E cadherin inhibits tumor cell growth by suppressing PI3KAkt

Published: Last Edited:

This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.

E-cadherin is a cell-cell adhesion protein and tumor suppressor that is silenced in many malignancies. E-cadherin is thought to suppress tumor cell growth by antagonizing -catenin signaling. However, the role of E-cadherin in ovarian cancer progression is still controversial. In this study, we demonstrated that loss of E-cadherin induced ovarian cancer cell growth and constitutive activation of phosphoinositide 3-kinase (PI3K)/Akt signaling by inhibition of PTEN transcription through downregulation of Egr1. In addition, immunofluorescence microscopy and TCF promoter/luciferase reporter assays showed that E-cadherin loss was associated with enhanced nuclear -catenin signaling. Constitutive activation of PI3K/Akt signaling reinforced nuclear -catenin signaling by inactivating glycogen synthase kinase-3 indicating cross talk between the PI3K/Akt and -catenin signaling pathways. Finally, we found that E-cadherin negatively regulates tumor cell growth, in part, by positively regulating PTEN expression via -catenin-mediated Egr1 regulation, thus influencing PI3K/Akt signaling. In summary, endogenous E-cadherin inhibits PI3K/Akt signaling by antagonizing -catenin-Egr1-mediated repression of PTEN expression. Thus, the loss of E-cadherin itself may contribute to dysregulated PI3K/Akt signaling through its effects on PTEN, or it may exacerbate the frequent activation of PI3K/Akt signaling that occurs as a result of overexpression, mutation and/or amplification .

Keywords: E-cadherin; PTEN; PI3K/Akt; -catenin; ovarian cancer

Introduction

Generally, cadherins have been studied for their vital roles in cell adhesion. They comprise a superfamily of transmembrane proteins that link adjacent cells via calcium-dependent homophilic interactions (Yagi and Takeichi 2000). E-cadherin is a tumor suppressor protein whose loss is thought to promote tumor growth and invasion via a mechanism involving -catenin (Wong and Gumbiner 2003).-catenin was originally identified as a cytoplasmic component of adherens junctions, where it associates with E-cadherin and, via -catenin, the actin cytoskeleton (Geiger et al. 1995; Kemler 1993).In addition, -catenin is the main effector of Wnt signaling in the nucleus, where it interacts with lymphoid enhancer factor /T cell factor (LEF/TCF) transcription factors to regulate the expression of genes involved in cell growth control, such as cyclin D1 (Conacci-Sorrell et al. 2003; van Noort and Clevers 2002). In the absence of Wnt signaling, cytosolic β-catenin is constantly phosphorylated by a degradation complex consisting of glycogen synthase kinase-3 (GSK3), axin adenomatous polyposis coli and casein kinase 1, thereby targeting -catenin for proteasomal degradation (van Noort and Clevers 2002). Upon activation of Wnt signaling, GSK3 is inhibited, which leads to the stabilization and nuclear translocation of -catenin, and the initiation of target gene transcription. E-cadherin-containing adherens junctions ensure that the cytoplasmic pool of -catenin is maintained at a low level. Thus, E-cadherin could antagonize -catenin signaling and induce growth inhibition (Gottardi et al. 2001; Shtutman et al. 1999). Indeed, β-catenin has been demonstrated to localize to the nucleus, following the loss of E-cadherin expression (Gottardi et al. 2001; Onder et al. 2008). However, the detailed mechanism by which the loss of E-cadherin contributes to enhanced -catenin signaling is not well understood.

Aberrant signaling of the PI3K/Akt pathway has been implicated in the pathogenesis of several human cancers, including epithelial ovarian cancer (Brugge et al. 2007; Woenckhaus et al. 2007). Phosphatase and tensin homolog (PTEN) acts as a tumor suppressor by dephosphorylating phosphotidylinositol-(3,4,5)-triphosphate (PIP3) produced by phosphoinositide-3-kinase (PI3K) (Myers et al. 1998). In human cancers, PTEN is one of the most common targets of mutation or downregulation resulting in the activation of the PI3K/Akt pathway (Blanco-Aparicio et al. 2007). Approximately 27 % of human ovarian cancers display reduced PTEN protein levels and loss-of-function mutations are found in 3-8 % (Bast, Jr. et al. 2009). Overexpression of PTEN suppresses the growth of tumor cells by up-regulating p27Kip1 (Furnari et al. 1998; Weng et al. 1999) and downregulating cyclin D1, in an Akt-dependent manner (Radu et al. 2003; Weng et al. 2001). PTEN also modulates migration and proliferation via interaction with cell adhesion molecules such as E-cadherin and -catenin (Hu et al. 2007; Subauste et al. 2005). It has been shown that PTEN transcription can be transactivated by early growth response gene 1 (Egr1) which binds directly to a consensus Egr1-binding motif in the PTEN promoter (Virolle et al. 2001). Recent studies suggest that E-cadherin modulates PTEN levels in breast cancer cells (Fournier et al. 2009; Li et al. 2007), however the exact mechanism by which E-cadherin regulates PTEN levels is unclear.

