N-Cadherin-Mediated Cell Motility Requires Cis Dimers

Abstract

Cadherins are expressed on the cell surface as a dimer in the membrane of one cell (cis dimer) that interacts with a cis dimer on an adjacent cell to form an adhesive trans dimer. It is well established that both cis and trans dimers must form for the cadherin to be an effective adhesion protein. In addition to their adhesive activity cadherins also play an important role in modulating cell behavior by regulating cell motility and signal transduction. Whether or not cis or trans dimers are necessary for the nonadhesive functions of cadherins has not been addressed. Here we show that N-cadherin cis dimers are necessary to induce cell motility in epithelial cells and that N-cadherin's ability to modulate the steady state levels of activated small GTPases requires both cis and trans dimers.

Keywords:

• Cadherins
• junctions
• cell motility
• cis dimers

INTRODUCTION

Cadherins comprise a family of calcium dependent type I glycoproteins that mediate cell-cell adhesion primarily through homotypic interactions. Cadherins contain an extracellular portion, a single-pass transmembrane segment, and a cytoplasmic tail. The extracellular portion, consisting of five homologous repeats (EC1–EC5), promotes cell-cell adhesion via protein-protein interactions between EC1 of cadherins on adjacent cells. The cytoplasmic tail of cadherins is connected to the cytoskeleton via several proteins known as catenins (reviewed in [1, 2, 3, 4]).

For cadherins to be functional as adhesion molecules they must first form cis dimers within the plane of the membrane of a single cell. It is thought that a cadherin cis dimer on one cell interacts with a cadherin cis dimer on an adjacent cell resulting in the formation of a trans dimer, which promotes cell-cell adhesion (reviewed in [5, 6]). Point mutations in E-cadherin or N-cadherin that change tryptophan #2 of the mature protein to alanine prevent cis dimer formation, which also prevents trans dimer formation [7, 8]. Alternatively, an aspartic acid #134 to alanine mutation in E-cadherin prevents trans dimer formation but still allows formation of cis dimers [7].

Recent studies have shown that cadherins act not only as glue between cells, but also influence cell behavior (reviewed in [1, 2, 3, 4]). In particular, expression of a cadherin that is not normally expressed by a cell can change the phenotype of that cell. For example, our lab and others have shown that expression of cadherins like N-cadherin, cadherin-11, and R-cadherin, by epithelial cells, increases cell motility and invasive capacity [9, 10, 11, 12, 13]. The mechanism whereby these cadherins promote motility in epithelial cells is still unknown. However, it has been suggested that N-cadherin mediated cell motility may involve growth factor receptor signaling.

Studies from the Walsh and Doherty labs have suggested that N-cadherin interacts with the FGF receptor, and this interaction promotes FGFR activation [14]. In their model, dimerization of N-cadherin brings together two FGFR molecules, leading to activation of the FGFR signal transduction pathway. Our lab showed that the motility of human mammary carcinoma cells (BT20) over expressing N-cadherin was abrogated by an inhibitor of the FGFR signal transduction pathway, further suggesting that N-cadherin may promote cell motility via an FGFR-dependent mechanism [12]. Of particular importance to the Walsh and Doherty model, complexes involving N-cadherin and FGFR have been reported. Peluso et al. showed coimmunoprecipitation of FGFR1 and N-cadherin from an ovarian surface epithelial cell line [15]. Cavallaro et al. showed an interaction between FGFR4 and N-cadherin from pancreatic beta cells that was dependent upon expression of NCAM [16]. Finally, Suyama et al. showed an interaction between FGFR1 and N-cadherin and identified the extracellular region of FGFR1 as being responsible for binding N-cadherin [13].

The present study was designed to further understand the mechanism whereby N-cadherin promotes cell motility. The Walsh and Doherty model predicts that dimerization of N-cadherin is necessary to activate the FGFR pathway. Thus, if FGFR activation is the mechanism whereby N-cadherin promotes cell motility, preventing N-cadherin dimerization in epithelial cells should reduce the motility of these cells to that of N-cadherin negative cells. To address this question a tryptophan #2 to alanine point mutation (W2A) in N-cadherin was made to prevent the formation of both cis and trans dimers. In addition, an aspartic acid #134 to alanine point mutation (D134A) in N-cadherin was made to allow cis dimer formation while preventing trans dimer formation. When these mutant cadherins were expressed in BT20 cells, the W2A mutant was significantly less effective than N-cadherin at promoting motility, while the D134A mutant was as effective as N-cadherin. Thus, our data support the idea that N-cadherin cis dimers are an important component of N-cadherin-mediated cell motility.

METHODS

Reagents

All reagents were from Sigma Chemical Company (St. Louis, MO) unless otherwise indicated.

Molecular Constructions

The polymerase chain reaction was used to assemble cDNAs or make the W2A and D134A point mutations. All of the final PCR products were sequenced to confirm they encoded the correct amino acid sequences. The LZRS-MS-neo viral expression vector was kindly provided by Dr. Al Reynolds [Vanderbilt University [17]]. The LZRS vector was modified by replacing the neo cassette with the pac gene from pBSpacΔ p [18] to make LZRS-MS-pac. Wild type N-cadherin and its mutants were modified at their C-termini to include either 2 birch profilin tags [19] or 2 myc tags. The myc and birch profilin tagged versions were inserted into LZRS-MS-pac or LZRS-MS-neo using standard procedures. Details are available upon request. The dominant negative cadherin, E-cad-M, has been previously described [20] and consists of the cytoplasmic tail of E-cadherin fused to the N-terminal myristoylation sequence from v-Src [21]. E-cad-M was ligated into LZRS-MS-neo.

