A novel protein complex, Mesh–Ssk, is required for septate junction formation in the Drosophila midgut

Epithelia play important roles as barriers that separate distinct compartments within the body. To accomplish these functions, epithelial cells have specialized intercellular junctions, designated as occluding junctions, that restrict the free diffusion of solutes across the cellular sheets through the paracellular pathway. In vertebrates, tight junctions (TJs) act as occluding junctions in all epithelia, including endothelial cells. These barrier/channel properties are determined primarily by membrane proteins of the claudin family (Anderson and Van Itallie, 2009; Angelow et al., 2008; Furuse, 2010).

In contrast to vertebrates, the epithelial cells of invertebrates generally lack TJs (although a few exceptions have been reported). Instead, they possess different membrane specializations, called septate junctions (SJs), which perform the role of occluding junctions (Lane et al., 1994a; Tepass and Hartenstein, 1994). In ultrathin-section electron microscopy, SJs are observed as parallel plasma membranes between adjacent cells with ladder-like septa spanning the intermembrane space. Morphological variants of SJs exist across the invertebrate phyla and some animals are reported to possess multiple types of SJs specific to different types of epithelial cells (Lane et al., 1994b; Green and Bergquist, 1982). However, the molecular architectures of these SJ types are largely unknown. In arthropods, two major classes of SJs have been described, based on morphological appearance: pleated SJs (pSJs) are observed in ectodermally derived epithelia and glia, while smooth SJs (sSJs) are found mainly in the endodermally derived midgut epithelium (Lane et al., 1994a; Tepass and Hartenstein, 1994). The outer epithelial layer of the proventriculus (OELP) and the Malpighian tubules also possess sSJs, although developmentally they originate from the ectoderm. The major criteria distinguishing these two types of SJs are the arrangement of the septa visualized in negatively stained membrane preparations and the appearance of intramembrane particles observed in freeze-fracture images. The septa in pSJs form regular undulating rows but those in sSJs are arranged in regularly spaced parallel lines. This structural difference seems to reflect differences in the molecular architecture of sSJs and pSJs. Among these two SJ types, the molecular and functional properties of pSJs have been extensively analyzed in Drosophila ectodermal epithelia. The molecular components of Drosophila pSJs include: the transmembrane proteins, Neurexin IV (Baumgartner et al., 1996), Neuroglian (Banerjee et al., 2010), Gliotactin (Schulte et al., 2003), Contactin (Faivre-Sarrailh et al., 2004), Fasciclin III (FasIII) (Woods et al., 1997), and Lachesin (Llimargas et al., 2004); an Na⁺/K⁺ ATPase (Paul et al., 2003); and the cytoplasmic proteins, Coracle (Cora) (Lamb et al., 1998), Discs large (Dlg) (Woods et al., 1996), Lethal (2) giant larvae (Lgl) (Bilder et al., 2000), Scribble (Scrib) (Bilder and Perrimon, 2000), and Varicose (Wu et al., 2007). Among them, it was recently reported that Dlg is unlikely to be a core pSJ component (Oshima and Fehon, 2011). The vertebrate homologs of Neurexin IV, Neuroglian Contactin and Cora are concentrated at the paranodal junctions (PJs) of vertebrate myelinated axons, which possess ladder-like structures similar to those observed in SJs (Bhat, 2003). Thus, the molecular organization and morphology of pSJs are similar to that of PJs. However, three claudin-like proteins, Megatrachea (Behr et al., 2003), Sinuous (Wu et al., 2004) and Kune-kune (Kune) (Nelson et al., 2010), have been identified as functional pSJ components, suggesting that pSJs also have some common features with TJs.

Morphological and physiological studies have suggested that sSJs function to restrict or regulate the diffusion of solutes through the paracellular pathway (Skaer et al., 1987) but detailed molecular and genetic analyses of sSJs are lacking. In Drosophila, Ankyrin, α/β-spectrin, FasIII (Baumann, 2001) and Dlg (Maynard et al., 2010) are localized at the apicolateral region of midgut epithelial cells, and Lgl is localized at the sSJs of the proventriculus (Strand et al., 1994).

