Myosin VI and vinculin cooperate during the morphogenesis of cadherin cell–cell contacts in mammalian epithelial cells

Cell–cell contacts are dynamic structures. This is evident during the biogenesis of cultured epithelial monolayers, in which discontinuous early contacts are characteristically replaced by continuous, linear cell–cell adhesions as the epithelia mature (Adams et al., 1998; Vaezi et al., 2002). The maturation of contacts is often accompanied by the appearance of specialized apical epithelial junctions and the establishment of apical–basal polarity (Vaezi et al., 2002). This implies that the transition to coherent linear contacts may represent an important stage in epithelial biogenesis. Cadherin cell adhesion molecules are key features of cellular contacts at all stages of epithelial biogenesis (Adams et al., 1998), and changes in cadherin function are often invoked to account for the dynamic regulation of cell–cell interactions. However, the precise molecular mechanisms responsible for this are not yet thoroughly characterized.

Classic cadherins such as epithelial cadherin (E-cadherin) exert their morphogenetic impact in close cooperation with the actin cytoskeleton (Mege et al., 2006). Actin integrity is essential both for cadherin surface adhesiveness and for cells to make and maintain cadherin-based contacts. Moreover, the capacity of the actin cytoskeleton to support diverse mechanical events, notably surface protrusion, contractility, and anchorage, makes it an ideal partner for cadherin adhesion to mediate dynamic changes at cell–cell contacts. However, the precise molecular mechanisms responsible for this functional interaction are far from clear. Passive scaffolding of cadherin adhesion complexes onto cortical actin filaments, although conceptually simple, has not been empirically verified (Yamada et al., 2005). It is also not evident how any single molecular mechanism for cadherin–actin interactions can readily encompass the diverse functional states that may occur at contacts during epithelial morphogenesis and maturation. Instead, there is increasing evidence that cadherins may interact with a range of different actin effector machinery, including actin nucleators and cross-linkers (Mege et al., 2006).

Actin-based motors are also major determinants of cytoskeletal organization and function. Indeed, to date, three members of the myosin superfamily—myosins II, VII, and VI—have been implicated in cadherin biology. Conventional nonmuscle myosin II has commonly been observed at cell–cell contacts (Dawes-Hoang et al., 2005). Its impact is complex (Ivanov et al., 2005), but, in epithelial cells, myosin II can be recruited and activated in response to E-cadherin ligation, where it serves to promote adhesion and the local accumulation of cadherin at cell–cell contacts (Shewan et al., 2005). Myosin VII was reported to associate indirectly with E-cadherin through the transmembrane protein vezatin (Kussel-Andermann et al., 2000). Myosin VII is involved in dynamic adhesive events in Dictyostelium discoideum (Tuxworth et al., 2001), but its physiological role in mammalian cadherin adhesion remains to be elucidated.

A role for myosin VI in cadherin function was first identified in the egg chamber of Drosophila melanogaster. Geisbrecht and Montell (2002) demonstrated that myosin VI was necessary for border cell migration, a morphogenetic process that requires Drosophila E-cadherin (DE-cadherin). Moreover, myosin VI and armadillo (Drosophila β-catenin) coimmunoprecipitated from ovarian extracts, and myosin VI and DE-cadherin each stabilized the expression of the other protein. Thus, myosin VI and DE-cadherin appeared to be biochemically and functionally interdependent. Myosin VI is unusual among myosins, as it is a minus end–directed motor (Wells et al., 1999; Walker et al., 2000). Accordingly, it has often been thought to participate in vesicle trafficking, both in endocytosis, in which its direction of movement is postulated to promote the internal movement of cargo (Buss et al., 2002; Aschenbrenner et al., 2003; Hasson, 2003), and in exocytosis, especially protein sorting in the biosynthetic pathway (Au et al., 2007). Membrane anchorage coupled with minus end–directed movement may also be a mechanism for myosin VI to exert protrusive force at the cell surface (Geisbrecht and Montell, 2002). However, myosin VI shows limited processivity, and its stepping behavior in vitro is stalled by resistant force, causing it to remain bound to actin filaments for prolonged periods (Altman et al., 2004). Thus, it has been alternatively suggested that myosin VI might act as an actin-based anchor under certain circumstances (Altman et al., 2004). Despite these interesting possibilities, it remains unclear how myosin VI may contribute to the biological activity of DE-cadherin or what its role might be in mammalian cells.

