Cell-Cell Dissociation upon Epithelial Cell Scattering Requires a Step Mediated by the Proteasome*

During development, tissue repair, and tumor metastasis, both cell-cell dissociation and cell migration occur and appear to be intimately linked, such as during epithelial “scattering.” Here we show that cell-cell dissociation during scattering induced by hepatocyte growth factor (HGF) or activation of the temperature-sensitive v-Src tyrosine kinase in MDCK cells can be blocked by inhibiting the proteasome with lactacystin and MG132. Although both proteins of the tight junction and the adherens junction redistributed during cell scattering, proteasome inhibitors largely prevented this process, resulting in the stabilization of Triton X-100-insoluble tight junction proteins as well as adherens junction proteins at sites of cell-cell contact. Proteasome inhibition also led to a decrease of E-cadherin turnover in 35S-labeled cells. In addition, proteasome inhibition partly preserved cell polarity, as determined by the subcellular distribution of Na+,K+-ATPase (basolateral marker) and gp135 (apical marker), and the structure of the subcortical actin ring, both of which are normally disrupted during scattering. However, cells were able to establish focal contacts, and single cell migration toward HGF was unaffected by proteasome inhibition in quantitative assays, indicating that cell-cell dissociation during scattering occurs independently of anchorage-dependent cell migration. Thus, a proteasome-dependent step during scattering induced by HGF and pp60v-Src appears to be essential for cell-cell dissociation, disassembly of junctional components, and (at least indirectly) it also plays a role in the loss of protein polarity.

During development, tissue repair, and tumor metastasis, both cell-cell dissociation and cell migration occur and appear to be intimately linked, such as during epithelial "scattering." Here we show that cell-cell dissociation during scattering induced by hepatocyte growth factor (HGF) or activation of the temperature-sensitive v-Src tyrosine kinase in MDCK cells can be blocked by inhibiting the proteasome with lactacystin and MG132. Although both proteins of the tight junction and the adherens junction redistributed during cell scattering, proteasome inhibitors largely prevented this process, resulting in the stabilization of Triton X-100-insoluble tight junction proteins as well as adherens junction proteins at sites of cell-cell contact. Proteasome inhibition also led to a decrease of E-cadherin turnover in 35 Slabeled cells. In addition, proteasome inhibition partly preserved cell polarity, as determined by the subcellular distribution of Na ؉ ,K ؉ -ATPase (basolateral marker) and gp135 (apical marker), and the structure of the subcortical actin ring, both of which are normally disrupted during scattering. However, cells were able to establish focal contacts, and single cell migration toward HGF was unaffected by proteasome inhibition in quantitative assays, indicating that cell-cell dissociation during scattering occurs independently of anchorage-dependent cell migration. Thus, a proteasome-dependent step during scattering induced by HGF and pp60 v-Src appears to be essential for cell-cell dissociation, disassembly of junctional components, and (at least indirectly) it also plays a role in the loss of protein polarity.
In largely epithelial tissues such as kidney and intestine, both the permeability barrier and the polarized sorting of proteins and lipids are intimately linked to the formation of intercellular junctions, such as adherens junctions (AJs) 1 and tight junctions (TJs) (1,2). In general, the structure and function of these intercellular junctions depends upon transmembrane proteins (e.g. E-cadherin, occludin/claudins) that are linked to nonmembrane proteins on the cytosolic face, which in turn are associated with the actin-based cytoskeleton (3)(4)(5). Under steady-state conditions, the assembly and maintenance of these intercellular junctions appear to be tightly regulated (3,6,7). However, during development, cell division, inflammation, and tissue repair, as well as invasion and metastasis of tumor cells, these structures are disassembled and sometimes internalized as cells diminish their contacts and become motile (2,8,9). Little is known of the molecular basis of junction disassembly and reassembly during the alterations in cell-cell interactions and motility that characterize these highly dynamic states of tissue remodeling, regeneration, and transformation, although in models such as the "calcium switch," a variety of signaling molecules, including protein kinase Cs, calcium, and GTP binding proteins have been implicated in junctional reassembly (10 -13).