Cell Culture

The human breast cancer cell line BT20 (American Type Culture Collection) was maintained in minimal essential medium (MEM) supplemented with 10% fetal calf serum (Hyclone Laboratories). Phoenix cells [22] were maintained in Dulbecco's modified Eagle's medium (DMEM) with 10% fetal calf serum. A431D cells, a derivative of the human cervical carcinoma cell line A431, have been described previously [23] and were maintained in DMEM with 10% fetal calf serum. HT1080 human fibrosarcoma cells (American Type Culture Collection) were maintained in DMEM with 10% fetal calf serum. Culture medium was supplemented with 50 units/l penicillin and 50 mg/l streptomycin. Infected populations were maintained in the appropriate selective antibiotic, either G418 or puromycin.

Antibodies

Mouse monoclonal antibodies 13A9, which recognizes the cytoplasmic domain of human N-cadherin, and 8C11, which recognizes the extracellular domain of N-cadherin, have been described [24, 12]. The monoclonal antibody against human E-cadherin (HECD1) was provided by Dr. Masatoshi Takeichi (Kyoto University). The mouse monoclonal antibodies against human P-cadherin (7B12), α -catenin (1G5), and β -catenin (5H10) have been described [25]. The mouse monoclonal antibody against the myc-epitope (9E10.2) was a gift from Dr. Kathleen Green (Northwestern University). The mouse monoclonal antibody against the birch profilin epitope (4A6) was a gift from Dr. Mannfred Rudiger (Technical University of Braunschweig). Secondary antibodies for immunofluorescence were FITC-conjugated anti-mouse secondary antibodies, and for western blots were horseradish peroxidase-conjugated anti-IgG (Jackson Immuno Research Laboratories). GAPDH antibody was purchased from Ambion.

Virus Production

Phoenix cells were transfected using a calcium phosphate kit (Stratagene) according to the manufacturer's instructions. Transfected cells were selected in puromycin and grown at 37°C until 50–80% confluence, at which time the medium was replaced with fresh medium without selective antibiotic. The cells were grown overnight at 32°C to increase virus stability [22]. Virus-containing media was filtered (0.45 μ m) and polybrene added to a final concentration of 4 μg/ml. Target cells were plated at low density and incubated in virus-containing medium at 32° overnight. Infected cells were then cultured in fresh media until 80% confluent and then selected in the appropriate antibiotic.

Immunofluorescence Microscopy

Cells were grown on glass cover slips overnight and then briefly washed in Hanks balanced salt solution buffered with 50 mM HEPES, pH 7.4. Cells were fixed with HistoChoice (Amresco) for 20 min and blocked in PBS containing 10% heat-inactivated goat serum for 30 min. Fixed cells were incubated in primary antibody for 1 h in a humid chamber, washed and incubated in secondary antibody for 1 h. Cover slips were washed and mounted in Vectashield mounting medium (Vector Laboratories). For immunofluorescence experiments with nonpermeabilized cells, the cells were fixed for 15 min in 3.7% formaldehyde-HEPES/Hanks, blocked for 10 min and incubated in primary antibodies followed by secondary antibodies for 20 min in a humid chamber. Phase and fluorescence pictures were taken with an ORCA-ER (Hamamatsu) digital camera mounted on a Zeiss Axiovert 200M microscope (Carl Zeiss) and images were collected using OpenLab software (Improvision Inc.).

Detergent Extraction of Cells

Confluent monolayers were rinsed three times with phosphate buffered saline and extracted in TNE (10 mM Tris acetate pH 8.0, 0.5% NP-40, 1 mM EDTA) containing 2 mM phenylmethylsulfonyl fluoride and 2 mM sodium orthovanadate or in the same buffer without EDTA. The cells were placed on ice, scraped, and triturated vigorously for 10 min. Insoluble material was pelleted by centrifugation at 14,000g for 15 min at 4°C, and the supernatant was used immediately or stored at −80°C.

Immunoprecipitation and Immunoblot

All polypropylene tubes were rinsed with 0.1% NP-40 and dried prior to use in immunoprecipitations. 300 μ l of hybridoma-conditioned medium was added to 50 μ l of packed anti-mouse IgG affinity gel (ICN Biochemical Co.), gently mixed at 4°C for 30 min and then washed to remove unbound antibody. 300 μ l of cell extract was added and mixed for 30 min. Immune complexes were washed five times with TBST (10 mM Tris HCl pH 7.5, 150 mM NaCl, 0.05% Tween-20) containing 2 mM sodium orthovanadate, resuspended in 50 μ l of 2X Laemmli sample buffer, boiled for 5 min, and the proteins resolved by SDS-PAGE [26]. Proteins were electrophoretically transferred overnight to nitrocellulose membranes and immunoblotted as previously described [27].

Aggregation Assays

Cells were trypsinized to generate single cell suspensions and resuspended at a density of 2 × 10⁵ cells/ml. Four thousand cells (20 μ l) were placed on the inside cover of a 100 mm dish and allowed to aggregate in hanging drops without rotation at 37°C for 18 h as described [28]. The cells were triturated and remaining aggregates observed and photographed as above.

