Soluble E-cadherin promotes cell survival by activating epidermal growth factor receptor

Abstract

High levels of the soluble form of E-cadherin can be found in the serum of cancer patients and are associated with poor prognosis. Despite the possible predictive value of soluble E-cadherin, little is understood concerning its patho-physiological consequences in tumor progression. In this study, we show that soluble E-cadherin facilitates cell survival via functional interaction with cellular E-cadherin. Exposure of cells to a recombinant form of soluble E-cadherin, at a concentration found in cancer patient’s serum, prevents apoptosis due to serum/growth factor withdrawal, and inhibits epithelial lumen formation, a process that requires apoptosis. Further, soluble E-cadherin-mediated cell survival involves activation of the epidermal growth factor receptor (EGFR) and EGFR-mediated activation of both phosphoinositide-3 kinase (PI3K)/AKT and ERK1/2 signaling pathways. These results are evidence of a complex functional interplay between EGFR and E-cadherin and also suggest that the presence of soluble E-cadherin in cancer patients’ sera might have relevance to cell survival and tumor progression.

Introduction

E-cadherin is a calcium dependent cell–cell adhesion molecule in epithelial cells and alterations in E-cadherin function have been linked to carcinogenesis in a variety of cancers [1]. Recent studies revealed that E-cadherin not only functions in cell–cell adhesion but also interacts with and modulates several signal transduction pathways [1]. In addition to its well-known interaction with β-catenin, a component of the Wnt signaling pathway, there is considerable evidence that E-cadherin functionally interacts with epidermal growth factor receptor (EGFR) at the sites of cell–cell contact [1–3]. The exact role of this association is unclear. While oncogenic activation of EGFR results in disruption of E-cadherin mediated cell–cell adhesion and E-cadherin internalization, alterations to E-cadherin’s cell adhesive functions also result in permissive activation of EGFR [1,4–7]. Establishment of E-cadherin mediated cell–cell adhesion can activate EGFR, triggering signaling pathways downstream of EGFR (e.g., MAPK and PI3K) [1,3,4,6–9]. These studies are indicative of a complex regulatory interaction between EGFR and E-cadherin proteins, and warrant further investigations to understand the patho-physiological relevance of the functional interplay between these two proteins.

Apoptosis or programmed cell death is induced by a diverse set of stimuli, including hypoxia, metabolic stress, or loss of contact with extracellular matrix (anoikis) and/or adjoining cells. E-cadherin appears to play a critical role in the latter, as inhibition of cadherin function can result in rapid cell death [10–12]. Further, maintenance of E-cadherin mediated cell–cell adhesion prevents cell death in transformed cells [11]. Additional evidence suggests that the cell survival function of E-cadherin may stem from its association with EGFR and EGFR related family members, implying an important physiological role for EGFR-E-cadherin association [9,11,13].

Clinical studies identified a soluble form of E-cadherin in sera of patients with different types of carcinoma [14–17]. Soluble E-cadherin results from cleavage of the E-cadherin extracellular domain by matrix metalloproteinases, releasing it from the cell membrane into the extracellular milieu (see schematic, Fig. 1A, 2.) [18–20]. It ranges from ~2 μg/ml in low-grade cancers to 10 μg/ml in patients with high grade cancers and appears to be predictive of patient survival [15,21]. Despite the wealth of clinical data available, little is known whether soluble E-cadherin has any patho-physiological significance or contributes to cancer progression.

Fig. 1 – Purification of recombinant SECAD and E-cadherin binding.

(A) Schematic overview comparing full length E-cadherin (1), physiological soluble E-cadherin (2) and recombinant SECAD (3) used in this study. TM, transmembrane domain; N, amino-terminal end; C, carboxy-terminal end; MYC, myc-tag epitope; 6× His, a stretch of 6 histidine residues used for protein purification via Ni column. Leader sequence to facilitate secretion is shown in gray. (B) Coomassie stain of purified SECAD and increasing amounts of bovine serum albumin (BSA) used for titration to quantify purified SECAD. 10 μl of purified SECAD (80 kDa) or collected media after passage over column (FT) was separated by SDS-PAGE and stained. (C) SECAD associates with E-cadherin. MDCK cells were incubated with or without 10 μg/ml SECAD before lysis and immunoprecipitation with cytoplasmic domain-specific E-cadherin antibody or control IgG antibody. After separation by SDS-PAGE, blots were incubated with antibody against E-cadherin (120 kDa), β-catenin (92 kDa) and myc-tag antibody to detect SECAD (80 kDa). β-catenin co-immunoprecipitation indicates that SECAD binding to cell-bound E-cadherin does not affect catenin association to E-cadherin. To confirm equal loading, respective inputs of the lysates used for immunoprecipitation were also blotted for E-cadherin. (D) SECAD binds to the cell surface of MDCK cells. Following biotinylation of cell surface proteins, MDCK cells were incubated with or without 10 μg/ml of SECAD. Biotinylated proteins were precipitated with streptavidin beads. Anti-myc tag antibody was used to identify SECAD. As a control for successful biotinylation, cell surface expression levels of Na,K-ATPase β-subunit are shown. Representative blots from three independent experiments are shown. To confirm equal loading, respective inputs of the lysates used for precipitation were blotted for annexin II.

Fig. 2 – SECAD prevents cell death due to serum withdrawal.

