Involvement of the Annexin II-S100A10 Complex in the Formation of E-cadherin-based Adherens Junctions in Madin-Darby Canine Kidney Cells*

E-cadherin and nectins are major cell-cell adhesion molecules at adherens junctions (AJs) in epithelial cells. When Madin-Darby canine kidney (MDCK) cells stably expressing nectin-1 (nectin-1-MDCK cells) are cultured at normal Ca 2 (cid:1) , E-cadherin and nectin-1 are concentrated at the cell-cell contact sites. When these cells are cultured at low Ca 2 (cid:1) , E-cadherin disappears from the cell-cell contact sites, but nectin-1 persists there. When these cells are re-cultured at normal Ca 2 (cid:1) , E-cadherin is recruited to the nectin-based cell-cell contact sites. We found here that this recruitment was dependent on protein synthesis, because a protein synthesis inhibitor, cycloheximide, prevented the accumulation of E-cad-herin. When nectin-1-MDCK cells, precultured at low Ca 2 (cid:1) in the presence of a proteasome inhibitor, ALLN ( N -acetyl-Leu-Leu-norleucinal), were re-cultured at normal Ca 2 (cid:1) , E-cadherin was recruited to the nectin-based cell-cell contact sites but the level of E-cadherin was reduced. Similar results were obtained when wild-type MDCK cells were used instead of nectin-1-MDCK cells. These results suggest that degradation of one or more protein factors and de novo synthesis of the same or different protein factor(s) are needed for the formation of the E-cadherin-based AJs. We biochemically identified the annexin II-S100A10 complex as such a candidate. Depletion of plasma membrane cholesterol, which abolished the localization of the annexin II-S100A10 complex at the plasma membrane, inhibited the re-con-centration of E-cadherin at the nectin-based cell-cell contact sites in the Ca 2 (cid:1) switch experiment. Knockdown of annexin II by RNA interference also inhibited the re-concentration of E-cadherin. These results indicate that the annexin II-S100A10 complex is involved in the formation of the E-cadherin-based AJs in MDCK cells. Other Procedures— as

somes (6). These junctional structures are typically aligned from the apical to basal sides, although desmosomes are independently distributed in other areas. These junctions are generally made between homotypic cells and mediated by homophilic interactions of cell adhesion molecules. The formation and maintenance of TJs and desmosomes are generally dependent on the formation of AJs (7). At AJs, the transmembrane protein E-cadherin functions as a Ca 2ϩ -dependent cell adhesion molecule (1,2,8). E-cadherin is a member of the cadherin superfamily, comprising over 80 members (5,9). Ecadherin forms homo-cis-dimers and then homo-trans-dimers (trans-interactions) through the extracellular region, causing cell-cell adhesion. The cytoplasmic tail is linked to the actin cytoskeleton through many peripheral membrane proteins (PMPs), including ␣-catenin, ␤-catenin, vinculin, and ␣-actinin (8,10,11). Cadherins directly bind ␤-catenin, which in turn binds ␣-catenin. ␣-Catenin then binds vinculin and ␣-actinin. Of these PMPs, ␣-catenin, vinculin, and ␣-actinin are actin filament (F-actin)-binding proteins (11). The association of Ecadherin with the actin cytoskeleton through these PMPs strengthens the cell-cell adhesion activity of E-cadherin (10,11).
Nectins, which constitute a family of four members, have recently emerged as Ca 2ϩ -independent Ig-like cell adhesion molecules at AJs (12,13). Nectins form homo-cis-dimers and then homo-and hetero-trans-dimers through the extracellular region, causing cell-cell adhesion. The cytoplasmic tail of nectins interacts with afadin, an F-actin-binding protein, which links nectins to the actin cytoskeleton. Nectins first form cellcell adhesion where cadherins are recruited, eventually forming AJs in epithelial cells and fibroblasts. Nectins are thought to recruit cadherins through the association mediated by their associating PMPs. The detailed molecular mechanisms for this association are not fully understood, but afadin and ␣-catenin are essential for this association and several proteins, which connect afadin and ␣-catenin, have been identified (12)(13)(14)(15)(16). The first putative connector unit for nectins and E-cadherin is a ponsin-vinculin unit (14). Ponsin is an afadin-and vinculinbinding protein, and vinculin is an F-actin-and ␣-cateninbinding protein. The second connector is an afadin dilute do-main-interacting protein (ADIP)-␣-actinin unit (15). ADIP is an afadin-and ␣-actinin-binding protein, and ␣-actinin is an ␣-catenin-binding protein. The third connector is a LIM domain only 7 (LMO7)-␣-actinin unit (16). LMO7 is also an afadin-and ␣-actinin-binding protein.
