Crumbs controls epithelial integrity by inhibiting Rac1 and PI3K

The polarized architecture of epithelial cells is crucial for the development and physiology of most organs. Epithelial polarity is controlled by evolutionarily conserved proteins, including the apical transmembrane protein Crumbs (Crb) (Bazellieres et al., 2009; Bulgakova and Knust, 2009). Fly embryos lacking Crb show epithelial polarity defects and lose epithelial tissue integrity (Bulgakova and Knust, 2009; Tepass et al., 1990). Overexpression of Crb leads to an expansion of the apical domain (Wodarz et al., 1995), illustrating that Crb is a central apical determinant. Identification of the molecular mechanisms involved downstream of Crb is important to understand epithelial polarity regulation and human diseases, including retinal dystrophies and tumor growth (Bazellieres et al., 2009; Bulgakova and Knust, 2009; den Hollander et al., 1999; Karp et al., 2008).

The phosphoinositide 3-kinase (PI3K) pathway is another important regulator of cell polarity (Gassama-Diagne et al., 2006; Martin-Belmonte and Mostov, 2007). PI3K produces mainly phosphatidylinositol 3,4,5-trisphosphate [PtdIns(3,4,5)P₃], which activates downstream effectors of the PI3K pathway, including Akt that is phosphorylated in response to production of PtdIns(3,4,5)P₃ (Yuan and Cantley, 2008). The tumor suppressor PTEN inactivates the PI3K pathway by dephosphorylating PtdIns(3,4,5)P₃ to generate PtdIns(4,5)P₂ (Leslie et al., 2008). In polarized epithelial cells, PTEN is enriched apically leading to concentration of PtdIns(4,5)P₂ at the apical membrane (Martin-Belmonte et al., 2007; von Stein et al., 2005). By contrast, PtdIns(3,4,5)P₃ is mainly associated with the basolateral domain (Gassama-Diagne et al., 2006; Martin-Belmonte et al., 2007). Forced insertion of PtdIns(4,5)P₂ in the plasma membrane generates ectopic apical membrane domains (Martin-Belmonte et al., 2007), whereas PtdIns(3,4,5)P₃confers a basolateral character to the plasma membrane (Gassama-Diagne et al., 2006). Therefore, the fine-tuning of the PtdIns(3,4,5)P₃ to PtdIns(4,5)P₂ ratio and their appropriate localization is crucial for epithelial polarity. However, the upstream regulators of PI3K and PTEN in the context of epithelial polarity need to be further defined. Here, we report that Crb controls epithelial integrity by counteracting a Rac1–PI3K module, which restricts Crb activity. Thus, the equilibrium between these mutually antagonistic polarity regulators is crucial to maintain epithelial organization.

PtdIns(4,5)P₂ and Crb are important regulators of apical membrane formation (Martin-Belmonte et al., 2007; Wodarz et al., 1995). By contrast, PtdIns(3,4,5)P₃ confers a basolateral character to the plasma membrane (Gassama-Diagne et al., 2006). This raises the intriguing possibility that Crb could promote apical membrane identity by increasing the concentration of PtdIns(4,5)P₂ and/or restricting the formation of PtdIns(3,4,5)P₃. To test this possibility, we assessed the effect of modulating Crb expression on Akt phosphorylation levels in vivo. Overexpression of Crb in Drosophila embryos reduced Akt phosphorylation (Fig. 1A,C). Moreover, Akt phosphorylation was increased in crb mutants compared with wild-type embryos (Fig. 1B,C), suggesting that Crb normally functions to repress the PI3K–Akt pathway. To verify the functional importance of the Crb-dependent inhibition of PI3K signaling on epithelial tissue integrity, we counteracted PtdIns(3,4,5)P₃ production in crb mutant embryos (null allele crb^11A22) by expressing a dominant-negative form of PI3K (Dp110^D954A) or overexpressing wild-type PTEN. Whereas loss of Crb resulted in epithelial breakdown and formation of isolated cysts of epithelial cells (Fig. 1E) (Tepass et al., 1990), crb mutant embryos expressing Dp110^D954A or increased amount of wild-type PTEN produced continuous sheets of well-organized epithelial cells (Fig. 1H,I). Rescue of the crb mutant phenotype was also evident in the organization of the cuticle, which is secreted by the underlying epidermis and reflects its integrity. The cuticle was severely fragmented and appeared as isolated granules in the absence of Crb (Fig. 1G) (Tepass et al., 1990), whereas uninterrupted layers of cuticle are formed upon reduction of PI3K activity or increased PTEN expression in crb mutant embryos (Fig. 1J,K). In addition, pten mutation enhanced epithelial collapse associated with the hypomorphic allele crb^S87-2 (Fig. 1L–O), further demonstrating the functional relationship between phosphoinositide metabolism and Crb. Collectively, these data show that the ability of Crb to increase the PtdIns(4,5)P₂ to PtdIns(3,4,5)P₃ ratio is critical for epithelial integrity in vivo.