Motility Assays

Motility assays were done essentially as previously described with some modifications [12]. Briefly, 3 × 10⁵ cells in 3 ml of medium supplemented with 10% FBS were plated in the top chamber of a 6-well tissue culture insert where the membrane had 8 μ m pores (Becton Dickson). 2 ml of medium supplemented with 10% FBS were placed in the bottom chamber of the six well dish. After 24 h, the top of the filter was scraped with cotton swabs three times to remove the cells that did not migrate to the bottom of the well. Cells on the bottom of the filter were stained with Diff-Quick stain set (Dade Diagnostics, Inc). Nine random fields of cells from each filter were counted at 100x magnification. The average number of cells in a field was calculated as well as the standard deviation. The data were collected from three independent experiments. Significant differences between two data sets were determined by one-way ANOVA followed by Fisher's F test.

Rac1 Pull Down Assays

Fusion protein was prepared and coated onto beads as described [29]. The beads were resuspended in an equal volume of wash buffer and stored at −80°C. Tissue culture cells were rinsed twice with cold Tris-buffered saline (50 mM Tris–HCl pH 7.5, 150 mM NaCl) and scraped in lysis buffer (50 mM Tris-HCl pH 7.5, 200 mM NaCl, 10 mM MgCl₂, 1 mM dithiothreitol (DTT), 1% (v/v) Nonidet P-40, 5% (v/v) glycerol, 1 μ g/ml leupeptin, 1 μ g/ml aprotinin, 1 mM PMSF). The mixture was incubated on ice for several min, clarified by centrifugation and used immediately. 10% of the cell lysate was used for the total protein control, while the remainder was incubated with beads coated with the GST fusion protein for 45 min to 1 h at 4°C. Beads were centrifuged, washed three times in lysis buffer and resuspended in 2X Laemmli sample buffer for SDS-PAGE. The relative activities for Rac1 pull down assays were calculated as follows: Adobe Photoshop was used to measure the intensity of the active and total Rac1 bands on the immunoblots. Active Rac1/total Rac1 was calculated and then the relative activity was calculated by setting the control BT20 cell activity to 1. Significant differences between two data sets were determined by one-way ANOVA followed by Fisher's F test.

RESULTS

Characterization of N-cadherin Mutants

The goal of this study was to determine if it was necessary for N-cadherin to form either cis or trans dimers in order to promote motility in epithelial cells. It has been previously shown that a tryptophan to alanine mutation in AA 2 of mature N-cadherin prevents the formation of both cis and trans dimers. In addition, it has been shown that an aspartic acid to alanine mutation in AA 134 of E-cadherin prevents trans dimers but allows cis dimers. An amino acid alignment of E-cadherin and N-cadherin suggested that an aspartic acid to alanine mutation in AA 134 of N-cadherin would also prevent trans dimers while still allowing cis dimers. Thus, we reckoned we could use N-cadherin constructs harboring either the W2A mutation or the D134A mutation to test whether cis and/or trans dimers were required for N-cadherin to promote motility in epithelial cells.

To test the efficacy of the mutations, we prepared two constructs of wild type N-cadherin and two constructs of each of the point mutations in N-cadherin and tagged one with myc epitope tags and the other with birch profilin tags (birch; Figure 1A). Each construct was cloned into the viral expression vector LZRS-MS-Neo or LZRS-MS-pac for production of infectious particles and infection into the cadherin null cell line, A431D. In addition, the myc and birch-tagged versions of wild type N-cadherin and of each mutation were coexpressed in the same cell (Figure 1B). Coculturing cells expressing myc-tagged wild type or mutant N-cadherin with cells expressing the birch-tagged version would allow us to determine if the cadherins could form trans dimers by immunoprecipitating with antibodies against one tag and immunoblotting back with antibodies against the other tag. In addition, the cell lines coexpressing both the myc tagged N-cadherin and the birch tagged N-cadherin would allow us to determine if the wild type or mutant cadherins were present as cis dimers by immunoprecipitating with antibodies against one tag and immunoblotting for the other tag.

Figure 1 N-cadherin constructs. A. Diagram of wild type N-cadherin and N-cadherin mutants showing the myc or birch tags and the location of the W2A and D134A mutations. B. Table indicating the construct name, cell line designation and constructs expressed in each cell line.

It was first necessary to show that each construct was expressed at approximately physiological levels and on the cell surface in A431D cells. Figure 2A shows an immunoblot of equal amounts of protein from extracts of A431D cells expressing each construct. The extracts were immunoblotted with anti-N-cadherin monoclonal antibody 13A9, so the levels of expression could be directly compared to one another. The amount of mature N-cadherin expressed by each cell line was somewhat higher than that expressed by the HT1080 control that endogenously expresses N-cadherin. In addition, there was a significant level of unprocessed N-cadherin. Cadherins are synthesized with a proregion that is processed before the protein is transported to the cell surface. When we express exogenous N-cadherin in cells, it is not unusual for the cells to retain unprocessed protein in the Golgi/ER [10, 12, 27, 30]. Most of the constructs showed some unprocessed protein, which ran above the mature protein and is indicated in the figure.

