Distinct Regions of the Cadherin Cytoplasmic Domain Are Essential for Functional Interaction with G (cid:1) 12 and (cid:2) -Catenin*

Heterotrimeric G proteins of the G 12 subfamily medi- ate cellular signals leading to events such as cytoskeletal rearrangements, cell proliferation, and oncogenic transformation. Several recent studies have revealed direct effector proteins through which G 12 subfamily members may transmit signals leading to various cellular responses. Our laboratory recently demonstrated that G (cid:1) 12 and G (cid:1) 13 specifically interact with the cyto- plasmic domains of several members of the cadherin family of cell adhesion molecules (Meigs, T. E., Fields, T. A., McKee, D. D., and Casey, P. J. (2001) Proc. Natl. Acad. Sci. U. S. A. 98, 519–524). This interaction causes (cid:2) -catenin to release from cadherin and relocalize to the cytoplasm and nucleus, where it participates in transcriptional activation. Here we report that two distinct regions of the epithelial cadherin (E-cadherin) tail are required for interaction with (cid:2) -catenin and G (cid:1) 12 , re-spectively. Deletion of an acidic, 19-amino acid region of E-cadherin abolishes its ability to bind (cid:2) -catenin in vitro , to inhibit (cid:2) -catenin-mediated transactivation, or to stabilize (cid:2) -catenin; causes subcellular mislocalization of (cid:2) -catenin; of E-cadherin Mutants— cDNA to the 98 C-terminal amino acids of human E-cadherin was amplified by polym- erase chain reaction (PCR). Deletion mutants within the E-cadherin cytoplasmic domain were generated by a two-step PCR procedure using primers corresponding to the N- and C-terminal ends of the domain, which introduced Bam HI and Eco RI restriction sites, respectively. For each mutant, these primers were used, along with internal primers flanking the region to be deleted, in two separate PCR reactions to amplify the N- and C-terminal regions of cDNA; products from the two reactions were then mixed and subjected to a second round of PCR using only the end primers. This reaction effectively “sewed” the N- and C-terminal regions together, producing constructs encoding the C-ter- minal domain of E-cadherin with the desired regions deleted. Constructs were then subcloned into pGEX-2T (Amersham Pharmacia Bio- tech, Piscataway, NJ) for production as glutathione S -transferase (GST) Full-length E-cad (cid:3) (cid:2) -cat (full-length E-cadherin lacking residues 819–837) was produced by first subcloning the full open reading frame of E-cadherin from pXEH2 into pcDNA1.1/CAT (which lacks a Bsp HI site), producing the plasmid E-cad-pcDNA. The region of E-cad (cid:3) (cid:2) -cat containing the deleted residues was then excised from the correspond- ing GST construct (described above) via digestion with Xho I and Bsp HI and subcloned into E-cad-pcDNA that had been digested in the identical manner. Finally, the open reading frame encoding the full-length version of E-cad (cid:3) (cid:2) -cat was subcloned into pcDNA3.1( (cid:1) ) (Invitrogen, Carlsbad, CA) using the Xba I and Eco RV sites present in both the pcDNA1.1/CAT and pcDNA3.1 ( (cid:1) ) vectors. Full-length wild-type E- cadherin was subcloned into pcDNA3.1( (cid:1) ) in a similar manner. All constructs were verified by sequencing. G (cid:1) 12 and (cid:2) -Catenin Protein— Recombinant G (cid:1) 12

Heterotrimeric guanine nucleotide-binding proteins (G proteins) 1 regulate cellular physiology by transducing extracellular signals to intracellular effector molecules (1). G proteins are composed of two functional signaling units, a nucleotide-bind-ing ␣ subunit and a tightly coupled ␤␥ subunit dimer. Based on the primary sequence of the ␣ subunits, G proteins have been classified into four subfamilies: G s , G i , G q , and G 12 (2). The G 12 subfamily, consisting of G␣ 12 and G␣ 13 , has been reported to mediate a variety of cellular processes, including regulation of Rho-dependent cytoskeletal rearrangements (3) and Na ϩ /H ϩ antiporter activity (4), activation of c-Jun N-terminal kinase (5), stimulation of phospholipase D activity (6), regulation of membrane depolarization (7), and conformational activation of radixin (8). Additionally, G 12 subfamily members have been implicated in pathways controlling cell proliferation and early developmental events, as well as oncogenesis (9 -12). Although Rho-dependent responses to G 12 signaling have been shown to involve specific guanine nucleotide exchange factors that directly interact with G␣ 12 and G␣ 13 (13)(14)(15)(16), signaling pathways involved in other G 12 -mediated responses, particularly cellular transformation, are not well understood.