In addition to its function of cell-cell adhesion, nectins also functions in small G protein signaling (13,17). Nectins first recruit and activate c-Src at the nectin-based cell-cell adhesion sites. c-Src then phosphorylates FRG, a Cdc42-guanine nucleotide exchange factor (18). In addition, c-Src induces activation of the Rap1 small G protein through the Crk-C3G complex, and Rap1 then induces activation of FRG, eventually causing activation of Cdc42 (19). The activation of c-Src and Cdc42 induces activation of Rac through Vav2, a Rac guanine nucleotide exchange factor (20). Rac is also activated by E-cadherin in both phosphatidylinositol 3-kinase-dependent and -independent manners in epithelial cells (21). Cdc42 then increases the number of the actin cytoskeleton-based filopodia and cell-cell contact sites. Rac induces formation of the actin cytoskeletonbased lamellipodia and efficiently seals the cell-cell adhesion between the filopodia like a zipper. Thus, nectins increase the velocity of the formation of AJs through the activation of these small G proteins.
HSC-39 cells, a human signet ring cell gastric cancer cell line, express E-cadherin but do not form AJs (22). The ␤-catenin gene is truncated at the N-terminal region, including the ␣-catenin-binding domain in HSC-39 cells (23,24), but overexpression of normal ␤-catenin fails to form cell-cell adhesion (25). HSC-39 cells express nectin-1, -2, and afadin, but not nectin-3 (25). Overexpression of nectin-3 or -2 causes the formation of cell-cell adhesion and the accumulation of E-cadherin, but not F-actin, at the cell-cell adhesion sites. Overexpression of a truncated form of nectin-2 incapable of interacting with afadin fails to induce cell-cell adhesion. However, the nectin-induced cell-cell adhesion is weaker than that observed in epithelial cells, such as MDCK and CaCo-2 cells. Co-expression of nectin-2 and normal ␤-catenin does not result in a stronger cell-cell adhesion. These observations suggest that another factor may be necessary for the robust cell-cell adhesion activity of E-cadherin in addition to the known components described above.
On the basis of this assumption, we attempted to identify the one or more missing factors. For this purpose, we used the Ca 2ϩ switch experiment using MDCK cells in the presence or absence of the inhibitors of protein synthesis and proteolysis, because such a factor or factors may be easily detected under these conditions. From these Ca 2ϩ switch experiments, we found that degradation of one or more protein factors and de novo synthesis of the same or different protein factor or factors were needed for the formation of the E-cadherin-based AJs. We biochemically identified the annexin II-S100A10 complex as such a candidate. Annexin II, also called calpactin I heavy chain, is a member of the annexin family of Ca 2ϩ -and phospholipid-binding proteins and forms a heterotetrameric complex with S100A10, also called calpactin I light chain (26). The annexin II-S100A10 complex has been implicated in the structural organization and dynamics of endosomal membranes, the organization of cholesterol-rich membrane microdomains, and the connection of lipid rafts with the actin cytoskeleton (26 -29). Depletion of plasma membrane cholesterol, which abolished the localization of the annexin II-S100A10 complex at the plasma membrane, inhibited the re-concentration of E-cadherin at the nectin-based cell-cell contact sites in the Ca 2ϩ switch experiment. Knockdown of annexin II by RNA interference (RNAi) also inhibited the re-concentration of E-cadherin. These results indicate that the annexin II-S100A10 complex is involved in the formation of the E-cadherin-based AJs in MDCK cells.

EXPERIMENTAL PROCEDURES
Antibodies, Chemicals, and Expression Vectors-A rabbit anti-nectin-1 polyclonal antibody (pAb) was prepared as described (30). A mouse anti-afadin monoclonal antibody (mAb) and a rabbit anti-afadin pAb were prepared as described (31). A rabbit anti-ADIP pAb was prepared as described (15). A rabbit anti-LMO7 pAb was prepared as described (16). A rat anti-E-cadherin mAb (ECCD-2) was supplied from Dr. M. Takeichi (Center for Developmental Biology, RIKEN, Kobe, Japan). A rabbit anti-nectin-3 pAb was supplied from Dr. K. E. Mostov (University of California, San Francisco, CA). Mouse anti-FLAG M1 and antihuman vinculin mAbs, cycloheximide, and methyl-␤-cyclodextrin were purchased from Sigma. Mouse anti-annexin II, anti-p120 ctn , anti-Ecadherin, and anti-␣-catenin mAbs were purchased from BD Transduction Laboratories. A mouse anti-␤-catenin mAb and a rabbit anti-␣actinin pAb were purchased from Santa Cruz Biotechnology. A rabbit anti-ERK1/2 pAb was purchased from Cell Signaling Technology. A mouse anti-actin mAb and secondary Abs for immunofluorescence microscopy were obtained from Chemicon. Calpain inhibitor I (ALLN) and calpain inhibitor II (ALLM) were purchased from Roche Applied Science. To generate a vector for the knockdown of annexin II, pBS-H1annexin II, a specific insert for annexin II was subcloned into pBS-H1. The insert used was as follows: canine annexin II gene-specific insert was a 19-nucleotide (nt) sequence corresponding to nt 404 -422 (5Ј-GGACCAACCAAGAACTTCA-3Ј) of canine annexin II cDNA, which was separated by a 9-nt noncomplementary spacer (5Ј-TTCAAGAGA-3Ј) from the reverse complement of the same 19-nt sequence.
