Modification of the E-cadherin-Catenin Complex in Mitotic Madin-Darby Canine Kidney Epithelial Cells*

One of the hallmarks of polarized epithelial cells undergoing mitosis is their rounded morphology. This phenotype correlates with a reduced cell-substratum adhesion, apparently caused by a modulation of integrin function. However, it is still unclear whether the cadherin-mediated cell-cell adhesion is affected as well. To address this question, the cadherin complex was analyzed in different cell cycle stages of Madin-Darby canine kidney cells. By immunofluorescence, mitotic Madin-Darby canine kidney cells showed an increased staining of E-cadherin and the catenins (α-catenin, β-catenin, plakoglobin, p120ctn) in the cytosol, suggesting a reorganization of the cadherin-catenin complex during mitosis. Biochemical analysis revealed that the overall amount of these components, as well as the proportion of the complex associated with the actin cytoskeleton, remained unchanged in mitotic cells. However, we found evidence for an internalization of E-cadherin during mitosis. In addition, the cadherin-catenin complex was analyzed for mitosis-specific changes in phosphorylation. We report a decrease in the tyrosine phosphorylation of β-catenin, plakoglobin, and p120ctn during mitosis. Moreover, we observed a mitosis-specific Ser/Thr-phosphorylation of p120ctn, as detected by the MPM-2 antibody. Hence, the cadherin/catenin complex is a target for different posttranslational modifications during mitosis, which may also have a profound impact on cadherin-mediated cell-cell adhesion.

Epithelial cells in culture or in growing tissues, such as the intestinal epithelium or the epidermal basal layer, are confined within the epithelium by extensive adhesion, both to surrounding cells and to the underlying extracellular matrix. During mitosis, epithelial cells undergo substantial changes in the distribution of all three cytoskeletal filament systems, together with a dramatic change in morphology. They generally round up, and when grown in culture, they often bulge above the monolayer. This phenotype may be at least partially caused by the loss of cell-substratum contacts during mitosis, as has been observed in epithelial cell cultures, such as Madin-Darby canine kidney (MDCK) 1 cells (1). In fact, for rat fibroblasts, it was shown that the rounded phenotype correlates with a modulation of integrin functions (2).
Epithelial cells are linked by a variety of cell-cell junctions, including tight junctions, adherens junctions (AJs), desmosomal junctions, and gap junctions. Little is known about the fate of these junctions during mitosis, but during cell division, some of these junctions may be modified as well. In this respect, components of the AJs are of special interest for several reasons: 1) AJs are associated with circumferential microfilament bundles, which are important components of the actin cytoskeleton in epithelial cells. Via AJs, the circumferential microfilament bundles of adjacent cells are interconnected, maintaining the integrity of the epithelial sheet. 2) AJs are main targets for posttranslational modifications, such as tyrosine phosphorylation, which induce the destruction of this region (3)(4)(5). 3) Specific proto-oncogenic tyrosine kinases, such as c-Src, c-Fyn, and the epidermal growth factor receptor (EGFR), and various tyrosine phosphatases, are concentrated in this region. Some of these have been found to be associated with components of AJs (6 -9).
␤-Catenin was found to be associated with tyrosine receptor kinases, such as the EGFR and c-ErbB-2, as well as with phosphatases of different families, such as protein tyrosine phosphatase and leukocyte antigen-related tyrosine phosphatase (6 -9, 17, 18). ␤-Catenin, plakoglobin, and p120 ctn have been shown to become tyrosine-phosphorylated in v-Src-transformed cells; this may be a key event for the observed disruption of AJs in these cells (4). Indirect evidence for a modulation of the E-cadherin-catenin complex by tyrosine phosphorylation during mitosis comes from results in mouse fibroblasts transfected with a cDNA encoding an E-cadherin-␣-catenin fusion protein (19). Whereas wild-type cells showed a rounded morphology and a reduced cell-cell interaction during mitosis, cells expressing the fusion protein remained morphologically indistinguishable from the rather flat interphase cells and retained cell-cell contacts. Because the ␤-catenin and plakoglobin binding sites had been deleted in the fusion protein, modifications of these proteins during mitosis may occur that play an essential role in the reconstruction of the actin-based cytoskeleton.
In the present study, we biochemically analyzed the E-cadherin-catenin complex in mitotic MDCK cells. Immunofluorescence revealed an enrichment of cytosolic E-cadherin-catenin-* 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 U.S.C. Section 1734 solely to indicate this fact.
