Localization and Function of Pals1-associated Tight Junction Protein in Drosophila Is Regulated by Two Distinct Apical Complexes

Background: Pals1-associated tight junction protein (PATJ) is an important regulator of Myosin-driven morphogenetic processes.

Results: PATJ associates with two different apical complexes and accomplishes its function by a redundancy of its protein interaction domains.

Conclusion: The correct targeting of PATJ to distinct apical domains is important for its function in Drosophila development.

Significance: Learning how adaptor proteins can be regulated by protein targeting to distinct cellular compartments.

Abstract

The transmembrane protein Crumbs (Crb) and its intracellular adaptor protein Pals1 (Stardust, Sdt in Drosophila) play a crucial role in the establishment and maintenance of apical-basal polarity in epithelial cells in various organisms. In contrast, the multiple PDZ domain-containing protein Pals1-associated tight junction protein (PATJ), which has been described to form a complex with Crb/Sdt, is not essential for apical basal polarity or for the stability of the Crb/Sdt complex in the Drosophila epidermis. Here we show that, in the embryonic epidermis, Sdt is essential for the correct subcellular localization of PATJ in differentiated epithelial cells but not during cellularization. Consistently, the L27 domain of PATJ is crucial for the correct localization and function of the protein. Our data further indicate that the four PDZ domains of PATJ function, to a large extent, in redundancy, regulating the function of the protein. Interestingly, the PATJ-Sdt heterodimer is not only recruited to the apical cell-cell contacts by binding to Crb but depends on functional Bazooka (Baz). However, biochemical experiments show that PATJ associates with both complexes, the Baz-Sdt and the Crb-Sdt complex, in the mature epithelium of the embryonic epidermis, suggesting a role of these two complexes for the function of PATJ during the development of Drosophila.

Introduction

Apical-basal polarization of epithelia is regulated by conserved complexes determining the apical versus the basolateral domain (1, 2): apical to the adherens junctions (AJ),³ the membrane-associated partitioning-defective (PAR)-atypical protein kinase C (aPKC) complex regulates assembly of the Crumbs (Crb) complex, which assembles more apically in the so-called subapical region. The activity of these two complexes is counterbalanced by proteins such as Scribble-Lethal (2) giant larvae-Discs large (Dlg), which localize to the basolateral domain. In the past, various studies have demonstrated that both apical complexes are rather dynamic and that their composition might be tissue-dependent and temporally and/or developmentally regulated (3–9).

In Drosophila, the multiple PDZ domain-containing protein PATJ has been described to function in a complex with Crb and Stardust (Sdt, the Drosophila homologue of Partner of Lin-7 one, Pals1) to regulate apical-basal polarity in follicle epithelial cells and photoreceptor cells (10–12). In mammalian and Drosophila epithelial cells, Pals1/Sdt is recruited by the cytoplasmic tail of Crb to the subapical region and, in turn, stabilizes Crb (13–17). The L27 domain of Pals1 has been shown to heterodimerize with the L27 domain of PATJ, thereby tethering PATJ to tight junctions (TJs) (18–20).

Recently, we and others reported that loss of PATJ in Drosophila does not affect apical-basal polarity in the embryonic epidermis or in follicle epithelial cells but, rather, modulates Myosin activity to support AJ stability (21–23). Exclusively in photoreceptor cells and, to some extent, in the follicular epithelium, PATJ seems to be essential for the correct subcellular localization of the Crb-Sdt complex, either directly by stabilizing this complex or indirectly by regulating photoreceptor morphology/development (21, 22).

Two mammalian orthologues of PATJ are expressed in epithelia: mammalian PATJ (mPATJ, encoded by INADL in mice) and multiple PDZ domain protein 1 (MUPP1). Both proteins are very similar to DmPATJ. In addition to an N-terminal L27 domain, they exhibit several PDZ domains (DmPATJ exhibits four, mPATJ ten, and MUPP1 thirteen) and localize to the TJ in mammalian epithelial cells (24). However, Adachi et al. (24) showed that, despite its domain similarity, mPATJ, but not MUPP1, regulates TJ stability (24). These data are in line with previous findings describing TJ formation delay or defects upon loss of mPATJ in cultured epithelial cells (25, 26). Other studies describe a role of mPATJ in Myosin-driven processes like apical constriction and cell migration (27–29).