Recently, our laboratory reported a specific interaction between G␣ 12 and G␣ 13 and the cytoplasmic domain of several cadherins (17), which are cell surface proteins involved in calcium-dependent cell-cell adhesion (18). This interaction was shown to cause ␤-catenin, a multifunctional protein involved in both cell adhesion and transcriptional activation (19), to release from its cadherin-bound state and relocalize to the cytoplasm and nucleus. Furthermore, in cells deficient in ␤-catenin degradation, expression of G␣ 12 or G␣ 13 was found to up-regulate ␤-catenin-mediated transcriptional activation (17). These findings have provided a mechanism that may explain the role of G 12 subfamily members in developmental processes as well as malignant transformation.
Identifying the regions of the cadherin tail necessary for interaction with both G␣ 12 and ␤-catenin is critical for understanding the mechanism of G␣ 12 -triggered release of ␤-catenin and for developing reagents to dissect the biological importance of the G␣ 12 /cadherin interaction. In the present study, we undertook a systematic analysis to identify the regions of Ecadherin that are required for interaction with G␣ 12 , and also expanded upon earlier studies by other researchers in which the region of cadherin critical for interaction with ␤-catenin was mapped. Our work demonstrates that distinct regions of the cadherin tail are required for interaction with G␣ 12 and with ␤-catenin and provides experimental systems that should allow dissection of the roles of these two key regulatory molecules in cadherin-mediated biological processes.

EXPERIMENTAL PROCEDURES
Miscellaneous Materials and Methods-The cDNAs for G␣ 12 Q229L and G␣ z Q205L were a gift of Henry Bourne (University of California, San Francisco). The plasmid pXEH2, encoding the full-length cDNA for human E-cadherin, was a gift of Yutaka Shimoyama (National Okura Hospital, Okura, Setagaya-ku, Tokyo, Japan). The G␣ 12 baculovirus construct was provided by Alfred Gilman (University of Texas South-* This work was supported in part by National Institutes of Health Grants GM55717 (to P. J. C.) and CA91159 (to P. J. C. and T. E. M.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 1 The abbreviations used are: G protein, guanine nucleotide-binding protein; E-cadherin or E-cad, epithelial cadherin; ␤-cat, ␤-catenin; wt, wild-type; TCF, T cell factor; G␣ 12 QL, constitutively active Q229L variant of G 12 protein ␣ subunit; GST, glutathione S-transferase; GFP, green fluorescent protein; FBS, fetal bovine serum; PBS, phosphatebuffered saline; PAGE, polyacrylamide gel electrophoresis; GTP␥S, guanosine 5Ј-3-O-(thio)triphosphate.
western Medical Center, Dallas). Baculovirus stock used for production of His-tagged ␤-catenin was from Barry Gumbiner (Memorial Sloan Kettering Cancer Center). Anti-G␣ 12 antibody was purchased from Santa Cruz Biotechnology, anti-G␣z antibody has been previously described (20), anti-E-cadherin and anti-␤-catenin antibodies were purchased from Zymed Laboratories Inc. (S. San Francisco, CA), and antigreen fluorescence protein (GFP) antibody was purchased from CLONTECH (Palo Alto, CA). K562 human chronic myelogenous leukemia cells and SW480 human colorectal carcinoma cells were from American Type Culture Collection (Manassas, VA). L mouse fibroblast cells were provided by Karl Willert (Stanford University). Protein concentrations were determined by the method of Bradford (21) or by staining with Coomassie Blue using bovine serum albumin standards.
