α-catenin associates the cadherin–catenin complex with the actin cytoskeleton. α-catenin binds to β-catenin, which links it to the cadherin cytoplasmic tail, and F-actin, but also to a multitude of actin-associated proteins. These interactions suggest a highly complex cadherin–actin interface. Moreover, mammalian αE-catenin has been implicated in a cadherin-independent cytoplasmic function in Arp2/3-dependent actin regulation, and in cell signaling. The function and regulation of individual molecular interactions of α-catenin, in particular during development, are not well understood. We have generated mutations in Drosophila α-Catenin (α-Cat) to investigate α-Catenin function in this model, and to establish a setup for testing α-Catenin-related constructs in α-Cat-null mutant cells in vivo. Our analysis of α-Cat mutants in embryogenesis, imaginal discs and oogenesis reveals defects consistent with a loss of cadherin function. Compromising components of the Arp2/3 complex or its regulator SCAR ameliorate the α-Cat loss-of-function phenotype in embryos but not in ovaries, suggesting negative regulatory interactions between α-Catenin and the Arp2/3 complex in some tissues. We also show that the α-Cat mutant phenotype can be rescued by the expression of a DE-cadherin::α-Catenin fusion protein, which argues against an essential cytosolic, cadherin-independent role of Drosophila α-Catenin.
Adherens junctions (AJs) are complex signaling and adhesion centers that make many contributions to development and tissue homeostasis (Halbleib and Nelson, 2006; Harris and Tepass, 2010). AJs associate with and organize the actin cytoskeleton. The zonula adherens of epithelial cells links to a circumferential actin belt and to actin cables that connect perpendicular to AJs. Contractile forces in these actin belts and cables make crucial contributions to morphogenesis (Lecuit and Lenne, 2007; Martin, 2010). α-catenin is a core component of the cadherin–catenin complex and operates at the cadherin–actin interface (Kobielak and Fuchs 2004; Pokutta and Weis, 2007; Benjamin and Nelson, 2008). How α-catenin functions, and how it contributes to tissue development is an area of intense investigation. Here we report the isolation of α-Catenin (α-Cat)-null mutants in Drosophila and characterize their phenotypic consequences.
E-cadherin and other classic cadherins bind to β-catenin and p120catenin, and β-catenin in turn links to α-catenin. αE-catenin, one of three mammalian α-catenin proteins, can bind to actin filaments directly, suggesting that it links cadherin to actin (Rimm et al., 1995). αE-catenin is also a known binding partner of several F-actin-binding proteins such as EPLIN, formin-1, spectrin, vinculin, afadin, α-actinin, ZO-1, and merlin (Gladden et al., 2010; Harris and Tepass, 2010), multiplying the possible ways that α-catenin could link cadherin to actin. Supporting the idea that α-catenin acts as a physical linker is recent evidence that αE-catenin undergoes a conformational change in response to tension generated by actomyosin. This allows vinculin to interact with αE-catenin and, consequently, enhance the cadherin–actin link (Yonemura et al., 2010). These physical linkage models contrast with the allosteric regulation model (Yamada et al., 2005; Drees et al., 2005). This model suggests that αE-catenin binding to β-catenin recruits αE-catenin at the AJs to enrich its local concentration. After dissociation from β-catenin, αE-catenin dimerizes, allowing it to bind to F-actin to modulate actin organization. Competitive binding with the Arp2/3 complex could be a mechanism through which αE-catenin might impact on actin organization. Displacement of Arp2/3 interferes with actin network formation, as seen in lamellipodia, and promotes actin bundling as observed at AJs (Drees et al., 2005; Weis and Nelson, 2006). Extending this model, it was recently proposed that αE-catenin has an AJ-independent cytosolic function in competing with Arp2/3, and consequently suppressing actin-based protrusive activity (Benjamin et al., 2010).
α-catenin is highly conserved across animals and has recently also been reported in Dictyostelium (Dickinson et al., 2011). Loss of HMP-1, the C. elegans homologue of α-catenin, does not disrupt epithelial intercellular adhesion per se. However, similar to the loss of C. elegans HMR-1/cadherin, loss of HMP-1 causes defects in ventral closure and body elongation that result apparently from a compromised linkage of cadherin to actin (Costa et al., 1998; Kwiatkowski et al., 2010). Mice lacking αE-catenin show similar defects to E-cadherin mutants, with a developmental arrest at the blastocyst stage due to defects in trophectoderm integrity (Larue et al., 1996, Torres et al., 1997). Similarly, depletion of maternal α-catenin in Xenopus results in loss of intercellular adhesion at the blastula stage (Kofron et al., 1997). Conditional ablation of mouse αE-catenin in the developing skin or the neuroepithelium causes defects in AJ formation and abnormal cell signaling, leading to developmental defects such as over-proliferation of skin cells (Stepniak et al., 2009; Schlegelmilch et al., 2011; Silvis et al., 2011). Disruption or misexpression of αE-catenin or E-cadherin has also been linked to many forms of human epithelial cancers (Jeanes et al., 2008; Benjamin and Nelson, 2008).
In Drosophila, α-Catenin function has been assessed through RNA interference (RNAi), which does not fully deplete α-Catenin (Magie et al., 2002; Cavey et al., 2008; Seppa et al., 2008). These experiments showed that Drosophila α-Catenin is required for AJ function and integrity. Our analysis of α-Cat-null mutations extends the range of α-Catenin-associated functions in cadherin-based cell adhesion in Drosophila through analysis of α-Catenin function in late embryos, imaginal discs and ovaries. We present evidence arguing that all essential functions of α-Catenin occur at the membrane and do not require dissociation from the cadherin–catenin complex or a cadherin-independent cytosolic pool of α-Catenin.
