on the basolateral membrane. This basolateral expression got lost in tumor cells of primary breast carcinoma and metastasis which showed clearly localization on the entire cell surface (Additional file 1: Figure S1). Primary epithelial cells from healthy breast tissue (HMECs) were analyzed for their phenotype by a panel of markers specific for myoepithelial (ASMA), progenitor (p63), basal (CK5/14) and luminal (CK18) epithelial cells. As expected, all HMECs lacked luminal (CK18-) or myoepithelial (ASMA-) markers, but displayed more a basal phenotype (p63-, CK5/14+). Interestingly, in vitro cultivated HMECs were negative for EpCAM in the immunofluorescence analysis ( Figure 1B), although low transcript levels could be detected by qPCR analysis (data not shown).

Adenoviral overexpression of EpCAM inhibited cell proliferation and migration in HMECs
Based on our observations that HMECs display low endogenous EpCAM expression in 2-dimensional cultures, we overexpressed the putative EpCAM oncogene and analyzed effects on cell proliferation and migration in vitro.
Using a multiplicity of infection (MOI) of 100 viruses/ cell we obtained a strong EpCAM expression in HMECs (Figure 2A) without any effects on cell viability (Additional file 2: Figure S2B). Noteworthy, next to the native EpCAM protein on plasma membrane (Additional file 2: Figure  S2A) we found a lot of immunoreactive EpCAM in cytoplasmic organelles in our immunofluorescence analysis (Additional file 1: Figure S1D). These high amounts of cytoplasmic EpCAM might originate by overload of the intracellular vesicular traffic system with EpCAM or by a preferential detection of cytoplasmic EpCAM isoforms in our immunofuorescence analysis.
A transient, about hundred-fold overexpression was obtained over the observed time period of 5 days in all HMEC cultures ( Figure 2B). EpCAM overexpression in HMECs was also confirmed on protein level by Western Blot analysis ( Figure 2C). Interestingly, proliferating HMECs produced predominantly glycosylated isoforms (days 1 to 3), whereas in confluent and contact-inhibited cultures most of EpCAM protein was not glycosylated ( Figure 2C). The presence of different EpCAM isoforms in HMECs was confirmed by enzymatic deglycosylation experiments with the enzyme PNGaseF and subsequent Western Blot analysis ( Figure 2D). Under optimal mitotic stimulation EpCAM overexpression inhibited cell growth in proliferating HMECs as determined by the Real Time Cell Proliferation System ( Figure 2E). In comparison to control cells, EpCAM transfected cells showed elevated expression of the tumor suppressor genes, p27 Kip1 and p53 ( Figure 2F). However, these changes were visible only as a post-transcriptional regulation, on the protein level. Gene expression levels of TP53 and p27 Kip1 did not significantly change after adenoviral transfection (Additionl file 2: Figure  S2C). EpCAM overexpression resulted also in a slight, but significant inhibition of cell migration as observed by the real time cell migration measurement ( Figure 2G).

EpCAM expression is not induced by polarization processes in HMECs
Although EpCAM expression was strictly basolateral in breast epithelia in vivo, it was not expressed in our in vitro cultures of HMECs ( Figure 1B). Therefore, we concluded, that maintenance of cell polarity with functional tight-and gap-junctions is necessary for the expression of EpCAM and for further overexpression studies. HMECs were grown as mitotic cultures on collagen type I ( Figure 3A) or as confluent, polarized monolayers on 0.4 μM transwell inserts coated with Matrigel ( Figure 3B). Polarization of HMECs was controlled after 10 days by measurement of transepithelial resistance (860 ±50 Ω) and by immunofluorescence stainings for the tight junction marker ZO-1, and cell-cell contacts mediated by E-cadherin and membranous β-catenin. Cell-cell contact proteins E-cadherin and β-catenin, molecular interaction partners of EpCAM, were strongly expressed in polarized HMECs cultures ( Figure 3C). However, we could not observe elevated EpCAM protein expression ( Figure 3C).

EpCAM overexpression does not alter gene expression profile of HMECs
HMECs grown as polarized cultures or under mitotic culture conditions were adenovirally transfected to overexpress EpCAM/GFP or GFP. As expected, transient transfection resulted in a strong overexpression of EpCAM in comparison to control cells ( Figure 3D). Despite equal multiplicities of infection (MOI = 100) used for all transfections, EpCAM overexpression was stronger in polarized cells than in standard culture conditions. Based on our data on EpCAM protein expression ( Figure 2C) we isolated mRNA 24 h after adenoviral transfection to identify genes directly regulated by EpCAM and not thereafter, by induction of the transcription factor p53. Apart from the clear overexpression of EpCAM, we did not observe any significant changes in the gene expression profile of HMECs under normal and polarized culture conditions ( Figure 3E/F, Additional file 3: Table S1). These microarray data indicate that EpCAM overexpression alone does not directly affect gene transcription in HMECs either cultured in a polarized, tissue resembling culture model or under mitotic standard conditions.