‡ Supported by the Deutsche Forschungsgemeinschaft. § Supported by the Alexander von Humboldt/Max-Planck Gesellschaft award given to James Nelson and R. K. To whom correspondence should be addressed. Tel.: 49-761-5108-475; Fax: 49-761-5108-474; Email: stappert@immunbio.mpg.de. 1 The abbreviations used are: MDCK, Madin-Darby canine kidney; AJ, adherens junction; EGFR, epidermal growth factor receptor; GST, gluthatione S-transferase; APC, adenomatous polyposis coli; Pipes, 1,4-piperazinediethanesulfonic acid; PAGE, polyacrylamide gel electrophoresis. complex in mitotic cells. However, neither changes in the overall amount of the complex nor changes in its composition were observed. Biotinylation experiments showed a reduction of cell surface-localized E-cadherin during mitosis, suggesting that the E-cadherin-catenin complex is partially internalized. Analysis for posttranslational modifications revealed a decrease of tyrosine phosphorylation for ␤-catenin, plakoglobin, and p120 ctn during mitosis, as well as a mitosis-specific Ser/Thr phosphorylation of p120 ctn . To our knowledge, this is the first biochemical analysis of the E-cadherin-catenin complex during mitosis. It demonstrates that components of the cell-cell adhesion complex are both relocalized and modified during mitosis.

Cell Lines and Tissue Culture
MDCK cells were cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum. To inhibit tyrosine phosphatases, 1 mM pervanadate was added directly to the medium 30 min before cell lysis.

Antibodies and Reagents
The mouse monoclonal antibody 3G8 against E-cadherin was kindly provided by Dr. J. Nelson (Stanford University). Mouse monoclonal antibodies against ␣-catenin, ␤-catenin, plakoglobin, p120 ctn (pp120), and phosphotyrosine (PY20) were obtained from Transduction Laboratories (Lexington, KY); monoclonal antibodies against ␤-actin were from Sigma. For immunoprecipitation and immunoblotting of adenomatous polyposis coli (APC), two APC antibodies (Ab-1 and Ab-5) were used (Oncogene Science, Manhasset, NY). The monoclonal antibody MPM-2 was kindly provided by the group of Dr. Marc Kirschner (Harvard Medical School). Fluorescein-conjugated secondary antibodies Lmimosine and nocodazole were purchased from Sigma.

Cell Synchronization
Nocodazole-MDCK cells were seeded at a density of 4 ϫ 10 5 cells/ 10-cm dish. Nocodazole (5 mg/ml stock in 100% Me 2 SO) was added directly to the medium at 0.33 M (0.1% Me 2 SO); after the cells were cultivated for an additional 11 h, they were lysed in CSK buffer (150 mM NaCl; 10 mM Pipes, pH 6.8; 3 mM MgCl 2 ; 300 mM sucrose; 0.5% Triton X-100; 10 g/ml phenylmethylsulfonyl fluoride; Complete TM (Boehringer Mannheim)). As a control, cultures were treated with 0.1% Me 2 SO alone. For cell cycle analysis, synchronized cells were analyzed by flow cytometry on a Becton Dickinson FACScan ® as described (20). To evaluate the percentage of mitotic cells, cells grown on coverslips were stained with Hoechst 33342 (Sigma), and the number of cells with condensed chromatin were quantified by counting under the fluorescence microscope.
L-Mimosine-Mimosine was added directly to the medium for 18 h, at a final concentration of 1 mM, to arrest the cells in early S phase. For biochemical analysis, either cells were directly lysed in CSK-buffer (G 1 phase) or mimosine was washed out and cells were cultured for additional 3 h (S phase), 6 h (S/G 2 phase) or 9.5 h (G 2 /M phase) in Dulbecco's modified Eagle's medium before lysis.

Immunofluorescence
MDCK cells were grown on collagen-coated coverslips in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum. The cells were washed with phosphate-buffered saline and fixed in 3% paraformaldehyde for 20 min at room temperature. Free aldehyde groups were blocked with 1 M glycine/phosphate-buffered saline, pH 8.5, for 5 min. After washing, cells were permeabilized with 0.5% Triton X-100 for 5 min. Alternatively, cells were fixed with 100% methanol for 4 min at Ϫ20°C. Fixed cells were incubated with primary antibodies at 2 g/ml for 1 h at 37°C and fluorescein-conjugated secondary antibodies (Sigma) for 1 h at 37°C. Cells were washed, and the nuclei were stained with Hoechst 33342 (1 g/ml) for 5 min at room temperature, washed, and mounted (50% glycerol; 50% phosphate-buffered saline; 100 g/ml 1,4-diazabicyclo-[2.2.2.]octone).