Ca 2ϩ Switch Assay-Ca 2ϩ switch experiments using nectin-1-MDCK or wild-type MDCK cells were done as described (33). Briefly, nectin-1-MDCK or wild-type MDCK cells were seeded on 18-mm glass coverslips in 12-well culture dishes. Forty-eight hours later, the cells were washed with phosphate-buffered saline (PBS) and cultured at 2 mM Ca 2ϩ in Dulbecco's modified Eagle's medium (DMEM) without serum for 1 h. The cells were then cultured at 2 M Ca 2ϩ (DMEM with 5 mM EGTA) in the presence or absence of 10 M cycloheximide, 50 M ALLN, or 50 M ALLM for 3 h. After the culture, the cells were washed with DMEM and cultured at 2 mM Ca 2ϩ in DMEM without serum in the presence or absence of 10 M cycloheximide, 50 M ALLN, 50 M ALLM, or 38 mM methyl-␤-cyclodextrin for 2 h.
Immunofluorescence Microscopy-Immunofluorescence microscopy was done as described (34,35). Briefly, the cells were fixed in the mixture of 50% acetone and 50% methanol at Ϫ20°C for 1 min or in PBS containing 1% formaldehyde for 15 min and PBS containing 0.2% Triton X-100 for 15 min at room temperature. After being blocked in Tris-buffered saline (TBS) containing 1% bovine serum albumin and 1 mM CaCl 2 for 1 h, the cells were incubated in the same buffer with various combinations of Abs for 1 h. The samples were washed three times with TBS containing 1 mM CaCl 2 for 5 min and incubated for 30 min in TBS containing 1% bovine serum albumin and 1 mM CaCl 2 with the secondary pAbs. The samples were then washed three times with TBS containing 1 mM CaCl 2 for 5 min and mounted in GEL/MOUNT (Biomeda). The samples were analyzed by a Radiance 2100 confocal laser scanning microscope (Bio-Rad Laboratories).
Cell Dissociation Assay-The cell dissociation assay was done as described (36). In brief, nectin-1-MDCK or wild-type MDCK cells (1 ϫ The cells were then washed with HEPES-buffered saline (HBS, pH 7.4) and treated with 0.01% trypsin supplemented with 1 mM CaCl 2 in HBS (TC treatment) or 0.01% trypsin supplemented with 1 mM EDTA in HBS (TE treatment) at 37°C for 2 h, followed by dissociation by 10times pipetting. The extent of dissociation of cells was represented by the index N TC /N TE , where N TC and N TE were the total particle number after the TC and TE treatments, respectively.
The samples were then subjected to SDS-PAGE, followed by Western blotting with the anti-nectin-1, anti-E-cadherin, and anti-ERK1/2 Abs.
Isolation of the Plasma Membrane Fraction-After the Ca 2ϩ switch, nectin-1-MDCK cells were washed with PBS, and then sonicated in Buffer A (10 mM HEPES-NaOH at pH 7.5, 100 mM KCl, 1 mM MgCl 2 , and 25 mM NaHCO 3 ) on ice for 15 s six times at 3-min intervals. The homogenate was centrifuged at 1,000 ϫ g at 4°C for 5 min. The supernatant was diluted with Buffer A into 8.0 mg/ml protein, and 0.45 ml each was applied on a 4.2-ml continuous sucrose density gradient (10 -40% sucrose in Buffer A) with 0.3 ml of 50% sucrose cushion, followed by centrifugation at 200,000 ϫ g at 4°C for 1 h with a swinging bucket rotor (P55ST2, Hitachi Ltd.). After the centrifugation, 0.3 ml of fraction each was collected. Each fraction was subjected to SDS-PAGE (8% polyacrylamide gel), followed by Western blotting with the anti-

FIG. 1. Requirement of protein synthesis for the concentration of E-cadherin at the nectin-based cell-cell contact sites after the Ca 2؉ switch in nectin-1-MDCK cells.