Isolation of null mutations for Drosophila α-Cat
The Drosophila homologue of α-catenin was first identified in 1993, shown to localize to AJs, and to form a complex with Armadillo (Arm, Drosophila β-catenin) and DE-cadherin (DEcad) (Oda et al., 1993; Oda et al., 1994). Genome sequencing revealed that α-Cat localizes to the left arm of the third chromosome in close proximity to the centromere. This region was genetically poorly characterized, which hampered the isolation of α-Cat mutations for many years. A P element insertion became ultimately available from Genexel (GE30561) and was located 1271 bp upstream of the predicted α-Cat transcription start site. GE30561 is a homozygous viable insertion that did not express its white (w+) marker. However, mobilization of the P element generated offspring that expressed w+, suggesting that the insertion of the P element at a different genomic location had occurred. We concluded that the expression of w+ at the α-Cat locus is probably suppressed by the nearby centromeric heterochromatin. We subsequently examined 150 lines in which w+ was expressed after P element mobilization, and isolated three independent genomic deletions that remove part of the α-Cat gene. These lines were named α-Cat1, α-Cat2 and α-Cat3, respectively (Fig. 1A).
The three deletions range in size from approximately 2.0 kb to 2.5 kb, and all remove the first exon of α-Cat that includes the translation start site, and much of the intergenic region between α-Cat and the predicted gene CG32230. The α-Cat1 deletion extends from the P insertion site into α-Cat, whereas α-Cat2 and α-Cat3 extend to both sides of the P element insertion site. If α-Catenin is still synthesized in α-Cat mutants we expect this protein to be truncated. Assuming that the next available ATG is used, the truncated α-Catenin protein would be 823 instead of 917 amino acids or ~10.5 kDa smaller than full-length α-Catenin. Immunoblot analysis of heterozygous animals did not detect a truncated protein. Homozygous α-Cat mutant embryos showed reduced levels of α-Catenin of normal size (Fig. 1B). Moreover, adult flies mutant for α-Cat1 that expressed a rescue construct with a tagged α-Catenin protein, which runs at a higher molecular weight, showed no detectable α-Catenin of normal size or at a smaller size (Fig. 1C). In addition, α-Catenin is reduced to background levels in α-Cat mutant cells in the follicular epithelium (FE) (supplementary material Fig. S1). These data suggest that our α-Cat mutations are protein negative and that the protein detected in mutant embryos is maternal gene product. All three mutants can be rescued by an α-Cat transgene to fertile adults, suggesting that all observed phenotypes are due to a disruption of α-Cat. Together, these findings suggest that we have isolated null mutations for α-Cat.
Loss of zygotic α-Cat expression causes defects in larval head morphogenesis
All three mutations, either homozygous or in heteroallelic combinations, cause embryonic lethality. The cuticle of α-Cat mutants showed severe defects in the head skeleton, which result from a failure in head involution (Fig. 1E), a phenotype reminiscent of weak alleles in the DEcad-encoding gene shotgun (shg) (Tepass et al., 1996), or of embryos that express a mutant Arm protein that does not bind to α-Catenin (Orsulic and Peifer, 1996). The ventral and dorsal cuticle appeared normal. The failure in head involution leaves a large anterior opening in the epidermis (visualized through examination of the cuticle) through which internal organs are expelled, and the epidermis/cuticle shrank as a result of muscle contractions (Fig. 1E). The residual amounts of α-Catenin protein detected in late mutant embryos (Fig. 1B,C) and the relatively weak embryonic phenotype are probably the result of maternal α-Cat gene product because α-Catenin is a component of AJs throughout embryogenesis and RNAi knockdown indicated that α-Catenin is required for AJ integrity in early embryos (Magie et al., 2002; Cavey et al., 2008). Like DEcad (Godt and Tepass, 1998; Gonzalez-Reyes and St Johnston, 1998) and Arm (Peifer et al., 1993), α-Catenin is required for the development of the female germline (data not shown), preventing us from examining embryos that lack the maternal and zygotic gene product.
Aside from defects in head involution, we did not see obvious abnormalities in other embryonic tissues in which defects had previously been described in embryos with reduced levels of DEcad or Arm, including the ventral neuroectoderm, the gonads, the Malphigian tubules, or the tracheal system (Tepass et al., 1996; Uemura et al., 1996; Jenkins et al., 2003). This suggests that maternal α-Catenin is sufficient to support normal development of most tissues except the larval head. This is consistent with a phenotypic series of embryos with reduced levels of DEcad (Tepass et al., 1996), which suggests that head morphogenesis is the aspect of Drosophila embryogenesis that is most sensitive to the reduction of cadherin–catenin complex function.
We followed head development in α-Cat mutant embryos labeled for Crumbs (Crb), which outlines apical cell circumferences of epithelial cells (Fig. 1F–M). α-Cat mutants showed gaps in the head ectoderm that covers the developing larval brain between stages 12 and 14. The strength of this defect showed a broad range from embryos with no apparent holes to those that lacked most of the dorsal head ectoderm covering the two brain hemispheres. In many embryos, smaller gaps in the head ectoderm were associated with clusters of apically constricted cells, identifying groups of delaminating neuroblasts or clusters of sensilla (Fig. 1 G,J,K) (Younossi-Hartenstein et al., 1996; Campos-Ortega and Hartenstein, 1997). These observations suggest that epithelial cells in α-Cat mutants detach from each other as a result of pulling forces exerted by neighboring cells that constrict apically.
Head involution, which takes place later during stage 15 and 16 in the wild type, failed in most α-Cat mutant embryos (Fig. 1L,M). During normal head involution, part of the dorsal epidermis located posterior to the brain moves forward over the epidermis to cover the clypeolabrum. The forward moving epidermal flap is the dorsal pouch and the space that is formed between the dorsal pouch and the underlying epidermis is the frontal sac (Fig. 1L) (Campos-Ortega and Hartenstein, 1997). In α-Cat mutants, the dorsal pouch does not form and the clypeolabrum remains exposed (Fig. 1M). Head involution defects are very consistent in α-Cat mutants and are also seen in embryos that show no apparent holes in the head ectoderm, suggesting that these two defects might be independent consequences of the reduction in α-Catenin function.