In this study, we report that, in the embryonic epidermis of Drosophila, PATJ assembles with the Crb-Sdt complex but additionally associates with the Baz-Sdt-complex we described previously (6). Notably, deletion of Baz and Sdt, but not of Crb, leads to mislocalization of PATJ during gastrulation and in fully differentiated epithelia of the embryonic epidermis. In contrast, localization of PATJ at the basal junction next to the tip of the invaginating plasma membrane during cellularization is independent of Baz/Sdt. Consequently, deletion of the L27 domain of PATJ leads to an abolished apical accumulation and impaired function of the protein. Studies with chimeric proteins further suggest that targeting of the PDZ domains of PATJ to the Baz-(Sdt) and Crb-(Sdt) complex are sufficient for the function of PATJ. Finally, we investigated the functionality of the four PDZ domains of PATJ and provide evidence that, under close to endogenous expression levels, none of these domains is essential.

Experimental Procedures

Drosophila Genetics

The following mutant alleles were used: PATJ^Δ1 (21), baz^815-8 (30, 31), baz^XR11 (32), sdt^K85 (33), and crb^11A22 (34). Germ line clones were generated with the mutant alleles recombined with FRT using the dominant female sterile technique (35). Homozygous mutant embryos were identified using antibody staining. Follicle cell clones were generated with the FRT-Flp system as described before (21). Ubi::PATJ-GFP (mutant/chimeric) constructs were generated using phiC31-mediated germ line transformation using attP40.

DNA and Constructs

The QuikChange site-directed mutagenesis kit (Stratagene) was used to generate domain deletions with full-length PATJ cDNA in pENTR (21) as a template. The following oligonucleotides were used for mutagenesis: PATJΔL27, 5′-GCGGATATTTCCAGCTCCAT-GTTGCCCAAC-3′; PATJΔPDZ1, 5′-GCCATA-GAGCTGGTCCGTCCCGTTGAGCAG-3′; PATJΔPDZ2, 5′-GAAACGGAGAAGCTTCGC-TACCTGAGGGGC-3′; PATJΔPDZ3, 5′-GGCT-CCGATGTGGAGTGCGGTCGCAACAGG-3′; and PATJΔPDZ4, 5′-ATGTGGTCGTCCCAACGC-ATTGGTGTGGCC-3′.

To generate truncated versions of PATJ, the following primers were used: PATJ-F, 5′-CACCATGCACCTCAGCGCGGA-3′; PATJ-151-R, 5′-CTCTATGGCCTGGATCTGAGC-3′; PATJ-256-R, 5′-CAGGGCGTACTGGGG-3′; and PATJ-449-R, 5′-TGATGGTGTAGTTGTGGC-3′.

For PATJ_{ΔL27 PDZ(Sdt)}, the PDZ domain of Sdt was amplified with Sdt-PDZ-F (5′-GCGGCCGCCCCCTTCACCATGCGTATCATCCAGATCGAG-3′) and Sdt-PDZ-R (5′-GCGGCCGCCGGTGGACTACCCGCTGG) and inserted with NotI (underlined) into PATJ_ΔL27 pEntry. Similarly, the PDZ domain of PAR6, the oligomerization domain of Baz, and the FERM domain of Yurt were cloned into NotI of PATJ_ΔL27 pEntry using the following oligonucleotides: PAR6-PDZ-F, 5′-GCGGCCGCCCCCTTCACCATGAGAAGAGTGCGGCTACTG-3′; PAR6-PDZ-R, 5′-GCGGCCGCCTTCACGGTGATTATCAGATTG-3′; Yrt-FERM-F, 5′-GCGGCCGCCCCCTTCACCATGGTGCTCGGAAAGGATGGC-3′; Yrt-FERM-R, 5′-GCGGCCGCTTTGACCGGCGCCCTAA-3′; Baz-CR1-F, 5′-GCGGCCGCCCCCTTCACCATGAAGGTCACCGTCTGCTTCGGC-3′; and Baz-CR1-R, 5′-GCGGCCGCATCTCCGCCTCCTTGC-3′. Baz_733–1221 was cloned into an endogenous SacII site (amino acid 633) of PATJ_ΔL27. All constructs were recloned into destination vectors (modified UWG, Murphy laboratory, DGRC) using Gateway technology (Life Technologies).