A key question concerning cell dissociation during cell movement is the metabolic fate of components within junctional complexes. One well studied experimental system in which this issue is clearly important is "scattering." Epithelial cell scattering requires the attenuation or dissolution of cell-cell adhesion before cell-cell dissociation and motility (14). To perform an analysis that might be applicable to a broader context, we employed two different systems: scattering induced by hepatocyte growth factor (HGF) and scattering resulting from the activation of temperature-sensitive v-Src tyrosine kinase (pp60 v-Src ) (15).
We now show that cell-cell dissociation during scattering induced by HGF or pp60 v-Src in MDCK epithelial cells can be blocked by inhibiting the proteasome without affecting cell migration, resulting in stabilization of junctional proteins at sites of cell-cell contact. We conclude that the proteasome plays an essential role as a regulator of cell-cell dissociation during cell movement.
For co-immunoprecipitation experiments, cells were lysed with a buffer containing 1% Triton X-100, 1% deoxycholate, 50 mM Tris-HCl, pH 7.4, 100 mM NaCl, 25 mM sodium fluoride, 10 mM sodium pyrophosphate, 2 mM sodium orthovanadate, 1.5 mM MgCl 2 , 2 mM EGTA, plus a protease inhibitor mixture (TN buffer) for 30 min at 4°C, and insoluble materials were separated by centrifugation at 4°C for 30 min at 14,000 ϫ g. The supernatant containing 1 mg of protein was clarified and incubated either with anti-ZO-1 or anti-E-cadherin antibody on a rocking platform at 4°C overnight. The immune complexes were collected either with anti-rat IgG-Sepharose (Organon Teknika, West Chester, PA) or protein A-Sepharose beads (Amersham Pharmacia Biotech) for 30 min at 4°C. The beads were washed three times with the lysis buffer, resuspended in 2ϫ sample buffer, and boiled for 5 min. Immunoprecipitated proteins were then analyzed by SDS-PAGE and visualized by Western immunoblotting using horseradish peroxidaseconjugated secondary antibodies (Jackson Labs, West Grove, PA) and the ECL kit (Pierce) as described (11,16).
Triton X-100 Extraction Assay-Cells were scraped into CSK-A buffer (0.5% Triton X-100, 25 mM Tris-HCl, pH 7.4, 300 mM sucrose, 25 mM sodium fluoride, 10 mM sodium pyrophosphate, 2 mM sodium orthovanadate) plus the protease inhibitor mixture and extracted for 20 min at 4°C on a gently rocking platform. The extract (E-fraction) was separated by centrifugation (14,000 ϫ g) for 10 min at 4°C, and the residue (R-fraction) was dissolved in 2ϫ sample buffer. Equal volumes of both fractions were separated by SDS-PAGE and subjected to Western immunoblot. Blots were quantified with NIH image software as described (11,16).
Metabolic Labeling and Immunoprecipitation-Cells were metabolically labeled with [ 35 S]methionine/cysteine (100 Ci/ml, Expre 35 S 35 S labeling mix; NEN Life Science Products) for 3 h. After the labeling period, cells were chased for 0, 3, 8, or 16 h in normal growth medium, lysed in CSK-B buffer (0.5% SDS, 0.5% Triton X-100, 50 mM Tris-HCl, pH 7.4, 25 mM sodium fluoride, 10 mM sodium pyrophosphate, 2 mM sodium orthovanadate, 300 mM sucrose, and a protease inhibitor mixture), and boiled for 5 min. Then the lysates were diluted with 5 volumes of CSK-A buffer. Equal amounts of protein in supernatants were subjected to immunoprecipitation. Immunecomplexes were collected with protein A-Sepharose beads, washed with CSK-A buffer, and separated by SDS-PAGE. Labeled proteins were quantified with a PhosphorImager (Molecular Dynamics).