Recently, we identified a novel protein with four membrane-spanning domains, Snakeskin (Ssk), which specifically localizes at sSJs and is required for the organization and function of sSJs (Yanagihashi et al., 2012). Here, we identify a previously uncharacterized putative membrane protein, which we have named ‘Mesh’. Mesh specifically localizes at sSJs, induces cell–cell adhesion in cultured cells, and is required for the formation and function of sSJs in Drosophila. We also found that Mesh and Ssk display mutually dependent localizations at sSJs and form a complex with each other. Therefore, we conclude that Mesh acts together with Ssk to organize sSJs.

To further identify sSJ-specific molecules, we generated monoclonal antibodies (mAbs) in rats against an sSJ-containing membrane fraction obtained from the midgut of silkworm (Bombyx mori) fifth-instar larvae, and finally isolated two mAb clones that specifically recognized the apical region of the lateral membrane of midgut epithelial cells, where sSJs occur (Fig. 1A). Immunoprecipitation of the midgut membrane fraction with these mAbs identified a ∼80 kDa protein (Fig. 1B). Mass spectrometry revealed this protein to be silkworm BGIBMGA009402-PA (supplementary material Fig. S1). The primary structure of this protein contains a domain characteristic of a transmembrane-spanning segment close to the C-terminus, a signal peptide, a NIDO domain, an Ig-like E set domain, an AMOP domain, a vWD domain, and a sushi domain (supplementary material Fig. S1). These extracellular domains are found in cell adhesion proteins playing important roles in cell–cell and/or cell–matrix adhesion (Bork et al., 1994; Ciccarelli et al., 2002; Colombatti et al., 1993; Ichinose et al., 1990; Mayer et al., 1998). To further investigate the function of this protein in Drosophila, we looked for its Drosophila ortholog by database searching and found the CG31004 gene (Fig. 1D; supplementary material Fig. S1), which is located on the right arm of the third chromosome. We named CG31004 protein ‘Mesh’ for its immunofluorescence staining images in Drosophila midgut (see below). Proteins characterized by similar domain compositions exist in other invertebrates, including Caenorhabditis elegans (K03H1.5) and sea urchins (LOC580458). In vertebrates, the mouse Susd2/Svs-1 ortholog is the sole protein containing the AMOP, vWD, and sushi domains (supplementary material Fig. S1), suggesting that Susd2/Svs-1 is a vertebrate ortholog of Mesh (Sugahara et al., 2007).

The Flybase predicts that Mesh transcripts are translated into three isoforms with different C-terminal cytoplasmic regions (Fig. 1C,D). A piggyBac insertion, pBac{WH}CG31004^f04955 is located in the region shown in the schematic drawing of the mesh gene and the protein (Fig. 1C,D). Embryos homozygous for the mesh^f04955 chromosome hatched into first-instar larvae but died at this stage. Df(3R)Excel6218 or Df(3R)tll-e, both of which lack the mesh locus, failed to complement the lethality of mesh^f04955. The lethality of mesh^f04955 homozygotes was rescued by precise excision of pBac{WH}CG31004^f04955 and the expression of a mesh-RNAi using the 48Y-GAL4 driver at 25°C caused lethality at the first-instar larval stage (data not shown), demonstrating that the lethality is attributable to the piggyBac insertion in the mesh gene. In addition, transheterozygotes for mesh^f04955 and Df(3R)Excel6218 showed identical phenotypes to the phenotype of mesh^f04955 homozygotes (see below; also supplementary material Fig. S4A,B), and expression of Mesh with a 48Y-GAL4 driver in mesh^f04955 homozygotes rescued their phenotype regarding sSJ organization (Fig. 3C′,F′). We confirmed that mesh^f04955 eliminated the immunostaining of Mesh and that the expression of a UAS-mesh in mesh^f04955 rescued the Mesh staining (Fig. 3B,C,E,F). Taken together, these observations indicate that mesh is an essential gene and that mesh^f04955 is a null or strong loss-of-function allele of mesh.