Immunoprecipitation and Western Blotting

For immunoprecipitations, w⁻ embryos from an overnight collection were dechorionated and lysed in lysis buffer (1% Triton X-100, 150 mm NaCl, 1 mm CaCl₂, 1 mm MgCl₂, and 50 mm Tris-HCl (pH 7.5)) supplemented with protease inhibitors. After centrifugation, 2 μl of rabbit anti-Baz (36) or 2 μl of the corresponding preimmune serum was added to the embryonic lysate, corresponding to 500 μg of total protein. Immune complexes were harvested using protein A-conjugated agarose (BioVision). GFP binder (Chromotek) was used to immunoprecipitate Crb-GFP. Wild-type flies served as a control. Beads were washed five times in lysis buffer and boiled in 2× SDS sample buffer before SDS-PAGE and Western blot analysis. Western blotting was done according to standard procedures. Primary antibodies used for Western blotting were as follows. Mouse anti-Crb (Cq4, 1:50, developed by E. Knust (37)) was obtained from the Developmental Studies Hybridoma Bank, created by the NICHD/National Institutes of Health and maintained at the University of Iowa, Department of Biology). Also used were guinea pig anti-PATJ (1:1000 21), mouse anti-Sdt (1:20 33), rabbit anti-Baz (1:2000 36), and mouse anti-GFP (B2, 1:500, Santa Cruz Biotechnology, catalog no. sc-9996).

Immunohistochemistry

Embryos were fixed in 4% formaldehyde and phosphate buffer (pH 7.4) as described previously (38). Primary antibodies used for indirect immunofluorescence were as follows: rabbit anti-Baz (1:1000 36), mouse anti Crb (Cq4, 1:50, Developmental Studies Hybridoma Bank), mouse Dlg (1:50, Developmental Studies Hybridoma Bank (39)), guinea pig anti-PATJ (1:500 21), mouse anti-Sdt (1:20 33), guinea pig anti-PAR6 (40), guinea pig anti-Yrt (41), rabbit anti GFP (catalog no. A11122, 1:1000, Life Technologies), rabbit anti-GFP (catalog no. sc-8334, Santa Cruz Biotechnology), and chicken anti-GFP (1:2000, Aves Laboratories). Secondary antibodies conjugated with Alexa Fluor 488, Alexa Fluor 568, and Alexa Fluor 647 (Life Technologies) were used at 1:400. Images were taken on a Zeiss LSM 710 Meta confocal microscope and processed using Adobe Photoshop.

Results and Discussion

PATJ Is Recruited by Sdt to a Complex with Baz at the Apical Junctions in the Embryonic Epidermis

Upon the formation of apical AJs in late cellularization/early gastrulation in Drosophila, PATJ is recruited to the apical cell-cell contact region and staining at the basal membrane ceases (21). Studies in Drosophila and cultured mammalian epithelial cells showed that PATJ associates with Sdt/Pals1, which, in turn, binds to the transmembrane protein Crb, which targets the complex to TJs in vertebrates and the corresponding subapical region in Drosophila (20, 42).

We recently found that, in the embryonic epidermis of Drosophila, Sdt is initially localized to the apical junctions in early gastrulation before Crb is expressed and even remains at the junctional region of epithelial cells when Crb is absent (6). This is accomplished by a direct interaction of the PDZ domain of Sdt with Baz. Upon phosphorylation of Baz by aPKC at serine 980, Sdt is released from Baz and becomes available to stabilize the Crb complex (6). We therefore tested whether the subcellular localization of PATJ is dependent on Crb or Baz or both. In maternal and zygotic crb-mutant embryos (crb^11A22 germ line clones), PATJ shows a correct localization not only during cellularization (data not shown) but also after gastrulation as long as apical-basal polarity is still intact (stage 6–11, Fig. 1B, compared with wild-type embryos in Fig. 1A). In later stages (from stage 11/12 on), apical-basal polarity is impaired because of the loss of Crb, finally resulting in a multilayered epithelium. In these cells, PATJ staining is cytoplasmic or in aggregates (Fig. 1C). Notably, the loss of cortical PATJ in these embryos is accompanied by a loss of membrane-associated Baz (Fig. 1C).