Immunocytochemistry-MDCK cells grown on a type I collagencoated coverslip were either directly fixed with 1% paraformaldehyde or after extraction with CSK-A buffer for 20 min at 4°C. Immunofluorescent analysis was as described previously, and samples were examined using a laser scanning confocal system (Bio-Rad MRC 1024) (17).
Chemotaxis Assay-Chemotaxis was assayed with the modified Boyden chamber technique as described (18,19). In brief, the lower chambers filled with serum-free DMEM containing HGF were covered with Nucleopore filters (Costar; pore diameter, 8 m) coated with type I collagen. Then freshly trypsinized MDCK cells suspended in serum-free DMEM (5 ϫ 10 5 /ml) were added to the upper chamber. Protease inhibitors were added to both sides. The chambers were incubated for 6 h at 37°C in an atmosphere of 95% air/5% CO 2 . Cells that migrated over the filter were fixed, stained, and counted under a microscope.

RESULTS AND DISCUSSION
Proteasome Inhibition Blocks Cell Scattering Induced by HGF and pp60 v-Src -MDCK cells (at the "cell island" stage) were pretreated with either potent proteasome inhibitors (lactacystin and MG132), neutral cysteine protease inhibitors (ALLN and E64), a calpain inhibitor (calpeptin), or inhibitors of lysosomal proteolysis (chloroquine and primaquine) for 30 min. Cells were then subjected to scattering conditions by adding either 20 ng/ml HGF or, in the case of the cells with temperature-sensitive pp60 v-Src , inducing a temperature shift from the nonpermissive (40°C) to permissive temperature (35°C). Of the inhibitors tested, only the proteasome inhibitors could markedly inhibit cell-cell dissociation in MDCK cells during scattering driven by HGF or pp60 v-Src (Fig. 1). Both MG132 and the highly specific agent lactacystin, which inhibit the proteasome through different mechanisms, gave similar results. ALLN showed a weak inhibitory effect on this cell-cell dissociation consistent with the compound's weak inhibition of the proteasome (20), whereas neither E64 nor lysosomal inhibitors affected scattering (data not shown). We have previously demonstrated that comparable concentrations of MG132 and ALLN inhibit the degradation of short and long-lived proteins in MDCK cells in a manner consistent with inhibition of the proteasome (20). Moreover, we have shown that reduction of the aldehyde group in MG132, which is the key moiety of this molecule necessary for proteasome inhibition, markedly diminishes its ability to inhibit protein degradation in MDCK cells, indicating that the action of this agent is largely on the proteasome (20).
Notably, the shape of the cell island after proteasome inhibition under scattering conditions was different from the normal cellular island: 1) cells at the margin appeared to spread into the cell-free space, resulting in the formation of long cytoplasmic protrusions, and 2) these marginal cells demonstrated extensive membrane ruffling at the cell-free side, whereas cellcell contact was maintained with neighboring cells.
The Total Amounts of TJ as Well as AJ Proteins Are Minimally Changed during Scattering in the Presence or Absence of Protease Inhibitors-To analyze the blockage of scattering by proteasome inhibitors biochemically, we sought to examine the overall amounts of TJ as well as AJ proteins in MDCK cells at the cellular island stage by Western immunoblotting (Ͼ90% of junctional proteins were dissolved in RIPA extraction buffer). As shown in Fig. 3a, in general, the amounts of TJ as well as AJ proteins were not significantly changed after treatment with various protease inhibitors, including proteasome inhibitors, during scattering induced by HGF in MDCK cells. The most notable change was an increase of ␤-catenin, in which case less mobile forms accumulated in the presence of MG132, lactacystin, or ALLN; however, this increase was also found in MDCK cells treated with lactacystin alone, indicating that this change in ␤-catenin is not specific to the blockage of cell-cell dissociation during scattering induced by HGF in MDCK cells ( Fig. 2A). Because ␤-catenin is known to be degraded by the proteasome through its ubiquitination (21,22), this increase is likely to be due to an accumulation of ubiquitinated ␤-catenin by proteasome inhibitors. In addition, a slight increase (Ͻ20%) in levels of total ␣-catenin, E-cadherin, and occludin was also observed after proteasome inhibition ( Fig. 2A). Similar results were obtained with temperature-sensitive v-Src MDCK cells treated with protease inhibitors at either nonpermissive or permissive temperatures (data not shown).
