R-cadherin influences cell motility via Rho family GTPases.

Classical cadherins are the transmembrane proteins of the adherens junction and mediate cell-cell adhesion via homotypic interactions in the extracellular space. In addition, they mediate connections to the cytoskeleton by means of their association with catenins. Decreased cadherin-mediated adhesion has been implicated as an important component of tumorigenesis. Cadherin switching is central to the epithelial to mesenchymal transitions that drive normal developmental processes. Cadherin switching has also been implicated in tumorigenesis, particularly in metastasis. Recently, cadherins have been shown to be engaged in cellular activities other than adhesion, including motility, invasion, and signaling. In this study, we show that inappropriate expression of R-cadherin in tumor cells results in decreased expression of endogenous cadherins (cadherin switching) and sustained signaling through Rho GTPases. In addition, we show that R-cadherin induces cell motility when expressed in epithelial cells and that this increased motility is dependent upon Rho GTPase activity.

Classical cadherins are the transmembrane components of the adherens junction. E-Cadherin is found in epithelial cell junctions (E-cadherin), 1 whereas other cadherins are found in similar structures in other cell types (1,2). Disruption of Ecadherin function through mutations in E-cadherin itself, through modifications to the E-cadherin promoter or through disruption of the cadherin/catenin complex, contributes to the aggressive behavior of epithelial-derived tumor cells (1). Studies from our laboratory and others have shown that expression of an inappropriate cadherin in epithelial cells is another way that tumor cells can alter their adhesive function (3)(4)(5)(6). In some cases, this is due to down-regulation of E-cadherin upon expression of the inappropriate cadherin (4). In other cases, neuronal cadherin (N-cadherin) can have a direct and dominant influence upon the phenotype of epithelial cells, despite their continued expression of E-cadherin. For example, expression of N-cadherin by oral squamous epithelial cells results in decreased expression of the endogenous cadherins, whereas expression of N-cadherin by some breast epithelial cells has no effect upon expression of endogenous cadherins, but none the less influences cell adhesion and behavior (4,5). Thus, exogenous expression of N-cadherin can modulate the expression of endogenous cadherins in some cell types but not in others. Interestingly, exogenous expression of vascular endothelial cadherin (VE-cadherin) in a variety of cell types does not influence the expression levels of any endogenous cadherins (7). However, VE-cadherin does prevent proper junctional localization of N-cadherin, but has no influence upon the localization of other cadherins (7). In the present study, we show that exogenous expression of retinal cadherin (R-cadherin) down-regulates the expression of all endogenous cadherins in every cell type examined. Thus, it is clear that cells have a mechanism for sensing cadherin expression and that different cell types respond differently when they express various cadherins.
R-cadherin resembles E-and N-cadherin structurally and is ϳ74% identical to N-cadherin. Shapiro and co-workers (8) showed that R-cadherin can form both trans and cis heterodimers with N-cadherin. In addition, the expression patterns of these two cadherins overlap in some tissues (9). Rcadherin is highly expressed in the brain and plays crucial roles in the segmentation of the developing brain and in neuronal outgrowth (10). Expression of R-cadherin is necessary for the epithelial to mesenchymal transition that occurs in developing kidneys and is important in striated muscle development (11). Whether or not R-cadherin plays a role in tumorigenesis similar to that played by N-cadherin has not been investigated. However, high levels of R-cadherin mRNA have been detected in several bladder and prostate carcinoma cell lines (12,13).
Recent studies have shown that E-cadherin-mediated cellcell contact can activate Rac1 and/or Cdc42, depending upon the experimental system, whereas RhoA activity decreases as cells become confluent (14 -18). Nelson and co-workers (19) showed that as individual cells make contact using E-cadherin, cell protrusions become focused on the area of cell-cell contact where Rac1 accumulates. Likewise, Yap and co-workers (16) plated cells on immobilized E-cadherin extracellular domain dimers to show that cadherin ligation recruits Rac1 to nascent contacts at the leading edges of cadherin-based lamellipodia. These investigators also showed that the ability to bind p120 catenin is critical to the ability of E-cadherin to rapidly activate Rac1 (20). Reynolds and co-workers (21) showed that cytosolic p120 catenin inhibits the activity of RhoA by acting as a guanine nucleotide dissociation inhibitor. The laboratories of Burridge and co-workers (22) and Bershadsky and co-workers (23) have suggested that p120 catenin activates Vav2, which has been reported to be a guanine exchange factor for RhoA, RhoG, Cdc42, and Rac1. Rac1 and Cdc42 seem to have similar functions in E-cadherin-mediated adhesion, whereas RhoA performs a separate function (2, 24 -26).