Digital images were taken with a computer controlled digital C4880 camera (Hamamatsu, Hamamatsu City, Japan) on an Axioskop microscope (Zeiss, Jena). Camera and microscope were controlled by the computer program Openlab (Improvision, Coventry, UK).

Immunoprecipitation and Immunoblotting
MDCK cells were lysed in CSK buffer, and the detergent-soluble and detergent-insoluble fractions were separated by centrifugation at 12,000 ϫ g for 10 min. The detergent-insoluble fraction was solubilized by heating to 100°C for 5 min in immunoprecipitation buffer (1% SDS; 10 mM Tris-HCl, pH 7.5; 2 mM EDTA) and further diluted to 0.1% SDS with CSK buffer. For immunoprecipitation, the supernatant and the solubilized cytoskeleton fraction were precleared with 10% protein A-Sepharose beads (Amersham Pharmacia Biotech, Freiburg, Germany) for 1 h at 4°C. Approximately 2 g of specific antibodies were coupled to protein A-Sepharose beads. The antigen-antibody complex was washed with CSK buffer, boiled in SDS sample buffer, and analyzed by PAGE as described (21). For immunoblotting of APC, cell extracts were prepared according to the manufacturer's description and separated by 5% SDS-PAGE. For metabolic labeling of synchronized cultures, cells were starved for 30 min and incubated with 100 Ci/ml [ 35 S]methionine/[ 35 S]cysteine (3000 Ci/mmol; Amersham Pharmacia Biotech, Braunschweig, Germany) for 2 h before lysis. For quantification, radioactive gels were quantified in a BAS 1000 bioimaging analyzer (Fuji). Western blots were quantified by Lumi-Imager (Boehringer Mannheim) or by analyzing the film using the NIH Image program version 1.59. Only differences in band intensities of a factor of at least 2 were considered significant and are described in the text.

Biotinylation of Cell Surface Proteins
After 24 h of growth in the presence or absence of nocodazole or L-mimosine, the cells were washed with Ringer's buffer (10 mM Hepes, pH 7.4; 154 mM NaCl; 7.2 mM KCl; 1.8 mM CaCl 2 ) and incubated with 1 mg/ml sulfo-NHS-biotin (20 mg/ml stock in Me 2 SO; Pierce) in Ringer's buffer for 30 min at 4°C. Subsequently, cells were washed with Trissaline (10 mM Tris-HCl, pH 7.4; 120 mM NaCl) and lysed in CSK buffer. The detergent-soluble and detergent-insoluble fractions were separated by centrifugation, and E-cadherin was precipitated from both fractions with the monoclonal antibody 3G8. Biotinylated E-cadherin was detected with peroxidase-conjugated streptavidin (Vectastain ABC peroxidase standard kit; Vectastain Laboratories, Burlingame, CA) after Western blotting.

GST-E-cadherin Affinity Precipitation
A recombinant gluthatione S-transferase (GST)-tagged cytoplasmic domain of mouse E-cadherin was expressed in E. coli and affinitypurified on glutathione S-transferase-agarose beads. Approximately 10 g of GST-E-cadherin were used for the affinity precipitation of ␤-catenin as described (22).

Subcellular Localization of the E-cadherin-Catenin Complex in Mitotic MDCK Cells-
The localization of the E-cadherincatenin complex was compared in mitotic versus interphase cells by indirect immunofluorescence. Subconfluent MDCK cells were stained with the monoclonal antibody 3G8 specific for E-cadherin or with monoclonal antibodies specific for ␣-catenin, ␤-catenin, plakoglobin, p120 ctn , and ␤-actin ( Fig. 1, A-F); nuclei were costained with Hoechst 33342 (Fig. 1, AЈ-FЈ). Optical sections were taken at 0.2-m intervals from the top to the bottom of the cell layer, and out-of-focus information was removed by a deconvolution algorithm. Typically, nonmitotic MDCK cells formed a flattened epithelial sheet, with E-cadherin predominantly localized at cell-cell boundaries, whereas the cytoplasm was apparently devoid of E-cadherin (Fig. 1, A and AЈ). Mitotic cells, instead, usually had a rounded phenotype, bulging above the monolayer, and differed markedly in the localization of the E-cadherin-catenin complex. In these cells, E-cadherin was not only restricted to sites of cell-cell contact, forming a peripheral ring-like structure, but was also substantially increased in the cytosol. Similar distribution patterns were also observed for the catenins, ␣-catenin, ␤-catenin, plakoglobin, and p120 ctn , in various stages of mitosis ( Fig. 1, B-E and BЈ-EЈ, respectively). The observed cytoplasmic staining is not based on an increase in the local concentration of the proteins due to a reduced cell size, because the same pattern was also observed for mitotic MDCK cells that were not com-pletely rounded but had the same size as interphase cells (Fig.  1E). This indicates that in mitotic cells, the subcellular localization of the cadherin-catenin-complex changes either by internalization or by cytoplasmic storage of newly synthesized complex.
