Identification of Four Distinct Pools of Catenins in Mammalian Cells and Transformation-dependent Changes in Catenin Distributions among These Pools*

Catenins are cytoplasmic proteins that were initially identified in a complex with cadherins, a superfamily of transmembrane glycoproteins important for cell adhesion in normal and disease states. We have used gel filtration to identify four complexes of catenins in extracts from normal and transformed epithelial cells. In normal Madin-Darby canine kidney epithelial cells, a significant fraction of α- and β-catenin and plakoglobin co-elute with cadherin in a high molecular weight complex (complex I). A portion of α-catenin and the remainder of β-catenin and plakoglobin co-elute in a high molecular weight complex that does not contain cadherin (complex II). The remainder of α-catenin elutes in a low molecular weight fraction (complex III). In extracts from two colon carcinoma cell lines, HCT116 and SW480, β-catenin elutes in an additional low molecular weight pool (complex IV) not present in Madin-Darby canine kidney cell extracts. In two subclones derived from SW480 cells, SW-E8 and SW-R2, β-catenin is distributed evenly between high and low molecular weight pools in SW-E8 cells, whereas it elutes primarily in the low molecular weight pool (complex IV) in SW-R2 cells. These changes in β-catenin elution profiles correlate with an increase in transformed phenotype and decreased cell-cell adhesion in the SW-R2 cells.

Cadherins comprise a superfamily of Ca 2ϩ -dependent transmembrane glycoproteins that play essential roles in the initiation and stabilization of cell-cell contacts (1). Regulation of cadherin-mediated intercellular adhesion is critical for normal embryonic development and the maintenance of mature epithelial tissues (2,3). Loss of this regulation has been implicated in disease processes such as invasion and metastasis (4 -6). For example, histological and biochemical studies have correlated increased invasiveness and tumor cell grade with decreased cadherin expression (7)(8)(9).
␣-Catenin, an actin-binding protein that shares 30% sequence identity with vinculin in conserved regions (17)(18)(19), is thought to stabilize the cadherin complex at the plasma membrane by linking it to the actin-based cytoskeleton (20,21). Cadherins cannot mediate strong cell-cell adhesion in cell lines that lack ␣-catenin, and tumor cell lines show increased invasiveness following mutation and/or down-regulation of ␣-catenin (22).
Both ␤-catenin and plakoglobin are members of the armadillo gene family (23)(24)(25). armadillo is a segment polarity gene in the Drosophila wingless signaling pathway (26). Expression of truncated forms of ␤-catenin, which can bind cadherin but not ␣-catenin, correlates with loss of cadherin function in cancer cell lines (27). Other tumor lines express unusually high levels of ␤-catenin, possibly due to mutations in the adenomatous polyposis coli (APC) 1 tumor suppressor gene (28,29). In addition, modulation of APC/␤-catenin association has been suggested to play a role in the regulation of both cadherinbased cell-cell adhesion and cell motility (30 -33).
Studies in Drosophila and Xenopus show that accumulation of cytoplasmic armadillo/␤-catenin, not bound to cadherin, correlates with changes in cell fate and transcriptional activation (34 -36). Because of the role of armadillo in the wingless pathway and the ability of ␤-catenin to bind to transcription factor LEF-1/Tcf family members (37)(38)(39), alterations in ␤-catenin levels may affect intracellular signaling in addition to, or in combination with, effects on cadherin-mediated adhesion. For example, mutations in APC protein that correlate with the accumulation of ␤-catenin may result in excess transcriptional activation by ␤-catenin-Tcf complexes (40 -42). In addition, Drosophila homologs of LEF-1/Tcf have been implicated as components of the wingless signaling pathway (43,44).
Consistent with the multiple functions assigned to catenins, studies in Madin-Darby canine kidney (MDCK) cells indicate that at least 50% of the catenins within a cell are not associated with cadherin (45). Hinck et al. (14) suggested that catenins may exist as monomers, may dimerize with other catenins, or may associate with other proteins within the cell. Recently, several proteins, including ZO-1, APC protein, EGF receptor, and fascin, have been found to interact with ␤-catenin (46 -49).