FIGURE 1.

A, endogenous PATJ colocalizes with Crb and Baz at the apical junctions in epithelial cells of the embryonic epidermis. Green, Crb; red, PATJ; Blue, Baz. B, endogenous PATJ localizes at the apical junctions in the absence of Crb expression during early embryogenesis. Green, Crb; red, PATJ; Blue, Baz. GLC, germ line clone. C, upon disruption of epithelial integrity in crb-mutant embryos (maternally and zygotically mutant, derived from germ line clones) of later stages, Baz as well as PATJ are mislocalized to the cytoplasm. Green, Crb; red, PATJ; Blue, Baz. D, PATJ localization during cellularization is not affected in baz mutant embryos derived from baz^815-8 germ line clones. Green, Baz; red, PATJ; blue, Dlg. E, loss of maternal and zygotic Baz in gastrulation (baz^815-8 germ line clones) results in disturbed apical-basal polarity and cytoplasmic PATJ localization. Green, Baz; red, PATJ; blue, Dlg. F, PATJ is not recruited to the apical junctions in the absence of Sdt. Green, Sdt; red, PATJ; blue, Baz. G, endogenous Sdt coimmunoprecipitates (IP) with PATJ-GFP but not with PATJ_ΔL27-GFP. H and I, endogenous PATJ coimmunoprecipitates with Baz in early stages (stage 1–9, H) and late stages (stage 12–17, I). Preimmune serum (pre) served as a negative control. J, coimmunoprecipitation of PATJ with Crb-GFP expressed from its endogenous promoter from embryonic lysates. Wild-type flies served as a negative control. K and L, high magnification of staining of wild-type embryos with PATJ, Crb, and Baz reveals a segregation of Baz and Crb that starts in earlier embryonic stages (stage 8, K) and becomes more pronounced in later stages (stage 12, L). Remarkably, PATJ overlaps with both proteins in both stages (arrows). Scale bars = 5 μm (A–F) and 2 μm (K and L).

In contrast, in maternal and zygotic baz-mutant embryos (baz^815-8 germ line clones), the accumulation of PATJ at the basal junction next to the tip of the furrow canal during plasma membrane invagination is not affected (Fig. 1D), but the targeting of the protein to the apical junctional region after cellularization is abolished (Fig. 1E).

Furthermore, we found endogenous PATJ and Sdt to coimmunoprecipitate with endogenous Baz from embryonic lysates (Fig. 1, H and I). Consequently, in embryos lacking Sdt, PATJ is correctly localized during cellularization (data not shown) but fails to relocalize to the apical junctions during gastrulation (Fig. 1F), indicating that PATJ is recruited by Sdt to the apical junctions. Furthermore, deletion of the L27 domain of PATJ results in a disturbed association with endogenous Sdt (Fig. 1G). This is consistent with studies in cultured mammalian cells demonstrating that PATJ directly binds to Pals1 via heterodimerization or hetero-oligomerization of its L27 domain with the (more N-terminal) L27 domain of Pals1 (18–20). Beside its association with Baz-Sdt, PATJ can also be coimmunoprecipitated with Crb-GFP expressed from its endogenous promoter (43, Fig. 1J), pointing to the existence of a second complex consisting of Crb-Sdt-PATJ. However, Crb cannot be found to coimmunoprecipitate with Baz (Fig. 1I), demonstrating that there is no quaternary Baz-Crb-Sdt-PATJ complex.

To test whether both PATJ-containing complexes are formed in a stage-dependent manner, Baz coimmunoprecipitation experiments were performed either with early-stage (stage 1–9) or late-stage (stage 12–17) embryos. Interestingly, PATJ and Sdt can be coimmunoprecipitated with Baz from lysates of early and late developmental stages (Fig. 1, H and I). This suggests that PATJ is initially recruited to the apical junctions by the Baz-Sdt-complex, whereas the Crb-Sdt-PATJ complex is formed later in development as soon as Sdt is released from Baz upon phosphorylation by aPKC (6). However, in late developmental stages, PATJ and Sdt can still be coimmunoprecipitated with Baz. This observation and the fact that PATJ still associates with Baz and remains correctly localized in the absence of Crb in later stages (stage 10/11, Fig. 1B) indicate that a portion of Baz remains unphosphorylated by aPKC and associates with Sdt-PATJ. Therefore, apically localized Baz can complement the function of Crb regarding the targeting of Sdt/PATJ throughout embryogenesis. Indeed, high magnification of Baz-Crb-PATJ staining in late-stage embryos (stage 12) shows a partial overlap of PATJ with both proteins (arrows), whereas Baz and Crb are more clearly separated (Fig. 1K). In earlier stages of epithelial development (stage 8), the segregation of Baz/Crb is less pronounced, and PATJ colocalizes with both proteins (Fig. 1K). These data are in line with previous studies describing the segregation of AJ-associated Baz from Crb, which localizes more apically to the subapical region (44, 45).