To determine the expression pattern and subcellular localization of Mesh in Drosophila, anti-Mesh antibodies were generated against the C-terminal cytoplasmic region. Western blot analysis revealed that Mesh was mainly detected as a protein of relative molecular mass 90,000 in embryos, in third-instar larvae, and in extracts of S2 cells expressing Mesh (supplementary material Fig. S2A). Mesh-PA/PB consists of 1431 amino acids with a calculated molecular mass of 162,400, suggesting that the protein is processed at a specific region. Indeed, higher-molecular-mass bands (∼200,000) were detected (supplementary material Fig. S2A), and a putative GDPH cleavage site, an autocatalytic proteolysis site in some mucins that cleaves between GD and PH residues (Hollingsworth and Swanson, 2004), is located in the vWD domain (a.a. 827–830 of Mesh-PA/PB) (Fig. 1D). Immunofluorescence microscopic analyses revealed that the expression of Mesh protein was first observed in the endodermally derived tissues at embryonic stage 12 (Fig. 2A). In late-stage embryos and third-instar larvae, Mesh was expressed in the midgut, OELP and Malpighian tubules (Fig. 2A–E), but was not expressed in the foregut and hindgut (Fig. 2B,D,E; supplementary material Fig. S2C), demonstrating that the expression of Mesh is specific for tissues bearing sSJs. The immunoreactivities of these antibodies were diminished in mesh mutant embryos and first-instar larvae, indicating the specificity of our anti-Mesh antibodies (supplementary material Fig. S2B,D). The expression pattern, timing, and subcellular localization of Mesh correspond with those of Ssk, a previously identified sSJ-specific protein (supplementary material Fig. S3A; Fig. 3A–A′″,D–D′″). To confirm Mesh localization at sSJs, we carried out immunoelectron microscopy using anti-Mesh antibody. As shown in Fig. 2F, immunolabels were detected at bicellular contacts where septa were observed, indicating that Mesh specifically localizes at sSJs in larval midgut epithelial cells. Expression of Mesh was also observed in the apicolateral region of epithelial cells in the adult midgut, OELP, and Malpighian tubules (supplementary material Fig. S3B), indicating that Mesh is a component of sSJs in Drosophila from the embryo through to adulthood.

Epithelia derived from ectoderm and endoderm possess pSJs and sSJs, respectively, raising an intriguing question of how the SJs at their boundary are organized. The specific localization of Mesh at sSJs enabled us to investigate this issue. Stage-16 embryos were double-stained with antibodies to Mesh and Kune as markers for sSJs and pSJs, respectively. Their localizations were closely examined in the proventriculus, which includes the boundary between the ectodermally derived foregut and endodermally derived midgut. As shown in Fig. 2G, we identified boundary cells expressing both Mesh and Kune (Fig. 2G–G″, asterisk). In these cells, Kune localized at the apicolateral membrane on the foregut side and Mesh localized on the midgut side (Fig. 2G″), suggesting that individual boundary cells possess both pSJs and sSJs depending on which cells they are adjacent to.

As described above, the mesh mutant animals hatched into the first-instar larvae, but died within 1 day. However, sSJs are not completed until late stage 17 (Tepass and Hartenstein, 1994). Therefore, in the present study, we analyzed sSJ formation in the first-instar larvae. We focused on the OELP and the anterior midgut epithelial cells because sSJ organization is clearest in these columnar cells. Several pSJ proteins including Dlg, Lgl, and FasIII have been reported to localize at the apicolateral region of bicellular contacts in Drosophila midgut (Baumann, 2001; Maynard et al., 2010; Strand et al., 1994). We confirmed that these proteins all colocalized with Mesh in the apicolateral region of wild-type OELP and midgut epithelial cells (Fig. 3A″,D″,G′,G″,I′,I″,O,P; and data not shown). We also checked whether other pSJ proteins localize at sSJs and observed that Cora colocalized with Mesh at the apicolateral region (Fig. 3K,L; and data not shown). These results indicate that Dlg, FasIII, Lgl, and Cora, at least, are both sSJ components and pSJ components.

To examine the role of Mesh in the molecular organization of sSJs, we analyzed the subcellular localization of the sSJ proteins in mesh mutants. Ssk was mislocalized to the apical and basolateral membranes of the OELP and midgut epithelial cells in mesh mutant larvae (Fig. 3B′,E′), and was often observed as aggregates in the cytoplasm (Fig. 3E′; supplementary material Fig. S4B). The intensity of Ssk signals in the apical membrane was much higher than that of the basolateral membrane. In contrast, Dlg was still localized at the apicolateral region (Fig. 3B″,E″,H″,J″), indicating that Mesh is not required for the localization of Dlg. Moreover, the polarized distribution of Dlg in mesh mutants suggests that Mesh does not have a significant role in specifying the apical-basal polarity of either OELP or midgut epithelial cells. Lgl was distributed along the lateral membrane with partial concentration in the apicolateral region in the mesh mutant OELP and midgut epithelial cells (Fig. 3H′,J′). Cora was observed in the apicolateral region in the wild-type OELP and midgut epithelial cells but it was spread more in the basal direction in the mesh mutant OELP and was distributed throughout the cytoplasm of the midgut epithelial cells (Fig. 3M,N). In the mesh mutant OELP, FasIII was localized in the apicolateral region but also mislocalized to the apical membrane (Fig. 3Q). In contrast, it was observed as large aggregates in the apicolateral region of the midgut epithelial cells (Fig. 3R). Expression of the UAS-mesh construct with 48Y-GAL4 rescued Ssk localization (Fig. 3C′,F′). In addition, mesh-RNAi (12074-R1 generated by NIG-FLY) induced by 48Y-GAL4 on mesh^f04955/+ backgrounds decreased the level of Mesh at sSJs and caused mislocalization of Ssk in the midgut epithelial cells (supplementary material Fig. S5). Taken together, these results indicate that Mesh determines the proper localization of several sSJ proteins.