Production of E-cadherin Mutants-cDNA corresponding to the 98 C-terminal amino acids of human E-cadherin was amplified by polymerase chain reaction (PCR). Deletion mutants within the E-cadherin cytoplasmic domain were generated by a two-step PCR procedure using primers corresponding to the N-and C-terminal ends of the domain, which introduced BamHI and EcoRI restriction sites, respectively. For each mutant, these primers were used, along with internal primers flanking the region to be deleted, in two separate PCR reactions to amplify the N-and C-terminal regions of cDNA; products from the two reactions were then mixed and subjected to a second round of PCR using only the end primers. This reaction effectively "sewed" the N-and C-terminal regions together, producing constructs encoding the C-terminal domain of E-cadherin with the desired regions deleted. Constructs were then subcloned into pGEX-2T (Amersham Pharmacia Biotech, Piscataway, NJ) for production as glutathione S-transferase (GST) fusion proteins. Selected constructs were also subcloned into pEGFP-C (CLONTECH) for generation of fusions to GFP.
Full-length E-cad⌬␤-cat (full-length E-cadherin lacking residues 819 -837) was produced by first subcloning the full open reading frame of E-cadherin from pXEH2 into pcDNA1.1/CAT (which lacks a BspHI site), producing the plasmid E-cad-pcDNA. The region of E-cad⌬␤-cat containing the deleted residues was then excised from the corresponding GST construct (described above) via digestion with XhoI and BspHI and subcloned into E-cad-pcDNA that had been digested in the identical manner. Finally, the open reading frame encoding the full-length version of E-cad⌬␤-cat was subcloned into pcDNA3.1(ϩ) (Invitrogen, Carlsbad, CA) using the XbaI and EcoRV sites present in both the pcDNA1.1/CAT and pcDNA3.1 (ϩ) vectors. Full-length wild-type Ecadherin was subcloned into pcDNA3.1(ϩ) in a similar manner. All constructs were verified by sequencing.
Production of G␣ 12 and ␤-Catenin Protein-Recombinant G␣ 12 protein was produced in Sf9 cells as previously described (22). His-tagged ␤-catenin, produced in Sf9 cells and purified as previously described (23), was a generous gift from Tim Fields (Duke University Medical Center).
In Vitro Protein Binding Studies-GST fusion proteins were produced in Escherichia coli strain BL21(DE3) and purified from cell lysates using glutathione-Sepharose 4B (Amersham Pharmacia Biotech). Bound GST fusion protein was washed with buffer containing 10 mM HEPES (pH 8.0), 1 mM dithiothreitol, and 150 mM NaCl and then stored in the same buffer.
Binding of G␣ 12 to GST-E-cadherin or GST-E-cadherin mutants was evaluated as described previously (17). Briefly, 350 ng of purified G␣ 12 was incubated in buffer containing 10 M GTP␥S for 2 h at 30°C, and the reaction was divided and then incubated with ϳ1 g of each glutathione Sepharose-bound GST fusion protein. Reactions were incubated with gentle agitation for 2 h at 4°C, followed by pelleting of glutathione-Sepharose and extensive washing. Bound G␣ 12 was eluted by heating for 10 min at 70°C in Laemmli sample buffer, subjected to SDS-PAGE, and subsequently analyzed by immunoblotting with 200 ng/ml G␣ 12 primary antibody followed by 200 ng/ml horseradish peroxidaseconjugated anti-rabbit secondary antibody. Binding of ␤-catenin (1 g per reaction) to GST fusion proteins was evaluated in the same manner, except that the nucleotide-loading step was omitted. ␤-Catenin primary antibody was used at a concentration of 250 ng/ml followed by horseradish peroxidase-conjugated anti-mouse secondary antibody at a concentration of 200 ng/ml.