Although the crucial function of PtdIns(3,4,5)P₃ in epithelial polarity regulation is well established, it is still unclear how this lipid acts in this context. It was previously shown that PI3K inhibition restricts lateral membrane growth, thus reducing epithelial cell height (Gassama-Diagne et al., 2006; Jeanes et al., 2009; Laprise et al., 2002). Interestingly, inhibition of Rac1 has a similar outcome (Jeanes et al., 2009). This suggests that Rac1 and PI3K could act in a common pathway to organize the lateral domain. In support of this hypothesis, PI3K signaling activates Rac1 in many mammalian cell types (Rivard, 2009). Similarly, an active form of PI3K (Dp110^CAAX) increased the amount of active GTP-bound Rac1, whereas expression of Dp110^D954A reduced Rac1 activation in Drosophila embryos (Fig. 2A). This suggests that PI3K activity is required for full activation of Rac1 in vivo. Our data further propose that Rac1 signals back to PI3K and activates this kinase, as expression of a dominant-negative form of Rac1 (Rac1^N17) suppressed Akt phosphorylation (Fig. 2B). Rac1 also activates PI3K during MDCK cell cyst morphogenesis (Liu et al., 2007). This shows that PI3K and Rac1 are part of a positive-feedback loop in embryos, and suggests that Crb-dependent inhibition of PI3K signaling impacts Rac1 activity. Accordingly, overexpression of Crb in embryos reduced the level of GTP-loaded Rac1 (Fig. 2C), showing that Crb is a negative regulator of Rac1 activation. Based on the rescue of the crb mutant phenotype by Dp110^D954A and the reciprocal positive relationship linking PI3K and Rac1 that we demonstrated, it is possible to predict that increased Rac1 activity is involved in the deleterious effects associated with the loss of Crb. To validate this hypothesis, we expressed Rac1^N17 in a crb mutant background. In contrast to what is observed for embryos lacking Crb (Fig. 1E), crb mutants expressing Rac1^N17 formed a well-organized epidermal epithelium and cuticle covering a large fraction of their surface (Fig. 2E,G). Similar results were observed when the crb mutation was combined with rac1 and rac2 mutant alleles to reduce cellular Rac activity (Fig. 2I,K), showing that the counteraction of Rac1 activation by Crb has a physiological role in maintaining epithelial integrity in vivo. Thus, our study establishes that Crb controls epithelial organization by counteracting a positive-feedback loop in which Rac1 and PI3K activate each other, and defines Crb as a membrane protein that regulates intracellular signaling pathways.

Next, we directly investigated the role of Rac1 and PI3K in epithelial cell polarity and organization in vivo. Staining of the apical marker Crb in the epidermis of stage 16 embryos expressing Rac1^N17 or Dp110^D954A revealed long protrusions (Fig. 3B,C, arrowheads) and deep invaginations (Fig. 3B,C, arrows), which indicate increased apical membrane length. Indeed, individual epithelial cells showed an enlarged dome-shaped apical membrane upon inhibition of Rac1 or PI3K (Fig. 3G,H, arrows). These features are typical of the phenotype resulting from Crb overactivation (Laprise et al., 2006; Laprise et al., 2009; Wodarz et al., 1995), suggesting that Crb is involved in Rac1^N17- and Dp110^D954A-induced phenotypes. Accordingly, loss of one copy of crb suppressed the enlargement of the apical domain associated with the inhibition of Rac1 or PI3K signaling (Fig. 3D–E,I–J). Additionally, Crb is severely reduced in stage 11 embryos expressing the constitutively active Rac1^V12, as shown by a surface view of the ectoderm (Fig. 3, compare K with L). Lateral view revealed that residual Crb forms tight aggregates corresponding to the contact point between apexes of several epithelial cells. These cells show a highly constricted apical domain compared with that observed in wild-type embryos (Fig. 3M,N). In addition, the basolateral protein Dlg colocalizes with Crb in cells showing contracted apexes (Fig. 3N, arrows) or is found around the periphery of cells where Crb expression is almost absent (Fig. 3N, arrowheads). This demonstrates that activation of Rac1 causes defects in the formation and organization of the apical domain. Together, these results show that the PI3K-Rac1 module represses Crb-dependent apical membrane growth. Expression of an activated form of Rac1 also decreases Crb level in salivary glands of Drosophila embryos (Pirraglia et al., 2010). In addition, it was recently shown that β_H-spectrin, which is physically and functionally linked to Crb (Medina et al., 2002; Pellikka et al., 2002), represses Rac1 activity (Lee and Thomas, 2011). Moreover, Pak1, an important Rac1 effector, decreases apical spectrin and Crb levels, thus destabilizing the apical domain (Conder et al., 2007). These observations further support the notion that the Crb complex and Rac1 signaling are mutually antagonistic to sustain epithelial polarity.