PATJ Localization in the Follicular Epithelium Depends on Sdt and on Crb but Not on Baz

Similar as in the embryonic epidermis, loss of Sdt in the epithelial cells surrounding the oocyte (follicular epithelium) abolishes apical accumulation of PATJ (Fig. 2A, mutant clones are marked by the absence of RFP). In contrast, in baz-mutant clones, Sdt, PATJ, and Crb are correctly localized to the apical junctions (data not shown and Fig. 2B, mutant clones are marked by the absence of Baz staining), which is in line with recent results showing that baz null alleles do not exhibit polarity defects (32).

FIGURE 2.

A and B, PATJ is lost from the apical junctions in sdt (A) but not in baz (B) mutant clones (arrows). C, PATJ localization is only partly retained in crb mutant clones (arrows). D, in crb mutant follicle cells, localization of Baz to the apical junctions is diminished (arrows). Mutant cells were generated using the FRT-Flp technique and marked by the absence of RFP (A), Baz (B), or GFP (C and D). Note the flat cells in sdt and crb mutant cells (A, C, and D). The anterior is left in all egg chambers. Scale bars = 10 μm.

In crb-defective follicle cells, apical Sdt and PATJ staining is drastically reduced (Fig. 2, C and D, arrows), which is partly in line with previous data (12). Notably, Baz localization is also affected upon removal of Crb in the follicular epithelium (Fig. 2D). Therefore, the follicular epithelium represents a phenotypic characteristic that differs from the epidermis (PATJ localization is only dependent on Baz-Sdt but not on Crb-Sdt) and resembles rather pupal photoreceptor cells in which PATJ localization depends on Crb (10, 46). Notably, in photoreceptor cells, PATJ seems to be crucial for the stabilization of the Crb-Sdt complex (10, 11, 22), whereas this phenotype is much weaker in the follicular epithelium. In follicle epithelial cells, loss of PATJ results in decreased apical junctional accumulation of Crb/Sdt but without subsequent disassembly of the complex and polarity defects. In the embryonic epidermis, loss of PATJ does not affect Crb/Sdt localization or apical-basal polarity (21–23).

The L27 Domain Is Essential and Sufficient for Apical Junctional Localization

To test which domains are crucial for the correct subcellular localization and function of PATJ, we generated deletion constructs of the N-terminal L27 domain and each of the PDZ domains as well as truncated versions of PATJ, all C-terminally tagged with GFP (Fig. 3A). To avoid artificially increased protein levels, we expressed the modified proteins under a ubiquitous promoter (ubiquitin) and used the PhiC31-Integrase system (47) to generate transgenic lines with an identical genomic background, ensuring comparable protein levels (Fig. 3B). Indeed, wild-type PATJ-GFP expressed by this system shows similar levels as endogenous PATJ (Fig. 3B). The exogenous protein is localized indistinguishably from endogenous PATJ (Fig. 3C) and it is capable of rescuing the PATJ^Δ1 null allele to a large extent (79% surviving flies, Fig. 3A).

FIGURE 3.

A, schematic of the different PATJ constructs tested in this study. The capacity to correctly localize to the apical junctions and to rescue a PATJ null allele (maternal and zygotic mutant PATJ^Δ1, n = 300) is indicated. B, Western blot analysis of embryonic lysates from Ubi::PATJ-GFP versus wild-type flies with anti-PATJ antibody indicates that PATJ-GFP is expressed at similar levels as the endogenous protein. An equal amount of total protein was loaded, as verified with anti-actin Western blot analysis. The results of three independent experiments were quantified using a chemiluminescence scanner and are depicted as mean ± S.E. C and D, PATJ-GFP (C) localizes to the apical junctions indistinguishably from endogenous PATJ, whereas deletion of the L27 domain (D) disrupts junctional accumulation. E, the isolated L27 domain is sufficient to localize to a far extent at the apical junctions. In all panels, GFP is depicted in green, Crb in red, and Baz in blue. Scale bars = 5 μm.