Figure 2 Expression of N-cadherin constructs in A431D cells. A. Equal amounts of protein from extracts of cells expressing each construct (see Figure 1B) were resolved by SDS-PAGE and immunoblotted with anti-N-cadherin (13A9) or GAPDH as a loading control. HT1080 cells were used as a positive control as they express endogenous N-cadherin. Visible are the mature cadherin proteins as the major bands and higher molecular weight unprocessed forms (immature forms). B. Proteins from cells expressing two differently tagged cadherins (myc tagged and birch tagged) were resolved by SDS-PAGE in two separate lanes and immunoblotted separately with antibodies against either the myc or birch tag to show that each construct was expressed in each cell line. C. Extracts of singly infected cells were immunoprecipitated (IP) with anti-N-cadherin (13A9) or control hybridoma supernatant (C) and immunoblotted for α-catenin and β-catenin. D. Extracts of doubly transfected cells were immunoprecipitated with antibodies against either the myc tag, the birch tag or control hybridoma supernatant as indicated at the bottom of the gel, and immunoblotted for α-catenin and β-catenin.

Immunoblotting with anti-N-cadherin allowed us to compare levels of expression among the cell lines, but not to distinguish the two tagged forms in cells expressing both cadherins (A431DN-myc + N-birch, A431DN-W2A-myc + N-W2A-birch and A431DN-D134A-myc + N-D134A-birch). To show that each form was expressed in these cells, we immunoblotted extracts of these cells with antibodies that detect the myc tag or the birch tag (Figure 2B). Each line expressed both versions of the tagged transgene. It was equally important to show that each protein was complexed with α-catenin and β-catenin. To do this we immunoprecipitated extracts of each singly transfected cell line with antibodies against N-cadherin (13A9) and immunoblotted with antibodies against α-catenin and β-catenin (Figure 2C). Each extract was also immunoprecipitated with control hybridoma supernatant (C) that did not contain specific antibody. There was little or no α-catenin or β-catenin in the control immunoprecipitations. To show that both cadherins were complexed with catenins in cells co-expressing myc-tagged and birch-tagged versions of the N-cadherin constructs, we immunoprecipitated the N-cadherin with antibodies against the tag and immunoblotted for α-catenin and β-catenin. In each case, both catenins coimmunoprecipitated with the cadherin and the control lanes were negative for both α-catenin and β-catenin (Figure 2D).

For these studies to be meaningful, we had to show that the mutant cadherins were expressed on the cell surface and determine if they were functional. We first tested the ability of each construct to mediate cell-cell interactions in aggregation assays (Figure 3A). As expected, A431DN cells aggregated, whether the protein was tagged with myc or birch. None of the mutant N-cadherin constructs were capable of mediating aggregation. This is consistent with the notion that they are not capable of forming trans (D134A) or cis + trans (W2A) dimers, which are required to mediate cell-cell interactions. It is possible the mutant cadherins did not mediate aggregation because they were not presented on the cell surface. To rule this out, we stained cells for immunofluorescence localization using nonpermeabilized cells and an anti-N-cadherin antibody that recognizes the extracellular domain of human N-cadherin (8C11). Figure 3B shows that each protein was present on the cell surface and thus available to mediate aggregation if it were capable of it. When nonpermeabilized cells were stained with the 13A9 antibody that recognizes the cytoplasmic domain of N-cadherin no signal was seen, confirming that indeed the cells were not permeable and the 8C11 antibody was staining protein on the outside of the cell (data not shown).

Figure 3 Assessment of adhesive function of N-cadherin mutant proteins. A. Mock infected A431D cells (A431D-neo) or cells infected with one of the tagged N-cadherin constructs were assessed for their ability to aggregate. Cells expressing either tagged version of wild type N-cadherin (A431DN-myc and A431DN-birch) formed large aggregates. Neither the A431D-neo cells nor cells expressing tagged versions of the mutant N-cadherins were able to form large aggregates. Although it appeared there were more cells in the fields where cells can fully aggregate (A431DN-myc and A431DN-birch), this is an artifact of the experimental procedure. The cells were photographed in a drop with a coverslip applied to the surface. Nonaggregated cells spread out more than did aggregated cells, giving the appearance of lower cell density. B. Cells expressing the neo control, tagged wild type N-cadherin, a single mutant N-cadherin or coexpressing two tagged versions of a mutant N-cadherin were processed for immunofluorescence without permeabilizing the cells. They were then stained with an antibody against the extracellular domain of N-cadherin (8C11). Each construct was effectively expressed on the surface of the cells.