Collectively, our data show that Rac1 and PI3K are part of a positive-feedback loop in fly embryos. Crb breaks this loop and represses activation of Rac1 as well as PI3K signaling. Reciprocally, the Rac1–PI3K module restricts Crb function and apical membrane growth, showing that the antagonistic functional relationship linking these proteins is bidirectional. In normal epithelial cells, Crb activity inhibits the Rac1–PI3K cassette at the apical membrane. In this way, epithelial cells establish proper apical domain and epithelial tissue integrity is preserved (Fig. 4). In crb mutant embryos, the Rac1–PI3K pathway is activated. This contributes to the crb mutant phenotype, as we have shown that reduction of Rac1 or PI3K activity rescues the phenotype associated with the loss of Crb. In addition, activation of Rac1 has a similar outcome to the absence of Crb, including defects in apical domain formation. By contrast, overexpression of Crb or reduction of Rac1 or PI3K activity results in a Crb-dependent increase of apical membrane size. Thus, the fine balance between Crb and the Rac1–PI3K module is crucial to establish proper apical to basolateral ratio in epithelial cells and to sustain epithelial tissue organization. Our study contributes to defining the molecular mechanisms that regulate epithelial polarity, which is crucial for the morphogenesis and physiology of epithelial tissues.

Fly stocks used in this study were: crb^11A22, crb^S87-2, rac1^j11, rac2^Δ, pten^dj189, UAS-rac1^N17, UAS-Dp110^D954A, UAS-PTEN2, UAS-crb^wt2e, da-GAL4 (all alleles and transgenic lines are described in Flybase; ><http://flybase.org/). Overexpression of Rac1^N17 in a crb^11A22 mutant background was achieved by crossing crb^11A22, da-GAL4 animals to crb^11A22, UAS-rac1^N17 flies at 18°C. Crb, Dp110^D954A, PTEN2, Rac1^N17 and Rac1^V12 were expressed in embryos by crossing the corresponding UAS lines with da-Gal4 flies at 25°C or ptc-Gal4 at 29°C. crb mutant embryos used were obtained from crb germ line clone females crossed to crb/+ males.

Drosophila embryos were fixed as described previously (Tepass et al., 1990). The primary antibodies anti-Fas3 (7G10) and anti-Dlg (4F3) (Developmental Studies Hybridoma Bank) were used at a dilution of 1:50 and 1:10, respectively. Anti-Crb antibody (Pellikka et al., 2002) was diluted 1:500. Secondary antibodies were conjugated to Cy3 (Jackson Immunoresearch Laboratories) or Alexa Fluor 488 (Molecular Probes).

Western blotting was performed as described (Laprise et al., 2002). Primary antibodies used were: rabbit anti-phospho Akt Ser473 (Cell Signaling), 1:1000 dilution; rabbit anti-Akt (Cell Signaling), 1:1000; mouse anti-Rac1 (clone 102; BD transduction Laboratories), 1:1000; rat anti-Crb (Pellikka et al., 2002), 1:5000; rabbit anti-GST (provided by J.-Y. Masson, Laval University), 1:10,000. HRP-conjugated secondary antibodies were from GE Amersham and used at a 1:1000 dilution.

The protocol used was based on the method described by Picard and colleagues (Picard et al., 2009). Briefly, embryos were homogenized in lysis buffer and part of extracts was kept to monitor the total amount of Rac1 in each sample (input). Then, GTP-bounded Rac1 was pulled down from 1 mg of the same samples with 20 μg of the CRIB domain of PAK fused to GST (45 minutes, 4°C). GST alone was used as negative control. The beads were then washed twice with washing buffer. Pulled-down Rac1 levels or the total amount of Rac1 present in cellular extracts was determined by western blotting.

We are grateful to S. Beronja for critical reading of the manuscript. We thank U. Tepass, A. Wodarz, E. Knust, S. Leevers, N. Lamarche-Vanne, J.-Y. Masson, the Bloomington Drosophila Stock Center and the Developmental Studies Hybridoma Bank for reagents.