To further characterize the nature of the sSJ defect in mesh mutants, ultrastructural analysis of the first-instar larvae was performed. In wild-type midgut epithelial cells, typical sSJs were observed at cell–cell contacts (Fig. 4A, brackets). In mesh mutants, which were transheterozygotes for mesh^f04955 and Df(3R)Excel6218, large gaps between the lateral membranes of adjacent epithelial cells were frequently observed compared with the wild-type (Fig. 4B,C, asterisks). However, a few septa were still observed at the cell–cell contacts of the mesh mutant midgut epithelial cells (Fig. 4B,C, brackets). pSJs in the mesh mutant epidermis (Fig. 4E, bracket) were indistinguishable from those in the wild-type (Fig. 4D, bracket). These results indicate that Mesh is specifically required for proper sSJ organization.

We speculated that Mesh is involved in the barrier function of the midgut epithelium. However, mesh mutant larvae fail to form the three-layered structure of the proventriculus (supplementary material Fig. S4D), although the structure is formed correctly in stage-16 embryos, suggesting that mesh mutant animals cannot maintain the proper structure of the proventriculus. Consistent with this observation, colored yeast fed to mesh mutant larvae did not accumulate in the gut, whereas it was observed throughout their gut in wild-type larvae (data not shown). This phenotype hampered the dye permeability assay used to examine the integrity of the paracellular barrier, by feeding with a fluorescent dye tracer. To overcome this problem, we generated mesh weak loss-of-function conditions using mesh-RNAi (12074-R1) induced by 48Y-GAL4 on mesh^f04955 heterozygous backgrounds. Uninduced control first-instar larvae (UAS-mesh-RNAi/mesh^f04955) and mesh-RNAi-induced first-instar larvae on wild-type (48Y-GAL4 >UAS-mesh-RNAi/+) or mesh^f04955 heterozygous (48Y-GAL4 >UAS-mesh-RNAi/mesh^f04955) backgrounds were fed fluorescent-labeled dextran of 10 kDa and observed by confocal microscopy. In control larvae, the midgut was well contrasted, with the fluorescent tracer confined within the midgut (Fig. 5, upper panel). In contrast, the tracer was detected in various parts of the body cavity in mesh-RNAi-expressing mesh^f04955 heterozygous larvae (Fig. 5, middle panel), indicating leakage of the tracer from the lumen of the midgut. These observations indicate that Mesh is required for the barrier function of the midgut epithelium in Drosophila. However, we did not observe significant leakage of the tracer on the mesh-RNAi-induced wild-type background (Fig. 5, lower panel), suggesting insufficient RNAi-mediated suppression of Mesh on the wild-type background.

Since Ssk was mislocalized in mesh mutant sSJs, we next investigated whether the localization of Mesh would be affected by suppression of Ssk. As described in our previous report (Yanagihashi et al., 2012), animals expressing ssk-RNAi with the 48Y-GAL4 driver exhibited a reduction in Ssk expression, while those homozygous for Df(3L)ssk showed no expression of Ssk in the midgut epithelial cells (Fig. 6B′,C′). In these cells, Mesh no longer localized to the apicolateral region but was distributed diffusely and formed some aggregates in the cytoplasm (Fig. 6B,C). Thus, Mesh and Ssk are mutually dependent on each other for their proper localization; Mesh is required for the accumulation of Ssk at sSJs, and Ssk is required for the translocation of Mesh from cytoplasm to sSJs. As observed in mesh mutants, Lgl (Fig. 6E), Cora (Fig. 6G) and FasIII (Fig. 6I) were mislocalized in Ssk-deficient cells. However, Dlg was still localized at apicolateral region (Fig. 6B″,C″). Taken together, these results suggest that Mesh acts together with Ssk to organize sSJs.