Although catenins have been implicated in a variety of cellular processes, the dynamics of cadherin/catenin assembly and the complexes in which they function are poorly understood, especially as they relate to changes in cell phenotype during differentiation or transformation. Therefore, we sought to frac-tionate different cadherin and catenin complexes from several cell types that represent a continuum of transformed phenotypes. Using gel filtration to separate proteins and immunoprecipitations to define complexes, we show that catenins are present in four distinct pools: complex I, a high molecular weight complex that contains cadherin; complex II, a high molecular weight complex without cadherin; and complexes III and IV, containing low molecular weight complexes of ␣-catenin or ␤-catenin, respectively. We show that while cadherin levels and elution profiles are similar among all cell types examined, both an increase in total levels of ␤-catenin and a shift in its distribution from high (complex I/II) to low molecular weight complexes (complex IV) correlate with an increase in transformed phenotype.

EXPERIMENTAL PROCEDURES
Cell Lines and Antisera-Human colon carcinoma HCT116 and human colon adenocarcinoma SW480 cells were obtained from ATCC. SW-E8 and SW-R2 cells were kindly provided by Dr. I. Bernard Weinstein (Columbia University) (50). MDCK cells have been described previously (51). Cells were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum. 3G8 mouse monoclonal antibody, specific for the extracellular domain of canine E-cadherin (52), and rabbit polyclonal antibodies against ␣-catenin, ␤catenin, plakoglobin (53), APC protein (32), and the cytoplasmic domain of mouse E-cadherin (54) have been described previously. A mouse monoclonal antibody specific for fascin (49) was kindly provided by Dr. Pierre McCrea (University of Texas, Houston, TX).
To determine the distribution of total protein in 0.5% Nonidet P-40-soluble and -insoluble cell fractions, cells were washed twice with Tris-saline, extracted in 1 ml of MEBC buffer (0.5% Nonidet P-40, 50 mM Tris-HCl, pH 7.5, 100 mM NaCl) containing protease inhibitors (0.1 mM Na 3 VO 4 , 50 mM NaF, 1 mM Pefabloc (Boehringer Mannheim), and 10 g/ml each of leupeptin, antipain, chymostatin, and pepstatin A) for 10 min at 4°C. Cell lysates were then sedimented in a Microfuge (Beckman Instruments, Inc., Fullerton, CA) for 10 min. Supernatants were removed, and pellets were solubilized in 200 l of SDS immunoprecipitation buffer at 100°C for 30 min with occasional vortexing. 1 ml of MEBC was then added to the solubilized pellets, and 200 l of SDS immunoprecipitation buffer was added to the supernatants to normalize the total volume of samples. Equal volumes of supernatant and pellet were analyzed by Western blotting and quantitated, as above.
Size Exclusion Chromatography-Cells were extracted with MEBC buffer as above, incubated at 4°C for 10 min, and sedimented for 10 min in a Microfuge. The supernatant was then transferred to a clean tube, centrifuged in a 100.3 rotor (Beckman Instruments, Inc.) for 30 min at 100,000 ϫ g, and then passed through a 0.22-m syringe filter (Millipore). 200 l of extract was loaded on a Superose 6 HR 10/30 column (10 mm ϫ 30 cm; Pharmacia Biotech, Inc.), which had been equilibrated in MEBC buffer containing 1 mM dithiothreitol and 0.1 mM Pefabloc. Proteins were eluted at a flow rate of 0.3 ml/min at 17°C, and 0.5-ml fractions were collected. Protein concentration for each fraction was determined as above. Fractions 7-30 were separated in 1.5-mm 7.5% SDS-PAGE gels, and gels were stained with Coomassie Brilliant Blue R. 15 g of recombinant ␣and ␤-catenin produced in Escherichia coli (recombinant proteins were kindly provided by Dr. Bill Weis, Stanford University, Stanford, CA) were diluted in MEBC buffer, centrifuged at 100,000 ϫ g, passed through a 0.22-m syringe filter, and analyzed by gel filtration and Western blotting, as above. For experiments using a mixture of recombinant proteins, 15 g of ␣and ␤-catenin were coincubated on ice for 8 h, diluted in MEBC buffer, and processed as above.
Immunoprecipitations-Following gel filtration, fractions 13-24 were preincubated with 5 l of preimmune serum and 35 l of pansorbin cells (Calbiochem) and then centrifuged for 10 min at 14,900 ϫ g. 10 l of antibody specific for cadherin or ␤-catenin was added to each fraction for 30 min followed by the addition of 65 l of a 50% slurry of protein A-Sepharose (Pharmacia), rotated for 2 h at 4°C, and washed as described previously (14). Immunoprecipitates were boiled in SDS sample buffer and then divided between replicate 7.5% SDS-PAGE gels, which were processed further for immunoblotting as above. For cadherin immunodepletion, antibody specific for cadherin was used to complete four consecutive immunoprecipitations from MDCK cell extracts, which removed Ͼ99% of cadherin from the samples as determined by Western blot analysis. Control immunoprecipitations were done in parallel using nonimmune rabbit sera. Immunodepleted and control samples were then fractionated and processed for immunoblotting as above.

