Wingless signaling modulates cadherin-mediated cell adhesion in Drosophila imaginal disc cells

The dynamic regulation of cell-cell and cell-substratum adhesion is of crucial importance for the control of cell shape and cellular movements that are the basis for morphogenesis in higher organisms. One of the best-understood cell adhesion molecules is E-cadherin, a protein present in most epithelial tissues and required for the formation of adherens junctions (Gumbiner, 1996). To function as a homophilic adhesion molecule, E-cadherin forms a complex with several cytoplasmic proteins. β-catenin, a member of a protein family characterized by internal Armadillo repeats, binds to the cytoplasmic domain of E-cadherin (Aberle et al., 1994; Hulsken et al., 1994; Jamora and Fuchs, 2002). α-catenin interacts with the N terminus of β-catenin and is thus indirectly bound to E-cadherin, via β-catenin (Aberle et al., 1994; Hulsken et al., 1994). α-catenin also locally regulates the organization of the actin cytoskeleton (Drees et al., 2005; Yamada et al., 2005).

In addition to its essential role in the structure of the adhesion complex, β-catenin has a major function in the regulation of cell-cell adhesion (Nelson and Nusse, 2004). β-catenin can be modified by phosphorylation on either serine/threonine or tyrosine residues (Behrens et al., 1993; Bek and Kemler, 2002; Brembeck et al., 2004; Hoschuetzky et al., 1994; Nelson and Nusse, 2004; Piedra et al., 2001; Shiozaki et al., 1995; Sommers et al., 1994). In v-src-transformed MDCK cells, loss of cell-cell adhesion and increased invasiveness of the cells correlate with an increase in tyrosine phosphorylation of β-catenin (Behrens et al., 1993). Similar observations have been made in a human esophageal cancer cell line that shows increased tyrosine phosphorylation of β-catenin after treatment with EGF (Shiozaki et al., 1995). Recently, Brembeck et al. have found evidence for a switch between the adhesive and transcriptional functions of β-catenin that is modulated by phosphorylation of Tyr 142 of β-catenin (Brembeck et al., 2004). Phosphorylation leads to binding of the BCL9-2 protein to β-catenin, which in turn interferes with α-catenin contact. These dynamic interactions suggest that Wnt signaling, by virtue of using β-catenin as a transcriptional effector, may also change the adhesive role of β-catenin.

In Drosophila, adherens junctions consist of the same components as in vertebrates: E-cadherin (DE-cadherin), β-catenin, which is encoded by the segment polarity gene armadillo (arm) and α-catenin (Cox et al., 1996; McCartney et al., 2001; Oda and Tsukita, 1999; Pai et al., 1996; Tepass et al., 1996; Uemura et al., 1996). Arm is also an essential component of the Wingless (Wg) signaling pathway. Wg is very pleiotropic and is involved in a large number of developmental processes, including segmentation of the embryo, development of the gut and the heart, and morphogenesis of wings, legs and eyes (Klingensmith and Nusse, 1994; Seto and Bellen, 2004).

In this work, we have addressed the question of whether there is any functional connection between Wg signaling and cadherin-mediated cell adhesion, or whether these two processes are independent of each other. Mammalian cells overexpressing Wnt1, the vertebrate ortholog of Wg, show effects of Wnt1 expression on cadherin-mediated cell adhesion (Bradley et al., 1993; Hinck et al., 1994; Seto and Bellen, 2004). These include an increase in the amount of β-catenin and plakoglobin (another member of the Armadillo protein family), an increase in cadherin protein levels and an overall strengthening of cell adhesion in the Wnt1-expressing cells. However, these observations were made in cells that constitutively express transfected Wnt genes, raising the possibility that the observed effects might not be the primary response of the cells to the Wnt signal. Therefore, we have used a Drosophila tissue culture system that is based on cl-8 wing imaginal disc cells (Peel et al., 1990). cl-8 cells respond to stimulation with Wg by an increase in the level of a cytosolic, hypophosphorylated form of Arm, which is caused by a posttranscriptional stabilization of the Arm protein (Van Leeuwen et al., 1994). The same response can be obtained by transfection of the cells with a cDNA encoding a temperature-sensitive allele of Wg or by overexpression of Dishevelled (Dsh), an intracellular component of the Wg signaling pathway (Van Leeuwen et al., 1994; Yanagawa et al., 1995). cl-8 cells have proved useful for Drosophila developmental and cell biological studies, in particular in recent RNAi screens for genes controlling signaling responses (Lum et al., 2003). We find that Wg signaling leads to an initial downregulation of the E-cadherin-Arm complex at cell-cell contacts, followed by transcriptional upregulation of DE-cadherin expression. We suggest that Wg signaling facilitates cell movement and division in epithelial tissues by transiently reducing cell-cell adhesion.