Gauthier-Rouviere and co-workers (27) used myogenic cells to show that RhoA, but not Rac1 or Cdc42, was activated upon N-cadherin-mediated cell-cell contact. Mege and co-workers (28) used the same system to show that inhibiting Rac1 did not influence initial N-cadherin cell-cell contact, but that Rac1 was important in connecting the cadherin to the cytoskeleton. Dejana and co-workers (29) showed that Rac1, but not RhoA, was activated in VE-cadherin-expressing endothelial cells, and Kouklis et al. (30) showed that non-junctional VE-cadherin induced Cdc42 activation, which led to membrane protrusions. In the present study, we show that R-cadherin expression by a variety of cell types results in increased steady-state activity levels of Rac1 and Cdc42, that activation of these GTPases correlates with increased cell motility, and that dominantnegative forms of these GTPases inhibit R-cadherin-dependent cell motility. These studies are significant because they highlight the importance of cellular context in cadherin-mediated GTPase activation and demonstrate that Rho GTPases and cadherins cooperate not only in cell adhesion but also in cell motility.

EXPERIMENTAL PROCEDURES
Cell Culture-The human breast cancer cell line BT-20 was obtained from American Type Culture Collection (ATCC) and maintained in minimal essential medium (MEM, Sigma) supplemented with 10% fetal calf serum (FCS, Hyclone Laboratories). BT-20R cells are BT-20 cells retrovirally transduced to express R-cadherin. A431 cells (ATCC), Phoenix cells (a kind gift from Dr. Albert Reynolds, Vanderbilt University), and A431Ds were maintained in Dulbecco's modified Eagle's medium (Sigma) with 10% FCS. A431 is a squamous epithelial cell line derived from a cervical cancer. A431D cells were derived in our lab by long-term treatment of A431 cells with dexamethasone. These cells have lost expression of all classical cadherins and have been described previously in more detail (31). A431DR and A431DE are A431D cells retrovirally transduced to express R-cadherin or E-cadherin, respectively. The human squamous carcinoma cell line UM-SCC-1 (SCC-1; a kind gift of Dr. Thomas Carey, University of Michigan) was maintained in MEM with 10% FCS. HT1080, HACAT, SkBr3, and MCF-7 (ATCC) were maintained in DMEM with 10% FCS. Culture medium was supplemented with 50 units/liter penicillin and 50 mg/liter streptomycin (Sigma). Where appropriate, cells that were infected with recombinant retroviruses were maintained in culture medium supplemented with 1 mg/ml G418 (Cellgro).
Aggregation and Motility Assays-Aggregation assays were performed as described (32). Briefly, trypsinized cells were resuspended in 50-l drops of medium containing 5000 cells/drop and pipetted onto the inner surface of the lid of a 100-mm Petri dish and incubated over culture medium for 24 h at 37°C. The drops were pipetted 10 times to disrupt weak aggregates prior to photography. Transwell filter motility assays were performed as described previously (5).