In mammalian epithelial cells, mPATJ has been shown to be targeted by Pals1 to the TJ via a heterodimerization of its L27 domains (17, 19, 20). Likewise, deletion of the L27 domain of Drosophila PATJ results in an abolished association with Sdt (Fig. 1G) and a cytoplasmic accumulation of the mutant protein in the embryonic epidermis as well as in follicle cells (Fig. 3D and data not shown). Consequently, PATJ_ΔL27-GFP is unable to rescue the PATJ-null allele, resulting in similar phenotypes as the null allele (PATJ^Δ1, pupal lethality).

In contrast to deletion of the L27-domain, the removal of any of the four PDZ domains alone does not impair the subcellular localization of the modified protein at the apical junctions (Fig. 4, A–D). Furthermore, ubiquitous expression of all single deletion constructs can complement the function of PATJ and can be maintained as a stable stock with the homozygous PATJ^Δ1 allele. However, quantification of the rescue capacity showed that the deletion of the first PDZ domain affects the functionality of the protein far more than deletion of PDZ2, PDZ3, or PDZ4 (34% in comparison with 58%, 55%, and 68%, respectively; Fig. 3A).

FIGURE 4.

A–H, PATJ deletion and truncation proteins expressed with a ubiquitous promoter were stained in the embryonic epidermis using a GFP antibody (green). Coimmunostaining with endogenous Crb (red) and Baz (blue) reveals a partial overlap of these proteins. I, expression of PATJ deletion and truncation proteins was tested in Western blotting using a GFP antibody. Scale bars = 5 μm.

Because overexpression of a truncated version of PATJ has been reported to be capable of rescuing a PATJ mutant to some extent (10, 11, 23), we determined which minimal region of PATJ is sufficient for the function of the protein. As expected, ubiquitous expression of the isolated L27 domain (PATJ_1–151) shows a mostly apical localization (Fig. 3E). This protein, lacking all PDZ domains, does not rescue the PATJ null allele. Experiments with flies lacking zygotic PATJ expression and ubiquitously expressed PATJ_1–240-GFP (L27 domain and the first PDZ domain, Fig. 4H) occasionally produced adult flies. The majority of homozygous flies died during late pupal stages, but, in contrast to the null allele, pupae in the PATJ_1–240 rescue underwent complete morphogenesis and died only shortly before hatching (or failed to hatch). Hatched flies were sterile and died after a few days, indicating that the truncated version exhibits sufficient functionality to overcome the pupal lethality of PATJ^Δ1, but it is not capable of fully replacing the wild-type protein. Overexpression of the same construct using arm::GAL4 resulted in increased rescue capacity, and the recued flies can be maintained as a stable stock. Therefore, only artificially increased levels of the protein consisting of the L27 domain and the first PDZ domain can accomplish the function of PATJ during development, which is in line with previous studies using overexpressed proteins (11, 23).

In contrast to PATJ_1–240, a protein consisting of the first 449 amino acids, including the L27 domain and the first two PDZ domains (PATJ_1–449, Fig. 4G), expressed at close to endogenous levels, can fully rescue the PATJ null allele (Fig. 3A). Rescued flies can be kept as a stable stock. Deletion of the first PDZ domain in this construct (resulting in PATJ_{1–449 ΔPDZ1}-GFP) results, again, in a loss of functionality, as seen in rescue experiments.

These results suggest that none of the PDZ domains are essential for the viability of the fly. Under overexpression conditions, the first PDZ domain is sufficient for viability of the fly. This is further supported by the observation that, upon deletion of the first two PDZ domains (PATJ_{ΔPDZ1 + 2}, Fig. 4E) or the first and the fourth PDZ domain (PATJ_{ΔPDZ1 + 4}, Fig. 4F), the mutated protein can still rescue PATJ^Δ1. However, survival rates (Fig. 3A) indicate that deletion of more than one PDZ domain strongly reduces the functionality of PATJ.