In order to study the effect of Wg signaling on cadherin-mediated cell adhesion, we used Drosophila cl-8 imaginal disc cells. cl-8 cells respond to Wg by an elevation of the cytoplasmic pool of Arm protein (Van Leeuwen et al., 1994), presumably because these cells express the Wg receptor, Dfz2 (Bhanot et al., 1996). cl-8 cells express low levels of DE-cadherin mRNA and protein (data not shown) and weakly adhere to each other. To restore E-cadherin-mediated cell adhesion, cl-8 cells were transfected with a cDNA encoding mouse E-cadherin under control of the metallothioneine promoter and a stable cell line was established (cl8mEcad). This experimental setup allowed us to distinguish potential effects of Wg signaling on transcription of endogenous DE-cadherin from posttranscriptional effects on mouse E-cadherin expressed under control of the metallothioneine promoter. cl8mEcad cells accumulated a low, baseline level of mouse E-cadherin protein (henceforth called E-cadherin) in the absence of Cu²⁺, due to the leakiness of the metallothioneine promoter. Addition of Cu²⁺ resulted in an approximately eightfold increase in E-cadherin levels (Fig. 1A). Expression of E-cadherin affected the level of endogenous Arm. Arm levels were low in untransfected cl-8 cells, increased ∼13-fold in cl8mEcad cells, and approximately threefold following induction of high E-cadherin expression with Cu²⁺. A slight increase in α-catenin levels was observed (a twofold difference between cl-8 and cl8mEcad cells; Fig. 1A). Both the hyperphosphorylated, slower migrating form (henceforth called phosphorylated) and the hypophosphorylated, faster migrating form of Arm (Peifer et al., 1994a) were elevated as a consequence of E-cadherin expression. This is in contrast to the increase in Arm after stimulation with Wg, in which only the hypophosphorylated form of Arm accumulates (Peifer et al., 1994a; Van Leeuwen et al., 1994). Northern Blot analysis showed that the increase in Arm protein levels was not caused by a change in the steady state levels of Arm mRNA (Fig. 1B); thus the increase in Arm is presumably the result of post-translational protein stabilization.

Expression of E-cadherin in cl-8 cells not only affected the levels of Arm and α-catenin proteins but also their subcellular localization. In untransfected cl-8 cells, Arm was about equally abundant in the cytosolic supernatant fraction and in a Triton X-100-insoluble pellet fraction, whereas α-catenin was present mostly in the cytosolic fraction (Fig. 1C). In cl8mEcad cells, almost all Arm and α-catenin were present in the insoluble fraction (Fig. 1C). This redistribution was caused by the formation of a protein complex containing E-cadherin, Arm and α-catenin, which could be immunoprecipitated from Triton X-100 lysates of cl8mEcad cells with antibodies against each of these three proteins (data not shown). cl8mEcad cells treated with Cu²⁺, to further increase levels of E-cadherin, showed a significant amount of Arm and α-catenin in the Triton X-100 insoluble pellet fraction, indicating a stable association of the cadherin-catenin complex with the cytoskeleton (Fig. 1C).

In untransfected cl-8 cells, Arm was hardly detectable by immunofluorescence microscopy and did not show any plasma membrane localization (Fig. 2). In cl8mEcad cells, by contrast, Arm was highly concentrated at cell-cell contact sites together with E-cadherin (Fig. 2). α-catenin accumulated at similar cell-cell contact sites (Fig. 2). These data indicate that expression of mouse E-cadherin in cl-8 cells initiates the formation of a functional cadherin-catenin complex at cell-cell contact sites.

We have shown previously that Wg signaling in cl-8 cells can be activated either by treatment with Wg-conditioned medium, or by transfection of wg^IL114, a temperature-sensitive allele (Van Leeuwen et al., 1994), or by overexpression of Dsh, an intracellular component of the Wg signaling pathway (Yanagawa et al., 1995). We generated stable cell lines that expressed Cu²⁺-inducible E-cadherin, combined with either wg^IL114 with a constitutive promoter (cl8Wg^ts/mEcad) or Dsh cDNA under control of the hsp70 heat shock promoter (cl8HSDsh/mEcad). Wingless signaling could then be activated by changing the temperature from the restrictive temperature (25°C) to the permissive temperature (16°C) in the case of the cl8Wg^ts/mEcad cells, or by a 30-minute heat shock at 37°C in the case of the cl8HSDsh/mEcad cells. In addition, the level of E-cadherin expression was controlled independently by addition of Cu²⁺ to the medium.