(A) Phase contrast images of serum starved MDCK cells treated with 10 μg/ ml of SECAD or without SECAD (Control). Note the vacuoles within the cytoplasm of control cells (arrows) but not in SECAD-treated cells. After 48 h of serum withdrawal SECAD-treated cells appear viable but control cells are present detached in the media. Bars, 20 μm. (B) Disruption of E-cadherin function prevents inhibition of cell death by SECAD. Cell viability was determined by measurement of LDH activity after 48 h of serum withdrawal (Control, SECAD) or from cells cultured in serum-containing medium (Serum). Cell viability in SECAD-treated cells in calcium containing medium (black bars) was 89.2%±3.5% compared to 37.4%±2.6% in control cells and similar to MDCK cells maintained with serum (98.9%±1.7%).Cell viability in SECAD treated cells in the presence of low calcium medium(white bars) (51.5%±1.0%) was similar to control cells (41.6%±1.8%). In the presence of serum, cell viability was 99.5%±1.3%, despite the absence of calcium. Standard error bars represent the average of three independent experiments done in quadruplicate. (C) Transmission electron microscopy of control and SECAD treated cells 24 h after serum withdrawal. Control cells show ruptured mitochondria (black arrows) and vacuoles (white arrow), which are not present in SECAD (10 μg/ml) treated cells. Bar, 0.5 μm. (D) Subcellular localization of cytochrome C. Immunofluorescence of cytochrome C (green) localizing to mitochondria (red) 24 h after serum withdrawal. DAPI (blue) marks nuclei. Note the co-localization of cytochrome C with mitochondria (yellow) in SECAD (10 μg/ml) treated cells. Bar, 15 μm. (E) Immunoblot of Caspase 3 and its cleaved form in control MDCK cells maintained under serum-free conditions, treated with SECAD (10 μg/ml), or cultured in the presence of serum. 48 h after serum withdrawal, only control cells show the cleaved, active form of Caspase 3. Blot represents data from two independent experiments. (F) Immunofluorescence of active Caspase 3 after 24 h of serum withdrawal. Note that cleaved, active Caspase 3 (green) is detected only in control, but not in SECAD (10 μg/ml) treated cells or in cells cultured with serum. Propidium iodide marks nuclei. Bar, 15 μm.

In this study we provide evidence that a clinically relevant dose of a recombinant form of soluble E-cadherin (SECAD) prevents cell death due to serum withdrawal and functions as a potent inhibitor of apoptosis. Moreover, at concentrations that had no effect on cell–cell adhesion, SECAD prevented normal development of polarized acini, a process that involves programmed cell death. We further show that SECAD mediates its anti-apoptotic effect through activation of EGFR and its downstream pathways. Thus, these findings for the first time demonstrate that SECAD acts as a survival factor and signals through EGFR to mediate its anti-apoptotic effects and suggest that soluble E-cadherin present in patients’ sera might contribute to cancer progression by inhibiting apoptosis in cancer cells.

Materials and methods

Cell lines and cell culture

MDCK and HEK-293 cell lines were purchased from American Type Culture Collection (Manassas, VA). MDCK-T151 cells were a kind gift from Dr. James Marrs, Indiana University. Cell lines were maintained in DMEM supplemented with 10% fetal bovine serum, MEM nonessential amino acid solution (Invitrogen, Carlsbad, CA) and penicillin/streptomycin.

Reagents

Antibodies against E-cadherin (DECMA-1 clone, Sigma, St. Louis, MO; clone 34, BD Biosciences, San Jose, CA), β-catenin (BD Biosciences), annexin II (BD Biosciences) phosphorylated tyrosine (clone 4G10), a sheep polyclonal antibody that recognizes canine EGFR (Millipore, Billerica, MA), Myc tag (clone 9B11), Caspase 3, Cleaved Caspase 3, ERK1/2, phosphorylated ERK1/2, AKT, phosphorylated (ser473) AKT and cytochrome C (Cell Signaling Technology, Danvers, MA), were purchased from the indicated vendors. Na,K-ATPase β-subunit antibody was a kind gift from Dr. William J Ball, Jr., University of Cincinnati, OH. Mitotracker^™, anti-mouse and anti-rabbit conjugated to Alexa 488^™, phalloidin-Alexa 546^™, and To-PRO3 were purchased from Molecular Probes/Invitrogen. LY294002, PD98059, AG1478 and CL-387,785 were purchased from EMD Biosciences (San Diego, CA) and diluted in DMSO as stock solutions (10 mM).