To study the stoichiometry of the individual components of the ZO-1 and E-cadherin containing complexes during scattering in the presence or absence of lactacystin, MDCK cells were lysed with TN buffer to preserve protein-protein interactions, and the supernatant was subjected to immunoprecipitation with anti-ZO-1 or anti-E-cadherin antibody, followed by immunoblotting (Fig. 2B). No significant differences in the amounts of co-precipitating proteins in the ZO-1 or E-cadherin containing complex were observed in scattered cells induced by both HGF and pp60 v-Src in the presence or absence of lactacystin. Although less mobile forms of ␤-catenin coprecipitated with E-cadherin, this was also found in cells treated with lactacystin alone (data not shown).  Fig. 2 indicated that the overall protein amount of TJ as well as AJ proteins is not markedly changed in MDCK cells treated with HGF and the activation of pp60 v-Src in the presence of proteasome inhibitors, this did not exclude the possibility that the subcellular localization and the cytoskeletal association of TJ and AJ proteins is affected by the treatment. To examine this possibility, we employed a Triton X-100 extraction assay in the cells (17). After overnight incubation with proteasome inhibitors under scattering conditions driven by HGF, cells were extracted with CSK-A buffer, which contains 0.5% Triton X-100, and processed for indirect immunofluorescence (Fig. 3). Indirect immunofluorescence revealed that after proteasome inhibition under scattering conditions, Triton X-100-insoluble ZO-1 was clearly detectable at the cell-cell contact site of HGF-stimulated MDCK cells treated with lactacystin (Fig. 3d), as was the case with a integral membrane TJ protein, occludin (Fig. 3c) (23); these staining patterns are similar to untreated cells (data not shown) (17,24). In addition, after proteasome inhibition under scattering conditions, the transmembrane AJ protein, E-cadherin (Fig. 3g), also colocalized with ␣-catenin (Fig. 3h), which links E-cadherin to the actin cytoskeleton via ␤-catenin at the lateral border of the cells (3). In marked contrast, in the absence of proteasome inhibition, the normal junctional localization of both TJ and AJ proteins, was disrupted, and these proteins lost their association with cytoskeletal elements in scattered cells, judging from immunocytochemistry of Triton X-100-insoluble junctional proteins (Fig. 3, a, b, e, and f), consistent with results of others (25). Similar results were obtained in temperature-sensitive v-Src MDCK cells (data not shown).