The next set of experiments was designed to demonstrate first that we could use the myc and birch tagged forms of wild type N-cadherin to identify cis and trans dimers and second that the W2A and D134A mutants behaved as predicted. We first used A431D cells coexpressing both N-myc and N-birch to show that the two proteins coimmunoprecipitated (Figure 4A left panel). We also used a culture that was a mixture of A431DN-myc cells and A431DN-birch cells (i.e. co-culture) to show that N-cad-myc coimmunoprecipitated with N-cad-birch (Figure 4A right panel) indicating that we could indeed identify trans dimers using this method. In the cells coexpressing both tagged versions of N-cadherin, both cis and trans dimers presumably contributed to the coimmunoprecipitation. When similar coimmunoprecipitation experiments were done using extracts of cells expressing the myc and birch-tagged W2A mutant N-cadherin constructs, the two tagged proteins did not coimmunoprecipitate whether they were expressed in the same cell or in different cells (i.e., in a coculture) indicating that neither cis nor trans dimers were formed (Figure 4B). When similar coimmunoprecipitation experiments were done using extracts of cells expressing the myc and birch-tagged D134A mutant N-cadherin constructs, the two tagged forms of the protein coimmunoprecipitated when they were expressed in the same cell but not when they were expressed in different cells, indicating that cis dimers did form but trans dimers did not (Figure 4C). When the immunoprecipitation reactions were immunoblotted with the same antibody used in the immunoprecipitation, the band was clearly evident, indicating that a significant amount of protein was pulled down in each immunoprecipitation reaction (data not shown). The above experiments show that we have generated mutant forms of N-cadherin that prevent it from making both cis and trans dimers (N-W2A) or prevent trans dimers while allowing cis dimers (N-D134A). Thus, we have the reagents necessary to determine whether the ability of N-cadherin to promote motility in epithelial cells depends on its ability to form cis and/or trans dimers.

Figure 4 Analysis of dimer formation by N-cadherin mutant proteins. Left hand panels: A431D cells expressing both myc-tagged and birch-tagged versions of wild type N-cadherin (A), the W2A mutant N-cadherin (B), or the D134A mutant N-cadherin (C) were extracted in buffer without EDTA to preserve cadherin interactions, and immunoprecipitations were done with either anti-myc tag or anti-birch tag. The immunoprecipitation reactions were resolved by SDS-PAGE and immunoblotted with antibodies against the other-tag, (i.e., lane 1 was immunoprecipitated with antibodies against the myc tag and immunoblotted with antibodies against the birch tag as noted above the lane). A. Antibodies against the myc tag effectively coimmunoprecipitated birch tagged wild type N-cadherin and vice versa, indicating that wild type N-cadherin can form cis and/or trans dimers as diagramed below the gel. B. Antibodies against the myc tag could not coimmunoprecipitate the birch-tagged W2A mutant N-cadherin and vice versa, indicating that the W2A mutant cannot make cis dimers or trans dimers. C. Antibodies against the myc tag effectively coimmunoprecipitated the birch-tagged D134A mutant N-cadherin and vice versa, indicating that wild type N-cadherin can form cis dimers as diagramed below the gel. We know this coimmunoprecipitation is not due to trans dimers because of the experiment shown on the right hand side of the figure and explained below. Right hand panels: A431D cells expressing myc-tagged wild type N-cadherin (A), the W2A mutant N-cadherin (B), or the D134A mutant N-cadherin (C), were cocultured with A431D cells expressing the birch-tagged version of the same construct. The cocultures were extracted and processed as described above. A. Antibodies against the myc tag effectively coimmunoprecipitated birch-tagged wild type N-cadherin and vice versa, indicating that wild type N-cadherin can form trans dimers as diagramed below the gel. In this case, we are certain that they are trans dimers, as this is the only way the two tags could coimmunoprecipitate. B. Antibodies against the myc tag could not coimmunoprecipitate the birch-tagged W2A mutant N-cadherin and vice versa, indicating that the W2A mutant cannot make trans dimers. C. Antibodies against the myc tag could not coimmunoprecipitate the birch-tagged D134A mutant N-cadherin and vice versa, indicating that the D134A mutant cannot make trans dimers. This experiment confirms that the D134A mutant can make cis dimers since the gel on the left shows coimmunoprecipitation of the tags. Diagramed below the gels are the dimers presumably seen in the blots.

N-cadherin Mutants that Cannot Form Dimers Retain the Ability to Down-regulate Endogenous Cadherins

Previous studies from our lab have shown that expression of N-cadherin increases cell motility and in some cases decreases the expression of endogenous E- and P-cadherins. Endogenous cadherins have been shown by others to be important regulators of cell motility, so analysis of changes in their expression level when we expressed exogenous N-cadherin constructs was important. We were interested in determining if it is the intrinsic characteristics of N-cadherin that produce phenotypic changes in epithelial cells. To address this question, we set out to determine if N-cadherin could produce these phenotypic changes in cells if it were mutated such that it could not form either cis or trans dimers. Thus, we expressed the mutant N-cadherin constructs in BT20 human breast cancer cells, which is the cell line we have previously used for motility studies. These cells express E-cadherin and P-cadherin but do not express endogenous N-cadherin. When BT20 cells were infected with constructs encoding N-cadherin (BT20N), N-W2A-cadherin (BT20N-W2A) or N-D134A-cadherin (BT20N-D134A), the cadherin was expressed at cell-cell borders or in a pattern that included some staining of cell-cell borders and some diffuse staining (Figure 5A). Immunofluorescence of nonpermeabilized cells with the anti-N-cadherin antibody directed against the extracellular domain (8C11) showed that a significant amount of the cadherin was localized on the cell surface (Figure 5B). When nonpermeabilized cells were stained with the 13A9 antibody that recognizes the cytoplasmic domain of N-cadherin no signal was seen, confirming that indeed the cells were not permeable and the 8C11 antibody was staining protein on the outside of the cell (data not shown). Immunoblots showed that each construct was efficiently expressed (Figure 5C).