Constructs-PCR products utilized in any of the constructions described below were sequenced and shown to contain only the intended changes. The mouse R-cadherin cDNA (GenBank accession no. D14888) was a kind gift from Dr. Masatoshi Takeichi (33). Using PCR, the R-cadherin stop codon was replaced with a ClaI site, allowing insertion into a derivative of pSPUTK (34) that added a C-terminal 2ϫmyc tag. The modified cDNA was then inserted into LZRS-MS-Neo/GFP (35). Two C-terminal myc tags were added to human E-cadherin using a similar strategy, and the modified cDNA was inserted into the LZRS vector. The Cdc42/Rac1 interactive binding (CRIB) domain of p21activated kinase 1 was amplified from total RNA of MDA-MB-435 cells. The resulting sequence was identical to the one used by Benard et al. (36). The primers PAK1.1 5Ј-CGTGGATCCAAGAAAGAGAAAGAGCG-GCCA-3Ј and PAK1.2 5Ј-CTAGAATTCAGCTGACTTATCTGTAAA-3Ј added 5Ј BamHI and 3Ј EcoRI sites for insertion into pGEX-2T (Amersham Biosciences), resulting in a glutathione S-transferase (GST)-CRIB fusion. The Cdc42-binding domain (CBD) of the Wiskott-Aldrich syndrome protein (15) was amplified from IMAGE cDNA clone 344979 (ATCC). The primers 5Ј-ATATATG GATCCCTCCCAGCACCTG-GACC-3Ј and 5Ј-CGCGCCGAATTCCTCAATGAAGTCGT-3Ј added 5Ј BamHI and 3Ј EcoRI sites for insertion into pGEX-2T, resulting in a GST-CBD fusion. The plasmid encoding the rhotekin Rho-binding domain-GST fusion consisting of amino acids 7-89 of mouse Rhotekin fused to GST was a kind gift of Dr. Richard Cerione (Cornell University, Ithaca, NY) and was provided by Dr. Robert Lewis (Eppley Cancer Institute). The N-terminal GFP-tagged constructs Cdc42N17, Rac1N17, and RhoAN19 (kind gifts of Dr. Klaus Hohn, Scripps Institute, La Jolla, CA) were subcloned into LZRS-MS-Neo.
Transfection, Virus Production, and Infection-Phoenix cells (37) were transfected using a calcium phosphate transfection kit (Stratagene) with LZRS-MS-Neo/GFP (35) and selected in 2 g/ml puromycin (Sigma). Cells were cultured until confluent, split, and stored in liquid nitrogen. Transfected Phoenix cells were grown at 37°C in selective medium until 50 -80% confluent, at which time the medium was replaced with fresh medium without puromycin and cells were grown overnight at 32°C. Virus-containing medium was removed, filtered through a 0.45-m filter, and supplemented with polybrene (4 g/l, Sigma). Target cells were plated overnight at low density to ensure formation of single cells and then incubated in the virus-containing media at 32°C for 2-6 h. Infected cells were cultured until ϳ80% confluent and then either selected in 1 mg/ml G418 or sorted by fluorescence-activated cell sorting for GFP expression. Cells infected with the empty LZRS-MS-Neo virus were used to set the background fluorescence. Cell sorting was performed by the University of Nebraska Medical Center Cell Analysis Facility.
Pull-down Assays for Rac1/Cdc42-Fusion protein was prepared and coated onto beads as described (15,36). The beads were resuspended in an equal volume of wash buffer and stored at Ϫ80°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 MgCl 2 , 1 mM dithiothreitol, 1% (v/v) Nonidet P-40, 5% (v/v) glycerol, 1 g/ml leupeptin, 1 g/ml aprotinin, 1 mM phenylmethylsulfonyl fluoride). 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, whereas the remainder was incubated with beads coated with the GST-CRIB or GST-CBD fusion proteins for 45 min to 1 h at 4°C. Beads were centrifuged, washed 3ϫ in lysis buffer, and resuspended in 2ϫ Laemmli sample buffer for SDS-PAGE (38). For RhoA pull-down assays, fusion protein was prepared and coated onto beads as described (39) with minor modifications. Tissue culture cells were treated the same as above except that the lysis buffer was 50 mM Tris-HCl, pH 7.2, 500 mM NaCl, 1% Triton X-100, 0.5% sodium deoxycholate, 0.1% SDS, 10 mM MgCl 2 .
Immunofluorescence, Cell Extraction, and Immunoprecipitation-Cells plated on glass coverslips for 24 -48 h were fixed and stained as described (42), viewed on a Zeiss Axiovert 200M equipped with an ORCA-ER (Hamamatsu) digital camera, and processed using OpenLab software (Improvision Inc.). Confluent cell monolayers were extracted and immunoprecipitated as described (42). The pellets were resuspended in 2ϫ Laemmli buffer and subjected to SDS-PAGE. Membrane and cytosolic fractions were prepared as described (43).