Aa and Ab, nectin-1-MDCK cells were cultured at 2 M Ca 2ϩ for 3 h and then incubated with 2 mM Ca 2ϩ for 2 h in the presence or absence of 10 M cycloheximide. After the incubation, the cells were fixed, followed by immunostaining for E-cadherin and nectin-1 using the anti-E-cadherin and anti-FLAG Abs, respectively. Aa, in the absence of cycloheximide; Ab, in the presence of cycloheximide; Low Ca 2ϩ , 2 M Ca 2ϩ ; Normal Ca 2ϩ , 2 mM Ca 2ϩ ; and bars, 10 m. Ac, quantitative analyses of Aa and Ab. Bars represent percentage of cell-cell contact sites with the signal for E-cadherin of the total cell-cell contact sites counted (n ϭ 50) and are expressed as means Ϯ S.E. of the three independent experiments. Ba, stage of the protein synthesis for the formation of the E-cadherin-based AJs. Left panels, nectin-1-MDCK cells were cultured at 2 M Ca 2ϩ for 3 h in the presence of 10 M cycloheximide. Next, the cells were washed with DMEM and incubated with 2 mM Ca 2ϩ for 2 h in the absence of 10 M cycloheximide. Right panels, nectin-1-MDCK cells were cultured at 2 M Ca 2ϩ for 3 h in the absence of 10 M cycloheximide. Then, the cells were incubated with 2 mM Ca 2ϩ for 2 h in the presence of 10 M cycloheximide. After the incubation, the cells were fixed, followed by immunostaining for E-cadherin and nectin-1 using the anti-E-cadherin and anti-FLAG Abs, respectively. Bars, 10 m. Bb, quantitative analysis of Ba. Bars represent percentage of cell-cell contact sites with the signal for E-cadherin of the total cell-cell contact sites counted (n ϭ 50) and are expressed as means Ϯ S.E. of the three independent experiments.
FLAG and anti-ZO-1 Abs. The fractions in which nectin-1 and ZO-1 were concentrated were collected and used as the plasma membrane fraction.
Two-dimensional PAGE-The aliquot of the plasma membrane fraction (0.33 mg of protein) was precipitated with trichloroacetic acid. The precipitated protein was dissolved in 150 l of Solution A (7 M urea, 2 M thiourea, 2% CHAPS, 0.8% Pharmalyte, and 10 mM dithiothreitol), to which 200 l of Solution B (8 M urea, 2% CHAPS, and 0.0025% orange G) was added. After centrifugation at 16,000 ϫ g for 5 min, 1.75 l of IPG Buffer (an ampholyte-containing buffer concentrate for Immobiline DryStrip; Amersham Bioscience) was added to the supernatant. A sheet of Immobiline Drystrip (18-cm long, pH 3-10, Amersham Bioscience) was rehydrated with this solution for 12 h and subjected to isoelectric focusing at 500 V for 1 h, 1000 V for 1 h, and finally 8000 V for 6 h using the IPGphor isoelectric focusing system (Amersham Bioscience). The electrophoresed strip was placed in an equilibration solution (6 M urea, 50 mM Tris-HCl at pH 6.8, 0.25% dithiothreitol, 30% glycerol, and 1% SDS) for 15 min. The strip was then incubated in a re-equilibration solution (6 M urea, 50 mM Tris-HCl at pH 6.8, 30% glycerol, 1% SDS, 4.5% iodoacetamide, and 0.01% bromphenol blue) for 15 min. The strip was placed on a 10% SDS-polyacrylamide gel without a stacking gel and subjected to electrophoresis at a constant current per gel of 10 mA for 30 min, 20 mA for 30 min, and finally 40 mA for 3 h. Protein spots were identified by silver staining as described (37).
Amino Acid Sequencing-Proteins separated by two-dimensional PAGE were stained with Coomassie Brilliant Blue R-250, the spots were excised, and amino acid sequence analysis was performed by the in-gel digestion method (38)  Met-Glu-His-Ala-Met-Glu-Thr-Met-Met-Phe-Thr-Phe) were 100% identical to those of human annexin II (amino acids 38 -47, GenBank TM accession number NP004030) and S100A10 (amino acids 2-16, Gen-Bank TM accession number NP002957), respectively.
Other Procedures-Protein concentrations were determined with bovine serum albumin as a reference protein (39). SDS-PAGE was done as described by Laemmli (40).