In agreement with previous results, the cortical actin belt was found to persist during mitosis (Fig. 1, F and FЈ), indicating that mitotic MDCK cells apparently retain AJs (23,24).
Cell Synchronization with Nocodazole or Mimosine-For a biochemical analysis of the cadherin-catenin complex during mitosis, MDCK cells were synchronized by nocodazole. After 11 h of incubation with 0.33 M nocodazole, about 95% of the cells showed a 4c DNA content by FACScan analysis (Fig. 2, A  and B, left panels). For further quantification of the number of mitotic cells, cells grown on coverslips were stained with Hoechst 33342 at each time point and analyzed under the microscope. After nocodazole treatment, nearly all cells were completely rounded and found to be in mitosis (Fig. 2, A and B, right panels). Nocodazole destabilizes microtubules and inhibits the formation of the mitotic spindle, causing an arrest in mitosis. Although no connections between the microtubule network and the E-cadherin cell adhesion complex have been described so far, some components of the complex, e.g. ␣-catenin, ␤-catenin, and plakoglobin, have been found to be associated with the tumor suppressor protein APC. APC is thought to be involved in the degradation of ␤-catenin and is also known to bind to microtubules (25)(26)(27). Therefore, to exclude unspecific effects of nocodazole, we also used mimosine for cell synchronization. Mimosine is a natural plant drug that arrests cells at the G 1 /S phase border in a concentration-dependent manner without affecting cytoskeletal structures (20,28,29). The inhibition of entry into S phase by mimosine treatment is reversed when the drug is removed, and the cells continue to progress into S phase. Hence, no drug was present when mitotic cells were analyzed.
For the mimosine experiments, MDCK cells were arrested at the G 1 /S phase interface in the presence of 1 mM mimosine for 18 h. FACScan analysis was performed to evaluate cell synchronization as described under "Materials and Methods." About 85% of the cells were at the G 1 /S phase transition (Fig.  2C, left panel). Upon release into complete medium without mimosine, the cells progressed into S phase. After 3 h, an average of about 90% of the cells had started DNA synthesis (Fig. 2D); 6 h after mimosine was removed, most of the cells had already replicated their DNA (considered as G 2 phase) (Fig. 2E), and at 9.5 h, about 82% cells showed a 4c DNA content (Fig. 2F). Whereas in nonsynchronized cultures only about 5% of the cells were in mitosis ( Fig. 2A, right panel), about 50% of the cells released from the mimosine block entered mitosis in a synchronized manner, as analyzed by microscopy (Fig. 2E, right panel).
Because nocodazole leads to a higher proportion of mitotic cells, its effects were more distinct. Thus, only the nocodazole experiments are presented here. However, we have observed comparable effects with both agents in all experiments described.
The Overall Amounts of E-cadherin and the Catenins Are Unchanged in Mitotic Cells-The immunofluorescence data suggested an overall increase in the cytosolic E-cadherin-catenin complex during mitosis. For a biochemical analysis, cells were arrested in mitosis by nocodazole treatment or at the G 1 /S phase border by mimosine. Both populations were compared with nonsynchronized cells and, as a control, to cells treated only with 0.1% Me 2 SO, used as solvent for nocodazole (Fig. 3). Cell extracts were separated into the detergent-soluble and detergent-insoluble cytoskeletal fraction and analyzed by Western blotting. No significant change in the overall protein amount of the E-cadherin-catenin complex could be observed in mitotic cells, in the detergent-soluble (Fig. 3A) and detergentinsoluble fraction (data not shown). This indicates that the intense cytosolic fluorescence staining in mitotic cells is not simply a result of an increased protein amount or the result of a translocation of the plasma membrane-localized, actin-associated complex to a cytosolic, non-actin-associated pool. In agreement to previous results, we found no increase in the overall actin amount in mitotic cells (30).
To determine the stoichiometry of the individual components of the cadherin-catenin complex during mitosis, MDCK cells were metabolically labeled with [ 35 S]methionine/[ 35 S]cysteine and arrested in the different cell cycle stages as described above. The complex was precipitated either by the 3G8 antibody or antibodies specific for ␤-catenin (Fig. 3B). No significant differences in the protein stoichiometry of the E-cadherincatenin complex were observed in a particular cell cycle stage. Additionally, no difference in the amount of E-cadherin-associated p120 ctn was observed during mitosis in E-cadherin immunoprecipitates (Fig. 3C).