Generation and purification of SECAD

To construct the recombinant form of SECAD (with His and Myc tags), the extracellular region of canine E-cadherin (positions 541 to 2169 of the predicted sequence: Genebank ID XM_536807) was amplified from canine E-cadherin cDNA, cloned into the pSEC-Tag2 mammalian expression vector (Invitrogen) and confirmed by DNA sequencing. The pSEC-Tag2-SECAD vector was transfected into HEK-293T cells and stable clones were selected with Zeocin (Invitrogen). SECAD secretion was confirmed by immunoblotting of conditioned media with Myc tag-specific antibody. For SECAD purification, SECAD-HEK-293 cells were grown in Ultra-DOMA-PF (Cambrex, Rutherford, NJ) media for 48 h. Cell debris was removed by centrifugation, Imidazole (5 mM) was added and the medium was equilibrated to pH 7.4. The 5 ml HisTrap (GE Healthcare, Piscataway, NJ) columns were equilibrated with 20 ml of phosphate buffer (160 mM phosphate, 4 M NaCl, 5 mM Imidazole, pH 7.4), followed by passage of conditioned medium. The columns were washed with50 mlphosphate buffer containing10 mMImidazole to remove non-specifically bound proteins. For the elution of SECAD, 100 ml of phosphate buffer containing 40 mM Imidazole was passed over the column and collected. The eluted SECAD was dialyzed overnight against sterile PBS at 4 °C, and concentrated by centrifugation using Centriplus centrifuge filter devices (Millipore). Concentration of the purified SECAD was determined by using the BioRad protein DC kit (Hercules, CA) and purity was assessed by Coomassie staining after SDS-PAGE. Purified, concentrated SECAD was aliquoted and stored at −70 °C until use.

Immunoblotting

MDCK cells were serum-starved overnight and treated with SECAD for indicated times. When required, samples were pre-incubated with PI3-kinase, ERK1/2 or EGFR inhibitor (LY294002, PD98059, AG1478, respectively) in serum-free media for 1 h before addition of SECAD. Total protein lysates were prepared in 20 mM Tris (pH 7.5), 150 mM NaCl, 1 mM EDTA, 1 mM EGTA, 1% Triton X-100, 2.5 mM sodium pyrophosphate, 1 mM β-glycerolphosphate, 1 mM sodium vanadate, 1 mM phenylmethylsulfonyl fluoride and 5 μg/ml of antipain, leupeptin, and pepstatin (protease inhibitor cocktail). 100 μg of cell lysate was separated by 10% SDS-PAGE, and transferred onto nitrocellulose membrane. For immunoblotting, membranes were blocked in 5% nonfat milk in tris-buffered saline/0.1% Tween 20 (TBST). Primary antibodies were diluted in either 5% nonfat milk/TBST (Myc Tag, DECMA-1, β-catenin, EGFR) or in 5% bovine serum albumin/TBST (P-AKT, P-ERK1/2, P-Tyr, ERK1/2, AKT, Caspase 3 and Cleaved Caspase 3) and incubated overnight at 4 °C. HRP-conjugated secondary antibodies were diluted in 5% nonfat milk/TBST and incubated at room temperature for one hour. Blots were developed with Enhanced Chemiluminescene Plus (GE Healthcare).

Immunoprecipitation

For SECAD-E-cadherin co-immunoprecipitations, MDCK cells were incubated with or without SECAD in serum-free DMEM for 1 h at 4 °C. Cells were washed and lysed in 10 mM Tris–HCl, 1% Triton X-100, 1 mM EGTA and protease inhibitor cocktail. 1 mg of total lysate was immunoprecipitated with an antibody specific to the cytoplasmic domain of E-cadherin (clone 34) or control IgG. Immunoprecipitates were separated by 8% SDS-PAGE and immunoblotted for E-cadherin, SECAD (anti-myc Tag), and β-catenin as described above. An aliquot of total cell lysate (100 μg) used for immunoprecipitation was loaded as input control and blotted for E-cadherin. For immunoprecipitation of active EGFR, MDCK cells were allowed to attach and serum-starved overnight. To inhibit EGFR activation, samples were pre-incubated with AG1478 in serum-free media for 1 h before incubation with SECAD for the indicated times. Protein lysates were prepared in a buffercontaining10 mM Tris, 150 mM NaCl,1 mM EGTA, 1 mM EDTA, 0.2 mM Na-vanadate, 1% Triton X-100, 0.50% IGEPAL, 0.1% SDS, 1% deoxycholic acid, and protease inhibitor cocktail. 1 mg of total protein was used for immunoprecipitation with anti-EGFR antibody. The precipitates were separated by SDS-PAGE and immunoblotted for phospho-tyrosine and EGFR.

Cell surface biotinylation

Cell surface proteins were cross-linked to sulfo-NHS-biotin (Pierce, Rockford, IL) on ice, as described previously [22]. MDCK cells were then washed in warm serum-free DMEM and incubated at 37 °C for 15 min. SECAD (10 μg/ml) in serum-free DMEM was added and incubated for an additional 30 min at 37 °C, wherever indicated. Cells were washed with PBS containing 1 mM CaCl₂ and 1 mM MgCl₂, before lysis in 10 mM Tris, pH 8.0, 1% Triton X-100, 1 mM EGTA and protease inhibitor cocktail. Biotinlyated cell surface proteins were precipitated with streptavidin beads (Pierce) from 1 mg of total protein lysate, separated by SDS-PAGE and immunoblotted for SECAD (anti myc-tag) or Na,K-ATPase β-subunit as described above. An aliquot of total cell lysate (100 μg) used for precipitation following biotinylation was loaded as input control and blotted for annexin II.

