Background
E-cadherin, encoded by the tumor suppressor gene
CDH1, is a homophilic cell-to-cell adhesion protein localized to the adherens junctions of all epithelial cells [
1]. Its cytoplasmic domain effectively creates a bridge between the cytoskeletons of adjacent cells by interacting with both cortical actin filaments and the microtubule network [
2]. These and other interactions [
3] extend E-cadherin’s functionality beyond cell-cell adhesion to roles in establishing and maintaining cell polarity, differentiation, stemness, cell migration and the mediation of signalling through various proliferation and survival pathways including WNT and EGFR [
1‐
5].
Abrogation of
CDH1 expression by mutation, deletion or promoter hypermethylation is a feature of many epithelial tumours, including prostate, ovarian, lung and hepatocellular carcinomas, and is the hallmark of both the sporadic and familial forms of diffuse gastric cancer (DGC) and lobular breast cancer (LBC) [
1,
6]. In both LBC and DGC,
CDH1 inactivation can be an early initiating event [
7,
8], whereas in other tumour types including prostate, lung, ovarian and colon, its downregulation is usually considered to be a late event that promotes an increase in invasive capacity [
9]. Increased invasiveness following
CDH1 downregulation is related, at least in part, to the central role played by E-cadherin in the de-differentiation process known as the epithelial-mesenchymal transition (EMT) [
10]. During the EMT, epithelial cells lose polarity and normal cell-cell adhesion, acquiring a mesenchymal phenotype with higher motility and an increase in cell-extracellular matrix (ECM) connections [
9,
11]. The EMT is associated not only with increased tumor invasion and metastasis, but also poor outcome, drug resistance and an increase in the number of cancer stem-like cells [
9,
12]. E-cadherin downregulation has been shown to be sufficient to induce an EMT in some [
4,
9,
10,
13], but not all [
14,
15], cancer cell lines/models. However, it remains unclear whether its loss can induce an EMT in cells which have not already undergone malignant transformation [
16].
Clues to the influence E-cadherin loss has on tumorigenesis and the initiation of the EMT come from study of the multifocal gastric signet ring cell carcinomas (SRCCs) that occur in Hereditary Diffuse Gastric Cancer (HDGC) families. HDGC is a familial cancer syndrome caused by germline mutation of the
CDH1 gene and is typified by highly penetrant DGC and an elevated risk of LBC [
17]. With few exceptions, mutation carriers develop tens to hundreds of gastric foci of SRCC, sometimes with enrichment in the transition zone between the antrum and body [
18]. LBC and lobular carcinoma
in situ (LCIS) are also observed to be multifocal in female mutation carriers (V.Blair, pers. comm). The multifocal gastric SRCCs are E-cadherin-negative and almost exclusively stage T1a tumours confined to the
lamina propria. Lineage markers suggest that the foci develop from mucous neck cells that have invaded through the basement membrane of the gastric gland [
19]. Invasion is likely to be triggered by inactivation of the wild-type
CDH1 allele through mechanisms including promoter hypermethylation [
6]. In one model [
20], E-cadherin loss creates instability in the orientation of the mitotic spindle, leading to a proportion of the cell divisions occurring out of the epithelial plane with subsequent displacement of daughter cells into the
lamina propria. The multifocal SRCC foci in the gastric mucosa are known to be relatively indolent, but show unpredictable progression to advanced disease. A small percentage of foci show characteristics of an EMT, and this change is associated with tumour progression [
19]. However, the absence of an EMT-like phenotype from the majority of SRCC foci suggests that E-cadherin loss alone is insufficient to induce an EMT in this relatively normal genetic background.
MCF10A is a spontaneously immortalized, non-transformed mammary epithelial cell line derived from human fibrocystic tissue. Although it does carry cytogenetic abnormalities associated with
in vitro cultured mammary epithelial cells, including
p16 and
p14ARF deletion and
MYC amplification [
21], MCF10A is considered a “normal” breast epithelial cell due to its near diploid, stable karyotype and characteristics of normal breast epithelium such as lack of tumorigenicity in nude mice, lack of anchorage-independent growth [
22] and ability to form mammospheres in culture [
21]. Here we have used cell-based assays and whole genome RNAseq to characterize an isogenic MCF10A cell line that is devoid of
CDH1 expression due to an engineered homozygous 4 bp deletion in
CDH1 exon 11. We show that E-cadherin loss disrupts the organization of the cell’s actin and microtubule cytoskeletons and modifies its adherence and migration characteristics but is insufficient to induce an EMT.