Taken together, our data revealed that none of the four PDZ domains are essential for survival of the fly, although they seem to play a more subtle role during Drosophila development, as suggested by the different rescue capacities. Taking into account that all four PDZ domains exhibit only 50–60% sequence similarity (the sum of identical and similar amino acids), these results are remarkable.

Association with Both Crb-Sdt and Baz-Sdt Complexes Rather than Apical Junctional Localization Is Essential for the Function of PATJ

To test whether the association of PATJ with junctional Baz/Crb is crucial for its function or whether an apical junctional accumulation is sufficient, we cloned the PDZ domain of Sdt to PATJ_ΔL27-GFP (PATJ_ΔL27-_PDZ(Sdt), Fig. 5A). This domain has been reported to bind to Crb and to Baz (6, 13, 14).

FIGURE 5.

A, Schematic of different PATJ constructs tested in this study. The capacity to correctly localize to the apical junctions and to rescue a PATJ null allele (maternal and zygotic mutant PATJ^Δ1, n = 300) is indicated. B–D, F, and G, subcellular localization of chimeric proteins (described in A) reveals only partial colocalization of the chimeric proteins with endogenous Crb and Baz. GFP is depicted in green, Crb in red, and Baz in blue. Scale bars = 5 μm. E and H, immunostaining with a PAR6 or Yrt antibody (both shown in red) demonstrates normal localization of these proteins in embryos lacking endogenous PATJ and expressing PATJ_ΔL27-_PDZ(PAR6) or PATJ_ΔL27-_FERM(Yrt), respectively. Chimeric PATJ protein is depicted in green, and Baz in blue. Scale bar = 5 μm. I, embryonic lysates (500 μg of total protein) from PATJ chimeric protein-expressing flies were subjected to immunoprecipitation using a GFP antibody, followed by Western blotting with the indicated antibodies. Wild-type flies served as a negative control. Notably, although all constructs were expressed with the same promoter (ubiquitin) from the same landing site (attP40), and the same amount of total protein was used for the Western blot, some chimeric proteins were expressed more weakly, suggesting either differences in mRNA stability/degradation or a posttranslational mechanism. J and K, The Myosin binding subunit myc (expressed via the UAS/GAL4 system with da::GAL4) and PATJ coimmunoprecipitate with endogenous Baz (J) and Crb (K).

We verified the interaction with both proteins in transgenic flies (Fig. 5I). Notably, the localization of this chimeric protein is more or less cytosolic, with only a minor fraction accumulating at the apical junctions (Fig. 5B). This might be due to the fact that the protein level of Sdt is restrictively controlled. Even a moderately increased protein amount leads to an entirely cytosolic localization of Sdt (data not shown). Nonetheless, PATJ_ΔL27-_PDZ(Sdt) restores, to some extent, the rescue capacity of the protein (19% hatching flies, Fig. 5A). The addition of the PDZ domain of PAR6, which is capable of directly binding to both Baz (48, 49) and Crb (4, 50) (Fig. 5I) to PATJ_ΔL27, results in an apical accumulation of the chimeric protein, although a substantial amount is still cytosolic (Fig. 5C). PATJ_ΔL27-_PDZ(PAR6) rescues the PATJ null allele similarly to PATJ_ΔL27-_PDZ(Sdt) (13% hatching flies, Fig. 5A). We verified that the low rescue capacity in the strains expressing the described chimeric proteins is not due to mislocalization of endogenous PAR6 (Fig. 5E) and Sdt (indirect evidence because Crb is not mislocalized as it would be if Sdt was lost from the subapical region; Fig. 5B).

In contrast, a protein composed of the three PDZ domains of PATJ and a fragment of Baz that accumulates at the apical junctions by direct binding to the plasma membrane (31), is, to a large extent, correctly targeted to the apical junctions (PATJ_ΔL27-_LB(Baz), Fig. 5D) but does not rescue the PATJ null allele (Fig. 5A). Remarkably, PATJ_ΔL27-_LB(Baz) protein expression is lower in comparison with PATJ_ΔL27-_PDZ(Sdt) and PATJ_ΔL27-_PDZ(PAR6) chimeric proteins. However, even upon overexpression of the chimeric protein, PATJ_ΔL27-_LB(Baz) cannot rescue the PATJ null allele (data not shown).