Adherens junction composition in α-Cat mutant embryos
Antibody staining of α-Cat mutant embryos showed strongly reduced levels of α-Catenin (Fig. 2B,F,K), consistent with immunoblot analysis (Fig. 1B,C). Quantification of immunofluorescent intensity (Fig. 2I–K) at AJs in epidermal cells of early stage 17 embryos indicated that α-Catenin staining intensity was reduced to approximately 5% of α-Catenin levels seen in normal siblings (Fig. 2K). Ultrastructural morphology of epithelial AJs in tissues of stage 17 α-Cat mutant embryos that had strongly reduced levels of α-Catenin were normal (supplementary material Fig. S2). We then asked how other components of the AJs were affected by the reduction of α-Catenin in epithelial cells that were morphologically normal. DEcad and Arm also showed strongly reduced staining intensities at cell contacts, but were reduced by only 80–85% rather than by ~95% as seen with α-Catenin (Fig. 2B,D,F,K). The other two AJ components we tested, Bazooka/Par3 (Baz) and Echinoid (Ed), were not reduced in α-Cat mutant embryos compared to wild-type siblings (Fig. 2B,F,H,K), indicating that levels of these proteins are not directly linked to the levels of the cadherin–catenin complex.
Function of α-Catenin in postembryonic development
To explore α-Catenin function in the development of postembryonic tissues we generated α-Cat mutant cell clones through FLP-FRT-mediated mitotic recombination. As no FRT recombination site was available between α-Cat and the centromere due to their close proximity, we generated an α-Cat rescue construct that expressed α-Catenin under the control of the Ubiquitin promoter (Ubi-α-Cat). Homozygous mutant α-Cat flies carrying one copy of Ubi-α-Cat were viable and fertile. The Ubi-α-Cat insertion was combined with an FRT element, and α-Cat mutant clones were generated through the recombination-induced loss of Ubi-α-Cat in an α-Cat1 mutant background. This strategy was combined with the MARCM system (Lee and Luo, 1999) to generate α-Cat mutant cell clones that were positively marked with mCD8::GFP (Fig. 3A).
We induced α-Cat mutant clones in third instar larval imaginal discs. Mutant clones did not develop, which suggests that α-Catenin is essential for cell survival in these tissues (supplementary material Fig. S3). Failure of imaginal disc cell clones to grow and to survive had previously been reported for cell clones that lack DEcad or Arm (Peifer et al., 1991; Dahmann and Basler, 2000).
We next examined α-Cat mutant cell clones in female ovaries. The development of Drosophila egg follicles has proven to be an excellent model for the analysis of cadherin function. Follicles consist of a cluster of 16 germline cells, including one oocyte that is encapsulated by the follicular epithelium (FE). DEcad and Arm were shown to contribute to a number of processes including the posterior positioning of the oocyte that results from a cell sorting process (Peifer et al., 1993; Godt and Tepass, 1998; Gonzalez-Reyes and St Johnston, 1998). DEcad also functions in border cell migration, in which a group of follicle cells migrates between nurse cells to the edge of the oocyte (Niewiadomska et al., 1999). A third cadherin-dependent process in oogenesis is follicle formation itself. Here, follicle cells surround and enclose the 16 germline cells. This process involves extensive cell movement during which follicle cells extend cellular protrusions that are highly enriched in DEcad. Failure in follicle formation or the related process of interfollicular stalk formation results in ‘fused’ follicles, which contain more than 16 germline cells (Godt and Tepass, 1998; Peifer et al., 1993). A final cadherin-dependent process that we investigated is the integrity of the FE. The apical surface of the FE faces the germline. The FE has a typical cadherin-based zonula adherens. Loss of Arm leads to severe defects in the integrity of the FE, whereas loss of DEcad has little effect on FE integrity during early to mid stages of oogenesis likely because the FE coexpresses a second classic cadherin, DN-cadherin (Peifer et al., 1993; Tanentzapf et al, 2000; Pacquelet and Rorth, 2005).
Ovaries carrying α-Cat mutant cells displayed defects in all aspects of oogenesis mentioned above. This included a compromised FE, in which cells lose contact (Fig. 3B), mispositioned oocytes (Fig. 3C), defects in follicle formation leading to fused follicles (Fig. 3D), and failure of border cell migration (Fig. 3E). α-Catenin, DEcad and Arm are reduced to background levels in α-Cat mutant cells (Fig. 3B; Fig. 4A; supplementary material Fig. S1). α-Cat mutant defects were also seen in ovaries that expressed double-stranded RNA against α-Cat (α-Cat-RNAi) under the control of tj-Gal4 in all follicle cells (tj>α-Cat-RNAi) (supplementary material Fig. S4 and data not shown). As α-Cat function is reduced in all follicle cells starting prior to follicle formation in tj>α-Cat-RNAi animals, these ovaries showed more consistent early defects in follicle formation than ovaries containing α-Cat mutant cell clones. Ovarioles frequently formed a single giant fused egg chamber that contained multiple germline cysts, which indicates a complete failure in follicle segregation (supplementary material Fig. S4). Similar defects were seen in tj>arm-RNAi flies (supplementary material Fig. S5). Collectively, our analysis of α-Cat mutants suggests that Drosophila α-Catenin is an essential component of the cadherin adhesion system in all the cadherin-dependent cellular processes that we have studied.
Breakdown of cell architecture in α-Cat mutant follicle cells
Follicle cells that lack α-Catenin function did not degenerate, in contrast to imaginal disc cells, but persisted throughout oogenesis for many days, providing an opportunity to study epithelial cells that had lost α-Catenin function in detail. α-Cat mutant cells remained confined between the basement membrane that surrounds each follicle and the germline cells. α-Cat mutant cells had lost cell–cell contact with neighboring mutant or wild-type follicle cells in most cases. These cells rarely formed multilayered clusters, but were usually flattened and tightly attached to the basement membrane (Fig. 4A–E). The distribution of molecular markers was investigated in young follicles (stages 5–9) and older follicles (stages 10–11).
α-Spectrin is highly enriched in the lateral cytocortex of FE cells where it forms the actin–spectrin membrane cytoskeleton. In α-Cat mutant cells, α-Spectrin was detached from the membrane and formed a prominent cytoplasmic aggregate present in both young and old follicles (Fig. 4). These α-Spectrin aggregates were also apparent in FE cells that lack Arm (supplementary material Fig. S5). Mutant cells that retained contact to wild-type cells showed less severe defects and often lacked α-Spectrin aggregates (e.g. Fig. 4A, arrowhead). F-actin was reorganized in α-Cat mutant cells. It became enriched in areas of the cell periphery where lateral cell contact was lost. This was also seen at the cell-contact-free surfaces of wild-type cells and probably represents protrusions extending on the basement membrane (Fig. 4B). In addition, F-actin accumulated in α-Spectrin aggregates, suggesting that the lateral membrane-associated cytoskeleton collapses into these clusters (Fig. 4B). In contrast to α-Spectrin, the lateral membrane marker Discs Large (Dlg) was largely retained at cell–cell contacts between α-Cat mutant cells in young follicles but accumulated at α-Spectrin clusters at later stages (Fig. 4C,D). DEcad also associated with α-Spectrin clusters, although only at late stages (Fig. 4E). Together, these findings suggest that lateral membrane components form large cytoplasmic aggregates in α-Cat mutant FE cells that associate with a vesicle population containing basolateral cargo such as DEcad.