Characterization of Cadherin and Catenin Expression in MDCK, HCT116
, and SW480 Cell Lines-Cells were plated at 85% confluency (ϳ2.5 ϫ 10 6 cells/p35), grown for 18 h at 37°C, and examined using phase contrast microscopy. MDCK cells grew in a tight monolayer in which colonies had well defined edges (Fig. 1). Cells were uniform in shape, which generated a characteristic cobblestone appearance of the monolayer. A few rounded cells were found, but they mostly represented dividing cells. HCT116 cells also grew as a monolayer, although the cells did not pack together as tightly as MDCK cells; translucent spaces between cells are evident even within the center of colonies. SW480 cells appeared highly transformed as they were rounded up and grew on top of each other, even before a monolayer of cells had established.
We considered whether differences in cell-cell interactions among these three cell lines were due to differences in total levels of cadherin and catenins. Cells were plated at a confluent density (ϳ3 ϫ 10 6 cells/p35) and grown for 36 h at 37°C. Cells were scraped off Petri plates into buffer containing 1% SDS and boiled for 10 min to solubilize proteins (for details, see "Experimental Procedures"). Equal amounts of total protein were separated by SDS-PAGE and then Western blotted with antibodies specific for ␣-catenin, ␤-catenin, plakoglobin, or the cytoplasmic domain of cadherin ( Fig. 2A). This cadherin antibody was generated against the cytoplasmic domain of mouse Ecadherin and recognizes cadherins other than E-cadherin due to the highly conserved amino acid sequence of cadherin cyto-plasmic domain (54). Western blots shown are representative of one experiment, and average protein levels from three independent experiments are plotted relative to the MDCK cell protein level. Quantitation of Western blots of whole cell lysates revealed that MDCK, HCT116, and SW480 cells express similar amounts of ␣-catenin and plakoglobin at steady state ( Fig. 2A). However, SW480 cells expressed approximately half the amount of cadherin and 3-4 times as much ␤-catenin as the other two cell types.
We also assessed the solubility of cadherin and catenins in buffer containing 0.5% Nonidet P-40. Previous studies have shown that the detergent insolubility of these proteins is an indication of complex association with the actin cytoskeleton (20,21). Confluent monolayers of cells were washed twice with ice-cold Tris-saline, scraped from the Petri dish into 1 ml of MEBC buffer at 4°C, incubated on ice for 10 min, and then centrifuged at 14,900 ϫ g for 15 min at 4°C. Equal volumes of supernatant and pellet were separated by SDS-PAGE and immunoblotted with antibodies specific for either cadherin or each of the catenins (Fig. 2B). Data shown are representative Western blots from one experiment, and the histograms represent the mean of three independent experiments. Approximately 80 -90% of cadherin, 90% of ␣-catenin, and 85% of ␤-catenin from all three cell types was solubilized in MEBC buffer. 65% of plakoglobin was solubilized from MDCK and SW480 cells, and 50% was solubilized from HCT116 cells under these extraction conditions.
Identification of Cadherin-dependent and -independent Pools

FIG. 2. Comparison of cadherin and catenin levels and solubility.
A, for total protein analysis, cells were plated at confluent density and grown for 36 h. Cells were solubilized in buffer containing 1% SDS, and protein concentration was determined for each sample. Equal amounts of total protein were separated by SDS-PAGE. Gels were transferred to Immobilon-P and immunoblotted with antibodies specific for cadherin, ␣-catenin, ␤-catenin, or plakoglobin. Secondary 125 I-labeled goat anti-rabbit IgG was used to detect primary antibodies, and signals were quantitated using a Phosphor-Imager. Blots shown are representative of one experiment. Protein levels in HCT116 and SW480 cells are plotted relative to levels in MDCK cells; bar graphs show mean results Ϯ S.D. from three independent experiments. B, to determine solubility characteristics of cadherin and catenins in buffer containing 0.5% Nonidet P-40, cells were plated an grown as in A and scraped from plates in MEBC buffer. Samples were sedimented in a Microfuge for 10 min, and the supernatant (Supnt) was removed to a clean tube. Pel of Catenins-To examine and compare cadherin-and catenincontaining complexes in different cell lines, we used FPLC Superose 6 gel filtration to separate protein complexes in the molecular weight range of 5-5000 kDa. Lysates from confluent cells were clarified by centrifugation and filtration and then loaded on a sizing column pre-equilibrated in extraction buffer (for details, see "Experimental Procedures").