Cell Culture-SW480 cells were grown in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum (FBS). L cells, stably expressing full-length E-cadherin or the corresponding E-cad⌬␤-cat, were selected in the same media supplemented with 1 mg/ml Geneticin (Invitrogen), and clonal populations were isolated and maintained in the same medium containing 500 g/ml Geneticin. K562 cells were maintained in RPMI 1640 medium (Life Technologies, Rockville, MD) supplemented with 10% FBS and 10 mM HEPES (pH 7.5). Transfections were performed using LipofectAMINE (Invitrogen) according to the manufacturer's protocol unless otherwise noted. For stable transfection of K562 cells, ϳ10 ϫ 10 6 cells were washed in RPMI 1640 medium to remove serum, combined with 30 g of DNA encoding either full-length E-cadherin or the corresponding E-cad⌬␤-cat, subjected to electroporation (340 V, 10 ms, 1 pulse, using a T820 square-wave electroporator (BTX, San Diego, CA)), and then plated in fresh medium. After 48 h, cells were seeded at varying dilutions into 96-well plates in the presence of 1 mg/ml Geneticin. After 2 weeks, plates with growth in Ͻ20% of wells were chosen, and wells were expanded and maintained in 500 g/ml Geneticin.
Immunofluorescence Microscopy-L cells stably expressing fulllength or mutant E-cadherin were grown on glass coverslips, washed with PBS, fixed in 4% paraformaldehyde, and incubated in blocking buffer (PBS containing 10% FBS and 0.2% saponin). Coverslips were incubated with 2.5-5 g/ml primary antibody in blocking buffer, followed by three washes in PBS containing 10% FBS and subsequent incubation with 7-14 g/ml fluorescein isothiocyanate-conjugated secondary antibody (Jackson ImmunoResearch, West Grove, PA) in blocking buffer. ␤-Catenin localization was observed using an LSM-410 laser scanning confocal microscope with a 63ϫ oil immersion objective (Carl Zeiss, Thornwood, NY). 12 and ␤-Catenin-A series of eight deletion mutants within the E-cadherin cytoplasmic domain, with each deleted region being defined by the primary sequence as proline-rich, serine-rich, acidic, basic, or mixed-charged ( Fig.  1), were constructed as fusions to GST. The fusion proteins were produced in a bacterial expression system, and the purified proteins evaluated in a cell-free system for their ability to interact with purified G␣ 12 and with purified ␤-catenin (Fig. 2). The wild-type E-cadherin cytoplasmic tail interacted with activated (GTP␥S-loaded) G␣ 12 and with ␤-catenin (Fig. 2C), as did six of the eight deletion mutants (data not shown). Deletion F, which lacks a highly charged region of 11 amino acids ( Fig.  2A), interacted poorly with G␣ 12 yet retained the ability to bind ␤-catenin (Fig. 2C). Based on these results, deletion F will subsequently be referred to as E-cad⌬G 12 ( Fig. 2A). Conversely, deletion D, which lacks a 19-amino acid stretch containing 5 acidic residues ( Fig. 2A), was completely unable to interact with ␤-catenin but maintained effective binding to G␣ 12 (Fig.  2C). Thus, deletion D will subsequently be referred to as E-cad⌬␤-cat ( Fig. 2A). Unexpectedly, deletion E, which lacks a serine-rich region (Fig. 1) that was previously implicated as essential for ␤-catenin binding (24,25), bound to both ␤-catenin and G␣ 12 (Fig. 2C); this finding will be discussed further below. The ability of each deletion mutant to bind at least one of the two proteins (G␣ 12 or ␤-catenin) diminishes the likelihood that these mutations produce misfolded proteins with a global loss of protein binding capacity. These data demonstrate that distinct regions of the cadherin cytoplasmic domain are required for interaction with G␣ 12 and ␤-catenin.