Figure 5 Expression and localization of mutant N-cadherin proteins in BT20 cells. BT20 cells were infected with wild type N-cadherin (BT20N), the W2A mutation (BT20N-W2A), or the D134A mutation (BT20N-D134A). A. Cells were processed for immunofluorescence under permeabilized conditions and stained with antibodies against the cytoplasmic domain of N-cadherin (13A9). Wild type N-cadherin is present at cell-cell borders as expected for a functioning cadherin. The two mutant cadherins are expressed at cell borders to a lesser degree, which is not surprising since they do not form effective junctions. B. To show the mutant N-cadherins are expressed on the cell surface, we did immunofluorescence of nonpermeabilized cells using antibodies against the extracellular domain of N-cadherin (8C11). Both the W2A mutant and the D134A mutant showed extensive staining with this antibody. C. Extracts of BT20 cells and BT20 cells expressing wild type N-cadherin or the mutant constructs were resolved by SDS-PAGE and immunoblotted with anti-N-cadherin (13A9). Each construct was effectively expressed.

To address the ability of these mutant N-cadherin constructs to down-regulate the expression of endogenous cadherins, we did immunofluorescence and immunoblots for the endogenously expressed E-cadherin and P-cadherin (Figure 6). Expression of wild type N-cadherin in BT20 cells (BT20N) resulted in significant down-regulation of both E-cadherin and P-cadherin. Likewise, expression of either N-W2A-cadherin or N-D134A-cadherin resulted in down-regulation of the endogenous cadherins. Interestingly, cells that expressed the D134A mutant, even though they expressed significantly decreased levels of endogenous cadherins, did put detectable amounts of the remaining endogenous cadherin at cell-cell borders. We do not have an explanation for this difference in cadherin localization. In contrast, the localization of endogenous cadherins in cells expressing the W2A mutant resembled that in cells expressing the wild type N-cadherin. These data suggest that the ability of N-cadherin to induce down-regulation of endogenous cadherins does not depend on its being an effective adhesion protein, and are consistent with reports from the Reynolds and Kowalczyk laboratories showing that competition for p120 catenin binding regulates cadherin levels [31, 32].

Figure 6 The W2A and D134A mutant N-cadherins down-regulate endogenous cadherins in BT20 cells. A. BT20 cells or BT20 cells expressing wild type N-cadherin (BT20N), the W2A mutant of N-cadherin (BT20N-W2A), the D134A mutant of N-cadherin (BT20N-D134A), or a dominant negative cadherin (BT20D/N) were processed for immunofluorescence with antibodies against E-cadherin (left hand row) or P-cadherin (right hand row). B. Extracts of the cells shown in (A) were resolved by SDS-PAGE and immunoblotted with antibodies against E-cadherin, P-cadherin, or GAPDH as a control. C. Immunoblots using E-cadherin antibodies were quantified and plotted, setting the value for BT20 cells at 1. The graph represents values and standard deviations of three experiments. D. Immunoblots using P-cadherin antibodies were quantified and plotted, setting the value for BT20 cells at 1. The graph represents values and standard deviations of three experiments.

Cis Dimer Formation is Essential for N-cadherin to Promote Motility in Epithelial Cells

Our main interest is in how N-cadherin functions to modulate the motility of epithelial cells, which we know from previous studies involves more than merely down-regulating expression of endogenous cadherins [12]. Thus, our next experiment was to compare the motility of mock infected BT20 cells to BT20 cells infected with wild type N-cadherin, N-W2A-cadherin or N-D134A-cadherin. Since there is data in the literature suggesting that E-cadherin acts to suppress motility in epithelial cells and one consequence of expressing N-cadherin in BT20 cells was to down-regulate the endogenous cadherins (Figures 6A, B), we used as a control for these studies a dominant negative cadherin construct that consists of the cytoplasmic domain of E-cadherin that is tethered to the plasma membrane by a lipid tail. We have previously shown that expression of this dominant negative cadherin results in significant down-regulation of endogenous cadherins [20]. When the dominant negative cadherin (D/N) was expressed in BT20 cells, E-cadherin and P-cadherin were down-regulated to a level equivalent to that seen with the N-cadherin constructs (Figures 6A, B). Thus, this construct provided an appropriate control for the motility assays.

The motility assay we performed was a transwell assay where the cells were plated on top of a filter and the number of cells traversing the filter was counted (Figure 7). The BT20 cells expressing D/N cadherin had a motility rate indistinguishable from the mock infected BT20 cells, indicating that merely down-regulating the endogenous E-cadherin and P-cadherin did not increase motility. As expected, the N-cadherin expressing cells showed significantly higher motility than the mock-infected cells (p < 0.0001). Cells expressing the D134A mutation had motility rates that were significantly higher than mock-transfected cells (p < 0.0001) but not significantly different from that of BT20N cells (p = .19). Interestingly, cells expressing the NW2A mutant had a motility rate intermediate between that of BT20 cells and BT20N cells. The motility was higher than mock transfected cells with a p value of < 0.001 but also significantly lower than that of BT20N cells (p = 0.001). These data are consistent with our hypothesis that cis dimers between N-cadherin promote cell motility possibly by causing dimerization of the FGF receptor.