Protein assays on column fractions from FPLC fractionation of MDCK cell extracts revealed that the void volume eluted in fraction 7, that all proteins eluted between fractions 7 and 35, and that approximately half of the proteins eluted between fractions 23 and 29 (Fig. 3A). The peak elution profiles of globular proteins of known molecular weights and Stokes radii are indicated. Coomassie Brilliant Blue staining of an SDSpolyacrylamide gel of proteins from a representative FPLC fractionation of MDCK cells showed that nearly all proteins eluted between fractions 7 and 30, with an increase in protein concentration in fractions 24 -28 (Fig. 3B).
We used immunoblot analysis to determine the elution profiles of cadherin and each of the catenins extracted from MDCK, HCT116, and SW480 cells. Immunoblots were quantified using a PhosphorImager, and values were normalized to the fraction of greatest intensity (Fig. 4). Immunoblots and corresponding graphs shown in Fig. 4 are representative of at least three independent experiments. Quantitation of blots revealed a single high molecular weight population of cadherin in MDCK cells that eluted in a peak at fraction 16 (Fig. 4). ␣-Catenin eluted as two distinct populations of protein; a major portion (70%) eluted with a peak at fraction 17, and the re-mainder eluted at fraction 23/24. ␤-Catenin appeared to elute as a single protein population with a peak at fraction 17. Plakoglobin eluted as a single protein population with a peak at fraction 16.
Elution characteristics of proteins and protein complexes during gel filtration are dependent on the effective column volume for a protein of a particular size and shape. As such, estimates of molecular weight by gel filtration can be affected by the structure of the protein or protein complex. Based upon the elution profiles of globular proteins with known molecular weights and Stokes radii, proteins eluting in a peak at fraction 16 have an elution profile consistent with a globular protein of M r Х 1400 ϫ 10 3 and R S Х 109 Å; proteins eluting at fraction 17 are consistent with M r Х 1100 ϫ 10 3 and R S Х 96 Å; proteins eluting at fraction 21 are consistent with M r Х 300 ϫ 10 3 and R S Х 59 Å; and proteins eluting at fraction 23/24 are consistent with M r Х 140 ϫ 10 3 and R S Х 43 Å.
After FPLC Superose 6 gel filtration of an HCT116 cell extract, a single high molecular weight peak of cadherin eluted at fraction 16. Approximately 85% of ␣-catenin eluted at fraction 16, and a second peak of ␣-catenin eluted at fraction 23/24. In HCT116 cells, ␤-catenin eluted in two discrete peaks. 85% of ␤-catenin eluted in a peak at fraction 16, and the remainder eluted at fraction 23. Plakoglobin eluted with peaks at fractions 16 and 23, similar to those of ␣and ␤-catenin.
Following fractionation of SW480 cell extracts, cadherin eluted as a single peak at fraction 15/16, similar to that of cadherin in MDCK and HCT116 cells. However, the elution profiles of catenins extracted from SW480 cells were very dif- that SW480 cells have a heterogeneous morphology (Fig. 1). SW480 cells have been subcloned into two lines, designated SW-E8 and SW-R2, that have flat epithelial-like and rounded appearances, respectively (Fig. 5) (50). We examined whether there were changes in cadherin and catenin complexes associated with differences in the transformed phenotype and apparent degree of cell-cell adhesion between these subclones. Cells from each clone had similar total levels of cadherin, ␣-catenin, and plakoglobin as MDCK and HCT116 cells and had the same 3-4 fold increase in total ␤-catenin present in SW480 cells compared with levels in MDCK and HCT116 cells (Fig. 2). Also, the solubility characteristics of cadherin and catenins were similar for each clone, and were similar to those in the other cell lines examined (Fig. 2).
The FPLC elution profiles of cadherin and each catenin from cell extracts of each subclone are shown in Fig. 6. Fractionation of SW-E8 cell extracts revealed a single peak of cadherin at fraction 15/16. ␣-Catenin from these cells had a complex elution profile with three separate peaks at fractions 16, 20/21, and 23. 50% of the ␣-catenin eluted at fraction 16, 30% at fraction 20, and 20% at fraction 23. ␤-Catenin from SW-E8 extracts was evenly divided between two peaks that eluted at fractions 16 and 23. The elution profile for plakoglobin showed two peaks at fractions 16 and 23/24.