Identification of Regions of the Cadherin Cytoplasmic Domain That Interact with G␣
E-cad⌬␤-cat Is Unable to Mediate Biological Functions of Cadherin That Require Interaction with ␤-Catenin-Expression of the cytoplasmic domain of cadherins has been shown to attenuate ␤-catenin-mediated transcriptional activation, presumably by sequestering ␤-catenin and preventing its interaction with target transcription factors such as TCF or lymphoid enhancer factor (26). To determine whether the lack of interaction between ␤-catenin and E-cad⌬␤-cat translates to functional changes within a cellular context, we fused the E-cad⌬␤cat construct to GFP, expressed this fusion protein in SW480 cells along with the TOPFLASH reporter plasmid, and measured ␤-catenin-mediated activation of the reporter gene. SW480 cells express a truncated form of the adenomatous polyposis coli protein that is unable to promote ␤-catenin degradation; this results in high levels of ␤-catenin and a robust ␤-catenin-mediated transactivation response in these cells. Consistent with previous results (17,26), we found that expression of a fusion between GFP and the wild-type E-cadherin tail significantly attenuated ␤-catenin-mediated transcriptional activation (Fig. 3A). However, the GFP-E-cad⌬␤-cat fusion protein was without effect (Fig. 3A), indicating that deletion of the acidic region of the E-cadherin tail abolishes its ability to bind ␤-catenin in a cellular context.
Formation of ␤-catenin/cadherin complexes has been demonstrated to stabilize ␤-catenin in cells by preventing its interaction with components required for its degradation (27)(28)(29). Thus, we examined the ability of both wild-type E-cadherin and E-cad⌬␤-cat to protect ␤-catenin from degradation. Full-length E-cadherin or full-length E-cad⌬␤-cat was stably expressed in K562 cells, which lack endogenous cadherins (30), and lysates from these cells were analyzed by immunoblotting for E-cadherin and ␤-catenin. Although mock transfected K562 cells had very low levels of ␤-catenin (Fig. 3B), cells stably expressing wild-type E-cadherin exhibited a dramatic stabilization of endogenous ␤-catenin (Fig. 3B). In contrast, essentially no stabilization of ␤-catenin was observed in cells stably expressing full-length E-cad⌬␤-cat (Fig. 3B).
In most cell types expressing cadherins, ␤-catenin is localized primarily at the cell periphery, whereas, in cells lacking cadherins, ␤-catenin is generally cytoplasmic or nuclear (31,32). We reasoned that a form of cadherin unable to bind ␤-catenin, such as E-cad⌬␤-cat, should fail to promote peripheral localization of ␤-catenin. To test this hypothesis, we first examined the localization of ␤-catenin in L cells, which, like K562 cells, lack endogenous cadherins (30). L cells are adherent and were used in place of K562 suspension cells to better facilitate growth on coverslips. Indirect immunofluorescence performed on these cells revealed a primarily cytoplasmic and nuclear staining pattern for endogenous ␤-catenin (Fig. 4A). As expected, in L cells stably expressing wild-type E-cadherin, ␤-catenin was now found predominantly at the cell periphery (Fig. 4B). However, no relocalization of ␤-catenin was observed in L cells stably expressing E-cad⌬␤-cat (Fig. 4C); the staining pattern instead was similar to that observed in control L cells. Preincubation of primary antibody with purified ␤-catenin ablated the immunofluorescence signal, whereas preincubation with the irrelevant protein farnesyltransferase did not (data not shown).
Having demonstrated that E-cad⌬␤-cat is unable to functionally interact with ␤-catenin in vitro or in cells, we next evaluated the impact of loss of ␤-catenin binding on cadherin function. A primary biological function of cadherins is to mediate cell-cell adhesion through homophilic interaction between cadherins on adjacent cells. Although the extracellular domain of cadherins is directly involved in adhesion, cadherins lacking the cytoplasmic domain are unable to promote cell-cell interaction (33). To evaluate the importance of the region of Ecadherin we defined as essential for ␤-catenin binding for cell adhesion, we utilized K562 cells, which do not normally exhibit cell-cell adhesion (30). Stable expression of wild-type E-cadherin in these cells resulted in the formation of large aggregates of cells (Fig. 5B). However, cells stably expressing equivalent levels (see Fig. 5D) of E-cad⌬␤-cat showed no such aggregation (Fig. 5C) and were indistinguishable from control K562 cells (see Fig. 5A). Taken together, these data clearly demonstrate that the 19-amino acid region deleted in E-cad⌬␤cat is essential for both ␤-catenin binding and for cadherin function. The absence of this domain abolishes the ability of the cadherin tail to inhibit ␤-catenin-mediated transactivation, eliminates cadherin's ability to localize and stabilize ␤-catenin, and disrupts cadherin-mediated cell adhesion.