EpCAM antagonizes TGF-β1 induced growth arrest
TGF-β1 acts on epithelial cells as potent growth inhibitory factor and promotes differentiation processes. Basal cells stimulated with TGF-β1 stop proliferation within 3 days. In contrast to untreated control cells, displaying a small cell body and a strong light refracting morphology, TGF-β1 treated cells changed morphology and acquired an enlarged and flat cell body ( Figure 4A/B, green signal). After EpCAM overexpression, TGF-β1 stimulated HMECs showed a higher percentage of cells with a small, strongly light-refracting morphology ( Figure 4A, arrows). Moreover, HMECs treated with TGF-β1 underwent a terminal growth arrest and stained positively for senescence-associated beta galactosidase (SA-β-Gal; Figure 4B), a marker of cellular senescence. In clear contrast, upon simultaneously EpCAM overexpression, we could observe many cell clusters that were negative for SA-β-Gal ( Figure 4B, arrows) indicating that cells were not growth arrested and maintained a longer capacity to proliferate.

EpCAM down regulates E-cadherin and prolongs proliferative lifespan of HMECs
Long term cultures of HMECs in culture medium containing TGF-β1 were analyzed for differences between EpCAM/GFP and GFP overexpression. In the presence of the differentiation factor TGF-β1 EpCAM overexpressing cells were still able to proliferate and formed bigger cell clusters after 6 days in vitro ( Figure 4C). GFP transfected control cells stopped cell divisions after 3 days and consisted predominantly of enlarged, flat and growth arrested cells ( Figure 4C). Analysis of cell numbers revealed a significant increase in cell counts in EpCAM overexpressing cells 6 days after transfection ( Figure 4D).
Additionally, we analyzed a panel of epithelial-to-mesenchymal transition (EMT) markers to define the phenotype of EpCAM overexpressing cells. In particular, HMECs showed down regulation of E-cadherin after TGF-β1 stimulation ( Figure 4E, 4F). In clear contrast to GFP control cells, EpCAM-overexpressing cells showed an additional down regulation of E-cadherin ( Figure 4E, 4F), which would explain the more spindle-shape phenotype of the cells ( Figure 4A). Another marker of EMT, vimentin expression, did not increase significantly after EpCAM overexpression in direct comparison to GFP transfected control cells ( Figure 4E, 4F).

EpCAM overexpressing HMECs form bigger xenografts consisting of p63 high progenitor cells and lack luminal structure formation
Based on in vitro findings we analyzed effects of EpCAM overexpression in our in vivo model. Therefore, we transplanted HMEC xenografts onto chicken embryos to analyze in vivo growth ( Figure 5A). Chicken embryos have only innate immune responses and thus, tolerate growth of human cells. Transfected HMECs were transplanted as growth factor-reduced matrigel drops containing TGF-β1. After 6 days in vivo growth, xenografts became well vascularized and human onplants could be visualized by expression of GFP ( Figure 5B). Macroscopically, there were no significant changes in the size of HMEC onplants, however immunohistochemical analysis of invading cell clusters in the chicken chorioallantoic membrane tissue revealed morphological and quantitative differences. Cell clusters of GFP controls were smaller, less frequent and displayed significantly more lumen formation ( Figure 5C, 5D, marked with asterisks). In contrast, EpCAM overexpressing HMECs formed bigger structures, with more frequent disseminating cell clusters and therefore, almost no lumen formation ( Figure 5C). Higher cell numbers of EpCAM overexpressing HMEC grafts were also correlating with more p63 high progenitor cells/high power field ( Figure 5D).