Because of the proportionately low amount of ␤-catenin coeluting with cadherin in the SW-R2 cell extracts, we were interested to test whether cadherin from these cells was still complexed with ␤-catenin. Immunoprecipitations with antibodies specific for cadherin were performed from fractions 13-20, and the immunoprecipitated complexes were analyzed by immunoblotting with antibodies specific for cadherin and ␤-catenin (Fig. 7). Cadherin immunoprecipitated from SW-R2 cell extracts eluted with a peak in fraction 16. A distinct peak of ␤-catenin co-immunoprecipitated with cadherin at fraction 16.
We compared the subcellular distributions of cadherin and catenins in SW-E8 and SW-R2 clones by immunofluorescence microscopy. In SW-E8 cells, cadherin, ␣and ␤-catenin appeared primarily at areas of cell-cell contacts with some intracellular staining (Fig. 8). Plakoglobin was diffusely distributed in these cells, but some staining in regions of cell-cell contact was present. In SW-R2 cells, cadherin staining appeared primarily at the periphery of cells, potential regions of cell-cell contact, while the catenins displayed abundant intracellular staining. While some cell borders stained for catenins in SW-R2 cells, this staining was due primarily to the rounded shape of individual cells rather than to specific accumulation of protein at regions of cell-cell contact. These results suggest that the protein pool containing both cadherin and ␤-catenin is large in SW-E8 cells but small in SW-R2 cells and are consistent with our FPLC data.
MDCK Cells Have Two High Molecular Weight Catenin Pools-Elution profiles of cadherins and catenins from MDCK cell extracts revealed that the peak of the high molecular weight catenin pool (fraction 17) eluted one fraction later than the peak of cadherin (Fig. 4). It is possible that the elution profile of catenins included two high molecular weight complexes of which one contains cadherin and one does not. To first confirm the presence of high molecular weight cadherin-catenin complex, we used an antibody against cadherin to immunoprecipitate complexes containing cadherin from fractions 13-24. Immunoblotting these cadherin immunoprecipitates with antibodies against cadherin and ␤-catenin revealed a complex containing cadherin and ␤-catenin, which eluted with a peak at fraction 16 (Fig. 9, A and B). Note that the fractions containing the cadherin-catenin complex are coincident with the elution profile of cadherin but not with that of total ␤catenin (see Fig. 4).
Next, we immunoprecipitated protein complexes from the same fractions with an antibody against ␤-catenin. Immunoblots of the co-immunoprecipitated protein complexes with cadherin antibody demonstrated the elution of a protein complex containing cadherin and ␤-catenin with a peak at fraction 16 (Fig. 9C). However, ␤-catenin Western blots from these ␤catenin immunoprecipitates showed a peak of ␤-catenin at fraction 17, indicating the presence of a high molecular weight, cadherin-independent pool of ␤-catenin (Fig. 9D).
To confirm the presence of this cadherin-independent pool, we examined the elution profile of catenins from a MDCK cell extract that had been immunodepleted of cadherins with an antibody against cadherin prior to fractionation by FPLC. Western blots of the FPLC fractions confirmed that cadherin had been removed from the extract (Fig. 10). Despite the absence of the cadherin-catenin complex, a population of ␣and ␤-catenin and plakoglobin co-eluted at fraction 17 in a high molecular weight complex.