Exogenous Expression of R-Cadherin Down-regulates Endogenous
Cadherins-To facilitate detection of R-cadherin in transfected cells, we myc-tagged mouse R-cadherin, cloned it into the viral expression vector LZRS-MS-Neo, and produced infective virus. Biological activity of the tagged construct was confirmed by stably expressing it in A431D cells, a cadherinnegative derivative of A431 cells (31). After infection, a population of cells was selected with G418 and designated A431DR. Control A431D cells infected with viral particles containing only the neomycin resistance gene were designated A431DNeo. Extracts of A431DNeo and A431DR cells were immunoblotted with monoclonal antibodies against myc, ␣-catenin, and ␤-catenin (Fig. 1A). A431DNeo cells did not exhibit a myc signal at the correct size for R-cadherin-2ϫ-myc whereas the A431DR cell extract had such a signal. We have shown previously that the transfection of a cadherin into A431D cells results in the stabilization of ␣and ␤-catenin (43). Thus, as expected, expression of R-cadherin in these cells resulted in increased signals for ␣and ␤-catenin. To mediate full adhesion, classical cadherins must be attached to the cytoskeleton via catenins. Extracts from A431DNeo or A431DR cells were immunoprecipitated with anti-myc, resolved by SDS-PAGE, and immunoblotted for ␣-catenin (Fig. 1B) or ␤-catenin (data not shown). The catenins co-immunoprecipitated with R-cadherin-2ϫ-myc, indicating that this cadherin was anchored to the actin cytoskeleton. To further show that R-cadherin-2ϫ-myc was functional, we performed aggregation assays and immunofluorescence localization. Fig. 1C shows that A431DR cells form large aggregates, whereas A431Neo cells do not aggregate. In addition, A431DR cells show cell border staining with anti-myc antibodies (R-cadherin) and with anti-␤-catenin antibodies (Fig. 1D). Together, the data presented in Fig. 1 show that R-cadherin-2ϫ-myc is correctly processed, localized at cell borders, interacts with catenins, and mediates cell adhesion.
Previous work from our laboratory (4,5) showed that exogenous expression of N-cadherin down-regulated expression of endogenous cadherins in some cells but not in others, suggesting that N-cadherin-mediated down-regulation of endogenous cadherins was cell-type specific. To determine whether R-cadherin influenced endogenous cadherin expression, we infected a number of cell lines with recombinant retrovirus encoding R-cadherin-2ϫ-myc. In every case, endogenous cadherins were down-regulated. Fig. 2A shows immunofluorescence localization of cadherins in control BT-20Neo cells (panels a-c) and in BT-20R cells (panels d-f). The control cells did not exhibit plasma membrane staining with the myc antibody, but the R-cadherin-expressing cells showed membrane localization with this antibody. Timed exposures of BT-20R cells and the neomycin counterparts showed that the R-cadherin-expressing cells had diminished signal for both E-and P-cadherin at cell-cell borders (compare panels d and e with panels a and b, respectively). Immunoblot analysis further showed that BT-20R cells had a decreased expression of E-cadherin and Pcadherin when compared with control BT-20Neo cells (Fig. 2B). In addition to the BT-20 cells, exogenous expression of R-cadherin-2ϫ-myc in A431 (E-and P-cadherin), HT-1080 (N-cadherin), HACAT (E-, P-, and N-cadherin), MCF-7 (E-cadherin), or SCC-1 (E-and P-cadherin) cells resulted in reduced expression of all endogenous cadherins (data not shown). These data suggest that R-cadherin differs from N-cadherin in its ability to regulate endogenous cadherin expression.