We have now examined the contribution of myosin VI to E-cadherin function in mammalian epithelial cells. We report that myosin VI is recruited to E-cadherin adhesions at a late stage in the maturation of cultured epithelial monolayers. There, it supports cadherin adhesive strength, the morphological integrity of epithelial apical junctions, and the perijunctional actin cytoskeleton. We further demonstrate that myosin VI exerts this effect by recruiting and cooperating with the actin regulator vinculin.

We began by analyzing the subcellular localization of myosin VI and E-cadherin in MCF7 mammary epithelial cells, which underwent a characteristic morphological transition during culture. 24 h after plating, cells were confluent but displayed E-cadherin staining that was punctate and serrated at the cell–cell contacts, which then became consistently linear and continuous by 48 h (Fig. 1 a). We quantitated this by counting the number of discontinuities in the contacts, which decreased between 24 (14.01 ± 0.81 [±SEM]; n = 30) and 48 h (3.00 ± 0.32; n = 30). Strikingly, myosin VI was preferentially detected at cell–cell contacts after 48 h, when it coaccumulated with linear E-cadherin (Fig. 1 a, 48 h), whereas relatively little coaccumulated at 24 h. Preferential coaccumulation at these linear (cohesive) contacts was also detected in cells transiently expressing GFP-tagged porcine myosin VI (Fig. S2). Similarly, in MDCK monolayers, myosin VI became detectable only when cells formed extensive, continuous contacts (unpublished data), although this process was more rapid than in MCF7 cells.

The preferential localization of myosin VI in older, cohesive contacts was not caused by changes in the total cellular expression of this protein (Fig. 1 c), nor did it simply reflect increased local amounts of E-cadherin at contacts. Although fluorescence intensity at contacts of both E-cadherin and myosin VI increased with time (Fig. S3 a), ratiometric analysis revealed that 48-h-old cultures possessed substantially more myosin VI at contacts relative to E-cadherin than did 24-h-old cultures (Fig. 1 b). Therefore, the comparatively late recruitment of myosin VI to cadherin contacts occurred independently of changes in the levels of E-cadherin at those contacts.

We then used coimmunoprecipitation analysis to test whether myosin VI formed a biochemical complex with E-cadherin. Myosin VI was detected in E-cadherin immunoprecipitates, and E-cadherin was identified in myosin VI immune complexes (Fig. 1 d), confirming a previous study of Drosophila (Geisbrecht and Montell, 2002). This interaction persisted when actin filament integrity was disrupted by 100 μM latrunculin A (Fig. 1 f). Moreover, a biochemical complex between these two proteins was only detected after 48 h of culture, which is coincident with the appearance of myosin VI at cadherin contacts (Fig. 1 e). In contrast, Mena and WAVE, which are other actin regulators found at cell–cell contacts (Scott et al., 2006; Yamazaki et al., 2007), did not coimmunoprecipitate with E-cadherin (unpublished data). These findings indicate that myosin VI is preferentially recruited into E-cadherin complexes as cells form linear, cohesive contacts with one another.

To test whether E-cadherin adhesion was responsible for recruiting myosin VI to cellular contacts, we treated 48-h-old MCF7 monolayers with a function-blocking anti–E-cadherin antibody (SHE78-7). As shown in Fig. 2 a, myosin VI staining was displaced from cell–cell contacts as early as 15 min after exposure to the antibody, before the integrity of the cell–cell contacts was overtly disrupted. This was confirmed by the measurement of myosin VI and E-cadherin fluorescence, which showed that myosin VI intensity fell rapidly relative to E-cadherin staining in cells treated with SHE78-7 antibody (Figs. 2 b and S3 b). Consistent with these immunofluorescence results, less myosin VI coimmunoprecipitated with E-cadherin 30 min after the addition of SHE78-7 antibody (Fig. 2 c). Together, these findings demonstrated that productive adhesion was necessary for E-cadherin to recruit myosin VI.

We then used RNAi to test the functional impact of myosin VI on E-cadherin. Two different siRNAs consistently reduced myosin VI expression by 80–100% for up to 7 d (Fig. 3 a). Because these duplexes yielded similar effects, data are only shown for one siRNA. Myosin VI protein expression was not affected by either scrambled oligonucleotides or a control duplex (Fig. 3 a, Cont) bearing a single nucleotide mismatch from the knockdown (KD) sequence, nor were β-tubulin or glyceraldehyde-3-phosphate dehydrogenase (GAPDH) levels affected by the siRNA (Fig. 3 a).