We first examined the distribution of Arm by cell fractionation. The cells were lysed in hypotonic buffer, followed by centrifugation, giving rise to a cytosolic supernatant fraction and a pellet fraction containing membranes, membrane-associated proteins and nuclei. In uninduced cl8Wg^ts/mEcad cells (Fig. 3, lanes 1 and 2) and cl8HSDsh/mEcad cells (Fig. 3, lanes 9 and 10) Arm was present almost exclusively in the pellet fraction and was absent from the cytosolic fraction, just as in cl8mEcad cells. Elevation of E-cadherin expression by addition of Cu²⁺ led to a further increase in the level of membrane-associated Arm, and to the accumulation of E-cadherin and Arm in the cytosolic fraction (Fig. 3, lanes 3, 4, 11 and 12). In the absence of Wg, the distribution of Arm was very similar to the distribution of E-cadherin, suggesting that most of the Arm was associated with E-cadherin. As in cl8mEcad cells, expression of E-cadherin led to the accumulation of two forms of Arm, as a result of a difference in phosphorylation on serine and threonine residues, as shown previously (Peifer et al., 1994a).

When we activated the Wingless pathway, either by incubation of the cl8Wg^ts/mEcad cells at 16°C (Fig. 3, lanes 5-8) or by a 30-minute heat shock of the cl8HSDsh/mEcad cells (Fig. 3, lanes 13-16), the distribution of Arm changed significantly. We found hypophosphorylated Arm in the cytosolic fraction (lanes 5 and 13), which did not contain detectable levels of E-cadherin, suggesting that this cytosolic, hypophosphorylated form of Arm was not associated with E-cadherin. Activation of Wg with concomitant induction of E-cadherin led to reduced levels of Arm in the cytosolic fraction (Fig. 3, lanes 7 and 15), indicating that E-cadherin competes with Wg signaling on the localization of Arm. Strikingly, activation of the Wg pathway not only led to an increase in the cytosolic pool of Arm, but also to a decrease in the membrane-associated pool of Arm (Fig. 3, lanes 6, 8, 14 and 16). Moreover, this reduction of Arm in the membrane fractions was accompanied by a reduction in the level of E-cadherin in those fractions. The amount of overexpressed E-cadherin in the cytosolic fraction was also reduced significantly at the permissive temperature for the wg^ts allele (Fig. 3, lanes 3 and 7), but less so upon overexpression of Dsh (Fig. 3, lanes 11 and 15).

To examine the effect of Wingless signaling on the cadherin-catenin complex in more detail, we analyzed the fractionation profiles of E-cadherin, Arm and α-catenin by FPLC gel filtration (Fig. 4). In cl8mEcad cells expressing baseline levels of E-cadherin in the absence of Cu²⁺ (Fig. 4A), Arm co-fractionated with E-cadherin, indicating that the majority of Arm was complexed with E-cadherin and that the pool of Arm not bound to cadherin was very small. For comparison, monomeric Arm expressed in a baculovirus system peaked in fractions 25-26, as expected for a protein of 105 kDa (data not shown). α-catenin had a very similar elution profile to Arm in cl8mEcad cells, with more than 90% of the total amount of α-catenin present in the same fractions as Arm and E-cadherin. To analyze the fractionation profile of Arm and α-catenin in cells with an activated Wingless pathway in the absence of cadherin, we used cl8Dshmyc cells (Yanagawa et al., 1995). In cl8Dshmyc cells induced to overexpress Dsh, Arm was exclusively found in low molecular mass fractions (Fig. 4B). A high percentage of α-catenin was found in the same fractions that contained Arm. Since cl8Dshmyc cells were not transfected with E-cadherin, the fractionation profiles of Arm and α-catenin reflect the fractionation profiles of pools of these proteins not complexed with E-cadherin.

We next examined the fractionation profiles of these proteins in E-cadherin-expressing cells in which the Wg pathway was activated. In cl8HSDsh/mEcad cells induced to overexpress Dsh (Fig. 4C), the fractionation profiles of Arm and α-catenin appeared as a combination of the profiles found in cl8mEcad cells and in cl8Dshmyc cells. The profile of E-cadherin remained similar to that from cl8mEcad cells (see Fig. 4A); note that both Arm and α-catenin fractionated into high and low molecular mass fractions. Thus, despite the presence of E-cadherin, a significant amount of Arm and α-catenin was not bound to E-cadherin upon activation of the Wg pathway.