Methods
Cell culture
MCF10A cells (product no: CRL 10317), a non tumorigenic mammary epithelial cell line, and the derived isogenic line with
CDH1 knock out (MCF10A
CDH1-/-) using CompoZr ZFN technology (product no: CLLS1042) were purchased from Sigma. The MCF10A isogenic lines were cultured in DMEM/F12: (1:1) (Invitrogen) with 5% horse serum (Invitrogen), 10 μg/ml Actrapid Penfil neutral insulin (Novo Nordisk Pharmaceuticals Ltd), 20 ng/ml human epidermal growth factor (Peprotech), 100 ng/ml cholera toxin, and 500 ng/ml hydrocortisone (Sigma) [
21]. Cells were grown at 37°C with 5% CO
2, seeded into T75 flasks at densities of 3.0 × 10
5 and 4.5 × 10
5, respectively and passaged at 90% confluency (~3 days) for a maximum of ten passages (
http://brugge.med.harvard.edu/protocols).
Western blot
MCF10A and MCF10A CDH1-/- cells were grown for 72 h to 90% confluency in T25 flasks and lysed using cell culture lysis reagent (Promega) containing cOmplete mini protease inhibitor (Roche). BCA assays (Thermo) were performed to equalize total protein loaded. Proteins were separated on 10% SDS-PAGE gel for 2 h, followed by blot transfer at 100 V for 1 h. Immunoblotting was performed using rabbit anti-E-cadherin antibody (Santa Cruz, SC7870) at 1:200 dilution overnight, or rabbit anti-α-actin primary antibody (Sigma) at 1:1,500 dilution overnight followed by anti-rabbit HRP-linked secondary antibody (Santa Cruz) at 1:5,000 dilution for 1 h. Chemiluminescence was performed using Pierce ECLplus reagent (Thermo) and imaged using LAS-3000 (Fujifilm).
Immunofluorescence
MCF10A and MCF10A CDH1-/- cells were seeded on Coverglass slides (Labtek) and grown to confluence for 72 h. Cells were fixed with 4% paraformaldehyde then permeabilized with 0.2% Triton-X100 in PBS for 5 min at room temperature. Cells were blocked with 10% FBS in PBS for 1 h at room temperature. E-cadherin primary antibody (Santa Cruz, SC-7870) used at 1:250. Anti-rabbit secondary antibody conjugated with AlexaFluor-488 (Invitrogen) used at 1:750. Immunofluorescence images were acquired with an Olympus IX71 microscope, under 40× objective.
Proliferation assay
MCF10A and MCF10A CDH1-/- cells were seeded at densities of 2.0 × 103 and 4.0 × 103 in three replicates in 96 well E-plates and incubated at 37°C in 5% CO2. The growth rate was monitored in real time at 15 min intervals for 96 h using the xCELLigence platform (Roche). Both cell lines were also seeded at the same densities into 96 well flat clear bottom black plates (Corning) and grown at 37°C in 5% CO2 and imaged every 2 h for 96 h using the IncuCyte 2011A FLR (Essen Bioscience). Confluency was determined using the IncuCyte software Confluence v1.5.
Cell adhesion assay
Cell adhesion assays were performed using the IncuCyte 2011A FLR. MCF10A and MCF10A
CDH1-/- cells were seeded in six replicates at 2.0 × 10
4 cells per well in 96 well flat clear bottom plates (Greiner Bio-one) with different surface coatings: no coating for the uncoated, 2 μg/ml collagen (Sigma), 2 μg/ml fibronectin (BD Bioscience), 8 μg/ml vitronectin (Invitrogen), 8 μg/ml laminin (Invitrogen) and grown at 37°C, 5% CO
2. Images were acquired every 2 h for 8 h using the automated image acquisition software. Cell numbers at each time point were also determined using the Cell Counter plugin (
http://rsbweb.nih.gov/ij/plugins/cell-counter.html) in ImageJ [
23].