Next, we investigated the localization of Mesh in dlg, lgl, cora and fasIII null mutants. In dlg^m52 and lgl⁴ zygotic mutants, Mesh and Ssk accumulated at the apicolateral region in the OELP and midgut epithelial cells (data not shown), suggesting that Dlg and Lgl are not required for the maintenance of sSJs. Since the maternally supplied Dlg and Lgl are thought to be adequate for the establishment of cell polarity and sSJs organization, we examined the phenotype of dlg^m52 and lgl⁴ maternal/zygotic mutant sSJs. When eggs from wild-type animals were allowed to develop for 24 h at 25°C, they hatched into first-instar larvae and their midguts developed a tube-like structure. In contrast, dlg^m52 and lgl⁴ maternal/zygotic mutants exhibited a hypertrophied midgut phenotype (Manfruelli et al., 1996). In the midgut epithelial cells of these mutants, Mesh and Ssk accumulated in the apicolateral region with faint leakage to the lateral membrane (Fig. 6J,K). In fasIII^E25 mutants, Mesh was localized to the apicolateral region in the midgut epithelial cells (Fig. 6L). Since cora⁵ mutant animals fail to hatch into larvae, we observed Mesh localization in the stage-16 OELP, by which time Mesh as well as Cora had accumulated in the apicolateral region of the wild-type (Fig. 6M, data not shown). As shown in Fig. 6N, Mesh was localized to the apicolateral region of the stage-16 OELP in cora⁵ mutants. Taken together, these results indicate that the accumulation of Mesh within the apicolateral region of the plasma membrane depends on Ssk, but not on Dlg, Lgl, Cora or FasIII.

Mesh and Ssk were mutually dependent for their localization at sSJs (Figs 3,6), raising the possibility that Mesh is physically associated with Ssk. When the embryonic and larval extracts of Drosophila were subjected to immunoprecipitation with anti-Mesh antibodies, Ssk coprecipitated with Mesh (Fig. 7A,B). Consistently, Mesh coprecipitated with Ssk during immunoprecipitation from embryonic extracts with anti-Ssk antibodies (Fig. 7C). Neither Mesh nor Ssk was precipitated by the pre-immune sera (Fig. 7A–C). These results indicate that Mesh forms a complex with Ssk in vivo.

To investigate the possible role of Mesh as a cell–cell adhesion molecule, we transfected Drosophila S2 cells with a Mesh–EGFP expression vector and carried out an aggregation assay to examine their adhesive properties. When S2 cells expressing Mesh–EGFP were co-cultured with those expressing mCherry, only Mesh–EGFP-expressing cells formed the cell aggregation (Fig. 8A–C). Since S2 cells appear to lack endogenous Mesh (supplementary material Fig. S2A), this experiment shows that Mesh expression leads to cell aggregation in a homophilic manner. Furthermore, Mesh accumulated at cell–cell contact regions between two cells expressing Mesh–EGFP (Fig. 8E,F, arrows). These results suggest that Mesh organizes sSJs by mediating cell adhesion via its homophilic interaction.

We have identified a novel membrane-spanning protein, Mesh, which is specifically localized at sSJs and has cell adhesion activity. Mesh is required for the formation of sSJs and paracellular diffusion barriers in the Drosophila midgut. This study, together with the recent identification of Ssk (Yanagihashi et al., 2012), whose interaction with Mesh was shown in the present study, provides a key starting point for understanding sSJs, which must play crucial roles in the gut and renal functions of arthropods, at the molecular level.

Electron microscopic observations have shown that sSJs and pSJs can be distinguished morphologically. Obliquely sectioned pSJs and sSJs are visualized as regular undulating rows and regularly spaced parallel lines, respectively (Lane et al., 1994b), while both types of SJs have ladder-like structures in the intermembrane space. Of the two sSJ-specific integral membrane proteins, Ssk is unlikely to be the structural element of the septa in sSJs, because its extracellular loops are both too short (25 and 22 a.a., respectively) to bridge the intercellular space. In contrast, Mesh induces cell–cell adhesion, implying that it may be one of the components of the septa observed in ultrathin section electron microscopy. Faint ladder-like structures were still observed in the mesh mutants, suggesting that other membrane proteins also contribute to the septal structures. FasIII is such a candidate because it shows cell–cell adhesion activity (Snow et al., 1989) and was still distributed to the apicolateral region, as well as the apical region, in the mesh mutants (Fig. 3Q,R). However, fasIII null mutant flies are viable (Whitlock, 1993) and both Mesh and Ssk are normally localized at their sSJs, indicating that FasIII is dispensable for sSJ formation. FasIII may provide robustness to the Mesh–Ssk-mediated sSJ organization via its cell–cell adhesion activity.