Requirement of Protein Synthesis for the Concentration of E-cadherin at the Nectin-based Cell-Cell Contact Sites after the
Ca 2ϩ Switch-We have previously shown that nectin-1 and E-cadherin are concentrated at the cell-cell contact sites of MDCK cells stably expressing nectin-1 (nectin-1-MDCK cells) cultured at 2 mM Ca 2ϩ (41,42). The sites of the signals for nectin-1 and E-cadherin correspond to AJs. When nectin-1-MDCK cells are cultured at 2 M Ca 2ϩ for 3 h, nectin-1 remains at the cell-cell contact sites, whereas E-cadherin disappears from these sites (41,42). When these cells are re-cultured at 2 mM Ca 2ϩ , E-cadherin is re-concentrated at the cell-cell contact sites where nectin-1 is concentrated, resulting in the formation of AJs (41,42). We first confirmed these earlier observations by staining of nectin-1 and E-cadherin (Fig. 1, Aa and Ac). We performed the Ca 2ϩ switch experiment in the presence of cycloheximide, an inhibitor of protein synthesis. When nectin-1-MDCK cells were cultured at 2 M Ca 2ϩ for 3 h in the presence of cycloheximide, nectin-1 remained at the cell-cell contact sites, whereas E-cadherin disappeared from these sites (Fig. 1, Ab and Ac). However, when these cells were re-cultured at 2 mM Ca 2ϩ in the presence of cycloheximide, E-cadherin was not re-concentrated at the cell-cell contact sites (Fig. 1, Ab and Ac). The staining pattern of nectin-1 was not affected. Similar results were obtained when wild-type MDCK cells were used instead of nectin-1-MDCK cells (Fig. 2, Aa and Ab). In this experiment, endogenous nectin-3 and afadin were stained instead of nectin-1, because endogenous nectin-1 was only faintly stained in wild-type MDCK cells as described (41,42). We then examined the cell-cell adhesion activity of E-cadherin in the presence or absence of cycloheximide by the cell dissociation assay. When nectin-1-MDCK or wild-type MDCK cells precultured at 2 M Ca 2ϩ were re-cultured at 2 mM Ca 2ϩ in the absence of cycloheximide, these cells formed aggregates (N TC /N TE ϭ 0.27 in nectin-1-MDCK cells; N TC /N TE ϭ 0.23 in wild-type MDCK cells). When nectin-1-MDCK or wild-type MDCK cells precultured at 2 M Ca 2ϩ were re-cultured at 2 mM Ca 2ϩ in the presence of cycloheximide, these cells formed fewer aggregates (N TC /N TE ϭ 0.55 in nectin-1-MDCK cells; N TC / N TE ϭ 0.66 in wild-type MDCK cells). These results suggest that de novo synthesis of one or more proteins is necessary for the formation of the E-cadherin-based AJs.
We next examined at which stage, during the culture at 2 M Ca 2ϩ or the re-culture at 2 mM Ca 2ϩ , the protein synthesis is necessary for the formation of the E-cadherin-based AJs. When nectin-1-MDCK cells, precultured at 2 M Ca 2ϩ for 3 h in the presence of cycloheximide, were then re-cultured at 2 mM Ca 2ϩ in the absence of cycloheximide, E-cadherin was re-concentrated at the nectin-based cell-cell contact sites (Fig. 1, Ba, left panels, and Bb). In contrast, when nectin-1-MDCK cells, precultured at 2 M Ca 2ϩ for 3 h in the absence of cycloheximide, were then re-cultured at 2 mM Ca 2ϩ in the presence of cycloheximide, E-cadherin was not re-concentrated at the nectinbased cell-cell contact sites (Fig. 1, Ba, right panels, and Bb). These results suggest that de novo synthesis of one or more proteins during the re-culture at 2 mM Ca 2ϩ is necessary for the formation of the E-cadherin-based AJs.
Requirement of Protein Degradation for the Concentration of E-cadherin at the Nectin-based Cell-Cell Contact Sites after the Ca 2ϩ Switch-The above results suggest that, during the culture at 2 M Ca 2ϩ , one or more proteins may be degraded and the same or different proteins may be synthesized during the re-culture at 2 mM Ca 2ϩ . We therefore examined the effect of ALLN, a proteasome inhibitor, on the re-concentration of Ecadherin at the nectin-based cell-cell contact sites in the Ca 2ϩ switch experiments. When nectin-1-MDCK cells, precultured at 2 M Ca 2ϩ for 3 h in the presence of ALLN, were re-cultured at 2 mM Ca 2ϩ in the presence of ALLN, E-cadherin was reconcentrated at the nectin-based cell-cell contact sites, but the level of the accumulated E-cadherin was reduced compared with that in the presence of ALLM, an inactive analogue of ALLN (Fig. 3, Aa and Ab). Although this effect of ALLN on the accumulation of E-cadherin was weaker than that of cycloheximide, the treatment with ALLN apparently impaired the reconcentration of E-cadherin in the Ca 2ϩ switch experiment. In contrast, the staining pattern of nectin-1 was not affected (Fig.  3, Aa). Similar results were obtained when wild-type MDCK cells were used instead of nectin-1-MDCK cells (Fig. 2, Ba and Bb).
We then examined the cell-cell adhesion activity of E-cadherin in the presence of ALLN or ALLM by the cell dissociation assay. When nectin-1-MDCK or wild-type MDCK cells precultured at 2 M Ca 2ϩ were re-cultured at 2 mM Ca 2ϩ in the presence of ALLM, these cells formed aggregates (N TC /N TE ϭ 0.31 in nectin-1-MDCK cells; N TC /N TE ϭ 0.27 in wild-type MDCK cells). When nectin-1-MDCK or wild-type MDCK cells precultured at 2 M Ca 2ϩ were re-cultured at 2 mM Ca 2ϩ in the presence of ALLN, these cells formed fewer aggregates (N TC / N TE ϭ 0.87 in nectin-1-MDCK cells; N TC /N TE ϭ 0.68 in wildtype MDCK cells). These results suggest that degradation of one or more protein factors is also necessary for the formation of the E-cadherin-based AJs.