Scratch wound assay
Scratch wound assay was performed using the IncuCyte 2011A FLR (Essen Bioscience). Briefly, MCF10A and MCF10A CDH1-/- cells were seeded in six replicates at densities of 2.5 × 104 and 3.5 × 104 cells per well, respectively, in 96 well Essen ImageLock Plate (Essen Bioscience) with different coating surfaces: no coating for the uncoated condition, 2 μg/ml collagen, 2 μg/ml fibronectin, 8 μg/ml vitronectin, 8 μg/ml laminin. Cells were incubated at 37°C and 5% CO2 and grown to 100% confluency. The usage of the Essen imageLock plates ensures wounds are automatically located and registered by the IncuCyte software and analyzed using wound confluence metrics. Precise and reproducible wounds were generated using the 96 PTFE pin WoundMaker (Essen Bioscience) on the confluent monolayer and cells returned to the incubator where images of cells were acquired every 1 h for 35 h under phase contrast microscopy. Wound confluence was graphed over time to quantitatively evaluate the characteristic of wound closing using the IncuCyte software, Wound Confluence v1.5.
Soft agar assay
An overlay of 2.0 × 104 MCF10A and MCF10A CDH1-/- cells and 2.0 × 103 MCF7 cells in 0.35% agar in medium were plated over a base layer of 0.5% agar (Applichem) and grown at 37°C with 5% CO2. Growth medium was added the next day and replenished twice a week. After 24 days, growth medium was removed and MTT (Sigma) solution was added (final concentration 2 mg/ml), and the plates further incubated at room temperature for 4 h with gentle shaking. The MTT solution was then removed and washed. Images were taken using Image Scanner 3 and colonies counted. The experiment was performed with at least two technical replicates for each cell line.
Immunofluorescence confocal microscopy
MCF10A and MCF10A
CDH1-/- cells were seeded on glass coverslips coated with fibronectin (Becton Dickinson) and allowed to grow to confluence for 48-72 h. Cells were fixed with ice-cold methanol for 5 min on ice for microtubule staining or fixed with 4% paraformaldehyde in cytoskeleton stabilization buffer (10 mM PIPES pH 6.8, 100 mM KCl, 300 mM sucrose, 2 mM EGTA, 2 mM MgCl2) on ice for 20 min and then permeabilized with 0.25% Triton-X100 in PBS for 5 min at room temperature for F-actin staining. Cells were blocked with 5% milk in PBS for 1 h at room temperature. Primary antibodies used: mouse monoclonal antibody (mAb) directed against the ectodomain of E-cadherin (HECD-1) (a gift from Peggy Wheelock, University of Nebraska, Omaha, NE; with the permission of M. Takeichi) 1:50; rabbit polyclonal Ab (pAB) against E-cadherin (generated in-house) [
24] 1:1000; rat monoclonal [YOL1/34] antibody against tubulin (Abcam, # ab6161); 1:100 rabbit polyclonal antibody against ZO-1 (Invitrogen, # 61-7300). F-actin was stained with AlexaFluor 488-phalloidin, 1:500 (Invitrogen). Secondary antibodies were species-specific antibodies conjugated with AlexaFluor-488, -594 or -647 (Invitrogen) for immunofluorescence (1:500). For immunofluorescence, confocal images were acquired with a Zeiss 710 Meta laser scanning confocal microscope, with a 60× objective, 1.4 NA oil Plan Apochromat immersion lens with 0.6-1.0 μm optical sections. Contrast adjustment and z-projections of raw images were done using ImageJ software (National Institutes of Health) [
23] and Illustrator (Adobe).
RNASeq
MCF10A and
MCF10A CDH1-/- cells were seeded at densities of 2.0 × 10
5 and 3.5 × 10
5 cells respectively in duplicate in a six well dish and grown until 70% confluency, with a medium change at 24 h. Total RNA was extracted at 48 h post seeding using quick-RNA Miniprep Kit (Zymo) according to the manufacturer’s protocol. RNA yield and purity were assessed using Qubit (Invitrogen) and the Agilent 2100 Bioanalyser. cDNA library preparation was performed by New Zealand Genomics Limited using Illumina TruSeq RNA preparation version 2.0. Each library had inserts of 200 bp and sequence reads were generated from one lane of an Illumina HiSeq™ 2000 run. Bowtie2 and Cufflinks version 2.0.1 software packages were used to align the read data to human genome build GRC37 and annotated with BiomaRt using Ensembl dataset ”hsapiens_gene-_ensembl”. Unannotated genes were removed and remaining count data was normalized using EdgeR [
25]. The per gene read counts were imported into the statistical software package R (
http://www.r-project.org), and analyzed using the functionality included in the edgeR and limma packages. Briefly, TMM (trimmed mean of M values) normalization was applied to generate normalized count data, and the lmFit command was used to fit a linear model to the data for each gene. Normalized data were converted to log-cpm (counts per million reads) prior to analysis using the voom command in limma. Differential expression results for MCF10A
CDH1-/- vs MCF10A were written to CSV files, viewable in Excel (limma moderated t- statistic produced for each comparison, per gene, with FDR p value adjustment applied). Gene Ontology (GO) functional enrichment analysis was carried out using GATHER [
26].