Depletion of myosin VI dramatically altered the morphology of cell–cell contacts. Whereas control cultures retained E-cadherin with a continuous linear distribution at the contacts, myosin VI KD cells showed punctate and serrated E-cadherin staining (Fig. 3 a) similar to that seen in 24-h-old control cultures before the appearance of myosin VI (Fig. 1). Quantitation confirmed an increased number of discontinuities in contacts from myosin VI KD cells (14.50 ± 0.77; n = 30) compared with control cells (2.60 ± 0.34; n = 30). Similarly, the transient expression of a dominant-negative myosin VI mutant lacking the motor head domain (p–myosin VI–tail-GFP) markedly disrupted cell morphology and cell–cell contacts (unpublished data). This suggested that myosin VI might critically influence the morphological transition of cell–cell contacts over time.

Consistent with this, the reconstitution of myosin VI using porcine myosin VI (p–myosin VI–GFP), which is resistant to KD (Fig. 3 b), rescued the phenotype of myosin VI KD cells. Myosin VI KD cells expressing p–myosin VI–GFP regained linear E-cadherin staining at their cell–cell contacts compared with the punctate E-cadherin distribution of untransfected neighboring cells (Fig. 3 b). However, the transient expression of p–myosin VI–tail-GFP failed to rescue the myosin VI KD phenotype (Fig. 3 b), indicating that the actin-binding motor head was necessary for myosin VI to affect junctional maturation.

We then directly tested the adhesive impact of myosin VI using laminar flow adhesion assays that measure the cellular resistance to detachment from substrata coated with recombinant cadherin ligand (hE/Fc). Myosin VI KD resulted in an ∼40% reduction in adhesive strength compared with cells transfected with control duplexes (Fig. 3 c). In contrast, integrin-based adhesion to fibronectin was unaffected by myosin VI depletion (Fig. S4), suggesting that myosin VI preferentially affected cadherin adhesion. Myosin VI KD did not affect the total or surface levels of E-cadherin (Fig. 3 d and Fig. S5, a and c), the cellular expression of catenins, or their binding to E-cadherin (Fig. S5, a and b).

Filamentous actin (F-actin), like E-cadherin, undergoes a distinct morphological transition as epithelial monolayers mature in culture. After 24 h, F-actin staining appeared relatively loosely organized in the perijunctional regions, whereas by 48 h, F-actin was densely accumulated at the cell–cell contacts (Fig. 4 a). This dense actin organization was lost in KD cells but not in control cells, most evidently at 48 h (Fig. 4 b). Expression of porcine myosin VI restored the dense perijunctional F-actin staining (Fig. 4 b). Interestingly, the total cellular levels of F-actin were not altered in KD cells (unpublished data), suggesting that myosin VI regulates the perijunctional organization of F-actin rather than affecting the global actin filament content of the cells.

E-cadherin adhesion is necessary for biogenesis of the specialized intercellular structures that comprise the apical junctional complex of epithelia (Gumbiner et al., 1988). Therefore, we next examined the impact of myosin VI depletion on the integrity of the junctional complex (Fig. 5). In control MCF7 monolayers, the tight junction marker ZO-1 displayed crisp, linear immunofluorescent staining throughout the apical-most regions of cell–cell contacts. In contrast, ZO-1 staining was discontinuous and fragmented in myosin VI KD monolayers. Similarly, staining for the desmosomal marker desmoplakin was distributed unevenly at contacts between myosin VI KD cells but was homogenous throughout contacts between control cells. However, despite this fragmentation, both ZO-1 and desmoplakin showed persistent staining in regions of contacts that retained E-cadherin (unpublished data), suggesting that the disruption of both tight junctions and desmosomes might reflect the loss of cohesive integrity between the cells.

In contrast, staining for vinculin, a marker of the zonula adherens (Geiger et al., 1980), was nearly completely lost from cell–cell contacts in myosin VI KD cells, even where E-cadherin persisted (Fig. 5). Vinculin also marks focal adhesions, but myosin VI depletion did not displace vinculin from these cell–matrix adhesive junctions, which is consistent with our observation that integrin-based adhesion was unaffected by myosin VI KD. Total cellular levels of vinculin were also not affected by myosin VI KD (Fig. S5 a), implying that myosin VI regulated the recruitment of vinculin to cadherin adhesions. As vinculin is reported to interact with the E-cadherin complex (Watabe-Uchida et al., 1998; Weiss et al., 1998), we investigated whether myosin VI participated in this biochemical interaction. Indeed, coimmunoprecipitation analysis identified an apparent complex containing E-cadherin, vinculin, and myosin VI in 48-h-old control MCF7 cells. Immune complexes generated with antibodies to any of these three proteins also contained the other two proteins (Fig. 6 a). However, the ability of vinculin to coimmunoprecipitate with E-cadherin was lost in myosin VI KD cells (Fig. 6 b). Therefore, myosin VI was necessary for vinculin to stably interact with E-cadherin in mature cell–cell contacts. The ability to bind both vinculin and E-cadherin resides in the myosin VI tail, as transiently expressed p–myosin VI–tail-GFP was able to coimmunoprecipitate these proteins (Fig. 6 c). Moreover, cross-linking by actin filaments cannot readily account for this interaction, as this myosin VI fragment lacks the actin-binding head domain.