Viability assay

MDCK cells were plated onto 12 well plates and allowed to attach. Cells were serum-starved in DMEM for 6 h prior to treatment. For low calcium experiments, cells were plated in SMEM with fetal bovine serum dialyzed against 0.1 mM EGTA to remove calcium, followed by starvation in serum-free SMEM. Where required LY294002, PD98059 and AG1478 were used at concentrations of 25 μM, 10 μM and 1 μM, respectively; SECAD and EGF were used at indicated concentrations. To assess the level of cell viability, levels of LDH released into culture media were compared relative to LDH levels of remaining viable cells after 48 h of treatment. LDH levels were determined using the Non-radioactive Cytotoxicity kit (Promega, Madison, WI) according to manufacturer’s instructions. Briefly, media were collected after 48 h and cell debris was pelleted by centrifugation (released LDH). 1 ml of fresh serum free medium was added back to wells and viable cells were lysed by incubation at −70 °C for 15 min, followed by thawing at 37 °C. Media (viable cell LDH) were collected and cell debris was pelleted by centrifugation. The percentage of viable cells was calculated by the equation: viable LDH/(viable cell LDH + released LDH) * 100. Assays were performed in quadruplicate.

Transmission electron microscopy

MDCK cells were serum-starved for 6 h and treated with 10 μg/ml SECAD or bovine serum albumin in serum-free conditions for 24 h. Cells were fixed in 2.5% glutaraldehyde in 0.1 M sodium cacodylate buffer, pH 7.4, for 2 h and processed for transmission electron microscopy by standard procedures [23].

Immunofluorescence and confocal microscopy

To detect cleaved caspase 3 and phosphorylated ERK1/2, MDCK cells were serum-starved for 6 h, treated with SECAD (10 μg/ml) or bovine serum albumin (10 μg/ml) for 24 h, and fixed in 2% paraformaldehyde/phosphate-buffered saline followed by cold methanol (−20 °C). For cytochrome C staining, cells were incubated with Mitotracker^™ at 37 °C for 30 min and fixed as described above. Primary antibodies were diluted in 1% bovine serum albumin/phosphate-buffered saline and incubated overnight at 4 °C, followed by Alexa-488^™ conjugated secondary antibodies. Nuclei were visualized with DAPI or propidium iodide. For phosphorylated ERK1/2 detection the cells were serum starved-overnight and pre-incubated with PD98059 for 1 h before SECAD treatment and processed as described above. Confocal microscopy was performed on a Zeiss LSM 5 Pascal laser-scanning microscope (Carl Zeiss, Oberkochen, Germany).

3D cultures

3D cultures of MDCK cells were grown and maintained according to published methods [24] with the following modifications. MDCK cells were maintained in DMEM. Rat-tail Collagen I (BD Biosciences) was used in lieu of bovine collagen I. Either SECAD or bovine serum albumin (control) was added to the collagen/cell mixture and allowed to harden at 37 °C. Fresh media containing SECAD or bovine serum albumin was then placed onto the collagen gels and incubated at standard cell culture conditions for 21 days. For EGF experiments, EGF (10 ng/ml; Sigma) was added to the culture medium. SECAD and EGF were refreshed every two days. Wherever indicated EGFR inhibitor CL-387,785 was added to the culture medium and replenished every two days. Immunofluorescence staining of 3D cultures in situ was performed according to published protocols [24]. 3D cultures were incubated with an antibody against β-catenin at a concentration of 1:100, followed by incubation with anti-mouse IgG conjugated to Alexa-488^™. Actin was stained with phalloiodin-conjugated to Alexa-546^™ and nuclei with TO-PRO3, and cysts were mounted in Prolong. Z-Stacks were obtained by confocal analysis with a Zeiss LSM 5 Pascal laser-scanning microscope (Carl Zeiss) and analyzed by Pascal software. For quantitative analysis, acini were assessed for normal localization of β-catenin (basolateral), cortical actin (apical) and well-established lumen. A total of 100 acini were counted from four independent experiments.

Results

Soluble E-cadherin prevents cell death following serum withdrawal

We generated a recombinant soluble form of E-cadherin (SECAD) to investigate on its biological effects. SECAD consists of an N-terminal secretion signal, the extracellular domain of canine E-cadherin (amino acids 541 to 2169) and C-terminal Myc- and His-epitopes for identification and purification of the recombinant protein (Fig. 1A, 3.). Coomassie blue staining of purified SECAD (Fig. 1B) or immunoblotting with anti-Myc antibody (Supplementary Fig. 1A) revealed a prominent band at approximately 80 kDa, corresponding to the N-glycosylated form of the purified recombinant SECAD (Supplementary Fig. 1B). Binding of SECAD to E-cadherin expressed on the plasma membrane of cultured cells was confirmed through co-immunoprecipitation of myc-tagged SECAD with E-cadherin using an antibody that recognizes only the cytoplasmic tail of E-cadherin and not SECAD (Fig. 1C). We utilized a cell surface biotinylation approach to further ascertain SECAD binding of E-cadherin on the plasma membrane. SECAD was found in the precipitates of cell surface proteins, suggesting that SECAD binds to E-cadherin expressed at the cell surface (Fig. 1D).

Fig. 3 – SECAD disrupts lumen formation in 3D cultures of MDCK cells.