Discussion
CDH1 is widely considered to be an ‘invasion suppressor gene’ whose inactivation is associated with tumor progression. However, the identification of large numbers of early stage gastric and breast cancers in
CDH1 germline mutation carriers [
18,
31] demonstrates that E-cadherin loss can also influence the initiation of cancer. To provide a better understanding of the impact of E-cadherin loss in a non-malignant genetic background, we have characterized an E-cadherin-deficient cell line that has been derived from the non-tumorigenic breast line MCF10A using zinc finger nuclease technology.
MCF10A
CDH1-/- cells largely retained an epithelial cobblestone morphology, although slightly more rounded cells and gaps were observed in the confluent monolayer when compared to wildtype cells. Our transcription analysis suggests that the retention of cell adhesion appears to be associated with compensatory changes in other cell-cell adhesion proteins localized at desmosomes, tight junctions and adherens junctions, including P-cadherin (
CDH3). Similarly, the targeted loss of E-cadherin from murine skin epithelium is also not associated with a significant loss of cell-cell adhesion, an effect attributed to compensatory upregulation of P-cadherin and desmosomal cadherin in the basal layer [
32,
33]. Despite the loss of E-cadherin from the adherens junction, a level of normal cell polarity is retained in the MCF10A
CDH1-/- cells, based on the apical ZO-1 localization (Figure
2b) and the ability of these cells to form luminal cores in 3D matrigel culture (data not shown). This is consistent with the polarity retention observed in the conditional knockout mouse whereby E-cadherin loss in the epidermis did not alter Par3 and Scribble localization [
34]. However, E-cadherin depletion disrupts cell polarization in MDCK cells [
35] and can lead to abnormal mitotic spindle orientation in different model systems [
20].
The cadherin adhesion complex interacts dynamically with the actin and microtubule cytoskeletons through various multiprotein complexes, allowing mechanosensing, force transmission, vesicle mediated transport of junctional proteins to the zonula adherens, and the regulation of microtubule stability and orientation [
2].
CDH1-deficient MCF10A cells lacked the radiating microtubule structure observed on the apical side of the wildtype cells (Figure
2a), consistent with a disruption of either plus-end or minus-end microtubule anchoring at the adherens junctions, [
36‐
38]. Not surprisingly, this microtubule reorganization was also associated with downregulation of genes involved in the adherens junction-microtubule axis including
KIFC3, a gene involved in microtubule minus-ends directed motor found to localize to both the zonula adherens and centrosome and also
NIN which encode ninein, involved in anchorage to the centrosome.
E-cadherin loss caused no evident difference in the organization of F-actin in the apical region of the MCF10A cells. However, in the basal region there was an increased prominence of stress fibers (Figure
2b) suggesting that basal contractility and traction forces (driven by Rho signalling) might be potentiated. Similar thickening and lengthening of actin stress fibres have been observed in mouse mammary epithelial, NMuMG cells following TGF-β induced EMT [
39]. Associated transcriptional changes in the E-cadherin-deficient cells, included upregulation of
RhoA,
RhoB and
RhoC, although their downstream effectors
ROCK1 and
ROCK2 were downregulated.