Interestingly, vinculin-deficient F9 embryonal carcinoma cells were reported to show defects in cell–cell cohesion reminiscent of those we observed upon the depletion of myosin VI (Watabe-Uchida et al., 1998). This led us to postulate that vinculin might be responsible for the impact of myosin VI on cell–cell integrity. To test this, we first examined the impact of depleting vinculin itself (Fig. 7). siRNA effectively reduced cellular vinculin levels (Fig. 7 a), most comprehensively at cell–cell contacts (not depicted). Importantly, junctional continuity was disrupted by vinculin KD (Fig. 7 b), with E-cadherin staining becoming punctate and serrated (14.40 ± 0.81 discontinuities/contact; n = 30) compared with control transfections (2.77 ± 0.45 discontinuities/contact; n = 30). Thus, the impact of vinculin KD on E-cadherin organization resembled that of myosin VI KD. However, neither myosin VI levels (Fig. S5 d) nor the ability of myosin VI to coimmunoprecipitate with E-cadherin were affected in vinculin KD cells (Fig. 7 c), nor were total cellular levels of E-cadherin or catenins perturbed by vinculin KD (Fig. S5 d).

Then, we asked whether the restoration of vinculin activity might affect the myosin VI KD phenotype (Fig. 8). Because full-length vinculin readily adopts an autoinhibited conformation (Johnson and Craig, 1995), we used membrane-directed fragments of vinculin that restored tight junction integrity in F9 cells (Watabe-Uchida et al., 1998). We transiently expressed chimeric constructs bearing the head (α-cat/vinHead) or tail (α-cat/vinTail) region of vinculin (Fig. 8 a) fused to the β-catenin–binding domain of α-catenin to target them to cadherin contacts and not to focal adhesions. Both proteins consistently localized to cell–cell contacts (Fig. 8 b). Strikingly, we found that both of these transgenes also restored the continuous linear organization of E-cadherin at contacts between myosin VI KD cells (Fig. 8, b and c). The α-cat/vinHead construct rescued contact integrity somewhat more efficiently than did the α-cat/vinTail construct, but both rescued much more effectively than did full-length α-catenin alone (Fig. 8, b and c).

Furthermore, we reasoned that if vinculin acts downstream of myosin VI, vinculin would be necessary for myosin VI to support the cohesive integrity of cell–cell contacts. As KD and reconstitution provides a strategy to isolate the activity of a specific protein, we tested whether exogenous myosin VI could rescue E-cadherin organization when vinculin as well as endogenous myosin VI were depleted (Fig. 9 a). Importantly, the transient expression of porcine myosin VI–GFP was unable to rescue linear cadherin contacts in double vinculin/myosin VI KD cells (Fig. 9, a and c). This indicated that vinculin was necessary for the reconstitution of myosin VI to be effective. Finally, we tested whether myosin VI could compensate for the loss of vinculin and found that the overexpression of myosin VI did not restore the integrity of contacts in vinculin single KD cells (Fig. 9, b and c). Together, these findings identify vinculin as a critical effector for myosin VI at cohesive cell–cell contacts.

Cadherin-based cell–cell contacts undergo distinct morphological transitions both in vivo and in vitro, even after cells have established contact with one another. The challenge is to define specific molecular mechanisms responsible for the morphogenesis of these cell–cell interactions. Our current work identifies myosin VI as an important factor that interacts with cadherin adhesion to control the process by which early discontinuous interactions become organized into cohesive, linear contacts as epithelial monolayers mature in culture. We found that myosin VI binds to E-cadherin and localizes at cell–cell contacts as they undergo the transition from discontinous to coherent and continuous. This appears to reflect a regulated recruitment process, as cellular levels of myosin VI do not change during this period. Importantly, the disruption of myosin VI function by RNAi or expression of a dominant-negative mutant perturbed the cohesive integrity of cell–cell contacts and reduced cadherin adhesion. Moreover, myosin VI depletion perturbed the integrity of tight junctions and desmosomes, which is consistent with the central role for E-cadherin function in junctional biogenesis (Gumbiner et al., 1988). In contrast, myosin VI had no effect on integrin-based cell adhesion to fibronectin. Thus, in cultured mammalian cells, myosin VI acts at a relatively late stage in epithelial maturation to generate cohesive cell–cell interactions.