Confocal analysis of MDCK acini after immunofluorescence labeling of the basolateral membrane marker β-catenin (green) and cortical actin (red), a marker of the apical membrane. Nuclei were stained with To-PRO3 (blue). (A) Control cells form well polarized cysts with a well-defined central lumen, while SECAD (10 μg/ml) treated cells grew as non-polarized cell clusters (B). (C) Control MDCK cells infected with empty vector form polarized hollow acini, while MDCK cells engineered to secrete SECAD fail to develop cysts with lumen (D). (E) MDCK acini treated with 10 ng/ml EGF. Although basolateral polarity develops (green), acini fail to develop a normal lumen. (F) MDCK-SECAD cells grown in the presence of 100 nM CL-387,785 results in lumen formation, as marked by cortical actin (red). MDCK-SECAD + CL cells still display defects in establishing basolateral polarity, as shown by β-catenin staining (green). Bar, 25 μm. (G) Quantification of polarized, lumen forming acini (SECAD 0%; MDCK-SECAD 4%±4%; EGF 58%±5%) when compared to control cells (100%). Standard error bars from 4 independent experiments.

MDCK cells are sensitive to serum withdrawal, with cells beginning to vacuolate after 24 h (Fig. 2A, arrows) and undergo cell death after 48 h [25]. Strikingly, in the presence of 10 μg/ml SECAD, a clinically relevant dose [15,21], most MDCK cells were viable even after 48 h of serum withdrawal as determined by phase contrast microscopy (Fig. 2A) and measurement of viable cells (Fig. 2B, black bars). SECAD at this concentration does not affect cell adhesion induced by E-cadherin (Supplementary Fig. 2). To confirm that the increased cell survival was due to SECAD binding to E-cadherin, we disrupted E-cadherin function by incubating MDCK cells in low calcium medium [26,27]. SECAD treatment did not prevent cell death following serum withdrawal in the absence of calcium (Fig. 2B, white bars; Supplementary Fig. 2B). In addition, SECAD-conditioned media from SECAD-expressing MDCK cells prevented MDCK cell death due to serum withdrawal (data not shown). To further demonstrate the role of E-cadherin in mediating the anti-apoptotic effects of SECAD, we utilized a MDCK cell line (MDCK-T151) expressing a dominant negative form of E-cadherin under the control of Tet repressor. In these cells, removal of doxycycline induces the expression of dominant negative E-cadherin, which is accompanied by downregulation of endogenous E-cadherin (Supplementary Fig. 3A) [28,29]. As expected, serum withdrawal severely reduced the cell viability of MDCK-T151 cells, regardless of doxycycline exposure (Supplementary Fig. 3B, lanes 2 and 5), and SECAD increased cell viability upon serum withdrawal in the presence of doxycyline (endogenous E-cadherin high) (Supplementary Fig. 3B, lane 3). However, in the absence of doxycycline, the anti-apoptotic effect of SECAD was compromised, reducing cell viability of MDCK-T151 cells under serum-free conditions despite SECAD treatment (Supplementary Fig. 3B, compare lane 3 to 6). While expression of dominant negative E-cadherin did not completely abrogate SECAD’s anti-apoptotic effects, perhaps due to incomplete suppression of endogenous E-cadherin levels (Supplementary Fig. 3A), disruption of E-cadherin function by dominant negative E-cadherin was sufficient to significantly (p < 0.05) reduce SECAD-induced cell survival. Thus, E-cadherin is necessary to mediate SECAD’s anti-apoptotic function. Taken together, these results suggest that SECAD binding to E-cadherin promotes cell survival in the absence of growth factors.

SECAD inhibits apoptotic cell death induced by serum withdrawal

Vacuolization of control cells after serum withdrawal is consistent with MDCK cells undergoing apoptosis [30]. Transmission electron microscopy confirmed ruptured mitochondria, a primary characteristic of cells undergoing apoptosis [31], (Fig. 2C, left top panel, black arrows) and vacuoles (Fig. 2C, right top panel, white arrow) in control cells 24 h after serum withdrawal. In SECAD-treated MDCK cells, mitochondria were intact and vacuoles were absent (Fig. 2C, bottom panels). Rupturing of the mitochondrial membrane releases cytochrome C into the cytoplasm, a key step in the activation of the apoptotic cascade [31]. In SECAD-treated serum-starved MDCK cells, cytochrome C co-localized to mitochondria, comparable to cells grown in the presence of serum (Fig. 2D, yellow). Conversely, in serum-starved MDCK cells without SECAD (Fig. 2D, Control) cytochrome C barely co-localized with mitochondrial staining. Cytochrome C release from mitochondria leads to the cleavage and activation of the pro-apoptotic protease, caspase 3. Control cells in serum-free medium clearly displayed a band corresponding to cleaved caspase 3 that was barely visible in SECAD-treated cells and not detected in cells cultured in the presence of serum (Fig. 2E). Furthermore, immunofluorescence staining confirmed the presence of cleaved/active caspase 3 in control (serum-free) but not in SECAD cells or in cells cultured with serum (Fig. 2F). Taken together, these data demonstrate that SECAD inhibits apoptosis due to serum withdrawal in MDCK cells.

SECAD prevents acini formation in 3D cultures

When grown in collagen matrix, single MDCK cells differentiate into multi-cellular hollow spheroids resembling more physiologic three-dimensional structures [32]. The cellular events responsible for acini formation include formation of cell–cell contacts, polarization into distinct apical and basolateral membranes, and responsiveness to apoptotic and proliferative signals. Apoptosis of centrally located cells is a critical event for the development of hollow lumen. Since SECAD inhibits apoptotic cell death, we hypothesized that SECAD might interfere with acinus formation in 3D cultures of MDCK cells. Under standard conditions, MDCK and vector-transfected MDCK cells formed polarized acini with distinct apical and basolateral domains (Figs. 3A and C), while MDCK cells grown in the presence of SECAD, failed to do so (Fig. 3B). Furthermore, stable clones of SECAD-secreting MDCK cells (MDCK-SECAD; see supplementary methods) failed to form hollow cysts (Fig. 3D), supporting the notion that SECAD disrupts apoptosis during lumen formation in MDCK 3D cultures.