Exogenous Expression of R-Cadherin Increases Cell Motility-To examine the effect of R-cadherin expression on cell motility, 5 ϫ 10 5 BT-20Neo or BT-20R cells were plated in three separate transwell inserts and incubated overnight with NIH3T3conditioned medium in the bottom chamber as a chemoattractant. Nonmotile cells were removed from the top of the insert, and the cells on the bottom of the filter were counted. Fig. 3A shows the average number of cells per field of view from the three independent experiments. BT-20 cells expressing R-cadherin were significantly more motile than the control BT-20Neo cells. A complicating factor in this experiment was that R-cadherin expression resulted in down-regulation of E-and P-cadherin in BT-20 cells. To determine whether R-cadherin could influence cell motility independently of its ability to decrease endogenous cadherins, we investigated the influence R-cadherin had on motility in cadherin-null cells. The cadherin-null cell lines A431D and SkBr3 were infected with control neomycin resistance virus or with R-cadherin-2ϫ-myc-expressing virus and used in motility assays. Expression of R-cadherin increased the motility of both A431D (Fig. 3B) and SkBR-3 (data not shown) cell lines. Interestingly, in the case of A431D cells, the motility rate was only two times that of the parental cell line. Because the increase in motility was roughly 4-fold higher in BT-20R cells than in BT20Neo cells, it is likely that the down-regulation of endogenous E-and P-cadherins in BT-20R contributed to the difference in motility. In any case, the experiments with A431D and SkBr3 cells showed that R-cadherin has an intrinsic ability to influence cell motility.
R-Cadherin-based Motility Is Due in Part to the Activation of Rac1 and Cdc42-A separate method to measure cell motility is a wound-healing assay, in which a confluent monolayer of cells is wounded and the cells are allowed to migrate into the denuded area. Such an assay is more difficult to quantify than FIG. 1. Functional analysis of R-cadherin-2؋-myc. Recombinant retrovirus was used to stably express R-cadherin-2ϫ-myc together with the neomycin resistance gene or the neomycin resistance gene alone in A431D cells (A431DR or A431DNeo, respectively). A, immunoblot. 40 g of total protein was resolved on 7% polyacrylamide gels, transferred to nitrocellulose, and immunoblotted with antibodies against myc (Rcadherin(myc)), ␣-catenin, or ␤-catenin. B, immunoprecipitation. 300 g of total protein from A431DNeo or A431DR cells were incubated with anti-myc antibody or control supernatant (C, no antibody control). Immunocomplexes were isolated with anti-mouse IgG-conjugated-Sepharose (IP, immunoprecipitation), resolved on 7% polyacrylamide gels and immunoblotted (IB) with anti-␣-catenin. Cell extract (E, 20 g of total protein) was run as positive control for the immunoblot. C, aggregation assays. 5000 cells were suspended in 50 l of medium and placed on the lids of tissue culture dishes to form hanging drops. Pictures were taken after overnight incubation at 37°C. D, immunofluorescence. A431DNeo cells (panels a-c) or A431DR cells were stained with antibodies against ␤-catenin or myc (R-cadherin(myc)). FITCconjugated anti-mouse secondary antibody was used to detect the primary antibodies. a and d, phase microscopy; b and e, R-cadherin localization; c and f, ␤-catenin localization. Equally timed exposures were used to take the photographs presented in panels b, c, e, and f. the assays used above, but it can provide information about how the cells move. When A431Neo cells were compared with A431R cells in a wound-healing assay, the R-cadherin-expressing cells not only migrated significantly further than the A431Neo cells but also displayed robust lamellipodia at the leading edge of the wound (Fig. 4, compare B with A). Similarly, the expression of R-cadherin in A431D cells caused these cells to display membrane protrusions (Fig. 1D), and immunofluorescence of BT-20 cells showed that E-cadherin contacts were more linear than R-cadherin contacts, which were jagged and suggestive of membrane protrusions ( Fig. 2A). Immunofluorescence localization of R-cadherin (Fig. 4D) and E-cadherin (Fig.  4C) in A431D cells using higher magnification suggested that R-cadherin-expressing cells interact through overlapping membranous structures, whereas E-cadherin interactions produce a smoother interface. Indeed, even when we expressed R-cadherin in A431 cells, which endogenously express E-and Pcadherin, the cell borders displayed long membrane protrusions (Fig. 4F), resembling filopodia or microspikes.