In the present study, we demonstrate that E-cadherin regulates tumor cell growth via the PI3K/Akt and -catenin signaling pathways in epithelial ovarian cancer cells. In the presence of E-cadherin, -catenin is localized at adherens junctions, PTEN mRNA and protein are high, and PI3K/Akt signaling is reduced. Loss of E-cadherin results in the nuclear translocation of -catenin, enhanced -catenin signaling and reduced Egr1 expression. Egr1 downregulation reduces PTEN, which enhances PI3K/Akt signaling and increases cell growth. Our results point to an interplay between adherens junction assembly and PTEN transcription mediated by the junctional control of -catenin signaling, and provide important insights into the role of cadherin complexes in cellular proliferation, anchorage-independent growth, and tumor progression.

Results

Loss of E-cadherin induces cell growth in human ovarian cancer cells

Initially, we tested whether E-cadherin mediates anchorage-independent growth. A2780, OVCAR-3 and SKOV-3 cells, which express different levels of endogenous E-cadherin protein (Figure 1A), were cultured in soft agar to test their capacities for anchorage-independent growth. Our results show that endogenous levels of E-cadherin are inversely correlated with the capacity for anchorage-independent growth in these cell lines (Figure 1B). To further confirm the role of E-cadherin in suppressing anchorage-independent growth, OVCAR-3 and SKOV-3 cells were stably transfected with pLKO.1 expression vectors encoding short hairpin sequences targeting human E-cadherin. As shown in Figure 2A, stable knockdown of endogenous E-cadherin increased anchorage-independent growth in OVCAR-3 and SKOV-3 cells. In contrast, the expression of exogenous murine E-cadherin (mEcad) suppressed anchorage-independent growth in A2780 and SKOV-3 cells (Figure 2B). Effects observed on anchorage-independent growth were proportional to the extent of E-cadherin downregulation or overexpression in SKOV-3 cells and A2780 cells, respectively (Figure 2A and 2B).

To confirm these results, stably transfected SKOV-3 cells were cultured in suspension in poly-HEMA-coated dishes to prevent cell substratum attachment. The viability of control cells declined slightly over a 2-day period, whereas mEcad cells exhibited a dramatic decrease in cell number and shEcad cells continued to grow (Figure 2C). Next, we examined the effects of E-cadherin modulation on the growth of adherent cells. Stably transfected SKOV-3/shEcad cells exhibited a faster growth rate than control cells, with significant differences observed at 48, 72 and 96 h (Figure 2D). In contrast, mEcad cells grew slower than control cells with significant differences observed at 72 and 96 h. Taken together, these results indicate that endogenous E-cadherin suppresses the growth of ovarian cancer cells in vitro.

Loss of E-cadherin promotes anchorage-independent growth via PI3K/Akt mediated -catenin/TCF signaling in human ovarian cancer cells

Previous studies have indicated that E-cadherin-mediated cell-cell adhesion activates the PI3K/Akt signaling pathway in ovarian carcinoma cell lines (De et al. 2009; Reddy et al. 2005). Thus, we next examined whether stable changes in E-cadherin expression modulate PI3K/Akt signaling by determining the phosphorylation level of Akt at Ser473. Depletion of endogenous E-cadherin increases basal phosphorylation of Akt (Figure 3A). PI3K/Akt signaling is known to lead to the phosphorylation and inactivation of glycogen synthase kinase-3 (Cross et al. 1995). Recent data have shown that E-cadherin loss inhibits GSK3 activation by inducing its phosphorylation (Onder et al. 2008). We hypothesized that E-cadherin-depletion-mediated activation of PI3K/Akt signaling may lead to the phosphorylation and inhibition of GSK3. Consistent with this hypothesis, we found that E-cadherin depletion increased the levels of phosphorylated GSK3in both SKOV-3 and OVCAR-3 cellsFigure 3A). In addition, E-cadherin-depleted cells exhibited increased levels of cyclin D1 and reduced levels of p27Kip1, known targets of Akt signaling involved in cell-cycle control. In contrast, expression of murine E-cadherin suppressed Akt and GSK3 phosphorylation, reduced cyclin D1 and increased p27Kip1 protein levels (Figure 3A). These data show that the Akt signaling pathway is activated in E-cadherin-depleted human ovarian cancer cells.

The observation that E-cadherin depletion induces Akt signaling and growth in soft agar suggested the possibility that loss of E-cadherin promotes anchorage-independent growth through induction of the PI3K/Akt signaling pathway. To test this hypothesis, we treated SKOV-3 control cells (shCtl) and E-cadherin-depleted cells (shEcad) with the PI3K inhibitor LY294002 (10 μM). Treatment with LY294002 abolished the shEcad-mediated increases in phosphorylated Akt and cyclin D1, but had no effect on the reductions in p27Kip1 (Figure 3B). Moreover, treatement with LY294002 completely abolished the increase in phosphorylated GSK3 observed in E-cadherin-depleted SKOV-3 cells (Figure 3B), thus suggesting a role for PI3K/Akt in GSK3 inactivation following the loss of E-cadherin. Functionally, LY294002 abolished the increase in cell growth in soft agar induced by E-cadherin depletion (Figure 3D). These data strongly indicate that loss of E-cadherin promotes the growth of SKOV-3 cells by activating the PI3K/Akt signaling pathway.