We next examined the distribution of apical markers in α-Cat mutant cells. βHeavy-Spectrin (βH-Spectrin), a spectrin isoform that specifically associates with the apical membrane of FE cells (Zarnescu and Thomas, 1999), largely colocalized with α-Spectrin in α-Cat mutant cells (Fig. 5A), and so did Crumbs (Crb, Fig. 5B), an apical transmembrane protein (Tanentzapf et al., 2000). By contrast, the cadherin Cad99C, which is a marker for the apical microvillus brushborder of FE cells (Fig. 5C) (D'Alterio et al., 2005), also labeled an apical tuft of microvilli in α-Cat mutant cells (Fig. 5D,E), suggesting that α-Cat mutant FE cells retained apical–basal polarity. However, the microvillus tufts pointed in different directions, indicating that the apical–basal axes of neighboring FE cells were not aligned when α-Catenin was lost. Many of the α-Spectrin aggregates were found in close proximity to the apical microvillus tufts (Fig. 5D). Consistent with the presence of DEcad and Crb in aggregates, we also found a strong association of Rab11 with α-Spectrin, suggesting that the vesicular compartments of the recycling endosome are enriched at these sites. As DEcad and Crb travel together in the biosynthetic pathway (Blankenship et al., 2007), but not in the endosomal pathway (Harris and Tepass, 2008) in Drosophila epithelial cells, this result suggests that DEcad- and Crb-containing biosynthetic vesicles accumulate at α-Spectrin aggregates and are not transported to the plasma membrane.
Genetic interaction of α-Catenin with Arp2/3 complex and its positive regulator SCAR
Recent reports have highlighted a potential role for mammalian αE-catenin in regulating the polymerization of actin through interference with the activity of the Arp2/3 complex (Drees et al., 2005, Benjamin et al., 2010). The Arp2/3 complex, which consists of seven protein subunits, nucleates actin polymerization to create a branched network of filaments (Goley and Welch, 2006). Members of the Wiskott–Aldrich Syndrome protein (WASp) and SCAR/WAVE family function as positive regulators of the Arp2/3 complex (Higgs and Pollard, 2001; Kurisu and Takenawa, 2009). Actin polymerization observed in the presence of Arp2/3 complex and WASp was suppressed by the addition of αE-catenin homodimer to in vitro assays (Drees et al., 2005). Furthermore, depletion of the cytosolic pool of αE-catenin in MDCK cells resulted in increased plasma membrane dynamics and enrichment of the Arp2/3 complex at the leading edge of lamellipodia (Benjamin et al., 2010). If the negative regulation of Arp2/3 by α-Catenin (either at cadherin adhesive contacts or independently of cadherin) makes a significant contribution to the observed phenotype of Drosophila α-Cat mutants, we would predict that the α-Cat mutant phenotype is ameliorated when the activity of the Arp2/3 complex is lowered.
To assess the interactions between α-Catenin and Arp2/3 genetically, we analyzed the phenotype of double mutants of α-Cat and Arp3 or Sop2 (Arpc1), which encode components of the Arp2/3 complex, or SCAR in embryos and during oogenesis. In Drosophila, SCAR rather than WASp is the primary regulator of Arp2/3 function during oogenesis and embryogenesis (Zallen et al., 2002, Hudson and Cooley, 2002). We first determined whether the α-Cat embryonic phenotype could be modified, which we would expect because α-Cat mutant embryos have a gradually declining maternal supply of α-Catenin that falls below a critical threshold in late embryos, resulting in the described defects in head morphogenesis. We found that even one copy of the weak shg/DEcad allele shgg119 causes a significant enhancement of the α-Cat mutant phenotype (Fig. 6A). To assess the phenotypes of single and double mutant embryos, we examined the embryonic cuticle and grouped embryos into different phenotypic classes as described in supplementary material Fig. S6. About 68% of α-Cat mutant embryos exhibited strong head defects and 32% showed weak head defects, but all embryos had a normal ventral cuticle (Fig. 1E and supplementary material Fig. S6). Heterozygous shgg119 animals showed no defects in development or survival. By contrast, introducing a copy of shgg119 enhanced the phenotype of α-Cat mutant embryos: >99% of embryos displayed a strong head defect and 21% of embryos showed defects in the ventral cuticle that were not seen in α-Cat single mutants (Fig. 6A and supplementary material Fig. S6).
Removing a single gene copy either of Sop2, Arp3 or SCAR from α-Cat mutants led consistently to an amelioration of the α-Cat mutant phenotype (Fig. 6B,C). Sop2, Arp3 or SCAR heterozygous mutants did not display any embryonic defects or lethality. In contrast to α-Cat mutants that always showed defects in head formation, α-Cat mutants with reduced Arp2/3 or SCAR activity showed less severe defects on average and included a fraction of embryos with normal heads. The observed phenotypic suppression was more pronounced when the SCAR or Sop2 alleles were introduced maternally rather than paternally, suggesting dose-dependent interactions. These data are consistent with the hypothesis that α-Catenin and the Arp2/3 complex act competitively in Drosophila embryonic head morphogenesis. We also observed a suppression of the phenotype of intermediate arm alleles (arm0403A1 and armXP33) when Arp2/3 activity was reduced (Fig. 6D, data not shown).