Proteasome Inhibitors Stabilize TJ as Well as AJ Proteins in the Triton X-100-insoluble Pool-Although the immunoblot analysis shown in
Immunoblot analyses of Triton X-100-soluble and -insoluble ("cytoskeletal") fractions were also consistent with the notion that proteasome inhibitors stabilized associations of TJ proteins with the cytoskeletal fraction (Fig. 4A). A significant portion of occludin and ZO-1 remained in the Triton X-100insoluble pool in the presence of inhibitors, whereas TJ proteins became Triton X-100 soluble after scattering in the absence of the inhibitors (Fig. 4B). The changes in solubility of these proteins became obvious after ϳ4 h of HGF treatment or the activation of temperature-sensitive pp60 v-Src and correlated well with cell-cell dissociation observed by light microscopy (data not shown). Moreover, the amount of a higher molecular weight form of occludin in the Triton X-100-insoluble pool, which appears to be localized at the TJ (Fig. 3c) (17,24), was increased by the inhibitors. The alteration of Triton X-100 extractabilities of ZO-2, E-cadherin, and ␣-catenin was less striking than ZO-1, occludin, and ␤-catenin (Fig. 4, B and C). The absolute increase in Triton X-100-insoluble ␤-catenin may be partly due to the total increase (Fig. 2); however, it is possible that the blockage of catenin degradation partly explains the inhibition of cell-cell dissociation upon scattering, because in addition to their key role in stabilizing the AJ (3), catenins may play a role in the stabilization of the TJ (26, 27). Thus, these results suggest that this blockage of cell-cell dissociation by proteasome inhibitors might stabilize not only Then cells were lysed in CSK-B buffer and boiled for 5 min. The lysates were diluted with five volumes of CSK-A buffer and homogenized by needle passing. The supernatant was separated by centrifugation and was subjected to immunoprecipitation (IP) with anti-E-cadherin antibody. Labeled E-cadherin was visualized by fluorography, and immunoprecipitated E-cadherin was analyzed by Western immunoblotting with the same antibody to E-cadherin. Note that densitometric analysis at the 16 h time point revealed that ϳ60% of radiolabeled E-cadherin was present in MG132-treated cells, whereas less than 10% of labeled E-cadherin was detectable in the absence of MG132. junctional protein complexes at the cell-cell adhesion site but also prevent disruption of links between junctional proteins and cytoskeletal elements (indicated by resistance to detergent extraction).

Proteasome Inhibitors Prolong the Half-life of E-cadherin under Scattering
Conditions-To gain further insight into the fate of junctional proteins during proteasome inhibition upon cell scattering, we used 35 S-labeled temperature-sensitive v-Src MDCK cells because v-Src has less mitogenicity for MDCK cells than HGF (15). During the last 30 min of the labeling period at the nonpermissive temperature, cells were treated with MG132 or vehicle and chased for various lengths of time in normal growth medium at the permissive temperature. Immunoprecipitation from 35 S-labeled cells at each time point revealed that E-cadherin is a long-lived protein: the half-life was ϳ8 h at the permissive temperature (Fig. 5). The proteasome inhibitor, MG132 as well as lactacystin (data not shown), significantly blocked the decrease of radiolabeled E-cadherin at the scattering temperature (Fig. 5B). We cannot exclude the possibility that those chemicals also inhibit cell proliferation (28 -30), and thereby radiolabeled E-cadherin might be diluted by newly synthesized E-cadherin molecule in the absence of the inhibitor. Nevertheless, this result supports our hypothesis that proteasomal proteolysis can alter the fate of junctional proteins upon epithelial cell scattering.
Neither Rearrangement of Focal Adhesions nor Single Cell Migration Induced by HGF Is Affected by Proteasome Inhibition-Cell scattering consists of at least two biological responses, which appear to occur simultaneously or synchronously in the cells: 1) cell-cell dissociation, resulting from breaking apart of intercellular junctional complexes, and 2) cell movement, driven by rearrangement of the cytoskeleton and formation of new cell-substratum contacts (focal adhesions). To address the question of whether the proteasomal proteolytic pathway might be involved in the latter step, we examined the immunocytochemical localization of paxillin, a major component of focal adhesions (31), upon cell scattering induced by HGF. As shown in Fig. 6A (panel a), paxillin mainly accumulated along the edge of normal MDCK cells in cellular islands. In scattered (or scattering) cells, paxillin accumulated at cell protrusions and the membrane ruffling site (Fig. 6A, panel b).