Because E-cadherin has been shown to inhibit -catenin signaling (Conacci-Sorrell et al. 2003; Gottardi et al. 2001; Stockinger et al. 2001), we therefore examined the effects of E-cadherin loss on -catenin signaling. Downregulation of E-cadherin in SKOV-3 cells resulted in the loss of -catenin from sites of cell-cell contact, as assessed by immunocytochemistry (Figure 3E). As inactivation of GSK3 has been shown to enhance -catenin protein stability and transactivation activity (Polakis 1999), we next investigated whether PI3K/Akt/GSK3 signaling was involved in E-cadherin-depletion-mediated changes in the subcellular localization of -catenin. A LEF/TCF promoter luciferase reporter system was used to confirm the nuclear translocation and transactivation activity of -catenin, and to examine the involvement of PI3K/Akt/GSK3 signaling (Figure 3F). LEF/TCF promoter activity was increased in SKOV-3/shEcad cells and was abolished by LY294002 treatment. To further demonstrate a vital role for PI3K/Akt-mediated GSK3 inhibition in the activation of -catenin/TCF-dependent transcription in SKOV-3 cells, we used a dominant negative Akt and a constitutively active form of GSK3 (GSK3-S9A) in which Ser9 was replaced with alanine, thus preventing phosphorylation and inactivation of the kinase (Eldar-Finkelman et al. 1996; Stambolic and Woodgett 1994). Overexpression of either dominant negative Akt or GSK3-S9A abolished the effects of E-cadherin loss on LEF/TCF promoter activity (Figure 3F). Taken together, these data strongly implicate the inactivation of GSK3 by the PI3K/Akt pathway in the enhancement of -catenin/TCF-dependent signaling in response to reduced levels of E-cadherin.

Several previous studies have suggested that E-cadherin suppresses tumor cell growth by antagonizing -catenin nuclear signaling (Gottardi et al. 2001; Stockinger et al. 2001). To determine whether -catenin mediates increased growth in response to E-cadherin-depletion, we used a short hairpin construct to stably knockdown -catenin expression in SKOV-3 and SKOV-3/shEcad cells. Downregulation of -catenin in SKOV-3/shCtl cells resulted in reduced levels of cyclin D1 and increased levels of p27Kip1 (Figure 3C). Importantly, -catenin knockdown in SKOV-3/shEcad cells inhibited the increases in cyclin D1, and upregulated p27Kip1 levels, associated with E-cadherin depletion (Figure 3C). We next determined whether -catenin is required for SKOV-3/shEcad tumor cell growth in soft agar. -catenin knockdown in SKOV-3/shEcad cells completely abolished E-cadherin-depletion-induced anchorage-independent growth (Figure 3D). These data suggest that -catenin signaling is required for the enhanced growth of SKOV-3 cells in response to E-cadherin-depletion.

Loss of E-cadherin inhibits PTEN transcription via Egr1 downregulation

To more precisely define the mechanism by which E-cadherin depletion induces Akt activation, we further examined the signaling upstream of Akt. In particular, we investigated the mRNA and protein levels of PTEN in control and shEcad transfected SKOV-3 and OVCAR-3 cells. As shown in Figure 4A, PTEN mRNA levels were downregulated by E-cadherin loss. In addition, PTEN protein levels were also reduced by E-cadherin loss and this effect could be reversed by overexpression of mouse E-cadherin (Figure 4B and 4C). Similarly, PTEN promoter activities were repressed by E-cadherin loss and could be restored by mouse E-cadherin overexpression in OVCAR-3 and SKOV-3 cells suggesting that E-cadherin can regulate PTEN at the transcriptional level (Figure 4D). Since transcription of PTEN can be transactivated by Egr1, via binding to an Egr1-binding site in the PTEN promoter (Virolle et al. 2001), we determined the protein levels of Egr1 in transfected SKOV-3 and OVCAR-3 cells. Egr1 protein levels were reduced in E-cadherin depleted cells and could be resorted by overexpression of mouse E-cadherin (Figure 4B and 4C), suggesting that Egr1 may mediate the effects of E-cadherin on PTEN transcription. To determine whether Egr1 is involved in E-cadherin mediated PTEN transcriptional regulation, we next examined the effects of E-cadherin downregulation on the activity of PTEN promoter constructs with serial deletions or mutation of the Egr1-binding site. As shown in Figure 4E, the luciferase activities of different 5' truncated PTEN promoter constructs that contain the Egr1-binding site were reduced by E-cadherin loss, whereas the luciferase activity of a full-length construct with a mutated Egr1-binding site, pGL3-PTEN2526/427(mutEgr1), was low in both control and shEcad transfected SKOV-3 cells. Since E-cadherin regulates PTEN, which has previously been implicated in the suppression of cell growth (Ramaswamy et al. 1999; Sun et al. 1999), we investigated whether the overexpression of PTEN regulates cyclin D1 and p27Kip1 protein levels. Transient transfection of SKOV-3 cells with PTEN decreased cyclin D1 and increased p27Kip1 protein levels (Figure 4F). Taken together, these results suggest that E-cadherin downregulation reduces PTEN transcription via the downregulation of Egr1, thus leading to reduced PTEN protein levels, enhanced Akt signaling and increased anchorage-independent growth.