The issue of how SJs are organized in cells at the boundary between pSJ- and sSJ-bearing epithelia is intriguing. Interestingly, we observed boundary cells in which the pSJ marker Kune and sSJ marker Mesh were concentrated in the anterior and posterior regions, respectively, of the apicolateral membranes. This result suggests that individual cells possess both pSJs and sSJs depending on the orientation of their plasma membranes. The proventriculus is originally derived from ectoderm (Tepass and Hartenstein, 1994). However, the OELP bears sSJs and expresses Mesh and Ssk, suggesting that the OELP has both ectodermal and endodermal characters. In fact, we observed weak Kune expression in the OELP but not in the midgut (Fig. 2G″). Therefore, the boundary cell may have the ability to form either sSJs or pSJs according to the SJ type of adjacent cells. The occurrence of such ‘SJ-boundary cells’ seems to be crucial because they connect the ectodermally and endodermally derived epithelia into a tandem tube while maintaining the continuity of the paracellular barrier. However, we cannot completely exclude the possibility that small amounts of pSJs and sSJs are also contained in the sSJs on the midgut side and pSJs on the foregut side of the SJ-boundary cells, respectively, to form hybrid junctions.

Our analyses of Mesh and Ssk have clarified their interaction, interdependency in their localizations, and requirements for the organization and barrier function of sSJs, suggesting that Mesh–Ssk is a key system for sSJ formation. In mesh mutants, Ssk failed to localize at sSJs, but mislocalized to the apical and basolateral plasma membrane domains. In ssk-RNAi and Df(3L)ssk fly, Mesh no longer localized at the sSJs, but was distributed in the cytoplasm. Ssk may translocate Mesh from the cytoplasm to sSJs or to the plasma membrane. However, how the Mesh–Ssk complex recognizes and localizes to sSJ regions remain elusive. Mesh expression in S2 cells leads to cell aggregation without Ssk expression, suggesting that there is a mechanism by which Mesh translocates to the cell membrane and induces cell–cell adhesion independently of Ssk in S2 cells. Detailed analysis of the dynamics of Mesh–Ssk distribution will shed light on the mechanisms of sSJ formation and the sorting systems for sSJ proteins.

By using Mesh and Ssk as specific markers for sSJs, we confirmed that Dlg, Lgl and FasIII localize at sSJs in the larval OELP and midgut epithelial cells. In addition, we found that Cora is also concentrated into sSJs. Among these proteins that are generally known as pSJ components, Lgl, Cora and FasIII were mislocalized in mesh mutants and ssk-RNAi lines. On the other hand, Lgl, Cora and FasIII were not required for the localization of Mesh and Ssk at the apicolateral membrane. These observations imply a possible hierarchy in the molecular constituents of sSJs; Mesh-Ssk might act as a platform for the assembly of Lgl, Cora and FasIII in endodermal epithelia. Such a feature in sSJs is in sharp contrast to that in pSJs where each molecular component is interdependent. Mutations in most of the genes encoding pSJ-associated proteins result in disruption of the barrier function and mislocalization of other pSJ proteins (Fehon et al., 1994; Baumgartner et al., 1996; Behr et al., 2003; Genova and Fehon, 2003; Paul et al., 2003; Schulte et al., 2003; Faivre-Sarrailh et al., 2004; Llimargas et al., 2004; Wu et al., 2004; Wu et al., 2007; Nelson et al., 2010).