Our findings support and extend earlier evidence from Drosophila that implicated myosin VI in cadherin-based cell–cell interactions. Myosin VI cooperates with DE-cadherin to support morphogenetic movement in the fly egg chamber (Geisbrecht and Montell, 2002). Specifically, myosin VI is expressed and forms a complex with DE-cadherin and armadillo in migrating border cells. Importantly, myosin VI is necessary for this cadherin-dependent form of cell-on-cell migration, which is consistent with our finding that myosin VI supports cadherin adhesion and aspects of cellular morphogenesis in mammalian epithelia. Similarly, the disruption of myosin VI perturbed intercellular cohesion, DE-cadherin localization, and dorsal closure during early fly morphogenesis (Millo et al., 2004). Collectively, our findings in mammalian cells, taken with these earlier precedents in invertebrates, suggest an important conserved contribution of myosin VI to cadherin function.

How might myosin VI regulate cadherin contacts and adhesion in mammalian epithelia? This unconventional motor has been implicated in two broad cellular processes: membrane transport and actin filament organization. Myosin VI is recruited to clathrin-coated pits and persists on the subsequent uncoated vesicles, likely through interaction with several adaptor proteins (Buss et al., 2001; Aschenbrenner et al., 2003). This, taken with its minus end–directed movement, has suggested a role in endocytosis (Buss et al., 2002). Myosin VI also associates with the Golgi apparatus and can support exocytotic transport (Warner et al., 2003). Indeed, in MDCK cells, myosin VI was necessary for the basolateral delivery of proteins that contain tyrosine-based motifs sorted by the AP-1B clathrin adaptor complex (Au et al., 2007). However, myosin VI did not affect protein sorting that depended on dileucine motifs, including E-cadherin (Bryant and Stow, 2004). Similarly, we found no substantial change in either the total or surface levels of E-cadherin in myosin VI KD cells, nor could we detect changes in the transport of E-cadherin to the cell surface (unpublished data). Thus, although myosin VI may play a more subtle role in cadherin trafficking, this pathway does not readily account for our results.

Instead, we favor the notion that myosin VI participates in cadherin–actin cooperation. We found that the perijunctional actin cytoskeleton was clearly disrupted by myosin VI depletion. As epithelial monolayers matured, reorganization of the perijunctional actin cytoskeleton accompanied the appearance of linear, cohesive cadherin contacts, an initially loose distribution of F-actin being replaced by dense staining concentrated in the immediate vicinity of cell–cell contacts. In myosin VI KD cells, the dense perijunctional packing of F-actin was replaced by a looser organization, without any concomitant change in total cellular F-actin levels. Similarly, earlier studies reported that myosin VI can stabilize actin filament networks (Noguchi et al., 2006) and potentially filaments themselves (Naccache and Hasson, 2006), characteristically promoting the dense packing and accumulation of filaments. For example, during spermatid individualization in Drosophila, myosin VI is necessary to organize the actin cones that separate the syncytial spermatids. Notably, the density of filaments in the actin cones is substantially reduced in myosin VI mutant testes without apparent changes in filament turnover, suggesting that myosin VI participates in packing and organizing actin filament networks (Noguchi et al., 2006). The impact of myosin VI on organization of the perijunctional actin cytoskeleton implies that a similar contribution may occur in epithelial cells.

Importantly, several lines of evidence identify the actin-binding protein vinculin as a downstream effector for myosin VI at cadherin adhesions. (1) Myosin VI was necessary for vinculin to stably associate with E-cadherin in mature epithelial monolayers, which was assessed by both coimmunoprecipitation and immunofluorescence analysis. Vinculin was also selectively lost from cell–cell contacts but not focal adhesions in myosin VI KD cells. However, the ability of myosin VI to interact with E-cadherin was not affected by vinculin KD, implying that vinculin recruitment to E-cadherin is downstream of myosin VI. (2) Vinculin KD disrupted the integrity of cohesive E-cadherin contacts in a manner similar to myosin VI KD. (3) Membrane-targeted vinculin fragments effectively restored the cohesive integrity of cadherin contacts in myosin VI KD cells. This implied that reconstitution of vinculin function at E-cadherin adhesions could compensate for the global loss of myosin VI. (4) Vinculin was necessary for myosin VI to regulate junctional integrity. Exogenous myosin VI could not rescue contact integrity in cells depleted of vinculin as well as endogenous myosin VI, indicating that vinculin was necessary for the reconstitution of myosin VI to be effective. (5) Myosin VI could not compensate for the loss of vinculin, as the overexpression of myosin VI did not rescue contact integrity in vinculin KD cells. This argues against the possibility that vinculin and myosin VI are in parallel pathways. Collectively, these findings indicate that vinculin is an important effector for myosin VI at cadherin contacts.