SECAD activates PI3K and ERK1/2 signaling

E-cadherin homophilic ligation activates PI3K/AKT signaling [2,8], a pathway intimately linked to cell survival [33]. Similarly, SECAD activated AKT signaling in serum-starved MDCK cells as determined by immunoblotting with antibodies that specifically recognize the active, phosphorylated form of AKT (Fig. 4A). SECAD activated AKT as early as 10 min (Fig. 4A) and up to 24 h (data not shown) upon treatment. Inhibition of PI3K with the specific inhibitor LY294002 abrogated SECAD-induced AKT phosphorylation indicating that SECAD activates AKT through activating PI3K signaling. However, while LY294002 completely inhibited AKT activation, serum-starved SECAD-treated cells still survived in the presence of the inhibitor (Fig. 4B, bar 4), suggesting that additional signaling pathways are activated by SECAD. Thus, we tested whether SECAD induces ERK1/2 signaling, another well-studied cell survival pathway [34]. In serum-starved MDCK cells, SECAD activated ERK1/2 within 10 min as determined by immunoblotting (Fig. 4C) and by an in vitro kinase assay (Supplementary Fig. 4). Inhibition of MEK, an immediate upstream activator of ERK1/2, with PD98059 prevented SECAD’s activation of ERK1/2 (Figs. 4C, D) but failed to prevent SECAD’s ability to inhibit apoptosis due to serum withdrawal (Fig. 4B, bar 5). Interestingly, when both PI3K and ERK1/2 signaling were inhibited, SECAD failed to prevent cell death due to serum withdrawal (Fig. 4B, bar 6), suggesting that activation of both pathways is required to inhibit apoptosis.

Fig. 4 – SECAD activates ERK 1/2 and PI3K-AKT signaling.

(A) SECAD activates AKT. Immunoblot of active, phosphorylated (Ser473) AKT after exposure to10 μg/ml of SECAD for 10, 20, and 30min. Blot was stripped and probed with a total AKT antibody as a loading control. To inhibit PI3K/AKT activation, MDCK cells were treated with 25 μM LY2954002 for 30min, followed by SECAD (10 μg/ml) and LY2954002 for an additional 30min. Blot represents data from two independent experiments. (B) SECAD (10 μg/ml) requires both PI3K and ERK1/2 activation to inhibit serum-induced cell death. Cells in serum-free medium were treated with LY294002 (10 μM) and/or PD98059 (25 μM) for 48 h. Treatment with both LY294002 and PD98059 inhibited SECAD function (bar 6), compared to SECAD alone (bar 3), SECAD+LY294002 (bar 4), SECAD+PD98059 (bar 5) and serum (bar 1). Inhibition of SECAD by both LY294002 and PD98059 was comparable to control (bar 2) and LY294002+PD98059 alone (bar 7). Standard error bars represent the average of the mean of three independent experiments done in quadruplicate. (C) Immunoblot of phosphorylated ERK1/2. SECAD (10 μg/ml) activates ERK1/2 after 10, 20, and 30 min of treatment. Pre-incubation with 10 μM PD98059, a specific inhibitor of the upstream ERK1/activator, MEK, abolished ERK1/2 activation by SECAD treatment for 30 min. Blot was stripped and probed for total ERK1/2 as a loading control. Data are representative of two independent experiments. (D) Immunofluorescence of phosphorylated ERK1/2. MDCK cells treated with SECAD (10 μg/ml), EGF (10 ng/ml) or control were fixed after 30min and stained for active, phosphorylated ERK1/2. ERK1/2 activation was prevented by pre-treatment with 10 μMPD98059 followed by treatment with SECAD (10 μg/ml) and 10 μMPD98059 for an additional 30min. Note the nuclear localization of active ERK1/2 in SECAD and EGF-treated cells (yellow) and its cytosolic localization (green) in control or SECAD+PD98059 treated cells. Nuclei (red) were stained with propidium iodide. Bar, 15 μm.

SECAD mediates cell survival through activation of EGFR

E-cadherin and EGFR have been shown to interact functionally and since EGFR can activate both PI3K and ERK1/2 signaling pathways [35], we tested whether SECAD binding to E-cadherin results in the transactivation of EGFR. Since MDCK cells are of canine origin and phosphorylation specific antibodies do not cross-react, we immunoprecipitated EGFR and probed with anti-phospho-tyrosine antibody. Indeed, SECAD treatment of serum-starved MDCK cells resulted in increased tyrosine phosphorylation of EGFR when compared to control cells (Fig. 5A). Inhibition of EGFR kinase activity with the inhibitor, AG1478, prevented SECAD-induced tyrosine phosphorylation of EGFR (Fig. 5A) as well as activation of both AKT and ERK1/2 (Fig. 5B), indicating that SECAD activates PI3K and ERK1/2 through the transactivation of EGFR.