As outlined above, Baz is essential to initially recruit Sdt to apical junctions. In later stages, this complex (partly) disassembles by phosphorylation of Baz by aPKC, resulting in apically enriched Sdt, which is capable of stabilizing Crb. To determine whether PATJ exhibits its function through the Baz-Sdt or via the Crb-Sdt complex, we established chimeric PATJ proteins lacking the Sdt-binding domain and exhibiting either a Crb-binding domain (FERM domain of Yrt (41)) or a Baz-binding domain (oligomerization domain CR1 (51, 52). Both chimeric proteins do not exhibit a dominant-negative phenotype by mislocalizing endogenous Baz or Yrt (Fig. 5, G and H).

Interestingly, although PATJ_ΔL27-_CR1(Baz) and PATJ_ΔL27-_FERM(Yurt) localize, at least to some extent, correctly to the apical junctions (Fig. 5, F and G), none of these chimeric proteins are capable of rescuing PATJ^Δ1 (Fig. 5A). Notably, PATJ_ΔL27-_CR1(Baz) and PATJ_ΔL27-_FERM(Yurt), in contrast to PATJ_ΔL27-_PDZ(Sdt) and PATJ_ΔL27-_PDZ(PAR6), seem to be stabilized posttranslationally (Fig. 5G). However, these elevated protein levels cannot be the reason for the lack of rescue capacity because expression of the chimeric proteins on a wild-type background does not result in increased lethality or obvious phenotypes, as would be expected for dominant-negative proteins.

These results suggest that an association with both apical junctional complexes is essential for the function of PATJ and that the targeting competence to these complexes is the most important feature of the L27 domain of PATJ. Our data further indicate that a transient association with certain apical polarity protein complexes (Crb and Baz complex), rather than a direct targeting to the apical-junctional compartment, is crucial for the function of PATJ during the development of Drosophila.

One possible explanation for these results is that the association of PATJ with both complexes, Baz-Sdt and Crb-Sdt, is essential for the function of PATJ. This might be explained by the implication of PATJ in the regulation of the Actin/Myosin cytoskeleton. By modulating Myosin phosphatase, PATJ regulates Myosin activity, which is essential for several morphological processes, including metamorphosis (21). We furthermore showed that PATJ associates with Myosin in vivo by direct interaction with the Myosin regulatory light chain (spaghetti-squashed, sqh in Drosophila) (21). This is in line with results from mammalian PATJ, suggesting that mPATJ regulates Myosin-driven processes such as cell migration and apical constriction (27–29). mPATJ associates with the RhoA-GTPase exchange factor (GEF) Syx to control RhoA activity in lamellipodia, controlling the migration of endothelial cells (29). Another Rho-GEF (p114RhoGEF) has been found to be recruited by mPATJ to regulate Myosin dynamics in the circumferential actomyosin belt during apical constriction (28). Strikingly, aPKC modulates p114RhoGEF activity by phosphorylating the adaptor protein Lulu2. aPKC and its modulator PAR-6, in turn, have been shown to associate with both Baz/PAR-3 and Crb (7, 44, 53). Therefore, association of PATJ with both complexes might serve as a scaffold to assemble different Myosin-modulating proteins in a defined compartment of the cell.

PATJ might also function in different processes in the two distinct complexes that are localized to different cellular compartments. Baz associates with the AJ (44, 54), which anchors Actin-Myosin filaments as well as Myosin-modulating enzymes (55, 56). Therefore, Sdt-PATJ recruitment to the AJ would provide a mechanism for Myosin modulation, as described before (21). On the other hand, Crb has been described to link the Actin cytoskeleton via the Moesin and β_heavy-chain spectrin to the plasma membrane in a compartment between the AJ and the free apical membrane (57). Interestingly, control of Myosin and Moesin activity by (de)phosphorylation is accomplished by the same set of enzymes (58, 59). Indeed, the Myosin binding subunit of Myosin phosphatase associates with both the Baz-Sdt-PATJ and the Crb-Sdt-PATJ complex (Fig. 5, J and K). Therefore, the association of PATJ with both apical complexes might be crucial to control (de)phosphorylation of Myosin and Moesin during morphogenetic events in the development of Drosophila.