Vinculin has long been recognized to accumulate at cell–cell contacts, where it is thought to incorporate into adherens junctions (Geiger et al., 1980). However, it is notable that previous studies implicated α-catenin in recruiting vinculin to cadherin adhesions (Watabe-Uchida et al., 1998; Weiss et al., 1998). We, too, found that vinculin stains in early cell–cell contacts (unpublished data) before myosin VI is readily detected, a process that therefore may entail α-catenin. Yet, in our experience, myosin VI was necessary for vinculin to stably incorporate into mature cadherin adhesions. This suggests that two mechanisms participate in localizing vinculin to cadherin adhesions, with myosin VI being dominant at a later stage than α-catenin.

Thus, our findings identify myosin VI and vinculin as part of a molecular apparatus responsible for generating the cohesive cell–cell contacts that distinguish epithelial biogenesis in vitro. We postulate that myosin VI and vinculin cooperate to reorganize the perijunctional actin cytoskeleton, leading to the generation of cohesive, linear cadherin contacts. Precisely how vinculin participates in this process has yet to be defined at a molecular level. Vinculin can both bind and bundle actin filaments (Jockusch and Isenberg, 1981; Menkel et al., 1994; Huttelmaier et al., 1997; Janssen et al., 2006), so may provide a mechanism to organize and compact actin filament meshworks at cadherin adhesions. The actin-binding site has been mapped to the tail region of the vinculin molecule (Johnson and Craig, 2000). Therefore, it was interesting to note that the head fragment could also rescue contact integrity. The mechanism for this effect remains to be determined.

However, our data further indicate that myosin VI does not act only by recruiting vinculin into a cadherin-based complex. Although the myosin VI tail alone can coimmunoprecipitate both E-cadherin and vinculin and localize to cell–cell contacts (unpublished data), this fragment did not rescue the cohesive integrity of those contacts in myosin VI KD cells. Therefore, the actin-binding activity of the myosin VI head domain must functionally cooperate with vinculin for myosin VI to support cohesive cell–cell contacts. The nature of this cooperation remains to be determined. Our data do not exclude the possibility that other proteins participate in this myosin VI–based effector pathway. One speculative possibility is that myosin VI may serve as an actin-based anchor (Altman et al., 2004), cooperating with vinculin's ability to organize actin in order to link cadherin complexes onto perijunctional actin filaments. Perhaps this cooperative interaction with actin also serves to stabilize vinculin at contacts, accounting for the role of myosin VI to stably localize vinculin in mature contacts.

Irrespective of the deep molecular mechanism, our findings highlight the concept that multiple actin regulators operate at cadherin adhesive contacts, and they reinforce the notion that the functional expression of individual mechanisms is likely to be tightly regulated by cellular context. Understanding the regulated recruitment of myosin VI will be important if we are to elucidate the varied molecular mechanisms that cadherin adhesion uses to regulate the actin cytoskeleton.

MCF-7 cells and CHO cells have been described previously (Goodwin et al., 2003; Helwani et al., 2004). For transient expression, cells were transfected with LipofectAMINE 2000 (Invitrogen) according to the manufacturer's instructions and analyzed 24–48 h after transfection. hE/Fc was prepared and used as previously described (Kovacs et al., 2002).

hE/Fc-based laminar flow adhesion assays were performed as described previously (Verma et al., 2004). Integrin-based adhesion assays were performed using longer term adhesion assays (Verma et al., 2004) with alterations to substrata coating. In brief, poly-l-lysine–coated six-well plates were incubated overnight at 4°C with 10 μg/ml fibronectin (Sigma-Aldrich) in PBS and were incubated with HBSS containing 5 mM CaCl₂ for 30 min at 37°C. Cells were isolated by incubation for 10 min in 0.01% (wt/vol) crystalline trypsin (Sigma-Aldrich) in HBSS containing 5 mM CaCl₂. Freshly isolated cells were allowed to attach to substrata for 90 min at 37°C in a CO₂ incubator and were subjected to detachment by systematic pipetting. For this, five regions in each well (the four quadrants and center) were washed three times with 200 μl HBSS/CaCl₂ delivered using a stand-mounted pipette. Cells remaining adherent to the wells were then incubated with 10 mg/ml MTT dissolved in dimethylsulfoxide and read at OD₅₉₅ in a microplate reader. Cellular content in wells after pipetting was compared with the cellular content of wells prepared under identical conditions but not subjected to pipetting (yielding the total number of cells plated in each well).