Experiments from other laboratories have shown a connection between cadherins and the activity of Rho GTPases (15,16,29,30,44). The increased motility of R-cadherin-expressing cells, together with the increased membrane ruffling observed in these cells, led us to hypothesize that expression of R-cadherin may alter the steady-state activity levels of one or more of the Rho GTPases Cdc42, Rac1, or RhoA. To assay for activated GTPases, we employed GST fused to the GTPase-binding domain of PAK1, WASP, and Rhotekin, which bind respectively to the active, GTP-bound forms of Rac1, Cdc42, and RhoA, and we performed pull-down assays with extracts of A431D cells, A431D cells expressing E-cadherin or R-cadherin, BT-20 cells, and BT-20 cells expressing R-cadherin.
The A431D cell line is a good model system because it lacks endogenous cadherin expression, allowing us to investigate the absolute influence of cadherin expression on GTPase activa-tion. 3 ϫ 10 6 cells were plated overnight, then lysed and incubated with fusion proteins conjugated to glutathione-Sepharose beads. A431D cells expressing neo-resistance alone or neo-resistance plus E-cadherin showed approximately equal levels of Rac1 activation, whereas expression of R-cadherin by these cells significantly increased the steady-state activity level of Rac1. There was a slight elevation in the level of activated Cdc42 in A431DR cells, but it was not statistically significant (Fig. 5A). Activated RhoA could not be detected in A431Dneo, A431DE, or A431DR cells (data not shown).
Of equal interest was determining if R-cadherin was capable of altering the steady-state activity levels of the Rho GTPases in the motility model cell line BT-20. The parental BT-20 cells express E-cadherin and P-cadherin, each of which may also influence the activity of small GTPases. If R-cadherin-mediated motility is driven by the activation state of these molecules, we would expect there to be elevated levels of the active forms of one or more of the small GTPases in BT-20R cells when compared with parental BT-20Neo cells. 2.5 ϫ 10 6 BT-20Neo or BT-20R cells were plated and processed for pull-down assays (Fig. 5B). There was a 1.5-fold increase in Rac1 activation and a 2.5-fold increase in the activation of Cdc42 in the R-cadherin-expressing cells. Interestingly, there was a decrease in steady-state RhoA GTP levels in the BT-20R cells compared with the parental cell line. This finding is consistent with the idea that Rac1 and Cdc42 activities tend to promote similar cellular traits, whereas RhoA opposes their action (24 -26).
To further investigate the role of Rac1 and Cdc42 in Rcadherin-dependent motility, a dominant-negative form of each GTPase was expressed in BT-20R cells. Recombinant viruses containing dominant-negative Rho GTPases fused to green fluorescent protein (GFP) or GFP alone were used to infect BT-20R cells. Cells expressing GFP or the GFP fusions were sorted from non-expressing cells by fluorescence-activated cell sorting. Fig. 6A shows phase microscopy of living cells and the

FIG. 2. R-cadherin-2؋-myc expression in epithelial cells down-regulates endogenous E-and P-cadherin.
Recombinant retrovirus was used to stably express R-cadherin-2ϫ-myc together with the neomycin resistance gene or the neomycin resistance gene alone in BT-20 cells (BT-20R or BT-20Neo, respectively). A, immunofluorescence. BT-20Neo cells (panels a-c) or BT-20R cells (panels d-f) were stained for E-cadherin, P-cadherin, or R-cadherin(myc). FITC-conjugated anti-mouse secondary was used to detect the primary antibodies. a and d, E-cadherin localization; b and e, P-cadherin localization; c and f, R-cadherin localization. B, immunoblots. 40 g of total protein were resolved by SDS-7% PAGE, transferred to nitrocellulose, and immunoblotted with antibodies against E-cadherin, P-cadherin, or R-cadherin(myc).