We next examined at which stage, during the culture at 2 M Ca 2ϩ or the re-culture at 2 mM Ca 2ϩ , the degradation of one or more protein factors is necessary for the formation of the Ecadherin-based AJs. When nectin-1-MDCK cells, precultured at 2 M Ca 2ϩ for 3 h in the presence of ALLN, were then re-cultured at 2 mM Ca 2ϩ in the absence of ALLN, the reconcentrated E-cadherin at the nectin-based cell-cell contact sites was reduced (Fig. 3, Ba, right panels, and Bb). In contrast, when nectin-1-MDCK cells, precultured at 2 M Ca 2ϩ for 3 h in the absence of ALLN, were then re-cultured at 2 mM Ca 2ϩ in the presence of ALLN, the re-concentrated E-cadherin at the nectin-based cell-cell contact sites was not reduced (Fig. 3, Ba, left panels, and Bb). These results suggest that degradation of one or more proteins during the culture at 2 M Ca 2ϩ is necessary for the formation of the E-cadherin-based AJs.
Amount of E-cadherin on the Plasma Membrane after the Ca 2ϩ Switch in the Presence of Cycloheximide or ALLN-We then examined whether E-cadherin diffusely remained on the plasma membrane but was not detected by immunofluorescence microscopy in nectin-1-MDCK cells, which were treated with cycloheximide or ALLN in the Ca 2ϩ switch experiment. The Ca 2ϩ switch experiment was performed in the presence of cycloheximide or ALLN, and then the extracellular regions of E-cadherin and nectin-1 of nectin-1-MDCK cells were labeled with sulfo-NHS-SS-biotin. Biotinylation of E-cadherin and nectin-1 was also performed using nectin-1-MDCK cells before the Ca 2ϩ switch experiment. After free sulfo-NHS-SS-biotin was removed by extensive washing, the detergent-soluble, surfacebiotinylated proteins on the plasma membrane were recovered on streptavidin beads and analyzed by SDS-PAGE, followed by Western blotting with the anti-nectin-1 and anti-E-cadherin Abs. The amount of nectin-1 or E-cadherin was not affected by the Ca 2ϩ switch experiment in the presence of ALLN (Fig. 4A,  lane 4). However, the amounts of nectin-1 and E-cadherin were reduced to 50% of the control in the presence of cycloheximide (Fig. 4A, lane 3). To exclude the possibility that the failure of the re-concentration of E-cadherin at the nectin-based cell-cell contact sites in the presence of cycloheximide was simply due to a decrease of the amount of E-cadherin on the plasma membrane, we performed the Ca 2ϩ switch experiment using MDCK cells overexpressing GFP-E-cadherin (GFP-E-cadherin-MDCK cells). The level of exogenous GFP-E-cadherin was 3-fold higher than that of endogenous E-cadherin (Fig. 5A). When GFP-Ecadherin-MDCK cells, precultured at 2 M Ca 2ϩ for 3 h in the presence of cycloheximide, were re-cultured at 2 mM Ca 2ϩ in the presence of cycloheximide, GFP-E-cadherin was re-concentrated at the nectin-based cell-cell contact sites, but the level of the accumulated GFP-E-cadherin was reduced compared with that in the absence of cycloheximide (Fig. 5B). These results suggest that the failure of the re-concentration of E-cadherin at the nectin-based cell-cell contact sites in the presence of cycloheximide was not simply due to the reduced level of E-cadherin on the plasma membrane. The amounts of the residual GFP-E-cadherin remained on the plasma membrane after the cycloheximide treatment was comparable to that of endogenous E-cadherin before the cycloheximide treatment, but the level of the accumulated GFP-E-cadherin at the cell-cell contact sites in the presence of cycloheximide was apparently reduced com- pared with endogenous E-cadherin in the absence of cycloheximide (Fig. 5, A and B). When GFP-E-cadherin-MDCK cells, precultured at 2 M Ca 2ϩ for 3 h in the presence of ALLN, were re-cultured at 2 mM Ca 2ϩ in the presence of ALLN, GFP-Ecadherin was re-concentrated at the nectin-based cell-cell contact sites, but the level of the accumulated GFP-E-cadherin was reduced compared with that in the presence of ALLM, an inactive analogue of ALLN (Fig. 5C). Taken together, these results suggest that one or more additional protein factors, different from E-cadherin, is necessary for the formation of the E-cadherin-based AJs.