We next examined genetic interactions between α-Cat and Sop2 or SCAR in oogenesis. A requirement of Arp2/3 and SCAR has been described in the Drosophila female germline for the formation of ring canals between germline cells (Hudson and Cooley, 2002), but little is known about Arp2/3 function in the FE. We found that Sop2 and SCAR mutant cell clones in the FE showed similar defects (Fig. 7A,C,D). Mutant cells were apically constricted, and AJs appeared jagged between mutant cells and between mutant and wild-type cells (Fig. 7A,D). These defects were very pronounced in stage 11 and older egg chambers. Clonal expression of Sop2-RNAi and SCAR-RNAi showed defects similar to, but less severe than those of genetically null mutant cells (Fig. 7B,E), suggesting that these RNAi lines strongly but not completely compromise Arp2/3 and SCAR function.
To investigate the effect of reducing Arp2/3 and α-Catenin function simultaneously we generated cell clones that expressed either Sop2-RNAi and α-Cat-RNAi or SCAR-RNAi and α-Cat-RNAi. These double knockdown cells displayed the same defects as cells expressing only α-Cat-RNAi (Fig. 7F–H), suggesting that the loss of Arp2/3 function cannot compensate for the loss of α-Catenin function in the FE.
DEcad::αCat fusion protein can replace endogenous α-Catenin
Recent studies in tissue culture cells have highlighted potential functions for mammalian αE-catenin outside of the cadherin–catenin complex. It was proposed that αE-catenin, when not bound to β-catenin, modulates actin polymerization (Drees et al., 2005; Benjamin et al., 2010) or regulates microtubule-based trafficking (Lien et al., 2008). To address the question of whether Drosophila α-Catenin has essential cytosolic functions when not bound to the cadherin–catenin complex we expressed a DE-cadherin::α-Catenin (DEcad::αCat) fusion protein in an α-Cat mutant background.
We first expressed DEcad::αCat ubiquitously in α-Cat mutant animals. As controls, we expressed α-Catenin or DEcad. Expression of α-Catenin in α-Cat mutant animals rescued α-Cat mutants to fertile adults whereas expression of DEcad showed no rescue. Expression of DEcad::αCat in α-Cat mutants rescued head morphogenesis and embryonic lethality in 89% of the tested animals (n=219), but animals died during larval stages. Expression of DEcad::αCat or DEcad in a wild-type background also causes larval lethality, indicating that the ubiquitous expression of DEcad::αCat and DEcad in larvae has some unknown adverse consequences. One possibility is that these constructs titrate Arm and disrupt Wnt signaling (Sanson et al., 1996).
We next expressed DEcad::αCat, α-Catenin, or DEcad in α-Cat mutant cell clones in imaginal discs and follicle cells. DEcad expression did not rescue α-Cat mutant cells: imaginal disc clones failed to develop (supplementary material Fig. S2), cells in the FE lost cell contacts, and border cells failed to migrate (Fig. 8B,E) as described for α-Cat mutant cells. By contrast, α-Cat mutant cells expressing exogenous α-Catenin or DEcad::αCat showed a full rescue in imaginal disc epithelia (supplementary material Fig. S3), the FE, and border cell migration (Fig. 8A,C,D,F). Unlike DEcad, which when expressed in α-Cat mutant cells showed a punctate cytoplasmic distribution (Fig. 8B), DEcad::αCat formed larger cytoplasmic aggregates in addition to being localized to the plasma membrane (Fig. 8C). Together, these data indicate that DEcad::αCat can compensate for the absence of endogenous α-Catenin in morphogenesis of the larval head, in imaginal discs and during oogenesis. These findings argue against models suggesting that the dissociation of α-Catenin from the cadherin–catenin complex is an essential aspect of α-Catenin function, or that a cytosolic pool of α-Catenin is required to regulate cell behavior in the tissues we have analyzed.
Drosophila is used intensively to investigate the function of AJs in development. This analysis is facilitated by the availability of mutations in three of the four core components of the cadherin–catenin complex: the classic cadherin (DEcad or DN-cadherin) (Uemura et al., 1996; Tepass et al., 1996; Iwai et al., 1997) and the catenins Arm/β-catenin (Peifer and Wieschaus, 1990) and p120catenin (Myster et al., 2003). Here we report the isolation of null mutations in Drosophila α-Cat, expanding the fly tool kit to address AJ function. α-Cat mutants display defects that are indicative of loss of cadherin-mediated adhesion in late embryos, imaginal discs and ovaries. Zygotic α-Cat mutant embryos display defects in head morphogenesis, which resemble those caused by weak mutations affecting shg/DEcad (Tepass et al., 1996; Uemura et al., 1996). This mild phenotype is probably due to the presence of maternal α-Catenin, which is still detectable in late embryos and supports most DEcad-dependent processes during embryogenesis.
Analysis of head morphogenesis revealed two defects in α-Cat mutants. First, tissue breaks in the head epithelium are apparent in close proximity to clusters of cells that undergo apical constriction. It is likely that the reduced levels of the cadherin–catenin complex in α-Cat mutants do not provide sufficient adhesive strength to withstand the mechanical force exerted by those apically constricting cells. A correlation between the degree of morphogenetic stress that the epithelium is exposed to and the level of expression of the cadherin–catenin complex needed to maintain tissue integrity has been well established in embryos with reduced DEcad (Tepass et al., 1996; Uemura et al., 1996) and in other systems (Chu et al., 2004). The second defect seen in α-Cat mutants is a failure in head involution. Despite any obvious defects in epithelial integrity, the dorsal fold fails to move forward and to envelope the anterior–dorsal head. As the mechanisms of head involution are unclear, we currently have no basis for speculation on how α-Catenin might contribute to this process. Breaks in the head epithelium are variable in frequency and strength, whereas the failure in head involution is a robust defect, suggesting that the breaks in the epithelium might not be the immediate cause for the problems in head involution.