To complement our cell fractionation and FPLC data, we examined the subcellular localization of Wg, Arm and E-cadherin in cl8Wg^ts/mEcad cells by triple label immunofluorescence staining (Fig. 5). At 25°C, the restrictive temperature for Wg^ts protein, Wg staining was diffuse and perinuclear, suggesting that the mutant protein was incorrectly folded in the ER (Fig. 5A). Arm and E-cadherin colocalized at cell-cell contacts and in intracellular vesicles (Fig. 5A), as in cl8mEcad cells (Fig. 2). At 16°C, the permissive temperature, Wg^ts protein exhibited a dot-like staining pattern in cells (Fig. 5B), very similar to that of wild-type Wg in Drosophila embryos (González et al., 1991; Van den Heuvel et al., 1989).

Activation of Wg led to a redistribution of Arm and E-cadherin. Arm was not concentrated at cell-cell contact sites but was enriched at the tips of cellular processes that appeared to be in close contact with the substratum. E-cadherin was almost completely absent from the cell surface and was found in a punctate, perinuclear distribution. There was little overlap between Arm and E-cadherin staining (Fig. 5B, merge). In general, the morphology of the cells was strikingly different at 16°C compared with 25°C (Fig. 5C,D). Cell-cell contacts appeared to be much less developed and cell-substratum contacts were much more prominent in cells with activated Wg (Fig. 5D). Very similar results were obtained with the cl8HSDsh/mEcad cells (data not shown).

To address the significance of our findings in cl-8 cells in vivo, we analyzed the subcellular localization of Armadillo and of endogenous DE-cadherin in Drosophila wing imaginal discs. In wild-type wing discs and in discs that overexpressed Dishevelled under control of the patched promoter, Armadillo and DE-cadherin were mainly cytoplasmic in cells that were exposed to high levels of Wingless (Fig. S1 in supplementary material) or overexpressed Dishevelled (Fig. S2 in supplementary material).

We examined whether loss of cell surface staining of E-cadherin upon Wg activation was due to increased turnover or degradation of E-cadherin (Fig. 6). cl8HSDsh/mEcad cells were heat shocked for 30 minutes in the presence of Cu²⁺ to induce overexpression of Dsh and E-cadherin. Cells were then metabolically labeled for 2 hours, followed by a chase for different times up to 8 hours. cl8mEcad cells were treated in the same way and served as a control for the stability of E-cadherin in the absence of Wg. E-cadherin is synthesized as a 135 kDa precursor polypeptide, which is processed to the mature protein of 120 kDa prior to its arrival at the cell surface (Shore and Nelson, 1991). We found that the amount of the 135 kDa precursor polypeptide was approximately the same in cl8mEcad and cl8HSDsh/mEcad cells (Fig. 6, 0 hour chase, lanes 1 and 6). During the chase, conversion of the 135 kDa precursor to the mature 120 kDa protein was clearly detected in the cl8mEcad cells (Fig. 6, lanes 2-5); note that a considerable amount of 120 kDa E-cadherin was detectable 8 hours after the beginning of the chase. By contrast, in the cl8HSDsh/mEcad cells, little of the 120 kDa mature protein was detected (Fig. 6, lanes 7-10). This result indicates that overexpression of Dsh may either lead to rapid degradation of newly synthesized E-cadherin before its arrival at the cell surface, or to degradation of mature E-cadherin after endocytosis from the plasma membrane. This result is in agreement with our immunofluorescence stainings, which showed staining of E-cadherin in a perinuclear region, but not on the cell surface of heat shocked HSDsh/mEcad cells (data not shown).

Next, we tested whether activation of Wg signaling had an effect on the expression of the endogenous DE-cadherin gene in cl-8 cells. As mentioned earlier, untransfected cl-8 cells express low levels of DE-cadherin. We found that prolonged overexpression of Dsh significantly increased the expression of DE-cadherin protein and led to the concomitant appearance of a phosphorylated form of Arm (Fig. 7A). Northern blot analysis showed that DE-cadherin protein expression was correlated with an increase in the steady state levels of DE-cadherin mRNA (Fig. 7B).

To examine the subcellular localization of DE-cadherin and Arm in cl8Dshmyc cells, we performed double label immunofluorescence staining. In cl8Dshmyc cells not treated with Cu²⁺, staining for DE-cadherin and Arm was very weak (Fig. 8). We found that a few individual cells showed high levels of Dsh expression in the absence of Cu²⁺; some of these cells were also positive for DE-cadherin (data not shown). After induction with Cu²⁺ for 24 hours, many, but not all cells showed intense staining for DE-cadherin at the plasma membrane (Fig. 8). We also found many cells that expressed high levels of Arm in cytoplasm and nuclei (Fig. 8). Remarkably, there were very few cells that expressed high levels of cytoplasmic Arm and high levels of DE-cadherin at the same time (Fig. 8, merge). We did not observe prominent colocalization of DE-cadherin and Arm at the plasma membrane in the presence of high levels of Dsh (Fig. 8, merge).