Figure 7 Cis dimers of N-cadherin are necessary to promote maximal cell motility. 3 × 10⁵BT20 cells or BT20 cells transfected with wild type N-cadherin (BT20N), the W2A mutant of N-cadherin (BT20N-W2A), the D134A mutant of N-cadherin (BT20N-D134A), or dominant negative cadherin (BT20D/N) were plated in the upper well of a Boyden chamber and the cells crossing the membrane were assessed by counting the number of cells per field of view. BT20N (P < 0.0001), BT20N-W2A (P = 0.0007), and BT20N-D134A (P < 0.0001) were more motile than BT20 cells (indicated by *). BT20N (P < 0.0001), BT20N-W2A (P = 0.0013), and BT20N-D134A (P < 0.0001) were more motile than BT20D/N cells (indicated by #). BT20N (P = 0.0003) was more motile than BT20N-W2A (indicated by ⁁). There were no significant differences between BT20 (P = 0.71) and BT20D/N, or between BT20N (P = 0.19) and BT20N-D134A cells. The graph represents relative motilities and standard deviation of three experiments.

We have previously shown that RHC80267 (which inhibits the activity of diacylglycerol lipase) inhibits N-cadherin-mediated motility in BT20 cells [12]. This inhibitor was shown by the Walsh and Doherty labs to effectively inhibit N-cadherin-dependent neurite outgrowth, which is dependent upon FGF receptor signaling [33]. When we treated cells expressing the mutant N-cadherins with RHC80267, their motility was inhibited in a dose dependent manner, similar to that of cells expressing wild type N-cadherin (data not shown). In addition, the Hazan lab showed that the FGF receptor coimmunoprecipitates with the W2A mutant, as it does with wild type N-cadherin [13]. Together, these data suggest that the mutant N-cadherins influence FGF receptor signaling in a manner similar to that of wild type N-cadherin.

N-cadherin Adhesive Activity Appears to be Necessary to Increase the Level of Activated Rac1

We have recently shown that when R-cadherin is expressed in epithelial cells, it also promotes motility [11]. We showed that expression of R-cadherin resulted in sustained signaling through Rac1 and that Rac1 activity was necessary for R-cadherin induced motility [11]. R-cadherin resembles N-cadherin structurally and is 74% identical to N-cadherin. R-cadherin is inappropriately expressed by some tumor cells, but inappropriate expression of N-cadherin by tumor cells is more widespread. Thus, characterization of the mechanism whereby N-cadherin promotes motility is important. Therefore, we investigated whether N-cadherin expression in BT20 cells also resulted in increased steady state levels of activated Rac1, and whether increased activation of Rac1 was dependent upon the ability of the N-cadherin to dimerize. Figure 8 shows that indeed, BT20N cells have a significantly higher level of activated Rac1 than do parental BT20 cells (p = 0.002). However, sustaining activated levels of Rac1 appears to depend on the adhesive activity of N-cadherin as neither the cells transfected with the NW2A mutant nor cells transfected with the D134A mutant had levels of activated Rac1 that were significantly different from mock transfected cells (p = .68 or .07, respectively). On the other hand, it is possible that there are pools of active Rac1 that are independent of N-cadherin adhesion since the D134A mutant had levels of Rac1 activity that are almost significant (p = .07). It is likely that multiple cellular components influence Rac1 activity and that cadherin-based adhesion only partially determines the level of active Rac1 in a cell.

Figure 8 Both Cis and trans dimers of N-cadherin are needed to increase the steady state level of activated Rac1. Pull down assays were used to evaluate GTP binding to Rac1 in BT20 cells or BT20 cells transfected with wild type N-cadherin (BT20N), the W2A mutant of N-cadherin (BT20N-W2A), or the D134A mutant of N-cadherin (BT20N-D134A). Cell extracts were made and 10% of the total volume was used for total GTPase levels. The remaining extract was subjected to pull down assays. Proteins were resolved by SDS PAGE and immunoblotted with antibodies against Rac1. A representative gel is shown indicating active Rac1 and total Rac1. B. The graph represents relative activities of five experiments and the standard deviation of Rac1 GTPase activity. For statistical analysis, one-way ANOVA followed by Fisher's F test was done to examine the differences between the cell lines. BT20N cells had significantly higher relative Rac1 activity than BT20 (P = 0.0015), BT20N-W2A (P = 0.0027), and BT20N-D134A (P = 0.03) cells (indicated by *). BT20N-W2A (P = 0.68) and BT20N-D134A (P = 0.07) were not significantly different than BT20 cells. Also BT20N-W2A (P = 0.14) was not significantly different than BT20N-D134A.

DISCUSSION

The expression of inappropriate cadherins like N-cadherin or R-cadherin in epithelial cells helps to promote late stage complications of tumorigenesis such as increased cell motility and metastasis. Understanding how expression of inappropriate cadherins accomplishes this will provide important information that may be useful in targeting treatments for metastatic disease.

Motility and invasion are complex cellular behaviors that are influenced by a number of factors [34, 35, 36, 37, 38, 39, 40]. One important inhibitor of tumor cell invasion is the interaction of individual tumor cells with one another. Since most cells of epithelial origin interact with one another using E-cadherin and E-cadherin is down-regulated in many metastatic tumor cells, it has been proposed that E-cadherin acts as a metastasis suppressor [34]. Obviously, since expression of an inappropriate cadherin by epithelial cells promotes down regulation of endogenous cadherins (e.g., E-cadherin) one must ascribe some aspects of altered cell behavior in response to expression of inappropriate cadherins to this decreased expression of E-cadherin. However, data from our previous studies [10, 11, 12] and from studies published by other investigators [9, 13, 42] together with the fact that expression of a dominant negative cadherin effectively down-regulates endogenous cadherins without increasing motility in an in vitro assay (Figures 6 and 7) points out that merely decreasing expression of E-cadherin is not sufficient to alter cell motility to a significant degree.