Loss of E-cadherin inhibits PTEN transcription via -catenin/TCF-mediated Egr1 downregulation

The observation that Egr1 and PTEN levels are decreased in SKOV-3/shEcad cells displaying nuclear -catenin and strong -catenin-mediated LEF/TCF transactivation, led us to investigate whether the low PTEN levels were the result of -catenin-mediated suppression of Egr1. Lithium chloride (LiCl), which mimics Wnt/-catenin signaling by inhibiting GSK3 activity and inducing GSK3 phosphorylation, was used to activate -catenin signaling (van Noort and Clevers 2002). Activation of -catenin signaling, which was confirmed by LEF/TCF promoter luciferase reporter (Figure 5A), suppressed PTEN and Egr1 protein levels (Figure 5B). Next, we used a constitutively active GSK3GSK3-S9Aconstruct to confirm the role of GSK3 inactivation in modulating Egr1 and PTEN levels. Expression of GSK3-S9A induced Egr1 and PTEN protein levels (Figure 5C). We also investigated the role of GSK3 inhibition on PTEN promoter activity and found that LiCl suppressed, whereas GSK3-S9A enhanced, promoter activity SKOV-3 cells (Figure 5D). Interestingly, LiCl and GSK3-S9A did not affect the activity of the Egr-1 mutant PTEN promoter construct (Figure 5D), indicating that GSK3 inhibition suppresses transcription of the PTEN gene via Egr1.

Next, we used SKOV-3/sh-cat and shEcad + sh-cat clones to assess whether -catenin regulates Egr1 and PTEN expression. Downregulation of -catenin resulted in increased Egr1 and PTEN levels in SKOV-3/sh-cat cells compared with control cells (Figure 6A, compare lane 2 to lane 1). Downregulation of -catenin in SKOV-3/shEcad cells abolished the suppression of Egr1 and PTEN levels induced by E-cadherin-depletion (Figure 6A, compare lane 1 to lane 3 vs. lane 2 to lane 4). To further examine the role of -catenin signaling in the inhibition of Egr1 and PTEN expression, we transfected overexpressed stabilized, constitutively active S33Y -catenin to activate -catenin signaling. Activation of -catenin signaling reduced Egr1 and PTEN protein levels in OVCAR-3 and SKOV-3 cells (Figure 6B). In addition, we transfected SKOV-3/sh-cat cells with S33Y -catenin and dominant negative TCF. Stabilized S33Y -catenin was able to reverse the induction of PTEN protein levels in SKOV-3/sh-cat cells, and this effect could be blocked by dominant negative TCF (Figure 6C). Similarly, stabilized S33Y -catenin reduced PTEN promoter activity, and this effect could be reversed by dominant negative TCF (Figure 6D). We also examined whether Egr1 is involved in -catenin/TCF-mediated PTEN suppression. Stabilized S33Y -catenin and dominant negative TCF did not affect the activity of the Egr1 mutant PTEN promoter construct (Figure 6D). Collectively, these data suggest that E-cadherin induces PTEN expression and suppresses the PI3K/Akt pathway by -catenin/TCF-mediated suppression of Egr1, a positive regulator of PTEN transcription.

Regulation of PTEN levels by cell density and E-cadherin-cadherin interactions

Previous studies in colon cancer cell lines have shown that E-cadherin levels increase in dense compared with sparse cultures, and that this depends on the junctional control of -catenin signaling (Conacci-Sorrell et al. 2003). Therefore, we next examined whether cell density influences E-cadherin levels and whether such changes in E-cadherin contribute, in turn, to the subsequent regulation of PTEN levels in a -catenin dependent manner. To test this hypothesis, OVCAR-3 and SKOV-3 cells seeded at different densities (sparse, 6 x103 cells/cm2; dense, 6 x 104 cells/cm2) and E-cadherin, Egr1, PTEN and pAkt levels were analyzed by Western blot. Both ovarian cancer cell lines showed increases in E-cadherin, Egr1 and PTEN levels in dense compared with sparse cultures (Figure 7A). Dense cultures also displayed a reduced activation of PI3K/Akt signaling (Figure 7A). We also tested whether enhanced -catenin signaling contributes to decreased Egr1 and PTEN levels in sparse cultures. -catenin depletion in sparse cultures of SKOV-3 cells resulted in increased levels of E-cadherin, Egr1 and PTEN, and reduced levels of PI3K/Akt signaling (Figure 7A), suggesting that -catenin signaling regulates E-cadherin, Egr1 and PTEN levels.

We also examined whether the assembly of adherens junctions in dense SKOV-3 cultures is involved in inducing Egr1 and PTEN expression. To inhibit E-cadherin-dependent adherens junction assembly, dense cultures were seeded in the presence of a monoclonal antibody against the extracellular domain of E-cadherin that is known to block E-cadherin-cadherin interactions. The localization of -catenin underwent a dramatic change, with -catenin relocalizing to the nuclei of cells, and with little -catenin remaining in adherens junctions (Figure 7B). Consistent with our shEcad findings, Egr1 and PTEN protein levels were reduced in SKOV-3/shCtl cells incubated with anti-E-cadherin antibody (Figure 7C). In contrast, treatment with the anti-E-cadherin antibody did not reduce Egr1 and PTEN protein levels in cells with reduced -catenin signaling (SKOV-3/sh-cat; Figure 7C). Taken together, these results suggest that E-cadherin, via the assembly of adherens junctions, regulates Egr1 and PTEN expression by modulating -catenin signaling.