We then examined the amounts of the nectin-and E-cadherin-associating PMPs, afadin, ␣-catenin, ␤-catenin, p120 ctn , ␣-actinin, vinculin, ADIP, and LMO7 in the presence of cycloheximide or ALLN. These nectin-and E-cadherin-associating PMPs are involved in the formation of E-cadherin-based AJs (15,16,41,42). The amount of afadin, ␣-actinin, vinculin, ADIP, or LMO7 was not affected by the Ca 2ϩ switch experiment in the presence of cycloheximide or ALLN (Fig. 4B). The amounts of ␣-catenin, ␤-catenin, and p120 ctn were slightly reduced in the presence of cycloheximide and somewhat increased in the presence of ALLN (Fig. 4B). These subtle effects of cycloheximide and ALLN on the amount of these PMPs are not likely responsible for the failure of the re-concentration of E-cadherin at the nectin-based cell-cell contact sites in the presence of cycloheximide or ALLN. These results suggest that degradation and de novo synthesis of one or more protein factors different from these PMPs are necessary for the formation of the E-cadherin-based AJs.
Identification of a Protein Factor(s) to be the Annexin II-S100A10 Complex-We next attempted to identify the protein factor(s) involved in the formation of the E-cadherin-based AJs. For this purpose, we first prepared the plasma membrane fractions of the nectin-1-MDCK cells treated with or without cycloheximide or ALLN by sucrose density gradient ultracentrifugation. The proteins were extracted from each plasma membrane fraction and subjected to two-dimensional gel electrophoresis, followed by protein staining with silver. We found two protein spots whose amounts were decreased when nectin-1-MDCK cells were cultured at 2 M Ca 2ϩ for 3 h (Fig. 6, A and  B). The amounts of these two protein spots were increased when nectin-1-MDCK cells were re-cultured at 2 mM Ca 2ϩ (Fig.  6C). The amounts of the two protein spots were decreased by the treatment with cycloheximide, but increased by the treatment with ALLN as compared with that with ALLM (Fig. 6, D  and E). Although the amounts of some additional proteins varied under these conditions, none of them behaved like these two protein spots, which were decreased at 2 M Ca 2ϩ and increased at 2 mM Ca 2ϩ . The two protein spots (#1 and #2) were separately excised out from the gel, digested by lysylendopeptidase, and their partial amino acid sequences were determined. The sequences (spot #1: Asp-Ala-Leu-Asn-Ile-Glu-Thr-Ala-Ile-Lys; spot #2: Pro-Ser-Gln-Met-Glu-His-Ala-Met-Glu-Thr-Met-Met-Phe-Thr-Phe) were 100% identical to those of human annexin II and S100A10, respectively. Annexin II was previously called calpactin I heavy chain, whereas S100A10 was previously called p11 or calpactin I light chain (25). These results suggest that at least one of the protein factors responsible for the formation of the E-cadherin-based AJs in the Ca 2ϩ switch experiment is the annexin II-S100A10 complex.
Inhibition of the E-cadherin-based Formation of AJs by a Cholesterol-sequestering Agent and Annexin II RNAi-In the last set of experiments, we examined whether the annexin II-S100A10 complex is indeed involved in the formation of the E-cadherin-based AJs in the Ca 2ϩ switch experiment. It has been reported that annexin II associates with the plasma membrane lipid raft microdomains in a cholesterol-dependent manner (27). We examined the effect of a cholesterol-sequestering agent, methyl-␤-cyclodextrin, on the re-concentration of E-cadherin at the nectin-based cell-cell contact sites in the Ca 2ϩ switch experiment. When nectin-1-MDCK or wild-type MDCK cells, precultured at 2 M Ca 2ϩ for 3 h in the absence of methyl-␤-cyclodextrin, were re-cultured at 2 mM Ca 2ϩ in the presence of methyl-␤-cyclodextrin, neither annexin II nor Ecadherin was re-concentrated at the nectin-based cell-cell contact sites (Fig. 7, A-C). In contrast, the staining pattern of nectin-1 or afadin was not affected (Fig. 7, A and B).