E-cadherin Remains Mainly Localized on the Cell Surface of Mitotic Cells but the Amount Is Decreased-The biochemical analysis indicated that the overall protein amount of E-cadherin is not changed during mitosis, but this does not exclude a change in the subcellular localization of the detergent-soluble (plasma membrane-associated and cytosolic) E-cadherin pool, e.g. by internalization. To determine the amount of cell surfacelocalized E-cadherin, cell surface proteins were biotinylated. E-cadherin was precipitated from cell extracts with the 3G8 antibody and visualized for biotinylated E-cadherin in Western blots by peroxidase-conjugated streptavidin. The amount of the detergent-soluble cell surface-localized E-cadherin was significantly (by a factor of 2.0) reduced in mitotic cells compared with non-mitotic cells or cells arrested in early S phase (Fig. 4). The smaller bands recognized by streptavidin-peroxidase may correspond to proteolytic degradation products of E-cadherin. Biotinylated E-cadherin was never observed in the detergentinsoluble protein fraction (not shown). We conclude that a small fraction of the detergent-soluble E-cadherin is internalized during mitosis.
The ␤-Catenin-APC Complex during Mitosis-In addition to its function in cell adhesion, ␤-catenin is also known to participate in signal transduction to the nucleus (for a review, see Ref. 31). A major requirement for the signaling competence of ␤-catenin is the existence of a cytoplasmic pool of ␤-catenin that is not associated with E-cadherin or any other protein that could prevent association with a member of the lymphocyte enhancing factor/T-cell factor family.
To analyze differences in the amount of the free, non-Ecadherin-associated pool of ␤-catenin, cell extracts from synchronized cells were subjected to in vitro binding assays with a recombinant E-cadherin-GST fusion protein. Cell extracts derived from nonsynchronized and synchronized cells were incubated with this recombinant GST-E-cadherin fusion protein.
The formed protein complex was precipitated by glutathione S-transferase beads and analyzed by Western blotting with ␤-catenin-specific antibodies. Because ␤-catenin that is already associated with cellular E-cadherin is inaccessible to bind to the recombinant GST-E-cadherin, only the free, non-E-cadherin-associated pool of ␤-catenin is detected (32). No difference in the amount of cytoplasmic ␤-catenin was observed in mitotic cells versus nonsynchronized cells (Fig. 5, A and B).
A molecule thought to regulate the cytoplasmic pool of ␤-catenin by participating in ␤-catenin degradation is the APC protein (for a review, see Ref. 33). Previous results have shown that APC is hyperphosphorylated in mitotic human osteosarcoma cells, resulting in a molecular weight shifted band when separated by SDS-PAGE (34). We observed a similar molecular weight shift for APC in mitotic MDCK cells, suggesting that in (C, right panel); 9.5 h after release of the mimosine block, 82% of the cells showed a 4c DNA content, whereas only 50% were found in mitosis (F, right panel). Fluorescence is shown in arbitrary units.

FIG. 2. FACScan analysis of MDCK cells after mimosine or nocodazole.
A, nonsynchronized cells; B, cells arrested in the mitotic phase by nocodazole; C, cells arrested in the early S phase by mimosine; D, E, and F, cells released from the mimosine block and analyzed 3, 6, and 9.5 h later, respectively. About 95% of the cells showed a 4c DNA content after synchronization with nocodazole. Nearly the same proportion of cells were found to be in mitosis, as judged from condensed chromosomes seen with Hoechst 33342 (B, right panel). In contrast, just 5% of the cells in a nonsynchronized MDCK culture were in mitosis (A, right panel). No mitotic cells were found in mimosine-arrested cultures these cells, APC is also hyperphosphorylated during mitosis (Fig. 5D). In contrast to previous reports describing an enrichment of ␤-catenin on hyperphosphorylated APC, no such in-crease was detectable in mitotic cells (Fig. 5C) (35).
Decreased Tyrosine Phosphorylation of ␤-Catenin, Plakoglobin, and p120 ctn during Mitosis-Cadherins and some catenins are targets for different tyrosine kinases, but the precise role of this tyrosine phosphorylation is largely unknown. It has been reported that tyrosine phosphorylation of ␤-catenin in cells expressing v-Src correlates with a decrease in cadherin-mediated cell-cell adhesion (4,5). In addition, ␤-catenin was found to be associated with the EGFR, the activation of which correlates with a partial dissociation of the complex (6,36).