Over the past years, abundant evidence has implicated β-catenin in two different cellular processes. β-catenin binds to the cytoplasmic domain of cadherins and plays an essential role in the structural organization and function of cadherin-mediated cell-cell contacts by linking cadherins to the cytoskeleton (Gumbiner, 1996; Jamora and Fuchs, 2002). At the same time, β-catenin is the main transcriptional effector of the Wnt signal, which leads to an accumulation of β-catenin protein and its nuclear activity as a transcription factor together with T-cell factor (TCF) (Willert and Nusse, 1998; Wodarz and Nusse, 1998). Are these two functions related to each other? In the context of Wnt signaling, cadherins may limit the availability of cytosolic β-catenin by sequestering it to the cell surface and making it unavailable for transfer to the nucleus (Heasman et al., 1994; Sanson et al., 1996). Moreover, tyrosine phosphorylation of β-catenin and its resulting binding to Bcl9-2 leads to loss of adherens junctional β-catenin and increased TCF-mediated gene transcription (Brembeck et al., 2004; Danilkovitch-Miagkova et al., 2001). But beyond these examples of the functional link between β-catenin's dual tasks (Nelson and Nusse, 2004), there is the question whether Wnt signaling influences the adhesive properties of cells.

We show in this paper that this can indeed occur, in a cell culture system in which Wg signaling can be controlled and its response be measured over time. We found two contrasting effects: Wg initially lowers the amount of the cadherin-catenin complex at cell-cell contacts, most probably by reducing the Arm pool interacting with E-cadherin, but later, Wg signaling leads to elevated DE-cadherin transcription. This later increase in DE cadherin may titrate the pool of Arm available for Wg signaling and therefore attenuate the transcriptional response to Wg. We have shown by metabolic labeling experiments that the reduction in E-cadherin levels was caused by a posttranscriptional mechanism that affected the stability of the E-cadherin protein. Although levels of E-cadherin precursors were similar in the absence or presence of Wg signaling, very little mature E-cadherin was detectable in cells with an activated Wg pathway, indicating that the majority of E-cadherin was either degraded before arrival at the cell surface or removed from the cell surface by rapid endocytosis, followed by degradation (Bryant and Stow, 2004). Our biochemical data are consistent with immunofluorescent stainings of these cells, which showed very little E-cadherin on the cell surface but strong staining in a perinuclear region. Since we have not determined whether this perinuclear region corresponds to the ER or to an endocytic compartment, we cannot decide at present at what step E-cadherin is degraded in response to Wg signaling. Loss of E-cadherin from the cell surface and increased staining in the cytosol was also observed in cells exposed to high levels of Wg or Dsh in intact imaginal discs, demonstrating the physiological relevance of our tissue culture results. Although we consider it very probable that the removal of the cadherin-catenin complex from cell-cell contacts causes reduced cell-cell adhesion and is the main cause of the cell shape changes that we observed, increased cell-substratum adhesion in response to Wg signaling could also contribute to the altered cellular phenotype.

What could be the reason for the increased degradation of E-cadherin in cells with an activated Wg pathway? We suggest that the stabilization of Arm upon E-cadherin expression is most probably caused by the formation of a complex between Arm and E-cadherin, which protects Arm from proteolytic degradation. The same might be true for E-cadherin itself. Since Wg signaling causes a shift in the subcellular localization of Arm from the membrane fraction to the cytosolic fraction, it is possible that Wg signaling generally reduces the affinity of Arm for E-cadherin. This would impair complex formation between E-cadherin and Arm, leading to the appearance of uncomplexed E-cadherin, which might be subject to degradation. Interestingly, we observed a stronger reduction of E-cadherin in the cytosolic fraction when the wg^ts allele was activated than when Dsh was overexpressed, indicating that the cytosolic pool of E-cadherin may react differently to these two modes of activating Wg signaling.

This scenario raises the question of how Wg signaling could alter the affinity of Arm for E-cadherin. The most obvious possibility is an alteration of the phosphorylation status of Arm. E-cadherin expression leads to the stabilization of both phosphorylated and hypophosphorylated Arm in the membrane fraction, whereas stimulation by Wg or Dsh overexpression leads exclusively to the stabilization of hypophosphorylated Arm in the cytosolic fraction. Zeste-white 3 (Zw3), a serine/threonine protein kinase in the Wg pathway that is genetically upstream of Arm (Peifer et al., 1994b; Siegfried et al., 1994) is involved in the regulation of Arm phosphorylation. It is not clear, however, whether the phosphorylation status of Arm is related to its subcellular localization. In this context it is worth mentioning that the intracellular region of E-cadherin required for binding of β-catenin is highly phosphorylated on serine and threonine residues, and mutation of these residues to alanines abolishes binding (Stappert and Kemler, 1994). It is therefore possible that not only the phosphorylation status of Arm affects complex formation between Arm and E-cadherin, but also the phosphorylation status of E-cadherin.