We finally examined whether annexin II was required for the formation of the E-cadherin-based AJs by knockdown of annexin II with RNAi. The expression level of annexin II was reduced in MDCK cells transfected with the vector expressing small interfering RNA oligonucleotides against annexin II (Fig.  8A). When the Ca 2ϩ switch experiment was performed using nectin-1-MDCK or wild-type MDCK cells in which annexin II was knocked-down, the re-concentration of E-cadherin at the nectin-based cell-cell contact sites was inhibited as compared with the control cells (Fig. 8, B-D). In contrast, the staining pattern of nectin-1 or afadin was not affected (Fig. 8, B and C). These results suggest that annexin II is indeed involved in the formation of the E-cadherin-based AJs. DISCUSSION We have found conditions where E-cadherin is not concentrated at the cell-cell contact sites where nectin-1 is concentrated in nectin-1-MDCK cells. When nectin-1-MDCK cells precultured at low Ca 2ϩ are re-cultured at normal Ca 2ϩ in the presence of cycloheximide, E-cadherin is not re-concentrated at the nectin-based cell-cell contact sites. When nectin-1-MDCK cells, precultured at low Ca 2ϩ in the presence of ALLN, a proteasome inhibitor, are re-cultured at normal Ca 2ϩ , the re- cadherin at the nectin-based cell-cell contact sites, because overexpression of exogenous E-cadherin does not restore the accumulation of E-cadherin at the cell-cell contact sites. In addition, the amount of E-cadherin is not reduced in the presence of ALLN, whereas the level of the re-concentrated Ecadherin at the nectin-based cell-cell contact sites is reduced. Thus, the failure of the re-concentration of E-cadherin at the nectin-based cell-cell contact sites in the presence of cycloheximide or ALLN is not simply due to the reduced levels of E-cadherin on the plasma membrane. The amounts of the PMPs associating with nectin and E-cadherin are also not changed dramatically in the presence of cycloheximide or ALLN. Thus, degradation and de novo synthesis of one or more protein factors, which is different from the known nectin-and E-cadherin-associating PMPs, are necessary for the formation of the E-cadherin-based AJs.
We have biochemically identified a candidate of such a protein factor(s) to be the annexin II-S100A10 complex. The amounts of both annexin II and S100A10 on the plasma membrane are decreased by the treatment with cycloheximide but increased by the treatment with ALLN. Depletion of plasma membrane cholesterol in nectin-1-MDCK and wild-type MDCK cells abolishes the localization of annexin II at the plasma membrane, causing the inhibition of the accumulation of Ecadherin at the cell-cell contact sites. This observation is consistent with the previous finding that depletion of plasma membrane cholesterol impaired the formation of AJs in endothelial cells (43). In addition, knockdown of annexin II by RNAi inhibits the formation of the E-cadherin-based AJs in nectin-1-MDCK and wild-type MDCK cells. These results indicate that degradation and de novo synthesis of the annexin II-S100A10 complex is involved in the formation of the E-cadherin-based AJs in MDCK cells. However, it remains unclear if the annexin II-S100A10 complex is the only cycloheximide-or ALLN-sensitive factor.
There are several possible mechanisms of the annexin II-S100A10 complex in the formation of the E-cadherin-based AJs. Because it has been shown that the annexin II-S100A10 complex is involved in the recycling endosomes (29), the annexin II-S100A10 complex might be involved in recycling of E-cadherin during the formation of the E-cadherin-based AJs. Because annexin II is implicated in the anchoring of lipid rafts with the actin cytoskeleton (27,28), the annexin II-S100A10 complex might function in the formation of the actin cytoskeleton at AJs. The nectin-based cell-cell adhesion is formed even in the presence of the actin filament-disrupting agents, cytochalasin D or latrunculin A, and is not disrupted by the agent in nectin-1-MDCK cells (44). In contrast, the E-cadherin-based cell-cell adhesion is abolished in the presence of cytochalasin D or latrunculin A (44 -47). Similarly, knockdown of annexin II by RNAi or depletion of plasma membrane cholesterol, which abolishes the localization of annexin II at the plasma membrane, does not affect the formation of the nectin-based cell-cell adhesion but abolishes that of the E-cadherin-based cell-cell adhesion. Thus, the annexin II-S100A10 complex might function in the formation of E-cadherin-based AJs through the organization of the actin cytoskeleton. Recently, it has been shown that the interaction of AHNAK with the annexin II-S100A10 complex regulates the organization of the cortical actin cytoskeleton (48). AHNAK might be involved in the accumulation of E-cadherin at the cell-cell contact sites.
It has been shown that the membrane-bound Rac is distributed in three complexes, the Rac-PAK complex, an 11 S complex, and a 16 S complex, in MDCK cells (49). Rac shifts from the 11 S complex to the 16 S complex during the formation of the E-cadherin-based AJs (49). Annexin II has been identified as a Rac binding partner in the 16 S Rac complex (49). The activation of Rac is induced by the trans-interactions of nectins and that of E-cadherin and enhances the formation of AJs (13,17,21). Rac activated by either the trans-interactions of nectins and/or that of E-cadherin induces formation of lamellipodia and efficiently seals the cell-cell adhesion between the filopodia like a zipper. Therefore, the annexin II-S100A10 complex might be recruited to the actin cytoskeleton formed by the action of the F-actin-binding proteins, which are associated with E-cadherin and nectins and also by the action of Rac through its downstream F-actin-binding proteins, such as IQ motif containing GTPase activating protein (IQGAP) and Wiskott-Aldrich syndrome protein family verprolin-homologous proteins (WAVEs). The annexin II-S100A10 complex then reorganizes the actin cytoskeleton, causing the association of the E-cadherin and nectin systems to form AJs. Further studies are necessary to appreciate the role of the annexin II-S100A10 complex in the formation of E-cadherin-based AJs.