To analyze changes in tyrosine phosphorylation of the individual components of the cadherin-catenin complex during mitosis, MDCK cells arrested in different cell cycle stages were lysed either directly after cell synchronization or preincubated with 1 mM pervanadate to inhibit tyrosine phosphatases before cell lysis. Immunoprecipitations were made with the anti-Ecadherin 3G8 antibody and analyzed by SDS-PAGE and Western blotting with anti-phosphotyrosine-antibodies (Fig. 6A). An anti-␤-catenin blot (Fig. 6B) was used as an internal control to verify that equal amounts of protein were loaded. The antiphosphotyrosine antibody specifically detected three bands, with molecular masses of approximately 90, 85, and 83 kDa, in cell lysates from cells preincubated with pervanadate (Fig. 6A). No reactivity could be observed on lysates from nontreated cells. To identify the three protein bands, the same blot was stripped several times and reprobed with antibodies specific for ␤-catenin (Fig. 6B), plakoglobin, or p120 ctn (not shown). The 83and 85-kDa bands were identified as tyrosine-phosphorylated plakoglobin. The weak 85-kDa band may be an additional or stronger posttranslational modification of plakoglobin (Fig. 6A, lane Nϩ). The 90-kDa band seems to represent tyrosine-phosphorylated ␤-catenin but may also include p120 ctn because in MDCK cells the two proteins migrate roughly at the same position when separated by SDS-PAGE. In agreement to previous results, we could not detect tyrosine-phosphorylated Ecadherin or ␣-catenin in MDCK cells (6). Interestingly, we observed a clear reduction in the tyrosine phosphorylation of ␤-catenin and plakoglobin during mitosis (Fig. 6A, lane M). In both cases, a conspicuously weaker band was observed in cell extracts from mitotic cells compared with those from nonsynchronized cells (lane N) and cells arrested in the early S phase (lane S). To unequivocally demonstrate a reduction of tyrosine phosphorylation of ␤-catenin and/or p120 ctn during mitosis, both proteins were immunoprecipitated under denaturing conditions (Fig. 6, C-E). Also, under these conditions, a clear reduction of tyrosine phosphorylation during mitosis was evident for ␤-catenin (by a factor of 2.8) as well as for p120 ctn (by a factor of 3.7) (Fig. 6, C and E, lane M). These results clearly demonstrate that the observed reduction in tyrosine phosphorylation of ␤-catenin is applicable to the whole cellular pool, i.e. for both E-cadherin-associated and nonassociated ␤-catenin. As expected, the antibodies directed against p120 ctn detected a band of exactly the same size as did the ␤-catenin antibodies, but this band was more diffuse and shifted to higher molecular weight isoforms; thus, the observed 90-kDa band in Fig. 6A represents ␤-catenin. This is consistent with previous reports describing that tyrosine-phosphorylated p120 ctn did not coprecipitate with E-cadherin (37). p120 ctn Is Specifically Phosphorylated during Mitosis-The MPM-2 antibody reacts specifically with a phosphoepitope that is phosphorylated in G 2 and dephosphorylated at the end of mitosis (38). A variety of proteins have been identified that contain a MPM-2 phosphoepitope, e.g. the microtubule-associated proteins 1 and 4 (MAP1 and MAP4) (39,40), DNA topoisomerase II (41), Xenopus CDC25 (42), Xenopus wee1-like kinase, and myt1 (43). Microinjection of MPM-2 antibodies into fertilized Xenopus eggs inhibits both the onset and completion of mitosis, indicating the importance of this mitosis-specific phosphorylation (44,45).
To see whether a component of the E-cadherin-catenin complex is specifically Ser/Thr-phosphorylated on a MPM-2 epitope during mitosis, MDCK cells were arrested in mitosis by nocodazole. E-cadherin was immunoprecipitated, and the precipitate was analyzed by Western blotting using MPM-2. We found one MPM-2-positive band, migrating at a molecular size equivalent  6. ␤-Catenin, plakoglobin, and p120 ctn are less tyrosinephosphorylated during mitosis. Synchronized MDCK cells were incubated with (ϩ) or without (Ϫ) pervanadate. A, the E-cadherincatenin complex was immunoprecipitated from cell lysates by the 3G8 antibody, separated by SDS-PAGE, blotted, and analyzed with an antiphosphotyrosine antibody. ␤-Catenin and plakoglobin showed a clear reduction of tyrosine phosphorylation during mitosis. B, the same blot was stripped and reblotted for ␤-catenin. C and E, to analyze the tyrosine phosphorylation of the total pools of ␤-catenin and p120 ctn and distinguish between the catenins, ␤-catenin and p120 ctn were immunoprecipitated under denaturing conditions, separated by SDS-PAGE, and analyzed in a phosphotyrosine blotback. Both proteins were also less tyrosine-phosphorylated in the cellular pool. D, loading of equal amounts of precipitates was verified by ␤-catenin blotback.