Fig. 5 – EGFR activation in SECAD-treated cells.

(A) Serum-starved MDCK cells were treated with SECAD (10 μg/ml) or EGF (10 ng/ml) for 30 min. EGFR was immunoprecipitated and immunoblotted for phosphorylated tyrosine (P-Tyr). EGFR kinase activity was inhibited by pre-treatment with the EGFR inhibitor, AG1478 (100 nM) followed by treatment with SECAD (10 μg/ml) and AG1478 for an additional 30 min. Blot was stripped and immunoblotted for EGFR as a loading control. Blot represents data from two independent experiments. (B) Inhibition of EGFR kinase activity prevents AKT and ERK1/2 phosphorylation by SECAD. MDCK cells were treated as in panel (A) and immunoblotted for activated AKT and ERK 1/2. Blots were stripped and probed for total AKT and ERK1/2 as a loading control. Blot is representative of two independent experiments. Numbers below each panel represent the densitometric quantitation of bands normalized to respective controls. (C) Inhibition of apoptosis by SECAD requires EGFR activation. Cells were treated with SECAD (10 μg/ml) or EGF (200 ng/ml) for 48 h and cell viability measured. EGFR kinase activity was inhibited with 1 μM AG1478. Inhibition of EGFR kinase function with AG1478 attenuated SECAD survival (bar 4), compared to SECAD alone (bar 3). EGF (200 ng/ml) partially prevented cell death due to serum withdrawal (bar 5). Bars represent the standard error of the mean from three independent experiments done in quadruplicate.

Both SECAD and EGF activated AKT and ERK1/2, but the level of activated AKT was 3.5-fold higher in SECAD-treated cells compared to EGF-treated cells (Fig. 5B). We tested whether SECAD requires activation of EGFR to inhibit apoptosis due to serum withdrawal. Upon inhibition of EGFR by AG1478, SECAD did not rescue cells efficiently from cell death upon serum withdrawal (Fig. 5C, compare bar 4 to bar 3). We also tested whether ligand-induced activation of EGFR by EGF prevents cell death. Activation of EGFR by EGF increased cell viability to 74%, when compared to control cells but did not reach the level achieved by SECAD alone (Fig. 5C, compare bar 5 to bar 3). Furthermore, in 3D cultures, EGF treatment significantly affected the lumen formation and the organization of the apical domain (Fig. 3E). However, the severity of the malformation of the lumen was less in EGF treated cells compared to SECAD treated cells (compare Fig. 3E to 3B). About 50% of the cysts in EGF treated cells appeared normal with distinct lumen which is consistent with about 50% reduction in the inhibition of cell death induced by EGF (Fig. 5C, bar 5), suggesting that increased EGFR activation affects lumen formation in MDCK cells. To further validate the role of SECAD and EGFR activation in lumen formation, we utilized a complementary approach in which EGFR activation was inhibited using an irreversible inhibitor of EGFR CL-387,785. This compound functions by covalently binding to EGFR’s ATP-binding pocket and blocking EGFR kinase activity [36]. MDCK-SECAD cells that normally do not form lumen (Fig. 3D) formed a partial lumen when EGFR kinase activity was inhibited (Fig. 3F) consistent with our notion that SECAD activation of EGFR and the subsequent inhibition of apoptosis prevents lumen formation. However, CL-387,785-treated MDCK-SECAD cells still displayed defects in apical-to-basolateral polarity, suggesting that SECAD may have other effects upon acini formation outside of EGFR kinase activity. Together, these data demonstrate that transactivation of EGFR by SECAD plays a critical role in the activation PI3K and ERK1/2 signaling and inhibition of cell death in MDCK cells.

Discussion

In this report, we present evidence that the soluble extracellular domain of E-cadherin acts as a potent anti-apoptotic protein, facilitating the survival of MDCK cells deprived of serum, as well as preventing formation of polarized epithelial acini, a process dependent upon apoptosis. This novel function of SECAD depends on its association with E-cadherin. We further show that SECAD mediates cell survival through activation of PI3K and ERK1/2 signaling pathways via transactivation of EGFR. These studies not only imply a physiological role for E-cadherin-EGFR mediated signaling, they also suggest that SECAD present in the sera of cancer patients might facilitate the survival of E-cadherin positive cancer cells and contribute to cancer progression. Moreover, presence of SECAD in the patients’ sera might have relevance for EGFR kinase inhibitor based therapeutic intervention of cancer.

We demonstrated that SECAD interacts with E-cadherin by three independent approaches. (1) We showed that SECAD (with myc-epitope tag) binds to E-cadherin by co-immunoprecipitation of myctagged SECAD with E-cadherin (Fig. 1C). (2) The requirement of calcium for E-cadherin’s homodimerization is one of the most well established facts about E-cadherin function. In a functional assay based on this fact, we showed that SECAD’s anti-apoptotic/pro-survival function is abolished under low calcium conditions (Fig. 2B). (3) Finally, we showed that SECAD significantly attenuated cell death in MDCK cells expressing dominant negative E-cadherin (Supplementary Fig. 3). Thus, these data demonstrate that SECAD associates with plasma membrane bound E-cadherin and that this association may be responsible for its anti-apoptotic activity.