corresponding GFP signal after sorting. The dominant-negative GTPases were both diffuse in the cytosol and at the plasma membrane, suggesting that at least a portion is in junctions, which is surprising because it is typically the GTP-bound form of these molecules that is junctional (14,25). Whether R-cadherin-containing junctions differ in this regard from E-cadherin-containing junctions has yet to be determined. 40 g of total protein from each cell line was resolved on a 7% polyacrylamide gel and immunoblotted for R-cadherin, E-cadherin, and P-cadherin (Fig. 6B). Importantly, dominant-negative Rho GT-Pases did not significantly influence the expression levels of the cadherins. Cell extracts were resolved on a 14% polyacrylamide gel and blotted with antibodies against Rac1 or Cdc42 (Fig. 6C). Expression of the dominant-negative Rho GTPases resulted in a slight down-regulation of the corresponding endogenous protein. To determine whether BT-20R cell motility was altered by the expression of dominant-negative Rho GTPases, BT-20R cells expressing dominant-negative (DN) Rho GTPase fusion proteins were subjected to motility assays. 5 ϫ 10 5 cells from each cell line (BT-20Neo, BT-20R, DNRac1, and DNCdc42) were plated overnight in a motility assay (Fig. 6D). As expected, the BT-20RGFP cells displayed a motility rate higher than that of the parental BT-20Neo cells. BT-20R cells expressing DNRac1 or DNCdc42 were less motile than the BT-20RGFP and were not significantly different from the parental cells.
A role for Rho GTPases in cell motility has been established (45). Thus, to rule out the possibility that dominant-negative forms of Rac1 and Cdc42 were inhibiting general BT-20 cell motility and not motility specifically due to R-cadherin expression, we compared the motility of BT20 cells with that of BT-20 cells expressing either DNRac1 or DNCdc42 (Fig. 6E). There was no significant difference in the motility of any of these cell lines, further supporting a role for Rac1 and Cdc42 in R-cadherin-mediated cell motility.
One link between cadherins and Rho GTPases is p120 catenin. A mechanism whereby cadherins could differentially activate Rho GTPases would be through differing affinities for p120 catenin. That is, each cadherin family member may differentially bind p120 catenin and as a result have different influences on Rho GTPase activity. To test the hypothesis that R-cadherin and E-cadherin have different affinities for p120 catenin, we transduced 2ϫ-myc-tagged E-cadherin or 2ϫ-myctagged R-cadherin into the cadherin-null A431D cells and selected populations of cells that expressed approximately equal levels of cadherin as judged by immunoblots using the myc antibody directed against the tag on each cadherin. The populations also expressed approximately equal levels of ␤-catenin and p120 catenin (Fig. 7A). When the cells were partitioned into membrane and cytosolic fractions, virtually all of the cadherin and p120 catenin were in the membrane fraction, whether the cells expressed E-cadherin or R-cadherin (Fig. 7B). However, when cell extracts were immunoprecipitated with antibodies against p120 catenin and immunoblotted for p120 catenin and myc, there was less R-cadherin than E-cadherin for an equal amount of p120 (Fig. 7C). Likewise, when approximately equal amounts of cadherin were immunoprecipitated with anti-myc antibodies, less p120 catenin co-immunoprecipitated with R-cadherin than with E-cadherin (Fig. 7D). DISCUSSION Animal development frequently employs "cadherin switching" to facilitate the epithelial to mesenchymal transitions that are necessary for normal tissue structure (46). Cells undergo a similar epithelial to mesenchymal transition during tumorigenesis (46,47). This sometimes occurs when the epithelial tumor cell turns on the expression of an inappropriate nonepithelial cadherin, which results in down-regulation of endogenous cadherins, decreased cell-cell adhesion, and increased motility and invasion (3-6, 48, 49). N-cadherin has been shown to be inappropriately expressed by cells derived from squamous epithelial tumors (4) and by breast cancer cells (5). Interestingly, exogenous expression of N-cadherin in squamous epithelial cells results in decreased expression of the endogenous cadherins, whereas breast cancer cells often continue to ex-press their endogenous cadherins when they are transfected with N-cadherin (4,5). R-cadherin plays an important role in epithelial to mesenchymal transitions in developing kidney and bladder and has been shown to be expressed in a number of cancer cell lines (12). In addition, R-cadherin is highly homologous to N-cadherin, which led us to hypothesize that transfection of R-cadherin into epithelial tumor cells may also influence expression of endogenous cadherins. Surprisingly, exogenous expression of R-cadherin consistently resulted in a drastic reduction of all endogenous cadherins in every cell line we checked. This was the case even for cells like HT1080 and HACAT that endogenously express N-cadherin. Thus, it seems that R-cadherin has an even higher capacity than N-cadherin to down-regulate expression of other cadherins.