to ␤-catenin or p120 ctn (data not shown). To clarify whether the observed band represents phosphorylated ␤-catenin or p120 ctn , sequential immunoprecipitations were carried out. First, Ecadherin was precipitated, and then, after the immunocomplex was dissolved, a second immunoprecipitation was done using either anti-␤-catenin or anti-p120 ctn antibodies. Only p120 ctn showed a mitosis-specific phosphorylation (Fig. 7, A and C). Whereas in nonsynchronized cells and cells arrested in the S phase, p120 ctn reacted only weakly with the MPM-2 antibody, a strong increase in the band intensity was observed in mitotic cells. The apparent decrease of the band intensity of E-cadherin-associated p120 ctn during mitosis is due to a difference in the loaded protein amount and should be considered not significant (Fig. 7D, lane M). DISCUSSION Our results indicate that the components of the E-cadherincatenin complex are subjected to alterations during mitosis. Whereas previous studies were predominantly focused on the localization of a few adherens junction proteins during mitosis (1,24), we present the first biochemical analysis of the Ecadherin-catenin complex in mitotic epithelial cells. Our immunofluorescence results confirm that all major components of the complex remained primarily localized at the cell surface during mitosis. Based on its function in cell-cell adhesion, the complex was exclusively localized at sites of cell-cell contact and was never seen in membrane regions not in contact with neighboring cells. However, we also found clear evidence for an increase of the E-cadherin-catenin complex in the cytosol of mitotic cells by immunofluorescence. This phenomenon was observed for E-cadherin and the catenins but not for ␤-actin, and it appeared already at the mitotic prophase (Fig. 1A). Consistent with this idea, the cytoplasmic staining was lost in mitotic cells permeabilized with Triton X-100 before fixation (not shown). Taken together, these results clearly demonstrate that the intense fluorescence staining of mitotic cells is caused by a cytoplasmic fraction of the E-cadherin-catenin complex not bound to the actin cytoskeleton.
However, in the biochemical analysis, we found no evidence for a change in the overall protein amount of the detergentsoluble or the detergent-insoluble E-cadherin-catenin complex during mitosis. This demonstrates that the strong cytoplasmic staining of mitotic cells is not simply the result of an increased protein amount. Earlier studies have illustrated that E-cadherin is not exclusively localized in the lateral membrane of mitotic MDCK cells but can also be found spread over the whole basolateral cell surface (1). Based on this data, it was hypothesized that mitotic cells make AJs on the entire basolateral surface, which may point to an increase in the total amount of the E-cadherin-catenin complex as well. However, from our data, we conclude that this observation does not correlate with an increase of the steady-state level of E-cadherin or the catenins and is also not due to a difference in the distribution of the actin-associated and non-actin-associated complex. We also observed a reduction of the cell surface-localized E-cadherin-catenin complex during mitosis, and we propose that the cytoplasmic localization of the complex of mitotic cells could be at least partially caused by an internalization. At present, we do not know the consequence of the reduction of the cell surface-localized complex on cell adhesion or its physiological relevance, because only the soluble fraction was successfully biotinylated. Similar results were obtained for ␣ 5 ␤ 1 integrin, where no change in the overall amount of this protein was observed in mitotic cells, but a variation of their cell substrate affinity to different extracellular matrices was observed (2). This result was explained by a reduction of the amount of cell surface localized integrin and a potential posttranslational modification, e.g. phosphorylation, that modulates integrin function.
The increased cytoplasmic pool of the E-cadherin-catenin complex in mitotic cells may also be partially caused by a storage of newly synthesized protein due to a lack of normal secretion in mitotic cells (47,48). However, we observed the intense cytoplasmic staining already during early mitosis (late prophase; Fig. 1A), suggesting that it is not a simple accumulation effect. Additionally, we were not able to detect a significant increase of the E-cadherin precursor form in mitotic cells as would expected if E-cadherin was trapped in the cytosol because of a general cessation of secretion.