.

Discussion

E-cadherin is known to suppress tumor cell invasion and re-expression of E-cadherin in E-cadherin-deficient carcinomas reverts cells to a less invasive, less aggressive phenotype (Gottardi et al. 2001; Soto et al. 2008; St Croix et al. 1998; Yanagisawa and Anastasiadis 2006). Loss of E-cadherin expression or function is a common event in tumor progression (Nollet et al. 1999; Yap 1998). However, the role of E-cadherin in ovarian cancer progression is still controversial. In the current study, we examined the mechanism by which E-cadherin suppresses tumor cell growth. Our data reveal that endogenous E-cadherin suppresses cell growth via a mechanism involving -catenin and PI3K/Akt signaling. In particular, we show for the first time that loss of E-cadherin induces -catenin signaling which represses Egr1-mediated PTEN transcription and leads to the activation of PI3K/Akt signaling in ovarian cancer cells.

Recent studies have suggested that E-cadherin suppresses cell growth by inhibiting -catenin signaling (Gottardi et al. 2001; Maher et al. 2009; Stockinger et al. 2001), we report here that shRNA-mediated depletion of E-cadherin resulted in relocalization of -catenin from the membrane to the nucleus and activation of -catenin-TCF signaling which in turn regulates cell growth in ovarian cancer cells. Theoretically, translocation of -catenin to the nucleus leads to its association with TCFs and results in regulated transactivation of genes containing the LEF-1/TCF-4 binding sequence near their promoter, such as cyclin D1 (Lin et al. 2000; Morin 1999; Shtutman et al. 1999). This theory is supported by our finding that the loss of E-cadherin in SKOV-3 cells resulted in increased activation of the TCF promoter-reporter construct, and was associated with increased levels of cyclin D1. In addition to the release of -catenin from cell adherens junctions, loss of E-cadherin led to the inactivation of GSK3 by phosphorylation. It is well known that the PI3K/Akt signaling can regulate nuclear -catenin through inhibition of GSK3 (Cross et al. 1995; Li and Sun 1998). In this study, we show that the loss of E-cadherin induces the activation of PI3K/Akt signaling resulting in the phosphorylation and inactivation of GSK3, thus reducing the degradation of -catenin and enhancing -catenin signaling. Dysregulated PI3K/Akt signaling as well as -catenin expression and signaling are crucial in tumorigenesis (Chalhoub and Baker 2009; Morin 1999). Our findings demonstrate that both -catenin signaling and PI3K/Akt signaling are required for increased cell growth in response to the loss of E-cadherin. In contrast, two earlier studies found that disruption of E-cadherin-mediated cell-cell adhesion reduces PI3K/Akt signaling and suppresses cell growth in ovarian carcinoma cell lines (De et al. 2009; Reddy et al. 2005). These differences may result from differences in experimental design or from using a calcium chelating agent (e.g. EGTA) which may non-specifically disrupt all Ca2+-dependent cadherin-mediated cell-cell adhesion.

Our results indicate that the loss of E-cadherin reduces PTEN levels in ovarian cancer cells, thus leading to increased PI3K/Akt signaling. In agreement, recent studies in breast cancer cells have implicated E-cadherin mediated cell-cell adhesion in the regulation of PTEN. Specifically, exogenous expression of E-cadherin increases, whereas function-blocking E-cadherin antibody or siRNA-mediated knockdown reduces, PTEN protein levels (Fournier et al. 2009; Li et al. 2007). PTEN transcription is regulated by numerous transcription factors. MEKK4 and JNK promote cell survival by suppressing PTEN transcription via direct binding of NFB to the PTEN promoter (Xia et al. 2007). In contrast, p53 activates PTEN transcription by directly binding to a p53-binding element in the PTEN promoter (Stambolic et al. 2001). Egr1 has been reported to be a crucial PTEN transactivator by directly binding to the PTEN 5'- untranslated region (Virolle et al. 2001). In the present study, we show that Egr1 provides a critical link between -catenin/TCF signaling, induced by the loss of E-cadherin, and PTEN transcription. First, mimicking the activation of β-catenin signaling by GSK3β inactivation using LiCl, reduced Egr1 and PTEN levels. Second, inhibition of -catenin signaling by constitutively active GSK-3 resulted in increased expression of Egr1 and PTEN. Third, the involvement of -catenin signaling was confirmed using constitutively active -catenin and -catenin shRNA. Finally, our PTEN promoter analysis showed that the Egr1-binding site is required for the suppression of PTEN transcription following the loss of E-cadherin, and is an important regulator of basal PTEN transcription. Interestingly, PTEN might also affect -catenin signaling through suppression of PI3K/Akt signaling, suggesting a reciprocal relationship between the PTEN/PI3K/Akt and -catenin signaling pathways (Persad et al. 2001).