E-cad⌬G 12 Is Impaired in G␣ 12 -mediated Release of ␤-Catenin-The inhibition of ␤-catenin-mediated transactivation resulting from expression of the GFP-E-cadherin tail (see Fig. 3A) can be attenuated by co-expression of mutationally activated G␣ 12 ; this effect is thought to be due to G␣ 12 disrupting the interaction between ␤-catenin and the E-cadherin tail and thus freeing ␤-catenin to function as a transcriptional co-activator (17). Hence, we hypothesized that the G␣ 12 binding region of cadherin identified by deletion analysis (see Fig. 2) should be required for G␣ 12 to elicit dissociation of ␤-catenin from the E-cadherin tail. To test this hypothesis, the E-cad⌬G 12 cytoplasmic tail was fused to GFP, and this construct was transfected into SW480 cells along with the TOPFLASH reporter plasmid and different mutationally activated G proteins. Consistent with our previous study (17), expression of activated G␣ 12 resulted in a significant increase in TOPFLASH reporter activity in cells expressing a GFP-wild-type E-cadherin tail fusion protein (Fig. 6). However, activated G␣ 12 was much less  12 and with ␤-catenin. GST-E-Cad fusion proteins were incubated with purified G␣ 12 (loaded with GTP␥S) or ␤-catenin protein (see "Experimental Procedures"). Bound proteins were separated by SDS-PAGE and identified by immunoblot analysis using anti-G␣ 12 or anti-␤-catenin antibody as indicated. Data are from a single experiment that is representative of six independent experiments. IB, immunoblot.
FIG. 3. E-cad⌬␤-cat does not inhibit ␤-catenin-mediated transactivation or protect ␤-catenin from degradation. A, effect of deleting the ␤-catenin interacting region on the ability of E-cadherin fusion proteins to suppress ␤-catenin-mediated transcriptional activation. SW480 cells were transfected with TOPFLASH, pRL-TK, and the indicated E-cadherin tail constructs fused to GFP (see "Experimental Procedures"). Cell lysates were analyzed for firefly luciferase (TOP-FLASH) activity and Renilla luciferase (pRL-TK) activity; the latter was used as an internal standard for transfection efficiency. TOP-FLASH activity was normalized for pRL-TK activity in each sample and is shown here as a percentage of the TOPFLASH activity in cells expressing GFP alone. Data represent means Ϯ S.E. of the mean of four independent experiments, each performed in duplicate. Inset, anti-GFP immunoblot from SW480 cell lysates. B, stabilization of endogenous ␤-catenin in K562 cells expressing full-length E-cadherin and the variant lacking the putative ␤-catenin binding region (E-cad⌬␤-cat). Stable cell lines were obtained as described under "Experimental Procedures." Cells were lysed, and equivalent amounts of lysate from each were separated by SDS-PAGE and subjected to immunoblot analysis using either anti-E-cadherin or anti-␤-catenin antibody as indicated. Data are from a single experiment that is representative of three independent experiments. IB, immunoblot. effective in stimulating ␤-catenin-mediated transactivation in cells expressing GFP-E-cad⌬G 12 (Fig. 6). Expression of activated G␣ z did not stimulate ␤-catenin-mediated reporter activation in the presence of either form of cadherin (Fig. 6). Our findings indicate that G␣ 12 indeed requires the 11-amino acid region deleted in the E-cad⌬G 12 construct to effectively trigger ␤-catenin dissociation from cadherin. DISCUSSION Inappropriate accumulation of ␤-catenin is believed to be a key event in the development of most colorectal tumors (34), as well as a variety of other human cancers (35). Additionally, the majority of epithelial tumors lose E-cadherin function as they progress to a malignant phenotype (36). Our recent finding, that G 12 subfamily proteins interact with the cytoplasmic domain of cadherins in a fashion that stimulates the release of ␤-catenin, suggests that G 12 subfamily members may be involved in ␤-catenin-mediated tumorigenesis (17). In the present study, we have demonstrated that distinct domains of the E-cadherin tail are required for interaction with G␣ 12 and with ␤-catenin. The mutant E-cadherin molecules generated in this FIG. 4. E-cad⌬␤-cat does not bind ␤-catenin at the cell periphery. A-C, subcellular localization of endogenous ␤-catenin, in either control L cells or L cells stably expressing the indicated fulllength E-cadherin construct, was analyzed by indirect immunofluorescence for ␤-catenin (see "Experimental Procedures"). ␤-Catenin staining was visualized using laser scanning confocal microscopy. D, anti-E-cadherin immunoblot of lysates from stable L cell lines indicating level of E-cadherin or E-cad⌬␤-cat expression. Data shown is from a single experiment that is representative of four separate experiments. The scale bar represents 10 m.