Interestingly, in mesh mutants and ssk-RNAi lines, Dlg still localized at the apicolateral region of the OELP and midgut epithelial cells, although sSJs were disrupted at the ultrastructural level. Furthermore, Mesh and Ssk were distributed to the apicolateral region in dlg mutants, suggesting that Mesh-Ssk and Dlg are independent in their localizations. This is consistent with a recent report that Dlg is probably not a core pSJ component (Oshima and Fehon, 2011). Nevertheless, a functional relationship exists between Dlg and Lgl in determining cell polarity in ectodermally derived epithelia. Therefore, in the absence of Mesh and Ssk, Dlg may be unable to function properly because of an inadequate level of Lgl in the apicolateral regions. In fact, dlg^m52 and lgl⁴ maternal/zygotic mutants exhibited a similar hypertrophied midgut phenotype (data not shown), suggesting that these proteins may function together in endodermal epithelia, as well as in ectodermal epithelia (Bilder et al., 2003; Tanentzapf and Tepass, 2003).

The functions of Dlg, Lgl, Cora and FasIII at sSJs remain unknown. Dlg may act together with Lgl to regulate the apical-basal polarity in the early stage of epithelial development. In the late developmental stage, compensation mechanisms for the Dlg function may rescue the apicolateral localization of Mesh, as noted in ectodermally derived epithelial cells of dlg^m/z and lgl^m/z mutants (Bilder et al., 2003; Tanentzapf and Tepass, 2003). As larval midgut sSJs are completed at the end of embryogenesis (stage 17) in Drosophila (Tepass and Hartenstein, 1994), the organization of sSJ may not be influenced by early polarity defects of dlg^m/z and lgl^m/z mutants. Alternatively, Dlg and Lgl may be important for the regulation of the epithelial cell shape change that induces the midgut tube-like structure (Manfruelli et al., 1996). In ectodermally derived epithelia, Cora acts together with Yurt to regulate the apicobasal polarity (Laprise et al., 2006; Laprise et al., 2009). Thus, Cora and a Yurt-like molecule may function together to organize sSJs and/or to regulate the endodermal epithelial polarity.

Homologous proteins, characterized by similar extracellular domains to Mesh, are present in vertebrates (e.g. mouse Susd2/SVS-1), implying that this family of proteins shares functions conserved across species. Mouse Susd2/SVS-1 has been suggested as a tumor-reversing gene product, because it inhibited the growth of cancer cell lines (Sugahara et al., 2007). Susd2/SVS-1 was distributed in the apical membrane of the epithelial cells in renal tubules and bronchial tubes, suggesting that it does not contribute to the cell–cell adhesion and/or paracellular barrier function in vertebrate epithelial cells. However, expressing Susd2/SVS-1 in HeLa cells induces the cell aggregation (Sugahara et al., 2007), implying that this protein family conserves the cell–cell adhesion activity. Further studies of the functions of Mesh–Susd2/SVS-1 family proteins in vertebrates and in invertebrates will lead to a better understanding of the conserved physiological functions in these proteins and of the evolution of intercellular junctions across species.

The fly strains mesh^f04955, Df(3R)Exel6218, fasIII^E25, lgl⁴ and 48Y-GAL4, were obtained from the Bloomington Stock Center, and the mesh-RNAi strains, 12074-R1 was obtained from NIG-FLY. We also used the strains dlg^m52 (a gift from P. J. Bryant), cora⁵ (a gift from R. G. Fehon), UAS-ssk-RNAi and Df(3L)ssk (Yanagihashi et al., 2012). Germline clones of lgl⁴ and dlg^m52 were made by the FLP-DFS technique (Chou and Perrimon, 1992). For the phenotype rescue experiment, pUAST vectors (Brand and Perrimon, 1993) containing mesh were constructed and a fly strain carrying this construct was established. 48Y-GAL4, which drives GAL4 expression in the anterior and posterior midgut primordium from embryonic stage 10 (Martin-Bermudo et al., 1997), was used to express UAS-mesh in mesh^f04955 for the rescue experiment.

The membrane fraction was prepared from midguts of silkworm 5th-instar larvae according to the method described previously (Yanagihashi et al., 2012).