To test further the ability of N-cadherin and the mutant N-cadherins to influence motility without the complication of other cadherins, we tested the motility of A431D (cadherin null) cells expressing the various N-cadherins shown in Figure 3A. N-cadherin and each of the mutant cadherins promoted motility when compared to the control A431D-neo cells (data not shown). However, cells expressing the W2A mutant were less motile than those expressing wild type N-cadherin or the D134A mutant, supporting our conclusion that cis dimers of N-cadherin contribute to its ability to modulate cell motility.

It has been proposed that N-cadherin cooperates with the FGF receptor to promote motility. This idea was first postulated by the Walsh and Doherty laboratories in studies of neurite extension [14]. These authors suggested that N-cadherin facilitates dimerization of the FGF receptor to initiate a growth factor-independent signal. They have shown that plating neurons on N-cadherin substrates can induce outgrowth of neurites in the absence of growth factor, and that this outgrowth is dependent on signaling downstream of the FGF receptor [43, 44]. A recent study by the Doherty laboratory showed that peptides designed to cluster N-cadherin also induced neurite outgrowth [45]. These studies suggested that dimerization of N-cadherin could activate an FGF receptor-dependent signal and demonstrated that N-cadherin-mediated signaling is distinct from its adhesive activity. The studies we present in this article demonstrate that cis dimerization of N-cadherin is critical to its ability to increase motility in cancer cells and draws additional parallels between epithelial cell motility and neurite extension, further implicating ligand independent FGF receptor signaling in N-cadherin-induced cell motility.

Interestingly, it has also been shown that N-cadherin is also involved in ligand-dependent FGF receptor signaling [9]. This study showed that N-cadherin interacts with the FGF receptor to stabilize it on the cell surface and thus prolong FGF receptor signaling. This report, as well as others, have shown the existence of a complex between the FGF receptor and N-cadherin using coimmunoprecipitation assays [13, 15, 16]. Our laboratory has attempted to demonstrate the existence of a complex that includes N-cadherin and FGF receptors using coimmunoprecipitation assays, and we have not been successful. We have used a number of commercially available antibodies, and we have expressed tagged versions of several receptor isoforms in N-cadherin-expressing cells and have not been able to convincingly coimmunoprecipitate a complex that includes N-cadherin and the FGF receptor. Thus, it is not possible for us to design experiments that would further define interactions between our mutant N-cadherin constructs and FGF receptors.

When we consider our data in light of the studies from the Walsh and Doherty labs together with that from the Hazan lab, it is clear that the relationship between the FGF receptor and N-cadherin is complex. This is further evidenced by the fact that there remains a significant increase in motility in cells expressing the W2A mutant of N-cadherin, suggesting a dimer-independent component to motility. Perhaps this is due to increased retention time of the receptor on the cell surface as suggested by the Hazan lab, who showed that the W2A mutant of N-cadherin still coimmunoprecipitated with the FGF receptor [13].

We have previously shown that Rho family GTPases play an important role in R-cadherin mediated motility [11]. In the present study, we extend these studies to show that the Rho family GTPase, Rac1 is elevated in cells that express N-cadherin. Experiments addressing the influence of cadherins on Rho GTPases have primarily focused on transient activation when cadherins engage [46, 47, 48, 49, 50]. Our results are unique because they show changes in the steady state activity levels of Rho GTPases in cells expressing N-cadherin (or R-cadherin), which could reflect the involvement of these molecules in the increased motility observed in cells expressing an inappropriate cadherin. BT20 cells express E-cadherin and P-cadherin, and yet expressing R-cadherin (our previous studies) or N-cadherin (this study) in these cells resulted in increased activation of Rac1, suggesting that cadherins may differentially activate Rho GTPases. The mechanism whereby cadherins might differentially activate Rho GTPases has yet to be determined. One possibility might be through their interactions with p120 catenin, a protein that has been shown by a number of laboratories to be involved in cadherin-dependent modulation of Rho GTPases [46, 47, 48, 49, 50]. The present study suggests that both cis dimerization and trans dimerization are important components of the ability of N-cadherin to promote an increase in the steady state level of activated Rac1 GTPase. It is not known what role p120 catenin plays in cadherin dimerization, but the fact that it plays a critical role in the stability of cadherins and the ability of cadherins to modulate the activity of Rho family GTPases indicates that this protein regulates multiple facets of cadherin metabolism and suggests it may also function downstream of cadherin dimerization.

It is interesting to note that the levels of active Rac1 in lysates of cells expressing wild type N-cadherin or the mutant N-cadherins shown in Figure 8 do not strictly agree with the motility shown in Figure 7. This further points out the complexity of cell motility and suggests there are multiple pathways that influence motility in these cells. However, it is clear that N-cadherin, the FGF receptor, and Rac1 influence some of these pathways.

ACKNOWLEDGEMENTS

The authors thank Drs. M. Rudiger, K. Green, M. Takeichi and A. Reynolds for reagents. This work was supported by DE12308, GM51188 and by RR018759 from the NIH.