Quantification of fluorescent intensities of α-Catenin at AJs of α-Cat zygotic mutant embryos suggested that the levels of α-Catenin had dropped to a small percentage of levels in wild-type embryos by late embryogenesis (early stage 17). DEcad and Arm were also strongly reduced but not to the same degree as α-Catenin. That the loss of α-Catenin does not lead to an immediate corresponding loss of other components of the cadherin–catenin complex was also seen during mesoderm formation in early Drosophila embryos (Oda et al., 1998). This suggests that the cadherin–β-catenin complex has significant α-catenin-independent membrane stability, perhaps mediated by the association with p120catenin, which regulates cadherin endocytosis (Xiao et al., 2007; Ishiyama et al., 2010). By contrast, normal protein levels of Ed and Baz were retained at AJs in α-Cat mutants, suggesting that the concentrations of these proteins at AJs do not depend on the levels of the cadherin–catenin complex. This is consistent with work in early Drosophila embryos that suggested that Baz acts upstream of the cadherin–catenin complex in AJ assembly (Harris and Peifer, 2004; McGill et al., 2009). Ultrastructurally, AJs can retain normal appearance and size even though the level of the cadherin–catenin complex is strongly depleted (supplementary material Fig. S2) (Tepass et al., 1996). Collectively, these findings suggest that although the adhesive strength of AJs correlates with the cadherin–catenin complex content, AJ maintenance is largely unaffected by variations in cadherin–catenin complex concentration.
α-Cat mutant cell clones in the adult ovary display several defects previously seen in shg/DEcad or arm mutant cell clones. These include a mis-localization of the oocyte, a block of border cell migration, and lack of follicle formation and separation. α-Cat mutant cells in the FE flatten and detach from each other and their wild-type neighbors, indicating a reduction or loss in lateral cell adhesion. Even in very large cell clones, α-Cat mutant cells retain a monolayered arrangement, being sandwiched between the germline cells and the basement membrane. Interestingly, α-Cat mutant cells display apical–basal polarity, as indicated by a tuft of microvilli positive for the microvillus cadherin Cad99C. However, apical–basal axis orientation between cells appears uncoordinated, suggesting that the cadherin–catenin complex is not required for all aspects of apical–basal polarity in follicle cells but is required for intercellular adhesion and axis alignment of neighboring cells to form a proper epithelium.
One interesting feature of α-Cat mutant follicle cells is the formation of prominent cytoplasmic clusters of α-Spectrin, probably the result of a collapse of the lateral cytocortex. Previous work with S2 cells suggested that the cadherin–catenin complex might not be engaged in recruiting and stabilizing cytocortical spectrin in Drosophila (Dubreuil and Grushko, 1999). However, our data and observations made in Drosophila embryos depleted of α-Catenin with RNAi (Magie et al., 2002) suggest that the cadherin–catenin complex plays a crucial role in the formation of the spectrin-based lateral cytocortex, similar to its role in mammalian cells (e.g. McNeill et al., 1990). Cytoplasmic spectrin clusters also become enriched in other basolateral and apical markers including DEcad and Crb. Notably, these clusters are also enriched in the recycling endosome GTPase Rab11. Rab11 and its effector, the exocyst, are required for surface delivery of DEcad and Crb (Blankenship et al., 2007; Roeth et al., 2009). AJs can act as docking stations for exocyst-mediated membrane delivery (Yeaman et al., 2004; Langevin et al., 2005). These observations raise the possibility that DEcad and Crb are retained in Rab11-positive compartments that fail to fuse with the plasma membrane due to the lack of a cadherin- or catenin-dependent docking mechanism.
Mammalian αE-catenin and the Arp2/3 complex can act as competitive regulators of actin polymerization (Drees et al., 2005). Three modes of interaction between the cadherin–catenin complex and Arp2/3 have been proposed. Model 1 proposes that the observed enrichment of αE-catenin at AJs, resulting from the interaction of αE-catenin with β-catenin, fosters local αE-catenin dimerization. Dimer formation and binding to β-catenin are mutually exclusive due to overlapping binding sites, so that dimerization of αE-catenin is thought to occur after dissociation from β-catenin (Pokutta and Weis, 2000; Yamada et al., 2005; Drees et al., 2005). αE-catenin dimers then bind to F-actin and prevent Arp2/3 interaction with F-actin (Drees et al., 2005; Benjamin et al., 2010). αE-catenin therefore promotes the formation of F-actin bundles that are normally associated with AJs, rather than Arp2/3-dependent actin networks, because αE-catenin itself has actin bundling activity and can interact with other actin bundling proteins such as formin (Kobielak and Fuchs, 2004). Model 2 proposes the existence of independent junctional and cytosolic pools of αE-catenin, and that the cytosolic pool of αE-catenin shows negative regulatory interactions with Arp2/3 and, consequently, counteracts lamellipodia formation and cell motility (Benjamin et al., 2010). A third model for functional interactions between Arp2/3 and the cadherin–catenin complex is based on evidence suggesting that Arp2/3 is required for cadherin endocytosis (Leibfried et al., 2008; Georgiou et al., 2008). Compromised Arp2/3 activity could enhance surface abundance of the cadherin–catenin complex and therefore counteract a genetic reduction of cadherin or catenin levels.
Each of these models predicts that defects arising from a reduction in α-Catenin function could be suppressed by a concurrent reduction in Arp2/3 activity. This genetic interaction was observed in embryos mutant for α-Cat in which Arp2/3 or SCAR function was reduced. We also observed that reduced Arp2/3 activity ameliorates defects seen in embryos mutant for intermediate arm alleles. This finding is consistent with models 1 and 3 but difficult to reconcile with model 2 because the loss of Arm should enhance the cytosolic pool of α-Catenin as less α-Catenin is recruited to the junction. However, the observed genetic interactions could be reconciled with model 2 by assuming that the Arm–α-Catenin interaction is required to make α-Catenin competent to interact with F-actin, either by promoting dimerization as suggested (Drees et al., 2005), or through promoting a post-translational modification of α-Catenin such as phosphorylation (Zhai et al., 2008).
To explore potential interactions between α-Catenin and Arp2/3 in follicle cells, we took advantage of RNAi and coexpressed α-Cat-RNAi with either Sop2-RNAi or SCAR-RNAi in follicle cell clones. Sop2 or SCAR mutant or knockdown cells showed defects similar to those reported for cells in the pupal wing disc epithelium that lacked Arp2/3 function, and which apparently disrupts cadherin endocytosis (Leibfried et al., 2008; Georgiou et al., 2008). Defects in follicle cells with compromised Arp2/3 are seen only at late stages of follicle development, suggesting that the early FE does not require Arp2/3 function. α-Cat-RNAi expression caused defects similar to those observed in mutant cells, characterized by a loss of cell contact and the formation of spectrin aggregates. Expression of either Sop2-RNAi or SCAR-RNAi in α-Cat-RNAi cells did not noticeably modify the α-Cat phenotype. One potential explanation for the lack of interactions in this context might be that the disruption of adhesion by α-Cat-RNAi is too strong to be overcome significantly by lowering Arp2/3 activity, assuming that the interaction between α-Catenin and Arp2/3 is one among several α-Catenin activities. Alternatively, α-Catenin might not functionally interact with the Arp2/3 complex in follicle cells.