We suppose that the early effects of Wg signaling on Arm and E-cadherin occur primarily at the posttranscriptional level, although we cannot rule out the possibility that the transcriptional regulation of specific target genes of Wg may be involved in regulating cadherin levels at the plasma membrane. We found that the endogenous DE-cadherin gene was transcriptionally activated in cl8Dshmyc cells only after prolonged overexpression of Dsh. Similar observations had been reported earlier (Yanagawa et al., 1997). Interestingly, in mouse hair follicles, Wnt signaling changes E-cadherin gene expression in the opposite manner, as it lowers E-cadherin transcription (Jamora et al., 2003). Such cell type-specific responses in gene expression are very common and underscore the range of effects that Wnt signals exert on cells.

The induction of DE-cadherin expression correlated with the appearance of a phosphorylated form of Arm, demonstrating again that expression of cadherins stabilizes phosphorylated Arm. The time course of DE-cadherin accumulation showed that it occurred considerably later than the initial effect on mouse E-cadherin protein stability. Nonetheless it might serve an important function during Wg signaling. Firstly, upregulation of DE-cadherin most probably leads to reestablishment of stable cell-cell contacts and secondly, it might be a means to attenuate the Wg signal by titration of the uncomplexed, cytosolic Arm. We observed that after prolonged overexpression of Dsh, cells expressed either high levels of cytosolic Arm or membrane-associated DE-cadherin, but not both at the same time, suggesting that DE-cadherin expression antagonizes the cytoplasmic accumulation of Arm. Strikingly, we did not observe significant colocalization of Arm and DE-cadherin at the plasma membrane of cl8Dshmyc cells overexpressing Dsh, supporting our hypothesis that the interaction between Arm and cadherins is less stable in cells with an activated Wg pathway. Two earlier reports found increased cadherin-mediated cell adhesion and increased cadherin levels in cells constitutively overexpressing Wnt1 (Bradley et al., 1993; Hinck et al., 1994). Since these results are very similar to our observations in cl8Dshmyc cells with prolonged overexpression of Dsh, we suggest that upregulation of cadherin and increased cell adhesion are late, adaptive responses to Wnt signaling.

How do these results relate to the known functions of the Wnt signaling pathway and to the phenotypes of mutations in wg during embryonic and larval development? It was recently reported that Wnt7a overexpression in the neural tube results in defective adherens junctions, with subsequent abnormalities in spinal neurulation (Shariatmadari et al., 2005). During zebrafish gastrulation, Wnt11 is required to modulate the cell surface localization of E-cadherin via a pathway involving the small GTPase Rab5, which regulates endocytosis (Ulrich et al., 2005). Signaling by another important growth factor TGFβ₁, leads to removal of E-cadherin from cell contacts, presumably by affecting the phosphorylation status of β-catenin (Vogelmann et al., 2005). It will be important to show at the biochemical level whether these changes in cell behavior are caused by the modulation of binding affinities between β-catenin and E-cadherin in response to signaling by secreted growth factors.

In Drosophila, Wg controls a large number of cell fate decisions, but Wg also plays an important role in more general processes, in particular mitosis, delamination and the formation of invaginations. Wg mutants have a reduction in the number of post-blastoderm mitoses in epidermis (Hartenstein et al., 1994). Also, delamination of a subset of neuroblasts is impaired, although cell fate of these neuroblasts appears to be unaltered (Hartenstein et al., 1994). Mitosis and delamination in the Drosophila ectoderm are correlated with prominent cell shape changes. Both dividing and delaminating epithelial cells change from a columnar to round shape, which requires a loosening of cell-cell contacts and is accompanied by a partial breakdown of the actin cytoskeleton and de-polymerization of microtubules (Hartenstein et al., 1994). Ultrastructural examination of adherens junctions in cells undergoing mitosis showed a reduction in the thickness of the junctional plaque, indicating partial dissociation of the protein complex constituting the adherens junction (Tepass and Hartenstein, 1994). Our findings point to a role for Wg in the transient reduction of cell-cell adhesion, which appears to be a prerequisite for both cell division and delamination. The transient nature of this effect should be emphasized here, since maintenance of the epithelial tissue structure requires the reestablishment of strong cell adhesion after division or delamination has occurred.