EpCAM overexpression enhances cell proliferation in immortalized MCF-10A cells
Based on our observation that EpCAM overexpression alone is not enough to reveal its oncogenic features and that tumorigenesis is a multistep process, we decided to use the immortalized breast epithelial cell line MCF-10A (p16 INK4a−/−) for additional investigations. MCF10A cells can be efficiently transduced with adenovirus to overexpress EpCAM, but loose EpCAM expression faster than HMECs ( Figure 6A). In comparison to HMECs ( Figure 2E), MC-F10A with EpCAM overexpression show an increased cell proliferation ( Figure 6B) and upregulation of c-myc gene expression ( Figure 6C). Changes of c-myc expression could also be monitored on protein level ( Figure 6D/E).
Moreover, MCF10A cell lines were generated by a lentiviral system to have a stable expression of a non-silencing control (ns/crtl) or an EpCAM specific (E#2) shRNA. Both cell lines, MCF10A ns/crtl and MCF10A E#2 , were adenovirally transfected to overexpress GFP or EpCAM/ GFP. In comparison to MCF10A ns/crtl cells MCF10A E#2 cells were significantly downregulating EpCAM transcript levels 24 and 48 h after adenoviral transfection ( Figure 6F). Real time cell proliferation of MCF10A E#2 cells was significantly lower than those of MCF10A ns/crtl after adenoviral EpCAM overexpression ( Figure 6G). These data clearly indicate that EpCAM overexpression can enhance proliferation and c-myc levels in immortalized human breast epithelial cells.

Discussion
EpCAM is a widely described tumor-associated antigen, stem cell and cancer stem cell marker [4,[22][23][24]. Cancer stem cells with a high EpCAM expression are considered to be more malignant and more prone to give metastasis than those with a low expression [24,25]. Although EpCAM overexpression in breast cancer is correlated with aggressive behavior and decreased overall survival of patients [13,[25][26][27], functions and effects of EpCAM overexpression in normal mammary epithelial cells, i.e. healthy tissue have not been described so far.
In normal breast epithelia EpCAM has a strict basolateral expression. Among all epithelial cell-types only myoepithelial cells are EpCAM-negative [28]. Tumor cells loosing cell-cell contacts and invading host tissue are also loosing the strict basolateral distribution of EpCAM and show more cytoplasmic and membranous staining [5,27]. Whether this is mediated by loss of cell polarity or by generation of translocated EpCAM isoforms is still under investigation [5,11,29]. Recent studies showed that glycosylation of EpCAM might affect stability and function of the protein [30]. Noteworthy, healthy tissue displays mainly weak expression of basic, not glycosylated EpCAM protein, whereas in tumor tissue, as well as in breast cancer cell lines, EpCAM is glycosylated and/or hyperglycosylated [30]. Differences in glycosylation we could also observe between highly mitotic cultures and growth arrested monolayers of transfected human mammary epithelial cells (HMECs).
In vitro cultivated HMECs showed no EpCAM protein expression, although gene transcripts could be detected by qPCR in a low abundance. Presumably, these cells loose expression under artificial in vitro conditions and loss of normal tissue polarity, since in vivo both basal/ progenitor as well as differentiated luminal cells are strongly positive for immunoreactive EpCAM. Moreover, cell-cell adhesions in our HMECs are primarily mediated by E-cadherin, which has been described to be a counter player of EpCAM [10,18,31]. Typically, HMEC cultures age under mitotic stress and induce p16 INK4A and/or p53 [32,33]. Aberrant expression of oncogenes has been shown to induce cellular senescence (oncogene-induced senescence) by activation of the p53, p16/Rb or p27 Kip1 checkpoint. These check-points of the cellular senescence program protect cells from oncogenic signaling, prevent immortalization and acquisition of genomic instabilities [34,35] and are very often inactivated in cancer cells. In comparison to control cells, overexpression of EpCAM led to inhibition of proliferation and migration in HMECs. This represents a frequently observed reaction of normal cells to an oncogenic stimulus. However, in contrast to effects described for oncogenic ras or the catalytic subunit of the telomerase (TERT) we did not observe a complete growth arrest mediated by induction of p16 INK4A [35,36]. EpCAM transfected HMECs are inhibited in cell proliferation, but do not undergo a terminal growth arrest. This might be due to simultaneous upregulation and accumulation of p53 and the cell cycle inhibitor p27 Kip1 . A crosstalk between EpCAM and p53 has already been reported [37]. EpCAM gene expression is downregulated by p53 and loss of p53 leads to increased EpCAM expression and a more invasive phenotype in tumor cells [36,37]. EpCAM did not affect p53 or p27 Kip1 gene transcription, upregulations were only visible on the protein level. Thus, EpCAM might induce changes in p53 protein by affecting posttranscriptional modifications processes or protein stability [38]. Moreover, p27 Kip1 has been shown to inhibit Rho-A driven cell migration processes [39]. Thus, our HMECs upregulating p27 Kip1 after EpCAM overexpression probably showed an inhibition of cell migration despite down regulation of the cell-cell adhesion molecule E-cadherin.
Against our expectations, EpCAM expression alone did not directly affect transcription of other genes in our HMEC culture models, although a signaling pathway, directly activated by EpCAM cleavage, has been previously described in pharyngeal cancer cells [35,36]. In fact, in HEK293 and FaDu tumor cell lines EpCAM has been reported to act directly on transcription of c-myc and cyclins [40,41]. We transfected growth arrested and polarized, as well as proliferating HMEC cultures and performed transcriptome analysis 24 h after overexpression of EpCAM. With this experimental approach we wanted to identify early genes directly regulated by EpCAM, before induction of the transcription factor p53 and its downstream genes. In fact, EpCAM is primarily acting on cell-cell adhesion proteins such as E-cadherin, claudins, tetraspanins and CD44 [42]. Changes on their protein levels, localization on the cell membrane and interactions, might affect intracellular signaling pathways and kinase activities. Indeed, it has been recently reported that EpCAM affects protein kinase C signaling and cell migration processes during gastrulation in xenopus embryos [43].
HMECs are very sensitive to the cytokine TGF-β1 treatment [36]. This cytokine is able to inhibit cell proliferation and induce EMT differentiation processes in healthy epithelial cells [44,45]. When HMECs are transfected to overexpress EpCAM many clones acquire resistance to TGF-β1 induced growth arrest and display more spindleshape phenotype. The underlying mechanism for increased resistance to TGF-β1 mediated growth arrest still remains to be investigated. Further, our in vivo studies support the concept of EpCAM overexpression as supportive factor for hyperplastic growth. EpCAM overexpression together with TFG-β1 and presumably other mitogenic factors present in Matrigel support hyperplastic growth and counteract growth arrest and terminal differentiation processes in vivo. We assume that HMECs with EpCAM overexpression gain longer proliferative capacities and acquire more resistances to growth inhibition due to activation of Wnt signaling. This increased stem cell signaling is supported by the observation that EpCAM overexpressing xenografts display an increased number of p63 + undifferentiated progenitor cells. This is of particular interest, since higher amounts of undifferentiated cells in mammary gland contribute to increased risk to develop breast cancer [46]. Moreover, EpCAM overexpression leads to stronger innate immune responses in vivo. EpCAM overexpressing xenografts attracts more neutrophils from host tissue, which would suggest that EpCAM is supporting migration processes of immune cells as described previously for dentritic cells [47]. However, further investigations are necessary to study effects of EpCAM expression on cancer cells in context of tumor immunology and microenvironment.
Thus, EpCAM overexpression might promote progression and metastasis of primary tumors. However, further studies are still needed to identify the underlying molecular mechanisms responsible for EpCAM overexpression in the context of TGF-β/Wnt signaling and breast cancer development. This background will allow us to understand the impact of EpCAM overexpression on transformation of breast epithelial cells and growth of breast cancer cells.