SECAD concentrations (10 μg/ml) used in our experiments are similar to the concentration of soluble E-cadherin found in patients’ sera [15,21] and did not affect E-cadherin mediated cell–cell adhesion (Supplementary Fig. 2). Only higher SECAD concentrations (50–100 μg/ml) inhibited the cell–cell adhesion function of E-cadherin (Supplementary Fig. 2). Thus, these results strongly indicate that SECAD’s role in cell survival does not require disruption of E-cadherin-dependent cell–cell adhesion. It is possible that SECAD binding to cell surface E-cadherin mimics E-cadherin engagement between neighboring cells resulting in the activation of EGFR, PI3K/AKT, and ERK1/2 signaling. These signaling pathways are also activated upon cell-to-cell E-cadherin adhesion [3,8,9] or upon interaction of cell surface E-cadherin with an immobilized E-cadherin extracellular domain [2]. However, in these studies the activation of EGFR, PI3K/AKT, ERK1/2 and EGFR signaling was associated with the formation of adherens junctions [2,3,8,9]. Our results suggest that SECAD-binding to E-cadherin mimics homophilic E-cadherin engagement between neighboring cells but the activation of subsequent signaling mechanisms and its role in cell survival are independent of adherens junction formation.

SECAD inhibited apoptosis in standard 2D culture conditions and disrupted the development of 3D polarized acini in collagen gels. A single polarized acinar structure develops through a highly regulated stepwise progression, which ultimately results in a hollow sphere with distinctive epithelial characteristics [32]. Apoptosis is a critical step for acinar development. Inhibition of apoptosis, such as by expression of anti-apoptotic proteins or oncogenes, can prevent lumen formation and leads to multicellular aggregates devoid of epithelial polarity [30,37,38]. Similar to these studies, we observed multi-cellular aggregates in MDCK cells grown in the presence of SECAD, as well as in SECAD-expressing MDCK cells. Since we used SECAD concentrations that do not disrupt cell–cell adhesion, the lack of proper cyst development in the presence of SECAD is likely due to SECAD’s anti-apoptotic function.

We show that SECAD induces cell survival by inhibiting serum starvation-induced apoptosis through the activation of EGFR, PI3K/AKT, and ERK1/2 signaling pathways. Inhibition of either PI3K or ERK1/2 did not inhibit SECAD-mediated cell survival but simultaneous inhibition of both pathways or inhibition of EGFR alone significantly reduced SECAD’s cell survival function. These results are consistent with the idea that SECAD acts as a ligand and binding to its receptor E-cadherin transactivates EGFR leading to the activation of PI3K/AKT and ERK 1/2 signaling. However, our co-immunoprecipitation analyses did not reveal association of SECAD with EGFR (data not shown). How SECAD transactivates EGFR is currently not known. Studies to understand how SECAD activates EGFR and AKT are currently under investigation in our laboratory.

Although SECAD potently inhibited cell death in an EGFR-dependent manner, surprisingly, EGF ligand-induced EGFR activation less efficiently inhibited apoptosis in both 2D (Fig. 5C) and 3D cultures (Fig. 3E). In addition, EGFR inhibition in 3D cultures of MDCK-SECAD cells showed incomplete restoration of polarized cyst formation (Fig. 3F). We consistently observed that SECAD activates AKT more effectively than EGF or ERK1/2, although, EGFR phosphorylation was higher in EGF-treated cells. Taken together, EGFR alone is not sufficient to mimic the effect of SECAD on cell survival. Therefore, other pro-apoptotic pathways including other growth factor receptors may be simultaneously activated by SECAD.

Apoptosis occurs within all eukaryotic organisms, during development of tissues and organs and, most importantly, in the removal of diseased or damaged cells, such as cancer cells. As a result, inhibition of apoptosis is a necessary and critical step in tumorigenesis [39]. Cancer cells escape apoptosis through a variety of mechanisms, such as mutational loss of key pro-apoptotic proteins, or increased activation of anti-apoptotic signaling. Thus, our finding that SECAD at a clinically relevant concentration protects cells from apoptotic cell death via activation of EGFR has important clinical relevance. It is possible that increased levels of soluble E-cadherin within the early tumor microenvironment allows diseased or damaged cells that might otherwise be removed via apoptosis, to survive and persist within the tumor. While our studies were ongoing, another study showed that growth factor withdrawal induced the production of soluble E-cadherin, followed by activation of the EGFR family member, HER2, in breast cancer cells [13]. Therefore, soluble E-cadherin present in the tumor microenvironment might induce survival of E-cadherin-positive cells via activation of EGFR and thus promote tumor progression. These findings also suggest that the presence of soluble E-cadherin in a patient’s serum could be indicative of EGFR or Her2 activation. Indeed, a recently published study indicated that serum levels of soluble E-cadherin correlated with patient response to treatment with a combination of an EGFR inhibitor and a COX-2 inhibitor in non-small cell lung cancer [40]. Thus, future clinical trials should validate the significance of soluble E-cadherin in cancer progression, treatment and therapy.

Supplementary Material

Supplementary materials related to this article can be found online at doi:10.1016/j.yexcr.2010.12.025.

Abbreviations:

EGFR

epidermal growth factor receptor

LDH

lactate dehydrogenase

MDCK

Madin–Darby canine kidney cells

PI3K

phosphoinositide-3 kinase

SECAD

recombinant form of soluble E-cadherin