The role of PTEN in tumor growth has been extensively studied. Reexpression of PTEN in PTEN-deficient cells has been shown to induce growth suppression (Ramaswamy et al. 1999; Sun et al. 1999). In a variety of cells, overexpression of PTEN suppresses tumor cell growth by up-regulating p27Kip1 and downregulating cyclin D1 (Furnari et al. 1998; Li and Sun 1998; Persad et al. 2001; Radu et al. 2003; Weng et al. 2001; Weng et al. 1999). We have demonstrated that overexpression of PTEN in SKOV-3 cells downregulates cyclin D1 and increases p27kip1 protein levels. In addition, we observed similar results by inhibiting PI3K/Akt signaling with the PI3K inhibitor LY294002. Interestingly, LY294002 had no effect on the reduction of p27Kip1 in E-cadherin-depleted cells, suggesting that a PI3K/Akt-independent pathway may mediate the suppression of p27Kip1 following E-cadherin loss. However, loss of -catenin, either in SKOV-3 or SKOV-3/shEcad cells, results in downregulation of cyclin D1 and induction of p27Kip1 protein levels. While our studies indicate that the alterations in cyclin D1 and p27Kip1 protein levels following E-cadherin depletion are very likely due to the regulation of -catenin signaling by PTEN/PI3K/Akt signaling, they also suggest a complex interplay between the pathways and/or the involvement of addiaitonal pathways.

E-cadherin is known to function in the density-dependent contact inhibition of cell growth. E-cadherin levels were increased in dense cultures and this was associated with increased levels of PTEN and reduced activation of the PI3K/Akt signaling pathway in OVCAR-3 and SKOV-3 ovarian cancer cells. Previous studies have suggested a link between reduced -catenin signaling and up-regulation of E-cadherin (Conacci-Sorrell et al. 2003; Weng et al. 2002). These studies support our observation that -catenin loss induces E-cadherin levels in sparse SKOV-3 ovarian cancer cells. Furthermore, we found that -catenin depletion induces Egr1 and PTEN levels in sparse cultures, indicating that -catenin signaling is important for the regulation of Egr1 and PTEN levels. We also demonstrated that disruption of E-cadherin-mediated cell-cell adhesion, by an inhibitory E-cadherin antibody, relocalizes -catenin to nuclei and reduces Egr1 and PTEN levels in SKOV-3 ovarian cancer cells. These findings are in agreement with recent studies suggesting that cadherin-mediated cell-cell interactions can regulate -catenin/TCF signaling in colon cancer cells (Conacci-Sorrell et al. 2003; Maher et al. 2009). Taken together, our results demonstrate the importance of -catenin signaling in the regulation of Egr1 and PTEN expression by E-cadherin-mediated cell-cell interactions.

In summary, we have demonstrated a role for E-cadherin in the transformed growth of ovarian cancer cells (Figure 8). The presence of E-cadherin acts to sequester -catenin and maintain PTEN, thus permitting its tumor-suppressive function through the inhibition of PI3K/Akt signaling. Upon E-cadherin loss during tumor progression, the nuclear translocation and activation of -catenin signaling leads to the suppression of Egr1, resulting in reduced PTEN transcription and activation of PI3K/Akt signaling. Under these conditions, the enhancement of PI3K/Akt signaling further stabilizes -catenin signaling, via the phosphorylation-dependent inhibition of GSK-3, and leads to increased transcription of oncogenic target genes that promote anchorage-independent growth (i.e. cyclin D1). Given that the absence of E-cadherin expression is associated with a poor prognosis and aggressive disease, our data could have significant implications for tumor biology and cancer treatment. Specifically, the combined inhibition of PI3K/Akt and -catenin signaling may block the transformed growth of such E-cadherin-deficient cells. Recent studies have shown that E-cadherin-deficient tumors are more resistant to treatment with epidermal growth factor receptor inhibitors (Black et al. 2008; Witta et al. 2006; Yauch et al. 2005). One possible explanation is that specific epidermal growth factor receptor inhibitors may fail to inhibit growth of E-cadherin-deficient cells, because -catenin signaling could still promote the constitutive activation of PI3K/Akt signaling via inhibition of PTEN expression, thus promoting transformed growth.

Legends

Figure 1 (A) Western blot analysis of E-cadhiern, -catenin, phosphoylated and total Akt, PTEN, and -actin levels in A2780, OVCAR-3 and SKOV-3 cells (B) A2780, SKOV-3 and OVCAR-3 cells were seeded in soft agar and analyzed for anchorage-independent growth. Results represent the mean ± SEM (n=3)

Figure 2 E-cadherin suppresses the growth of human ovarian cancer cells. (A) OVCAR-3 and SKOV-3 cells were stably transfected with scramble shRNA vector (shCtl) or E-cadherin shRNA vector (shEcad) and the ability of the cells to grow in soft agar was tested (*, P < 0.05; **, P < 0.001). Data represent the mean ± SEM (n=3). Representative immunoblot of E-cadherin protein levels in the various cell lines are shown in the lower panels. (B) A2780 and SKOV-3 cells were stably transfected with pIRES control vector (pIRES) or murine E-cadherin expression vector (mEcad) and the ability of the cells to grow in soft agar was tested. (C) Stably transfected SKOV-3 cells were placed in suspension and seeded in poly-HEMA coated plates, cultured for 1 or 2 days and viable cells were counted with Trypan blue. (mean ± SEM, n =3). (D) The ability of stably transfected SKOV-3 cell lines to grow on plastic was also determined by cell counting. Results represent the mean ± SEM (n=3; *, P < 0.05; **, P < 0.001).