It is clear that, in addition to their role in cell adhesion, cadherins play important roles in other cellular characteristics including cell motility, cellular polarity, cytoskeletal organization, and cellular signaling (1, 2). Investigations into the role were used to show that L cells transfected with empty vector were capable of closing the wound in 18 h, whereas L cells expressing E-cadherin were not, further substantiating that E-cadherin activity decreases the motility of cells (51). However, studies from other laboratories, including our own, have shown that inappropriate expression of non-epithelial cadherins such as N-cadherin increase motility and invasive behavior of tumor cells (4,5). These studies further suggested that N-cadherin increases cell motility through interactions with cell-surface receptors (5,52,53). In the present study, we showed that expression of R-cadherin, like N-cadherin, results in increased cell motility. Thus, there may be a number of cellular attributes that are influenced by cadherins. E-cadherin may suppress motility by increasing cell-cell interactions, whereas other cadherins like N-cadherin and R-cadherin may increase cell motility through signal transduction pathways or other mechanisms intrinsic to the cadherin molecule itself. This is evidenced by previous studies showing that exogenous expression of N-cadherin in breast cancer cells did not alter E-cadherin expression but did increase motility (5,52), and by the present study showing that R-cadherin expression increases the motility of the cadherin-null A431D cells.
Prior experiments addressing the influence of cadherins on Rho GTPases have primarily focused on transient activation when cadherins engage. Our results are exciting because they show changes in the steady-state activity levels of Rho GT-Pases, which could reflect the involvement of these molecules in the increased motility observed in R-cadherin-expressing cells. Thus, Rho GTPase family members may play an important role in the signaling events that regulate cadherin-dependent motility in addition to the role they play in cadherinmediated adhesion. Other laboratories have suggested links between Rho GTPases and cell motility (54). This is the first suggestion that increased cell motility via differential cadherin expression was due to altered activity of Rho GTPases. Importantly, R-cadherin-induced cell motility was reduced to that of parental BT-20Neo cells by expression of dominant-negative Rho GTPases. In addition, the levels of endogenous cadherins in BT-20R cells did not increase when dominant-negative Rho GTPases were expressed, suggesting that the dominant-negative effect was truly due to diminished GTPase activity and not to restoration of endogenous cadherins. These data support the idea that R-cadherin-induced cell motility is at least partially due to increased Rho GTPase activity.
BT-20 cells express E-cadherin and P-cadherin, and yet expressing R-cadherin in these cells resulted in increased activation of Rac1 and Cdc42, suggesting that cadherins may differentially activate Rho GTPases. A number of recent studies have shown that cadherins activate Rho GTPases; however, this is the first indication of differential activation of Rho GTPases by different cadherin family members. The mechanism whereby cadherins might differentially activate Rho GT-Pases has yet to be determined. One possibility might be through their interactions with p120 catenin, a protein which has been shown by a number of laboratories to be involved in cadherin activation of Rho GTPases (20 -23, 25, 44). In addition, the Dejana (55) lab has shown that VE-cadherin and N-cadherin have different affinities for p120 catenin, suggesting that cadherin-p120 catenin interactions do differ between different cadherins and lending credence to this hypothesis. The experiments shown in Fig. 7 indicate that almost all of the p120 catenin in A431DE and A431DR cells is membrane-associated and that there seems to be no differences in the behavior of the two p120 catenin isoforms expressed by these cells. Because less p120 catenin co-immunoprecipitates with R-cadherin than with E-cadherin (Fig. 7C), the simplest explanation is that p120 catenin is lost during the immunoprecipitation washes. However, an alternative possibility is that, although p120 catenin is membrane-associated in A431DR cells, it interacts with membrane-associated proteins in addition to R-cadherin. Perhaps the overall consequence is that R-cadherin and p120 catenin form more dynamic complexes than do E-cadherin and p120, and it is this dynamic aspect that contributes to the increased steady-state activity level of activated Rho family GTPases in cells expressing R-cadherin. An additional possibility is that when cells make contact with E-cadherin, there is eventually a reduction in the levels of activated Rho family members that contributes to contact inhibition. In contrast, this does not happen when cells make contact using R-cadherin, resulting in sustained membrane ruffling or other activities at sites of cell-cell contact. Our lab is currently investigating these possibilities.