To reveal a potential function for a cytosolic pool of α-Catenin, we expressed a DEcad::αCat fusion protein in α-Cat mutants. Previously, it was shown that a mammalian E-cadherin:: αE-catenin fusion protein could function in cell adhesion (Nagafuchi et al., 1994), and that in Drosophila the DEcad::αCat protein can fully substitute for the loss of DEcad or Arm in all cadherin-dependent processes during oogenesis (Pacquelet and Rorth, 2005), and for the loss of DEcad (but not Arm) during dorsal closure of the embryo (Gorfinkiel and Martinez-Arias, 2007). However, the Drosophila rescue experiments were carried out in the presence of endogenous α-Catenin, leaving open the possibility that α-Catenin has a cytosolic function or interacts with the DEcad::αCat protein through the α-Catenin dimerization domain, which could create a dimer that interacts with actin. We found that expression of DEcad::αCat rescued head morphogenesis and embryonic lethality of α-Cat mutants, α-Cat mutant cell clones in imaginal discs, the FE, and border cell migration. These effects were similar to those for expression of exogenous α-Catenin. It was suggested that the expression of DEcad::αCat increases the surface abundance of cadherin, which could restore cell adhesion (Weis and Nelson, 2006). However, overexpression of DEcad in α-Cat mutants did not show any rescue activity, arguing against this notion. We also considered the possibility that DEcad::αCat undergoes proteolytic cleavage to release α-Catenin. However, no cleavage product was detected on immunoblots (data not shown). Moreover, whereas expression of a truncated form of α-Catenin that lacks the N-terminal β-catenin-binding and dimerization domain (deletion of amino acids 1–233) does not result in any rescue of α-Cat mutant defects, expression of the same truncated form of α-Catenin fused to DEcad rescues α-Cat mutants similarly to DEcad::αCat (R.S., R.R.P. and U.T., unpublished data).
Collectively, our data indicate that a cytosolic form of α-Catenin is not required for α-Catenin function in several Drosophila tissues that we have investigated, and that all essential aspects of α-Catenin function during morphogenesis are executed in the immediate vicinity of the plasma membrane. Our data do not rule out the possibility that α-Catenin directly interferes with the Arp2/3–actin interaction, but confines the potential for this interaction to the immediate submembranous space.
Materials and Methods
α-Cat mutations and other Drosophila lines
α-Cat1, α-Cat2 and α-Cat3 were generated by imprecise P element excision of GE30561 (Genexel, Korea). α-Cat1 deletes 2403 bp (23273158–23270755), a-Cat2 deletes 2521 bp (23273929–23271408) and a-Cat3 removes 2083 bp (23273929–23271846). Other mutant lines were armYD35, arm0403A1, armXP33 (obtained from Mark Peifer, University of North Carolina, Chapel Hill, NC). shgg119 (Tepass et al., 1996). Arp3EP(3)3640, SCARΔ37 FRT40A, Sop2/Arpc1Q25Sa FRT40A, Sop2/Arpc1Q25St FRT40A, and hsFLP; Ubi-GFP33 Ubi-GFP38 FRT40A were obtained from Lynn Cooley, Yale University, New Haven, CT (Hudson and Cooley, 2002). tj-GAL4 (Tanentzapf et al., 2007). α-Cat-RNAi (19182), arm-RNAi (7767), Sop2/Arpc1-RNAi (100573) and SCAR-RNAi (21908) were obtained from the Vienna Drosophila RNAi Center (Vienna, Austria). Act5c<CD2<GAL4, tj-Gal4, Act5c-GAL4, da-GAL4, Tub-GAL80 FRT40A, hsFLP, and UAS-mCD8::GFP were obtained from the Bloomington (IN) or Kyoto (Japan) Stock Centers.
Clonal analysis and rescue experiments
Flies were grown at 25°C unless specified otherwise. Homozygous mutant clones for α-Cat in ovaries were generated by heat-shocking flies of 0–2 days old at 37°C for 1 hour each on two consecutive days (Fig. 3A). To induce large follicle cell clones, pupae were heat-shocked as above and shifted to 29°C until the pharate adult stage, and then shifted back to 25°C. Freshly eclosed females were transferred to yeasted vials with males and then dissected 2 days later. Rescue experiments in imaginal discs and ovaries were carried out with a modified version of the MARCM system (Lee and Luo, 1999). Clones were induced in flies of the genotype Ubi-α-Cat Tub-GAL80 FRT40A/hsFLP FRT40A; α-Cat1 Act5c-GAL4 UAS-X/α-Cat1 da-GAL4 UAS-mCD8::GFP, in which UAS-X refers to either UAS-α-Cat, UAS-DEcad or UAS-DEcad::αCat.
arm mutant clones were generated with the genotype armYD35 FRT101/Ubi-GFP FRT101; MKRS, hsFLP/+.
Sop2/Arpc1 and SCARΔ37 mutant clones were generated in flies with the following genotype hsFLP; Sop2 FRT40A/hsFLP; Ubi-GFP33 Ubi-GFP38 FRT40A; SCARΔ37 FRT40A/hsFLP; Ubi-GFP33 Ubi-GFP38 FRT40A.
To make clones expressing RNAi constructs, UAS-α-Cat-RNAi, UAS-Sop2-RNAi, UAS-SCAR-RNAi and double RNAi combinations were crossed to hsFLP; Act<CD2<GAL4 (Struhl and Basler, 1993) and pupae of the resulting progeny were heat-shocked at 37°C for 25 minutes.