The presence of a high molecular weight cadherin-independent pool of catenins suggested that this complex might contain other catenin-binding proteins. To initially analyze this complex, we immunoblotted FPLC fractions from MDCK cell extracts with antibodies specific for APC protein and fascin, two proteins shown previously to bind catenins (see Introduction) (Fig. 11). APC protein eluted with a peak in fraction 9, separate from the peak of ␤-catenin at fraction 17 (Figs. 4 and 11). Fascin eluted with a peak in fraction 28 (Fig. 11); however, upon overexposure of the blot, a small amount of fascin could be detected in fraction 18, partially overlapping the peak of ␤catenin. 2 While both APC protein and fascin fractionated separately from the major peak of ␤-catenin, these proteins bind only a small portion of total cellular ␤-catenin, and levels of fascin-␤-catenin complexes have been demonstrated to be relatively low in MDCK cell extracts in comparison with those from A-431 and HeLa cells. 3 Recombinant Catenins Co-fractionate with the Low Molecular Weight Catenin Pool-Our analysis of the elution profiles of catenins showed that both ␣and ␤-catenin elute in a low molecular weight peak at fraction 23. We sought to determine the molecular organization of catenins in this fraction by comparing the elution profiles of pure, recombinant ␣and ␤catenin; recombinant proteins were fractionated under buffer conditions identical to those used for extraction and gel filtration of proteins from MDCK, HCT116, and SW480 cells. Recombinant ␣-catenin purified from E. coli extracts lacked the 50 N-terminal amino acids due to proteolytic degradation during purification. Recombinant ␣-catenin eluted in fraction 23 (Fig. 12A), similar to that of ␣-catenin found in the low molecular weight fractions from whole cell extracts. The major peak of recombinant ␤-catenin eluted at fraction 23, and this peak co-fractionated with the ␤-catenin in the low molecular weight fractions in the HCT116 and SW480 lines. Small amounts of higher molecular weight aggregates of ␤-catenin occasionally appeared in fractions 12-20. After incubation at 4°C for 8 h, a mixture of recombinant ␣and ␤-catenin eluted identically to recombinant proteins fractionated individually, with peaks at fraction 23 (Fig. 12A). Western blots of immunoprecipitations specific for ␤-catenin from fractionated samples of recombinant ␣and ␤-catenin mixtures revealed that heterodimers of ␣and ␤-catenin eluted at fraction 23 (Fig. 12B).

DISCUSSION
Cadherin and catenins play important roles in cell-cell adhesion, migration, development of cellular polarity, and intracellular signaling (1,35,(55)(56)(57). Incorrect regulation of cadherin-catenin complex formation and functions is frequently associated with disruption of cell-cell adhesion, changes in cell morphology, and the progression of cell transformation (5,58,59). This range of cellular functions indicates that these proteins comprise multiple, distinct complexes within cells. In the present study we have examined cadherin-and catenin-containing complexes from cells representing normal (MDCK), moderately transformed (HCT116, SW-E8), and highly transformed (SW-R2) phenotypes.
All cells examined had similar levels of cadherin, which fractionated as a single, high molecular weight complex. We found that catenins co-fractionated and co-immunoprecipitated with this pool of cadherin. However, the relative amount of catenins complexed with this cadherin pool differed between these cell lines. In MDCK cells, which form tight cell-cell contacts, a substantial portion of both ␣and ␤-catenin co-fractionated with cadherin and could be co-immunoprecipitated in a complex with cadherin. However, in SW-R2 cells, which exhibit poor cell-cell adhesion, relatively little ␤-catenin co-eluted with the high molecular weight cadherin complex, although the total amount of ␤-catenin in SW-R2 cells was 2-3 times greater than that in MDCK cells. In SW-R2 cells, Ͼ95% of ␤-catenin fractionated in a low molecular weight complex. Since all cells were treated identically, it is unlikely that ␤-catenin simply dissociated from cadherins during protein extraction and fractionation. Also, some ␤-catenin co-immunoprecipitated with cadherin in fractionated SW-R2 extracts. The mechanism that causes ␤-catenin to accumulate in the low molecular weight complex may involve mutations in APC protein that do not target ␤-catenin for degradation (28). We note that our previous studies have shown that this low molecular weight pool of ␤-catenin is competent to bind the cytoplasmic domain of cadherin (16).
The estimated molecular mass for the cadherin complex (ϳ1400 kDa) is much higher than that calculated for a monomeric cadherin-␤-catenin-␣-catenin complex based upon globular protein standards (ϳ300 kDa). This apparent high molecular weight may have been due to the presence of multimers, the presence of additional proteins in the complex, or aberrant migration during gel filtration due to unusual shape. It is noteworthy that the cadherin complex extracted from SW-R2 cells eluted in the same fraction as that of cadherin from adherent MDCK cells, although the SW-R2 cadherin complex FIG. 7. ␤-Catenin from SW-R2 cells co-immunoprecipitates with cadherin. Protein complexes from fractions 13-20 were immunoprecipitated with antibodies specific for the cytoplasmic domain of cadherin. Immunoprecipitated complexes were washed stringently and then separated by SDS-PAGE. Proteins were transferred to Immobilon-P membranes, which were subsequently immunoblotted with antibodies specific for cadherin and ␤-catenin.