Porcine myosin VI–GFP and p–myosin VI–tail-GFP constructs were both gifts from T. Hasson (University of California, San Diego, La Jolla, CA; Aschenbrenner et al., 2003). Plasmids encoding α-cat/vincHead, α-cat/vincTail, and α-catenin–HA were gifts from M. Watabe-Uchida and M. Takeichi (Kyoto University, Kyoto, Japan; Watabe-Uchida et al., 1998). To construct α-catenin bearing a C-terminal T7 tag, the HA tag of pCA-huaEcat-HA-pA was replaced by cloning annealed oligonucleotides encoding the T7 tag into a unique Sal1 site. GFP-C1 control empty vector was purchased from CLONTECH Laboratories, Inc.

Primary antibodies used in this study are listed as follows: (1) a rabbit pAb (pAbMVI) directed against the whole tail domain of myosin VI was generated using a GST fusion protein (provided by F. Buss, University of Cambridge, Cambridge, UK; Buss et al., 1998). Approximately 250 μg of protein was used with Freud's adjuvant per injection to immunize rabbits. Immunoblotting cell lysates from several cell lines, including MCF-7 cells, with anti–myosin VI sera detected an intense band at the expected molecular mass of 140 kD as well as several other minor bands at lower molecular masses (that are possible degradation products), whereas naive rabbit sera from the same rabbit did not detect this band (Fig. S1)). Additionally, RNAi depletion of myosin VI resulted in loss of the band at 140 kD, and the reexpression of pMVIGFP restored this band at the expected molecular mass of 170 kD (Fig. 3 b). The other antibodies were obtained as follows: (2) mouse (mAb) HECD1 against human E-cadherin (a gift from P. Wheelock, University of Nebraska, Omaha, NE; with permission from M. Takeichi); (3) mouse mAb against GFP (Roche); (4) mouse mAb against human β-tubulin (Transduction Laboratories); (5) rabbit pAb against the ectodomain of E-cadherin (pAbE-cad; Helwani et al., 2004); (6) mouse mAb SHE78-7 against human E-cadherin (Zymed Laboratories); (7) mouse mAb against the cytoplasmic domain of E-cadherin (Transduction Laboratories); (8) rabbit pAb against human GAPDH (R&D Systems); (9) mouse mAb against human α-catenin (Transduction Laboratories); (10) mouse mAb against human p120-catenin (Transduction Laboratories); (11) rabbit pAb against human desmoplakin (a gift from K. Green, Northwestern University School of Medicine, Chicago, IL); (12) rabbit pAb against human ZO-1 (Zymed Laboratories); (13) mouse mAb against human vinculin (hVIN-1; Sigma-Aldrich); (14) mouse mAb against T7 tag (Novagen); and (15) rabbit pAb against GFP (Invitrogen). Secondary antibodies were species-specific antibodies conjugated with AlexaFluor350, 488, or 594 (Invitrogen) for immunofluorescence or were conjugated with HRP (Bio-Rad Laboratories) for immunoblotting.

To identify myosin VI, cells were incubated with prepermeabilization buffer (0.5% Triton X-100, 10 mM Pipes, pH 6.8, 50 mM NaCl, 3 mM MgCl₂, and 300 mM sucrose containing 1× complete protease inhibitors [Roche]) for 5 min on ice followed by fixation with 4% PFA in cytoskeletal stabilization buffer (100 mM KCl, 300 mM sucrose, 2 mM EGTA, 2 mM MgCl₂, and 10 mM Pipes, pH 7.2) for 10 min on ice. Otherwise, cells were fixed in PFA on ice for 10 min and were permeabilized. Fixed and stained specimens were mounted in either Mowiol (Calbiochem) or polyvinylalchohol mounting medium containing N-propylgallate (Sigma-Aldrich).