Protein expression and immunoblotting
For the data shown in Fig. 1B, stage 16–17 embryos were homogenized in 1% Triton-X 100 buffer [50 mM Tris-HCl pH 8.0, 100 mM NaCl, 1.0% Triton-X 100 containing incomplete protease inhibitor tablet (Roche Applied Science, Canada)]; 25 μg of total protein was loaded in each lane. For the experiment shown in Fig. 1C, dechorionated wild-type embryos of stages 12–15 or stages 15–17, α-Cat1 mutant embryos and α-Cat1 mutant embryos expressing UAS-α-CatΔCTD::HA (R.R.P. and U.T., unpublished results), or heads from α-Cat1 mutant adult flies that were rescued with UAS-α-CatΔCTD::HA were homogenized in SDS sample buffer (62.5 mM Tris-HCl pH 6.8, 2.3% SDS, 10% glycerol, 5% β-mercaptoethanol and 0.005% bromphenol blue); 15 μg of total protein was loaded in each lane. SDS-PAGE and immunoblots were done as described (Laprise et al., 2002). Primary antibodies were rat mAb anti-α-Catenin (DCAT-1; Developmental Studies Hybridoma Bank, DHSB, Iowa City, IA), mouse mAb anti-Armadillo (1:1000, N2-7A1; DSHB), rat mAb anti-DEcad (1:100, DCAD1; DSHB), guinea pig pAb anti-α-Catenin (1:5000, p121), mouse mAb anti-β-Tubulin (1:1000, E7; DSHB).
Immunohistochemistry and histological techniques
Embryonic cuticle was prepared following standard methods. Antibody stainings on Drosophila embryos and ovaries followed published protocols (Tepass et al., Tepass, 1996). Embryos used for anti-DEcad staining were prepared under methanol-free conditions and devitellinized in an ethanol–heptane mixture. Ovaries were fixed in 5% formaldehyde for 10–12 minutes in PB buffer. For Crb and βH-Spectrin, the ovaries were fixed in 5% formaldehyde in PBS for 12 minutes and then treated with methanol for 5 minutes before washing with PBS containing 0.03% Triton-X 100.
Primary antibodies used were: rabbit pAb anti-Baz (TH1) (1:4000; a gift from Tony Harris, University of Toronto, Toronto, Canada), mouse mAb anti-Arm (N2-7A1) (1:50; DSHB), rat mAb anti-DEcad (DCAD2; 1:50; DSHB), guinea pig pAb anti-α-Catenin (p121; 1:1000), and rat pAb anti-Crb (F1; 1:1000; Pellikka et al., 2002), mouse anti-Crumbs (Cq4, 1:15) (Tepass and Knust, 1993) mouse anti-α-Spectrin (3A9, 1:25; DSHB), mouse anti-Dlg (1:250; DSHB), rabbit anti-α-Spectrin (1:500, gift from Daniel Branton, Harvard University, Cambridge, MA), anti-βH-Spectrin (1:500, gift from Graham Thomas, Pennsylvania State University, Pennsylvania, PA), guinea pig anti-Cad99C (1:3000) (D'Alterio et al., 2005), rabbit anti-GFP (1:500; Invitrogen), mouse anti-GFP (1:1000; Abcam) and mouse anti-ratCD2 (1:400; AbD Serotec, Raleigh, NC). Fluorescent secondary antibodies were used at a dilution of 1:400 (Jackson ImmunoResearch Laboratories and Invitrogen). Phalloidin–rhodamine (Invitrogen) was used at a dilution of 1:30. Cell nuclei were visualized with PicoGreen (Invitrogen). TEM analysis was carried out as described (Tepass and Hartenstein, 1994).
For Ubi-α-Cat, a NcoI restriction site was inserted into the KpnI site 5′ of the α-Cat complementary DNA (cDNA) in pBS-DCT-Nc (Oda and Tsukita, 1999). Using NcoI and XbaI, α-Cat was cloned downstream of the Ubiquitin promoter in pUP2M (Lee et al., 1988). Ubi-α-Cat was then cloned into KpnI/XbaI of pCaSpeR4 and transgenic lines established through P-element-mediated transformation. Ubi-α-Cat4A was found to rescue α-Cat mutants to fertile adults, and was mapped to the chromosomal site 29F by inverse PCR.
For UAS-α-Cat, UAS-DEcad and UAS-DEcad::αCat, an attB sequence was introduced into the NsiI site of pUASP and pUASP2 to generate pUASP-attB and pUASP2-attB. α-Cat cDNA was fused to a double hemagglutinin (HA) tag and subcloned into pUASP-attB. shg/DEcad cDNA was inserted into pUASP-attB. To generate UAS-DEcad::αCat, we inserted an attB sequence into pUASP2-DEcad::αCat (Pacquelet and Rorth, 2005). Constructs were inserted into the attP2 site (Groth et al., 2004).
Measurement of fluorescent intensities
Imaris 4.0 software (Bitplane, Switzerland) was used to measure fluorescent staining intensities. Confocal Z-stacks spanning the apico–lateral cell contact sites were loaded into the Imaris program and regions of interest were defined manually, as illustrated in Fig. 2I′,J′ in the lateral epidermis of thoracic segments T1, T2 and T3. The fluorescence intensity of each voxel within the defined three-dimensional space was calculated by the software as an intensity sum. To determine the average intensity per cell, the intensity sums for each region were subsequently divided by the number of cells in each region as determined by Cell Count in ImageJ (NIH, Bethesda, MD). In order to perform the statistical analysis we calculated the average intensity per cell in 15 distinct regions of either wild-type or α-Cat1 mutant early stage 17 embryos.
We are grateful to Pernille Rorth, Lynn Cooley, Tony Harris, Graham Thomas, the Vienna Drosophila Research Center, the Developmental Studies Hybridoma Bank, and the Bloomington Drosophila Stock Center for supplying reagents. We thank Tony Harris for critical reading of the manuscript.
↵* These authors contributed equally to this work
↵‡ Present address: Li Ka Shing Faculty of Medicine, The University of Hong Kong, Hong Kong SAR, China
This work was supported by an Ontario Graduate Scholarship (to R.R.P.), an operating grant from the Canadian Institutes of Health Research (to D.G and U.T), and an operating grant from the Canadian Cancer Society Research Institute (to U.T.).
Supplementary material available online at http://jcs.biologists.org/lookup/suppl/doi:10.1242/jcs.096644/-/DC1
- Accepted August 23, 2011.
- © 2012.