Conclusions
EpCAM revealed oncogenic features in normal human breast cells, inducing resistance to TGF-β1-mediated growth arrest and supporting a cell phenotype with longer proliferative capacities in vitro. EpCAM overexpression resulted in hyperplastic growth and enhanced innate immune responses in vivo. Thus, we suggest that EpCAM acts as a prosurvival factor counteracting terminal differentiation processes in normal mammary glands.

Ethical standards
This study on commercially available tissue sections and primary cells and cell lines did not need approval of the local ethic committee of the Medical University of Innsbruck. Handling of animals was conducted in compliance with Austrian State laws. Since the CAM assay, i.e. chicken embryo is an alternative model to replace animal experiments according to Austrian law an ethical approval is not required.

Additional files
Additional file 1: Figure S1. Representative immunohistochemical staining of EpCAM in healthy tissue (A), primary invasive ductal carcinoma (B) and corresponding lymph node metastasis (C) Note: The strict basolateral expression of EpCAM in healthy glandular tissue gets lost in tumor cells in favor to a signal on the entire cell membrane. (D) Immunofluorescence analysis of EpCAM transfected HMECs. Adenoviral transfected cells were fixed, permeabilized and stained with an EpCAM specific antibody (green signal); nuclei were counter-stained with ToPRO-3 (red signal). Bars indicate 100 μm.
Additional file 2: Figure S2. Flow cytometry analysis of EpCAM expression on cell membranes of viable HMECs after adenoviral transfection with EpCAM or GFP. EpCAM high cells were quantified in direct comparison to GFP transfected controls (A). Apoptosis/necrosis of HMECs was analyzed 48 h after transfection by staining of Annexin V/ propidium iodide and flow cytometric analysis (B). Relative gene