Epiillumination fluorescence microscopy of fixed specimens was performed at room temperature using a microscope (IX81; Olympus) with 100× or 60× 1.40 NA plan Apochromat objectives (Olympus) and imaged with cameras (Orca-1 ER; Hamamatsu) driven by MetaMorph imaging software (Universal Imaging Corp.). Confocal microscopy was performed at room temperature using a spinning disk confocal system (Ultra-View; PerkinElmer) mounted on a microscope (IX81; Olympus) with 100× or 60× 1.40 NA plan Apochromat objectives (Olympus) and a camera (Orca-1 ER; Hamamatsu) run with MetaMorph. Background correction and contrast adjustment of raw data images were performed with ImageJ (National Institutes of Health) or Photoshop (Adobe).

The intensity of E-cadherin and myosin VI fluorescence at cell–cell contacts was measured using the ImageJ tracing tool to identify contacts and quantitate the mean pixel intensity. Background was subtracted using a region with identical area for each contact quantitated. Experiments were performed four times with 40 contacts analyzed per experiment. To quantitate the differences in the continuity of contacts, the number of breaks within a contact were counted (contacts being defined as the regions between the vertices of two contiguous cells). Data shown are representative of four independent experiments.

MCF7 monolayers were incubated with function-blocking antibody SHE78-7 directed against the E-cadherin ectodomain at a 2-μg/ml concentration for 15–30 min. The surface expression of E-cadherin was measured by the sensitivity to surface trypsinization as described previously (Verma et al., 2004). Surface biotinylation was performed as previously described (Stehbens et al., 2006).

Cells were lysed in 1 ml of lysis buffer (50 mM Tris, pH 7.4, 2 mM CaCl₂, 150 mM NaCl, and 0.5% NP-40). Protein complexes were immunoprecipitated with anti–E-cadherin pAb, anti–myosin VI pAb, anti-GFP pAb, or a monoclonal vinculin mAb bound to protein A–agarose beads and were separated by SDS-PAGE. To coimmunoprecipitate α- and β-catenin with E-cadherin, the lysis buffer was 50 mM Tris, pH 7.4, 2 mM CaCl₂, 150 mM NaCl, and 1% NP-40; for p120-catenin, the buffer was 20 mM Hepes, 50 mM CaCl₂, 0.1 mM EDTA, and 1% NP-40.

MCF7 monolayers seeded to 25% confluency were transfected with validated human myosin VI RNAi duplexes (5′-GGUUUAGGUGUUAAUGAAGtt-3′) and control duplexes (5′-GGUUUAGGUGUGAAUGAAGtt-3′; Ambion) at a 50-nM concentration using LipofectAMINE 2000 (Invitrogen) according to the manufacturer's guidelines. Cells were then replated at 25% confluency on day 3 after transfection and were allowed to incubate overnight before retransfection with 50 nM RNAi duplexes to ensure sustained KD. Cells were studied 24–48 h after the second plating. Vinculin RNAi was performed similarly with 100 nM of validated human vinculin RNAi duplexes (5′-GGCAUAGAGGAAGCUUUAAtt-3′) and mismatched controls (5′-GGCAUAGACCTTGCUUUAAtt-3′; Ambion).

Fig. S1 shows a characterization of the myosin VI antibody. Fig. S2 depicts the localization of myosin VI–GFP at cell–cell contacts. Fig. S3 quantitates the fluorescence intensity of myosin VI and E-cadherin at cell–cell contacts as cultures mature or when treated with cadherin-blocking antibodies. Fig. S4 shows the effect of myosin VI RNAi on cell adhesion to fibronectin. Fig. S5 depicts the effects of myosin VI KD on the expression of cadherin, catenins and vinculin, the cadherin–catenin complex, and the surface expression of cadherin. It also shows the effect of vinculin KD on levels of cadherin, catenins, and myosin VI.

We thank the many colleagues who provided reagents for this project, especially Tama Hassan, Folma Buss, Honami Naora, Mitsuko Watanabe-Uchida, Masatoshi Takeichi, and Robert Kypta. Our colleagues in the laboratory have, as always, been unfailing in their support, advice, and encouragement; we particularly thank Nicole den Elzen for her punctilious reading of the manuscript.

This work was supported by the Human Frontiers Science Program. M.P. Maddugoda is a Dora Lush Research Scholar of the Australian National Health and Medical Research Council (NHMRC), and A.S. Yap is an NHMRC Senior Research Fellow. A.S. Yap is a Research Affiliate of the Australian Research Council Special Research Centre for Functional and Applied Genomics, which provided infrastructure support for this work. Confocal microscopy was performed at the Australian Cancer Research Foundation (ACRF)/Institute for Molecular Bioscience Dynamic Imaging Facility for Cancer Biology, which was established with the support of the ACRF.