Of interest was that paxillin also accumulated at cell protrusions of marginal cells in lactacystin-treated MDCK cell islands in a distribution similar to that of scattering cells treated with HGF alone (Fig. 6A, panel c). Moreover, a chemotaxis assay using a modified Boyden chamber technique did not reveal any effect of lactacystin, even at higher concentrations (5 M), on cell migration toward HGF (Fig. 6B). In agreement with a previous report (32), a calpain inhibitor, calpeptin, partially inhibited cell migration induced by HGF, whereas calpeptin could not block cell scattering induced by HGF and pp60 v-Src (data not shown). These data suggest that the proteasomal proteolytic pathway is not critical for rearrangement of focal adhesions and cytoskeleton induced by HGF as well as cell motility in this context, but it is essential for disruption of cell-cell adhesion and intercellular junctions.
Proteasome Inhibitors Partially Prevent Loss of Cell Polarity under Scattering Condition-To analyze cell polarity, we examined the immunocytochemical localization of Na ϩ ,K ϩ -ATPase, which is normally present at the basolateral domain of polarized MDCK cells (33), and gp135, a glycoprotein associated with apical microvilli of MDCK cells (34). As shown in Fig.  7A (panel a), the ␤ subunit of Na ϩ ,K ϩ -ATPase was clearly localized at the lateral border of control MDCK cells, whereas Na ϩ ,K ϩ -ATPase was widely redistributed in scattered cells (Fig. 7A, panel b). In the presence of lactacystin, Na ϩ ,K ϩ -ATPase was still found at the lateral border after HGF stimulation; however, this staining at the lateral border of the cells was weaker than in control cells, and faint diffuse intracellular staining was present (Fig. 7A, panel c). Analysis of gp135 revealed clear punctate staining characteristic of apical microvillar staining in control MDCK cells (Fig. 7, A, panel d, and  B, panel a). After treatment with HGF, this punctate staining diminished significantly, and gp135 redistributed diffusely in scattered cells (Fig. 7, A, panel e, and B, panel b). Lactacystin inhibited the loss of the apical staining of gp135 (although the punctate staining was finer) on the apical surface of the cells (Fig. 7, A, panel f, and B, panel c), consistent with partial preservation of cell polarity.
Because polarized epithelial cells possess a subcortical cytoskeleton that undercoats the AJ as well as TJ (35,36), we also examined the distribution of F-actin in the cells. Phalloidin staining revealed that loss of the cortical actin bundle was largely prevented by lactacystin in HGF-stimulated MDCK cells (Fig. 7, C, panel a, and C, panel c). This structure was markedly disrupted in scattered cells (Fig. 7C, panel b). These results are consistent with the stabilization of junctional proteins at the lateral border in MDCK cells treated with proteasome inhibitors upon cell scattering (Fig. 3), because junctional complexes are tightly associated with cortical actin bundles that lie under the complexes in epithelial cells. Apical microvillar actin staining also appeared to be partly preserved but was obscured because of other alterations in the actin cytoskeleton accompanying changes in cell shape and motility. Thus, the partial preservation of cell polarity may be not only due to the blockage of cell-cell dissociation but also the preservation of cell-cell junctions and the subcortical actin cytoskeleton.
It is worth noting that several junctional proteins can serve as proteasome substrates (21,37) and that a deubiquitinating enzyme, Fam, has been localized to the junctional complex (38). Our results demonstrate that proteasome inhibition blocks cellcell dissociation by HGF and pp60 v-Src in MDCK epithelial cells; however, it is still unknown whether this blockage of cell-cell dissociation by proteasome inhibition affects a constitutive degradation pathway of junctional proteins or a pathway enhanced only in the context of scattering, resulting from the activation of a particular tyrosine kinase cascade.
Assembly and disassembly of junctional complexes appear to involve multiple signaling pathways (7, 10 -13, 17); it is thus possible that, apart from playing a key role in the turnover of junctional proteins such as E-cadherin, the proteasome may regulate the turnover of a key protein involved in signaling. Further examination of the degradation pathway of junctional proteins will lead to a better understanding of how epithelial cell-cell contact is regulated not only under physiological and/or developmental conditions but also in disease states.