FIG. 5. E-cad⌬␤-cat is unable to promote cell-cell adhesion. K562 cells were stably transfected with either wildtype E-cadherin or E-cad⌬␤-cat, and clonal populations were isolated as described under "Experimental Procedures." A-C, cells were visualized under phase-contrast microscopy using a light microscope with 10ϫ objective, and images were captured using a digital camera. D, cells were analyzed for expression of E-cadherin or E-cad⌬␤-cat by immunoblotting. Clonal lines that expressed approximately equal amounts of the wildtype and mutant E-cadherin were selected for analysis. Data shown is from a single experiment that is representative of three separate experiments. study should be of great utility for dissecting the roles of G␣ 12 and ␤-catenin in cadherin-mediated functions.
Since the discovery that the cytoplasmic domain of E-cadherin is required for cell-cell adhesion (37) and the identification of ␤-catenin as a protein associated with the cadherin cytoplasmic domain (38), numerous studies have analyzed the cadherin cytoplasmic domain to identify the region to which ␤-catenin binds. Early studies demonstrated that deletion of the 72 C-terminal residues of E-cadherin abolished interaction with ␤-catenin (37)(38)(39)(40). Stappert and Kemler (24) later demonstrated that the mutation of eight serines to alanines between residues 838 and 853 of E-cadherin abolished interaction with ␤-catenin, as determined by co-immunoprecipitation. Conversely, a study by Jou et al. (41) presented data indicating that deletion of a 25-residue acidic region, corresponding to amino acids 815-839 of human E-cadherin, eliminated the ability of ␤-catenin to interact with E-cadherin in a yeast two-hybrid system. Near the completion of our study, the crystal structure of the murine ␤-catenin/E-cadherin complex was reported (42). The crystal structure indicates that E-cadherin binds to ␤-catenin as an extended polypeptide with as many as 100 residues of the E-cadherin tail interacting with ␤-catenin (42). However, the authors concluded that 18 residues of the cadherin cytoplasmic tail form a "core" binding region that is essential for interaction with ␤-catenin, whereas the other residues involved in the extended interface serve to modulate the affinity of the interaction between the two proteins. The core domain identified in the crystal structure corresponds to amino acids 821-838 of human E-cadherin, which is nearly identical to the 19 residue acidic region (amino acids 819 -837) we identified as essential for interaction with ␤-catenin in the current study. Thus, our results support the conclusion from the structural study that the core ␤-catenin-binding region of cadherin is essential for the biologically functional interaction between cadherin and ␤-catenin.