FIG. 8. Subcellular localization of cadherin and catenins in SW-E8 and SW-R2 cells.
Cells were grown on glass coverslips, washed thoroughly, fixed with 1.75% formaldehyde, and permeabilized in buffer containing 1% Triton X-100. Samples were individually incubated with antibodies specific for cadherin, ␣-catenin, ␤-catenin, or plakoglobin, followed by rhodamine-conjugated sheep anti-rabbit secondary antibody. Note the large intracellular pools of catenins in SW-R2 cells in comparison with SW-E8 cells. Bar, 15 m.
FIG. 9. ␤-Catenin exists in cadherin-and non-cadherin-containing complexes in MDCK cell extracts. Following Superose 6 gel filtration chromatography of proteins extracted from MDCK cells extracted with MBEC buffer, fractions 13-24 were processed for immunoprecipitation with antibodies specific for either cadherin (A and B) or ␤-catenin (C and D) under nondenaturing conditions. Immunoprecipitated complexes were separated by SDS-PAGE, transferred to Immobilon-P, and immunoblotted with antibodies specific for either cadherin (A and C) or ␤-catenin (B and D). In cadherin immunoprecipitates, both cadherin and ␤-catenin eluted with a peak at fraction 16. In ␤-catenin immunoprecipitates, however, cadherin eluted with a peak at fraction 16, while ␤-catenin eluted with a peak at fraction 17. The relative electrophoretic mobilities of molecular weight markers are shown to the left of each immunoblot. had relatively little ␤-catenin (see above). This result indicates that the apparent high molecular weight of the cadherin-catenin complex might be due to multimers of cadherin rather than due to the association of catenins, per se. In support of this notion, it has been reported recently that cadherins may form dimers and higher ordered structures (60).
In addition to the pool of cadherin-associated catenins, we found evidence for multiple pools of catenins that did not coelute with cadherin. One population of catenins had an elution profile similar to that of the cadherin-catenin complex but did not contain cadherin (complex II). The apparent molecular mass of this complex was high (ϳ1100 kDa) relative to the elution patterns of pure, recombinant ␣and ␤-catenin. It is possible that other proteins associate with this population of catenins. However, the elution profiles of APC protein and fascin, which have been shown previously to bind ␤-catenin independently of cadherin (46,49), did not overlap with this population of catenins. The investigation of this high molecular weight catenin complex will be the subject of future studies.
The other major peak of catenins eluted at fraction 23/24 (complex III/IV). It is noteworthy that pure, recombinant ␣and ␤-catenin eluted individually at fraction 23. Therefore, cellular catenins that co-elute in these fractions could be monomers, heterodimers, or homodimers. To attempt to resolve the organization of catenins in these fractions, we separated a mixture of pure, recombinant ␣and ␤-catenin. We found that these proteins co-eluted in fraction 23/24 and could be coimmunoprecipitated as an ␣/␤-catenin heterodimer. Analytical   FIG. 10. Cadherin-independent catenins from MDCK cell extracts fractionate in both high and low molecular weight complexes. Cadherin was immunodepleted from MDCK cell extracts prior to fractionation by Superose 6 gel filtration. Fractions 7-30 were separated by SDS-PAGE, transferred to Immobilon-P, and immunoblotted with antibodies specific for E-cadherin, ␣and ␤-catenin, and plakoglobin. Primary antibodies were detected using 125 I-labeled goat anti-rabbit secondary antibody. ␣-Catenin eluted with peaks at fractions 17 and 23/24. Both ␤-catenin and plakoglobin elute as a single population with a peak at fraction 17.
FIG. 11. APC protein and fascin do not co-fractionate with the high molecular weight peak of catenins. MDCK cell extracts were fractionated by Superose 6 gel filtration. Fractions 7-30 were separated by SDS-PAGE, transferred to Immobilon-P, and immunoblotted with antibodies specific for APC protein and fascin. Primary antibodies were detected using horseradish peroxidase-conjugated sheep anti-mouse secondary antibody. APC protein elutes with a peak at fraction 9, and fascin elutes with a peak at fraction 28.