According to the current knowledge about the function of tyrosine phosphorylation of ␤-catenin, our finding that E-cadherin-associated ␤-catenin and plakoglobin in mitotic cells is less tyrosine-phosphorylated than in interphase cells implies that the connection between the E-cadherin-catenin complex and the actin cytoskeleton, as well as between the complex and the E-cadherin mediated cell-cell adhesion, may be stronger during mitosis. The decrease in tyrosine phosphorylation cannot explained just by assuming that tyrosine kinases are generally inactive during mitosis. c-Src, known to be concentrated in AJs and to phosphorylate cadherins/catenins, is activated rather than repressed in mitosis (49,50). Another explanation for the observed decrease in Tyr phosphorylation may be an inactivation of the EGFR. ␤-Catenin and plakoglobin are known to interact with EGFR, and this linkage may be phosphorylation-dependent. In a normal human breast epithelial cell line, non-phosphorylated ␤-catenin was no longer found in association with EGFR (36). Thus, mitosis-specific inactivation of the EGFR tyrosine kinase may lead to the dissociation of ␤-catenin and/or plakoglobin from EGFR, and this could possibly stabilize cell-cell junctions. In fact, it was shown that in cells expressing normal amounts of EGFR, the kinase activity of EGFR is tightly suppressed during mitosis (51). FIG. 7. p120 ctn reacts with the MPM-2 antibody during mitosis. To distinguish between ␤-catenin and p120 ctn associated with E-cadherin, the E-cadherin-catenin complex was precipitated sequentially. The complex was precipitated first with the E-cadherin-specific 3G8 antibody and disrupted by 1% SDS, and then the precipitates were split and re-precipitated either with anti-p120 ctn or anti-␤catenin antibodies. The two immunocomplexes were separated by SDS-PAGE, blotted and analyzed with the MPM-2 antibody (A and B). The Western blots were stripped and reprobed for ␤-catenin (C) or for p120 ctn (D). N, nonsynchronized cells; D, Me 2 SO-treated cells; S, S phase cells; M, nocodazole-arrested cells.
In addition to tyrosine phosphorylation, we analyzed the complex for mitosis-specific Ser/Thr phosphorylation with the monoclonal antibody MPM-2, made against mitotic HeLa cell extracts (52). This antibody reacts with epitopes that are specifically phosphorylated during mitosis. Several proteins detected by this antibody have already been identified, but so far p42 mapk is the only protein for which the functional significance of the MPM-2 epitope has been established (53). Several protein kinases have been implicated in the phosphorylation of MPM-2 epitopes, including mitogen-activated protein kinase kinase, mitogen-activated protein kinase, and cdc2 (53)(54)(55)(56). However, the exact epitope that is recognized by MPM-2, as well as the function of this phosphorylation and the corresponding phosphorylation mechanism, is still obscure. Here, we describe that p120 ctn also reacts with the MPM-2 antibody specifically during mitosis, whereas no other E-cadherin-associated protein was detected. p120 ctn was identified as a Src substrate, the tyrosine phosphorylation of which correlated with cell transformation (57,58), but it was also found to be phosphorylated on serine and threonine residues (59). The kinase responsible for this phosphorylation, as well as the function of this phosphorylation event, is still unknown. At present, we do not know whether there is any correlation between the MPM-2 reactivity to p120 cat and the described serine and threonine phosphorylation. Remarkably, p120 ctn is the first MPM-2 reactive protein found to be associated with a cell surface protein such as E-cadherin, whereas most other MPM-2 positive proteins are directly involved in regulating mitosis and are localized cytoplasmically in mitotic cells.
In addition to its function in cell-cell adhesion, ␤-catenin is also involved in the Wnt-signaling pathway. A central event in this pathway seems to be the accumulation of cadherin-unbound cytosolic ␤-catenin. Down-regulation of the unbound pool of ␤-catenin appears rather important, because many cells derived from colon carcinomas or melanomas show a high amount of non-cadherin-bound ␤-catenin (46). The observed increase of ␤-catenin is presumably caused by a truncated APC protein found in many of those cell lines. However, we did not find any evidence for an increase in the amount of the cytosolic, non-cadherin-bound pool of ␤-catenin during mitosis. Instead, we observed a hyperphosphorylation of APC during mitosis, a phenomenon that has been described recently for a human osteosarcoma cell line (34). In addition, phosphorylation of a specialized domain in APC by glycogen synthase kinase-3 was found to increase the binding of ␤-catenin to APC (35), although we found no increase in the amount of APC-associated ␤-catenin in mitotic cells.
In summary, from our data, we conclude that during mitosis, the E-cadherin-catenin complex is primarily modulated by differences in phosphorylation. One consequence of these modifications may be the observed internal relocalization of the cell surface-localized complex. Further studies should clarify whether these modifications also influence the cadherin-mediated cell adhesion during mitosis.