Data from Drosophila and vertebrates show that Wnts may regulate cell-cell junctions during morphogenetic movements in development. Based on our results, the Wnt-induced, transient weakening of cell adhesion might play an important permissive role in enabling the cells to translate changes in their transcriptional program into cell shape changes, movement or division.

The full-length mouse E-cadherin cDNA construct pSUM (Stappert and Kemler, 1994) was cut with XbaI and HindIII, and the resulting fragment filled in with Klenow and inserted into the EcoRV site of the vector pMK33 (Koelle et al., 1991), which drives expression of the insert from the Drosophila metallothioneine promoter. The construct carrying the temperature-sensitive wg allele wg^IL114 has been described previously (Van Leeuwen et al., 1994). Although this construct is also in the pMK33 vector, its expression is constitutive and not inducible by Cu²⁺, probably because of a cis-effect of the wg DNA on the metallothioneine promoter. The full-length Dsh construct in pMK33 has been described previously (Yanagawa et al., 1995). To allow expression of Dsh under control of the hsp70 heat shock promoter, the full-length Dsh-Myc construct in Bluescript (Yanagawa et al., 1995) was first cut with SalI, the resulting 5′ overhang was filled in with Klenow and the fragment was then cut with XbaI. The resulting fragment was inserted between the XbaI and StuI sites of pCaSpeR-hs.

cl-8 cells were maintained and transfected as described previously (Van Leeuwen et al., 1994; Yanagawa et al., 1995). Expression of mouse E-cadherin and Dsh from the metallothionein promoter was induced by addition of CuSO₄ to a final concentration of 1 mM, usually for 6 hours, unless otherwise indicated. Induction of Dsh expression from the hsp70 heat shock promoter was induced by a 30-minute heat shock at 37°C, followed by incubation at 25°C for 5.5 hours. The temperature-sensitive Wg protein was activated by incubation of the cells at 16°C for 6 hours. In general, cells that were analyzed under different induction conditions were taken from the same flask prior to the experiment.

Cells were grown to 80% confluency, washed in PBS and lysed in TNT buffer (50 mM Tris-HCl pH 8.0, 150 mM NaCl, 1% Triton X-100) supplemented with protease and phosphatase inhibitors (0.5 mM Pefabloc, 0.02 mg/ml leupeptin, 1 mM Na₃VO₄, 0.5 μM microcystin) on ice for 15 minutes. 50 μg of total protein were loaded in each lane of a 7.5% SDS-polyacrylamide gel. Western immunoblot analysis was performed as described previously (Van Leeuwen et al., 1994). The following primary antibodies were used: rabbit polyclonal anti-mouse E-cadherin (Marrs et al., 1993) diluted 1:10,000; mouse monoclonal anti-Arm N27A1 (Peifer, 1993) diluted 1:1,000; rat monoclonal anti-α-catenin DCAT-1 (Oda et al., 1993) diluted 1:1,000; rat monoclonal anti-DE-cadherin DCAD-2 (Oda et al., 1994) diluted 1:50; rat polyclonal anti-Dsh region 1 (Yanagawa et al., 1995) diluted 1:1,000. Quantification of western blots treated with ECL reagent (Amersham) was performed by densitometric scanning of exposed films.

The cell fractionation experiment shown in Fig. 1C was performed exactly as described previously (Yanagawa et al., 1995). For the fractionation experiment shown in Fig. 3, a simplified procedure was applied: cells were washed in PBS and lysed in hypotonic buffer (10 mM Tris-HCl pH 7.4, 0.2 mM MgCl₂) supplemented with protease and phosphatase inhibitors (0.5 mM Pefabloc, 0.02 mg/ml leupeptin, 1 mM Na₃VO₄, 0.5 μM microcystin) on ice for 10 minutes. The cells were disrupted using an Eppendorf Dounce homogenizer and 0.25 volumes of 1.25 M sucrose and 5 mM EDTA were added to the lysate. The lysate was centrifuged for 10 minutes in a microfuge at 20,000 g. The supernatant of this centrifugation is the cytosolic fraction and the pellet contains nuclei, membranes and membrane-associated proteins. Equivalent amounts of the cytosolic fraction and the pellet fraction were loaded on a 7.5% SDS-polyacrylamide gel and processed for western blotting as described above.