The ␤-catenin/E-cadherin crystal structure was solved with both unphosphorylated and phosphorylated E-cadherin. The authors of the study found that the serine-rich region of the E-cadherin tail, corresponding to residues 838 -853 of human E-cadherin, interacted with ␤-catenin only when phosphorylated. This finding provides an explanation for earlier reports demonstrating that phosphorylation of serine residues in this region of cadherin increases the affinity of its interaction with ␤-catenin (25,43). The structural data suggest that, although the serine-rich region is involved in modulating the affinity of the interaction between the two proteins, it is not essential for the interaction to occur. Our finding that E-cadherin missing this precise serine-rich region (deletion E, see Fig. 1) still binds effectively to ␤-catenin (Fig. 2C) supports this conclusion.
Another very recent study reported identification of a minimal region of the Drosophila E-cadherin cytoplasmic tail that was able to interact with ␤-catenin (44). This analysis found that a polypeptide encoding as few as 23 residues of the Drosophila E-cadherin tail could bind to ␤-catenin in a yeast twohybrid system; this 23-residue region corresponds to amino acids 828 -850 of the human E-cadherin tail. This region overlaps substantially with the domains identified as essential for ␤-catenin binding in both the present study and the structural study noted above, although only about half the residues in this region are conserved from fly to mammals.
The present study validates the G␣ 12 /cadherin interaction as a key event leading to destabilization of the cadherin/␤-catenin complex in cells. Identification of residues 854 -864 of human E-cadherin as critical for its interaction with G␣ 12 allowed testing of the hypothesis that this interaction leads to release of ␤-catenin from the E-cadherin tail (see Fig. 6). In the ␤-catenin/ E-cadherin crystal structure, these residues of E-cadherin are seen to form part of a cap that protects hydrophobic residues of the first armadillo repeat region of ␤-catenin from solvent exposure (42). Earlier studies have reported that deletion of the cap region reduces interaction of E-cadherin with ␤-catenin (45,24). However, because robust binding to ␤-catenin was still observed when the N-or C-terminal half of the E-cadherin cap region was deleted (see "Results"), it is unlikely that disruption of the interaction between the hydrophobic cap of cadherin and ␤-catenin could account for the ability of G␣ 12 to stimulate ␤-catenin release from cadherin. Rather, we would speculate that binding of G␣ 12 to cadherin disrupts additional regions involved in formation of the ␤-catenin/cadherin complex, which likely include interactions involving the acidic core region of cadherin that is essential for ␤-catenin binding.
An intriguing study that may shed light on additional mechanisms by which G 12 subfamily members can affect ␤-catenin signaling was recently reported by Fujino et al. (46). They demonstrated that stimulation of FP B prostanoid receptors, which are G protein-coupled receptors thought to signal at least in part through G 12 and G 13 proteins (47), leads to activation of a ␤-catenin/TCF signaling pathway, thereby providing additional evidence of a link between heterotrimeric G proteins and ␤-catenin signaling.
The specific mechanism by which G␣ 12 stimulates ␤-catenin release is not yet apparent, but our data clearly demonstrate that these two proteins bind to distinct regions of the cadherin tail. Therefore, a simple competition model cannot adequately explain G␣ 12 -stimulated ␤-catenin release, and the possibility exists that additional proteins may be involved in this process. We are currently attempting to identify such proteins to investigate whether G 12 subfamily members may serve to recruit FIG. 6. G 12 is ineffective in eliciting ␤-catenin release from E-cad⌬G 12 . ␤-Catenin release from GFP fusions of the E-cadherin or E-cad⌬G 12 tail was evaluated by measuring ␤-catenin-mediated reporter activation. SW480 cells were transfected with TOPFLASH, pRL-TK, and the indicated combinations of GFP-E-cadherin tail fusions and mutationally activated (QL) G protein ␣ subunits. TOPFLASH activity from cell lysates was normalized for pRL-TK activity in each sample and is shown here as percent increase over cells expressing the corresponding GFP-E-cadherin fusion alone. Data represent means Ϯ S.E. of the mean of four independent experiments, each performed in duplicate. These errors for basal luciferase activity of cells expressing only GFP-E-cadherin or GFP-E-cad⌬G 12 were 11.6% and 5.4%, respectively. them to the cadherin/␤-catenin complex. These and other studies noted above should result in elucidation of the complex interplay between the G 12 and ␤-catenin signaling pathways.