Figure 3 Loss of E-cadherin promotes anchorage-independent growth via the PI3K/Akt mediated -catenin/TCF signaling pathway in human ovarian cancer cells (A) Western blot analysis of phosphoylated and total Akt, phosphorylated GSK3 cyclin D1, p27Kip1 and -actin levels in A2780, OVCAR-3 and SKOV-3 cells (shCtl, shEcad or mEcad). (B) Stably transfected SKOV-3 cells were treated with DMSO or 10 µM LY294002 for 24 h and total cellular levels of phosphoylated and total Akt, phosphorylated GSK3 cyclin D1, p27Kip1 and -actin were analyzed by Western blot. (C) Immunotblots showing E-cadherin, -catenin, cyclin D1, p27Kip1 and -actin levels in shCtl, sh-cat, shEcad and shEcad + sh-cat cells. (D) Stably transfected SKOV-3 cells were seeded in soft agar in the presence of DMSO or 10M LY294002 and analyzed for anchorage-independent growth. Results represent the mean ± SEM (n=3; a, P < 0.001, as compared with the SKOV-3/shCtl controls [DMSO]; b, P < 0.001, as compared with SKOV-3/shEcad controls). (E) SKOV-3 cells (shCtl or shEcad) were immunostained for -catenin (green), cell nuclei were stained with DAPI (blue) and analyzed by fluorescence microscopy. Note the absence of -catenin staining at cell-cell junctions and its nuclear localization in shEcad cells. Scale bar: 20 m (F) TCF activity was analyzed using the TOPFLASH and FOPFLASH luciferase reporters. Cells were transfected with either the TOPFLASH or FOPFLASH luciferase reporter, along with pcDNA 3.1, dominant negative Akt (DN-Akt), or constitutively active GSK3 (GSK3-S9A). -galactosidase vector was cotransfected for normalization of transfection efficiency. 10 M LY294002 or DMSO was added for 24 h before harvesting the cells for the measurement of luciferase and -galactosidase activities. Values are normalized luciferase activity (as described in the Materials and methods section) and are shown as mean ± SEM of three independent experiments performed in triplicate (a, P < 0.001, as compared with the SKOV-3/shCtl controls [DMSO or pcDNA3.1]; b, P < 0.001, as compared with SKOV-3/shEcad controls).

Figure 4 Loss of E-cadherin reduces PTEN mRNA and protein levels. (A) PTEN protein levels were analyzed by Western blot in stably transfected OVCAR-3 and SKOV-3 cells. Results represent the mean ± SEM (n=3; **, P < 0.001). (B) Relative PTEN mRNA levels in stably transfected cells were analyzed by RT-qPCR. Results represent the mean ± SEM (n=3; **, P < 0.001). (C) Stably transfected SKOV-3 cells were transient transfected with pIRES empty vector or murine E-cadherin expression vector (mEcad) for 24 h and subjected to immunoblotting for E-cadherin, PTEN and -actin. . Results represent the mean ± SEM (n=3; a, P < 0.001, as compared with SKOV-3/shCtl control; b, P < 0.001, as compare with SKOV-3/shEcad controls [pIRES]). (D) Stably transfected OVCAR-3 and SKOV-3 cells were transient transfected with PTEN promoter construct and -galactosidase plasmid. Twenty-four hours after transfection, cells were transfected with pIRES empty vector or murine E-cadherin expression vector (mEcad) for a further 24 h and subjected to luciferase and -galactosidase assays. The luciferase activity of each sample was normalized with the -galactosidase activity. Results represent the mean ± SEM (n=3; a, P < 0.001, as compared with SKOV-3/shCtl control; b, P < 0.001, as compared with SKOV-3/shEcad controls [pIRES]). (E) Illustration of PTEN promoter reporters used for luciferase assay (left). Stably transfected SKOV-3 cells were transient transfected with pGL3-basic vector (pGL3-basic), truncated pGL3-PTEN promoters, Egr1 mutant promoter construct (mutEgr1) and -galactosidase plasmid for 48 h and subjected to luciferase and -galactosidase assays. The luciferase activity of each sample was normalized with the -galactosidase activity. Results represent the mean ± SEM (n=3; **, P < 0.001, as compared with SKOV-3/shCtl control). (F) SKOV-3 cells were transiently transfected with pcDNA-GFP (GFP) or pcDNA-PTEN-GFP (PTEN-GFP; 1-2 g) for 24 h and subjected to immunoblotting for PTEN, cyclin D1, p27Kip1, and -actin. The total amount of plasmid DNA transfected in each group was balanced with pcDNA-GFP.