Cells were extracted on ice for 10 minutes in MEBC buffer (50 mM Tris-HCl pH 7.5, 100 mM NaCl, 0.5% Nonidet P-40) containing 0.5 mM Pefabloc, 0.1 mM Na₃VO₄, 50 mM NaF, and 0.01 mg each of chymostatin, leupeptin, antipain and pepstatin A per ml. Extracts were centrifuged at 12,000 g for 15 minutes, the resulting supernatant was cleared by centrifugation at 100,000 g and then passed through a 0.22 μm filter (Millipore). 200 μl of cleared extract were separated by FPLC (Pharmacia) on a Superose 6 HR column equilibrated in MEBC buffer containing 0.5 mM dithiothreitol and 0.1 mM Pefabloc. 0.5 ml fractions were collected and equal amounts of each fraction were subjected to western blot analysis as described above.

Cells were grown on poly-L-lysine-coated slides, fixed in 3.5% paraformaldehyde/CGBS (55 mM NaCl, 40 mM KCl, 15 mM MgSO₄, 5 mM CaCl₂, 10 mM tricine pH 6.9, 20 mM glucose, 50 mM sucrose) for 15 minutes, permeabilized in 0.3% Triton X-100/TBSC (50 mM Tris-HCl pH 7.5, 150 mM NaCl, 1 mM CaCl₂) for 15 minutes and then incubated with primary antibodies in TBSC/5% normal donkey serum for 2 hours at room temperature. Imaginal discs were fixed in 4% formaldehyde and stained according to standard procedures. The following antibodies were used: rat monoclonal anti-mouse E-cadherin DECMA-1 (Vestweber and Kemler, 1985) diluted 1:20; mouse monoclonal anti-Arm N27A1 (Peifer, 1993) diluted 1:4; rat monoclonal anti-α-catenin DCAT-1 (Oda et al., 1993), diluted 1:50; rabbit polyclonal anti-Wg (Van den Heuvel et al., 1989), affinity purified, diluted 1:20; rat monoclonal anti-DE-cadherin DCAD-2 (Oda et al., 1994), diluted 1:20. After several washes in TBSC, cells were incubated with secondary antibodies in TBSC/5% normal donkey serum. All secondary antibodies (Jackson Immunoresearch) had been preadsorbed by the manufacturer against immunoglobulins from a variety of species to avoid crossreactivity. For triple stainings the following secondary antibodies were used: Cy3-conjugated donkey anti-rabbit, diluted 1:400; FITC-conjugated donkey anti-mouse, diluted 1:200; Cy5-conjugated donkey anti-rat, diluted 1:200. Cells were washed in TBSC, mounted in Vectashield mounting medium (Vector) and viewed on a BioRad MRC 1000 confocal microscope attached to a Zeiss Axioscope.

Cells were grown to 80% confluency and expression of mouse E-cadherin was induced with 1 mM CuSO₄ for 30 minutes. At the same time, cells were heat shocked at 37°C for 30 minutes to induce overexpression of Dsh in the cl8HSDsh/mEcad cells (the cl8mEcad cells were also heat shocked as a control). Immediately after heat shock, cells were starved in serum-free methionine-free medium in the presence of 1 mM CuSO₄ for 20 minutes. Cells were labeled in serum-free methionine-free medium containing 100 μCi/ml [³⁵S]methionine and 1 mM CuSO₄ for 2 hours, followed by a chase in complete medium with an excess of unlabeled methionine and without CuSO₄. At different time points during the chase, cells were harvested and immediately frozen at -80°C until samples from all time points had been collected. Subsequent steps were performed on ice or at 4°C. Cells were lysed in TNT buffer (50 mM Tris-HCl pH 8.0, 150 mM NaCl, 1% Triton X-100) supplemented with protease and phosphatase inhibitors (0.5 mM Pefabloc, 0.02 mg/ml leupeptin, 1 mM Na₃VO₄, 0.5 μM microcystin). The lysates were preadsorbed for 1 hour against protein A Sepharose and then incubated with polyclonal rabbit anti-mouse E-cadherin antibody (Marrs et al., 1993) on a rotator for 6 hours. Immunoprecipitates were collected with protein A Sepharose, washed four times in TNT buffer, boiled in SDS sample buffer and subjected to SDS-PAGE and autoradiography. Quantification of band intensities was performed by densitometric scanning.

Total RNA was isolated from subconfluent cells using RNAzol reagent (TEL-TEST, Friendswood, TX). Electrophoresis, transfer of RNA to nylon filters and radioactive labeling of DNA was performed according to standard procedures. A 415 bp XbaI-EcoRI fragment from the C terminus of the Arm cDNA (Riggleman et al., 1989) was used as a hybridization probe for Arm, and a 4 kb XbaI-BglII fragment of the DE-cadherin cDNA (Oda et al., 1994) was used as a probe for DE-cadherin.

These studies were supported by the Howard Hughes Medical Institute, a postdoctoral research fellowship from the Deutsche Forschungsgemeinschaft to A.W., and by NIH grant GM35527 to W.J.N.