One striking characteristic of the
CDH1-deficient MCF10A cells was the reduction in cell-substrate adhesion. Multiple transcriptional changes consistent with this phenotype were observed, including the downregulation of genes encoding ECM proteins [
40], integrin subunits and focal adhesion proteins involved in linking integrins with actin stress fibers such as talin, paxillin, tensin and P130Cas. The most marked integrin downregulation was of
ITGA1, ITGA4, ITGA5, ITGAV, ITGB1 and
ITGB2, genes that encode subunits of the α1β1, α2β1, α3β1, α4β1, α5β1, αvβ1, α1β2 integrin receptors, respectively [
27]. The crosstalk between E-cadherin and integrin has been observed previously in major cellular functions including adhesion, migration, proliferation and apoptosis [
41]. This is not surprising as both transmembrane adhesion receptors share some common signaling effector molecules, scaffold and cytoskeletal proteins, hence the combined ability to influence coordinated regulation of cell-cell and cell-substrate adhesion crucial in normal cell growth and disease state [
10,
41]. The decreased substrate adhesion also translated to reduced cell motility in MCF10A
CDH1-/- compared to wildtype cells. Another E-cadherin depletion study of MCF10A also showed no gain in cell migration speed [
30]. The compromised cell-substrate adhesion and migration could be partially restored when ECM proteins were coated onto the growth surface, however the downregulation of talin (encoded by
TLN1 and
TLN2) and other genes involved in focal adhesion assembly and disassembly like
TNS1, TSN3 and
UTRN has probably further compromised traction force [
42,
43].
The EMT incorporates a series of coordinated events which involve altered cell-cell and cell-ECM interactions, reorganization of the cytoskeleton and the adoption of a new transcription program to induce and maintain a mesenchymal phenotype. We found little support for E-cadherin loss being able to initiate an EMT in a non-malignant genetic background; deletion of
CDH1 from MCF10A cells was not associated with the reduction in epithelial markers and the coordinated increase in mesenchymal markers such as
CDH2[
44], nor were the EMT regulators
TWIST1,
TWIST2,
SNAI1,
SNAI2,
ZEB1 and
ZEB2 upregulated [
9,
11,
45]. Treatment of MCF10A cells with either TGF-β or FN caused an EMT without downregulation of E-cadherin [
39,
46], supporting our observations that
CDH1 loss does not drive the EMT in these cells. A recent study done in a panel of 38 breast cell lines also indicated that E-cadherin loss is not causal for EMT in human breast cancer [
15].
The only EMT feature that was clearly activated in the MCF10A
CDH1-/- cells was increased expression of several metalloprotease genes (
MMP9,
MMP14,
MMP15,
MMP17 and
MMP28). Increased expression of
MMP2 and
MMP9 has been shown previously in MCF10A cells following TGF-β and ERBB2 induced EMT [
47]. We predict that the impact of elevated MMP expression would be detectable in invasion assays using 3D matrices like Matrigel, although such assays were not part of our analysis. In addition, several genes from the S100A calcium-binding protein family,
S100A7,
S100A8 and
S100A9, were also strongly upregulated (≥2 fold) in the MCF10A
CDH1-/- cells.
S100A7, while not expressed in normal epithelia, is frequently seen to be expressed in pre-invasive ductal carcinoma
in situ[
48] and is associated with EMT.
Acknowledgements
We would like to thank Dr. Michelle McConnell and Ms Clare Fitzpatrick (Department of Microbiology and Immunology, University of Otago) for assistance with the xCELLigence realtime system, Dr. Adele Woolley and Mr Michael Algie (Department of Pathology, University of Otago) for the use of the IncuCyte 2011A FLR, Dr. Sofie Van Huffel for technical assistance with the soft agar assay and Andrew Single, James Frick and Tom Brew for their assistance in cell counting for the cell adhesion assay.
Guilford lab funding: “This work is supported by the Health Research Council of New Zealand (11/513). HB and BJT are supported by the University of Otago Doctoral Scholarship”.
Yap lab funding: “Work in the Yap group was supported by project grants (1010489) and fellowships (1044041) from the National Health and Medical Research Council of Australia. RP is supported by a UQI (UQ International) Ph.D. Scholarship and an ANZ Trustees Ph.D. Scholarship in Medical Research. Confocal microscopy was performed at the ACRF/IMB Cancer Biology Imaging Facility established with the generous support of the Australian Cancer Research Foundation”.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made.
The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder.
Competing interest
The authors declare that they have no competing interest.
Authors’ contributions
AC participated in the design of the study, performed the initial characterization experiments and the drafting of the manuscript. PJG* participated in the design of the study and writing of the manuscript. HB performed the cell adhesion, scratch and soft agar assays. HB and JG performed the RNAseq experiment and MAB performed the bioinformatics analysis. RP performed the confocal immunofluorescence. ASY supervised RP and participated in the writing of the manuscript. BJT contributed to the layout and editing of the figures. JG, BJT, GARW, TDG contributed to the analysis of the cell adhesion assay. All authors read and approved the final manuscript.