Figure 5 Loss of E-cadherin inhibits PTEN transcription via GSK3 inactivation (A) TCF activity was analyzed using the TOPFLASH and FOPFLASH luciferase reporters. Cells were transfected with either the TOPFLASH or FOPFLASH luciferase reporter, -galactosidase vector was cotransfected for normalization of transfection efficiency. Twenty-four hours after transfection, 20mM LiCl was added for 24 h before harvesting the cells for the measurement of luciferase and -galactosidase activities. Values are normalized luciferase activity (as described in the Materials and methods section) and are shown as mean ± SEM of three independent experiments performed in triplicate (**, P < 0.001). (B) OVCAR-3 and SKOV-3 cells were cultured for 24 h in the presence or absence of 20 mM LiCl and PTEN, Egr1 and -actin protein levels were analyzed by Western blotting. (C) SKOV-3 cells were transiently transfected for 48 h with pcDNA 3.1 (pcDNA), or constitutively active GSK3 (GSK3-S9A), and Western blots of PTEN and -actin protein levels were analyzed. Results represent the mean ± SEM (n=3; **, P < 0.001). (D) SKOV-3 cells were transient transfected with wild type PTEN (WT) or Egr1 mutant promoter construct (mutEgr1) and -galactosidase plasmid. Twenty-four hours after transfection, cells were treated with 20mM LiCl, or transfected with pcDNA 3.1 (pcDNA), or constitutively active GSK3 (GSK3-S9A) for a further 24 h and subjected to luciferase and -galactosidase assays. The luciferase activity of each sample was normalized with the -galactosidase activity. Results represent the mean ± SEM (n=3; *, P < 0.05; **, P< 0.001).

Figure 6 Loss of E-cadherin inhibits PTEN transcription via -catenin/TCF-mediated Egr1 downregulation (A) Protein levels of PTEN and -actin were determined in SKOV-3/shCtl, sh-cat, shEcad, and shEcad + sh-cat cells. (B) OVCAR-3 and SKOV-3 cells were transiently transfected with pcDNA 3.1 or constitutively active -catenin (S33Y; 0.5-1 g) for 24 h and subjected to immunoblotting for -catenin, PTEN, and -actin. The total amount of plasmid DNA transfected in each group was balanced with pcDNA 3.1. Results represent the mean ± SEM (n=3; *, P < 0.05; **, P< 0.001). (C) -catenin, PTEN and -actin protein levels were examined in SKOV-3/sh-cat cells transiently trasfected with pcDNA 3.1 (pcDNA), constitutively active -catenin (S33Y), or dominant negative TCF (DNTCF4). The total amount of plasmids transfected in each group was balanced with pcDNA 3.1. Results represent the mean ± SEM (n=3; a, P < 0.001, as compared with SKOV-3/shCtl control; b, P < 0.05, as compare with SKOV-3/sh-cat controls [pcDNA]). (D) SKOV-3 cells were transient transfected with wild type PTEN (WT) or Egr1 mutant promoter construct (mutEgr1) and -galactosidase plasmid. Twenty-four hours after transfection, cells were transfected with pcDNA 3.1 (pcDNA), constitutively active -catenin (S33Y), or dominant negative TCF (DNTCF4) for a further 24 h and subjected to luciferase and -galactosidase assays. The luciferase activity of each sample was normalized with the -galactosidase activity. The total amount of plasmids DNA transfected in each group was balanced with pcDNA 3.1. Results represent the mean ± SEM (n=3; a, P < 0.001, as compared with pcDNA 3.1 control; b, P < 0.001, as compared with S33Y).

Figure 7 Regulation of PTEN levels by cell density and E-cadherin-cadherin interactions. (A) OVCAR-3, SKOV-3/shCtl, and SKOV-3/sh-cat ovarian cancer cells were grown as sparse (6 x 103 cells/cm2) or dense (6 x 104 cells/cm2) cultures, and the levels of E-cadherin, PTEN, phosphoylated and total Akt, and -actin were determined by Western blot. (B) SKOV-3 cells were seeded as dense cultures in the presence of monoclonal mouse anti-E-cadherin antibody or control antibody, and the localization of -catenin was examined by immunofluorescence microscopy. Scale bar: 20 m (C) The levels of PTEN and -actin were determined by Western blot analysis of lysates from SKOV-3 cells (shCtl or sh-cat) incubated with anti-E-cadherin antibody or control antibody. Results represent the mean ± SEM (n=3; **, P< 0.001).

Figure 8 Proposed model of E-cadherin action. The presence of E-cadherin inhibits PI3K/Akt signaling, reduces cyclin D1, and promotes increased levels of p27Kip1, thus inhibiting cell growth. E-cadherin decreases the accumulation of -catenin in the nucleus leading to increased levels of Egr1 which induces PTEN expression and result in reduced PI3K/Akt signaling. However upon E-cadherin loss during tumor progression, the accumulation of -catenin in the nucleus leads to -catenin/TCF transactivation and the suppression of Egr1 and PTEN levels, resulting in reduced negative regulation of the PI3K/Akt signaling pathway. Under these conditions, the activation of PI3K/Akt signaling further stabilizes -catenin signaling by inhibiting GSK3, thus leading to increased cell growth.