FIG. 12. A, Superose 6 gel filtration chromatography of pure, recombinant ␣and ␤-catenin. Purified, bacterially expressed recombinant ␣and ␤-catenin were fractionated separately (rows 1 and 2) or as a mixture (rows 3 and 4) under conditions identical to those used to fractionate cell extracts. Fractions 7-30 were immunoblotted with antibodies specific for either ␣-catenin (rows 1 and 3) or ␤-catenin (rows 2 and 4). Recombinant ␣and ␤-catenin eluted with a peak at fraction 23, when fractionated either separately or as a preincubated mixture. B, immunoprecipitation of ␣-catenin-␤-catenin heterodimers. A mixture of recombinant ␣and ␤-catenin were fractionated by gel filtration. Fraction 23 was processed for immunoprecipitation with antibodies specific for ␤-catenin. Immunoprecipitated complexes were washed stringently and then processed for immunoblotting with antibodies against ␣-catenin (a) and ␤-catenin (b). The relative electrophoretic mobilities of molecular weight markers are shown to the left of the immunoblots. ultracentrifugation of pure, recombinant ␣-catenin indicates that ␣-catenin can also form homodimers. 4 A large pool of ␣-catenin in MDCK cells in the absence of corresponding pools of ␤-catenin or plakoglobin indicates that this low molecular weight pool comprises either monomers or homodimers. That cellular ␣-catenin may be present as homodimers in vivo is supported by our observation that ␣-catenin from MDCK cells eluted with a peak at the same fraction as ␣/␤ heterodimers.
The amount of catenins in complexes I-IV varies relative to the state of cellular transformation, as assessed by cell organization and degree of cell-cell adhesion. HCT116 cells are slightly transformed compared with MDCK cells. However, analysis of ␤-catenin elution profiles shows that extracts from both cells contain high molecular weight complexes of ␤-catenin (complex I/II), and the only difference is the presence of a small pool of ␤-catenin from HCT116 cells that elutes at fraction 23 (complex IV). In contrast, SW480 cells are highly transformed; cells are rounded, with little cell-cell adhesion. In extracts from SW480 cells, Ͻ10% of ␤-catenin elutes in a high molecular weight complex with cadherin (complex I/II). The elution profiles of catenins are also different between the SW-E8 and SW-R2 clones derived from SW480 cells. The most striking difference is in the elution profiles of ␤-catenin. In SW-E8 cells, ␤-catenin is divided evenly between high (complex I/II) and low molecular weight pools (complex III/IV), but in SW-R2 cells Ͼ95% of ␤-catenin elutes in the low molecular weight fraction. Significantly, SW-E8 cells exhibit a cell phenotype that is more similar to that of MDCK or HCT116 cells, with extensive cell-cell contacts, while SW-R2 cells are rounded and do not form an epithelial monolayer.
At present, we do not understand the mechanisms involved in regulating the distribution of catenins between high (complex I/II) and low (complex III/IV) molecular weight pools. Steady state levels of ␤-catenin have been shown to be affected by APC protein. Some tumor cell lines with mutant APC protein accumulate elevated levels of ␤-catenins in a low molecular weight pool (28). Transfection of these cells with APC protein decreased the level of ␤-catenin, primarily from a low molecular weight pool, but decreases in a high molecular weight pool were also detected (28). This suggests that a balance between high and low molecular weight complexes containing ␤-catenin is closely regulated and that APC protein may play a role in limiting the accumulation of ␤-catenin in low molecular weight complexes.
Post-translational regulation, perhaps by serine/threonine or tyrosine phosphorylation, may be responsible for the distribution of catenins in these different pools. Alternatively, binding of catenins to other proteins may be regulated in this way, thus affecting complex formation. The phosphorylation status of ␤-catenin can also be affected by intracellular membrane-associated tyrosine kinases such as Src and receptor tyrosine kinases such as the EGF receptor (47,61,62). In response to transfection with v-Src, both ␤-catenin and cadherin are phosphorylated, and phosphorylation may interfere with cadherin function and disrupt adherence junctions (62,63). In addition to phosphorylation by Src, ␤-catenin can bind EGF receptor and is a substrate for tyrosine phosphorylation by EGF receptor upon ligand binding (47). Thus, specific pools of ␤-catenin may play a role in interactions between EGF signal transduction pathway and cadherin function.
In summary, we have identified four distinct pools of catenins in extracts of epithelial cells. Significantly, we have shown that the extent of cell-cell adhesion and transformed cell phenotype correlates with the relative amounts of ␤-catenin in high and low molecular weight pools. Furthermore, we have identified a high molecular weight complex of catenins that is not bound directly to cadherins. The identification of these pools of catenins provides a foundation for future studies on mechanisms that determine the distribution of catenins among these different pools. An understanding of these mechanisms may provide new insights into metabolic pathways involved in cellular differentation and transformation.