Rat mAbs against membrane fractions of silkworm fifth-instar larval midguts were generated as described previously (Yanagihashi et al., 2012). For identification of the antigens, the membrane fractions (∼500 mg) were centrifuged at the maximum speed in a microcentrifuge for 20 min and the pellet was resuspended in 500 ml of lysis buffer [25 mM Tris-HCl, pH 8, 27.5 mM NaCl, 20 mM KCl, 25 mM sucrose, 10 mM EDTA, 10 mM EGTA, 1 mM DTT, 10% (v/v) glycerol, 1% NP40 and protease inhibitor cocktail (Nakarai, Kyoto, Japan)] for 30 min at 4°C. The lysates were centrifuged at the maximum speed for 20 min, and the supernatants were used for immunoprecipitation with protein G sepharose (GE Healthcare) conjugated with the mAbs. The sepharose preparations were incubated with the supernatants for 4 h at 4°C and were washed five times in lysis buffer. Bound proteins were separated by SDS-PAGE and analyzed by Coomassie Brilliant Blue G-250 (Wako) staining. Mass spectrometry analyses of the tryptic peptide mass data were carried out by the Integrated Center for Mass Spectrometry (Kobe University Graduate School of Medicine). The resulting tryptic peptide mass data were matched against the NCBInr database using the Mascot program.

A region of the Mesh PA/PB protein (amino acids 1211–1431) was cloned into pGEX-6P (GE Healthcare) to produce a GST-fusion protein. The proteins were expressed in Escherichia coli. Polyclonal antibodies were generated in rabbits (995-1 and -2) and rats (8002) by MBL (Nagoya, Japan).

Embryos were fixed with 3.7% formaldehyde in PBS for 20 min. Larvae were dissected in Hanks' Balanced Salt Solution and fixed with 3.7% formaldehyde in PBS with 0.4% Triton X-100. The following antibodies were used: rabbit and rat anti-Mesh, rabbit anti-Ssk (6981-1; 1∶1000) (Yanagihashi et al., 2012), rabbit anti-Kune (1:1000) (Nelson et al., 2010), mouse anti-Dlg 1∶50 [Developmental Studies Hybridoma Bank (DSHB)], mouse anti-coracle C615.16 1∶50 (DSHB), mouse anti-FasIII 1∶20 (DSHB), rabbit anti-Lgl 1∶1000 (provided by F. Matsuzaki, RIKEN CDB). Alexa Fluor 488-conjugated (Invitrogen), and Cy3- and Cy5-conjugated (Jackson ImmunoResearch Laboratories) secondary antibodies were used at 1∶400. Samples were mounted in Vectashield (Vector Laboratories). Images were acquired with a confocal microscope (model TCS-SPE; Leica) with its accompanying software using HC PLAN Apochromat 20×NA 0.7 and HCX PL Apochromat 63×NA 1.4 objective lens (Leica). Images were processed with Adobe Photoshop®.

First-instar larvae of wild-type or mesh mutants were dissected and fixed overnight at 4°C with a mixture of 2.5% glutaraldehyde and 2% paraformaldehyde in 0.1 M cacodylate buffer (pH 7.4). The specimens including the midguts were prepared as described previously (Yanagihashi et al., 2012). For immunoelectron microscopy, first instar larvae were dissected and fixed for 2 h at room temperature with 4% paraformaldehyde in 0.1 M sodium phosphate buffer (PB) (pH 7.4). The specimens were washed three times with 50 mM glycine in PB and incubated with 0.1% saponin in PB. After blocking with 10% normal goat serum in PB for 1 h, they were incubated for 2 days at 4°C with anti-Mesh antibody (995-2; 1∶1000) diluted in the blocking solution. After six washes with PB, the specimens were incubated for 2 h with a secondary antibody that had been conjugated with both the 1.4 nm NANOGOLD particles (1∶100; Nanoprobes, Inc.), followed by six washes. The specimens were fixed for 15 min with 2.5% glutaraldehyde in PB, washed with 50 mM glycine in PB, and again four times with 50 mM HEPES, pH 5.8 for 15 min. Signals were silver-enhanced by use of an HQ-silver kit (Nanoprobes, Inc.) for 14 min in the dark. After thorough washing with distilled water, they were fixed with 0.5% osmium oxide in PB for 1.5 h on ice and washed again with distilled water. Subsequently the specimens were embedded with Epon 812. The ultrathin sections (50–100 nm) were stained doubly with 4% hafnium (IV) chloride and lead citrate, and observed with a JEM-1011 electron microscope (JEOL) at an accelerating voltage of 80 kV.

Wild-type fly embryos and third-instar larvae were mixed with a 5-fold volume of lysis buffer [25 mM Tris-HCl pH 8, 27.5 mM NaCl, 20 mM KCl, 25 mM Sucrose, 10 mM EDTA, 10 mM EGTA, 1 mM DTT, 10% (v/v) glycerol, 0.5% NP40 and protease inhibitor cocktail from Sigma] and homogenized using a pestle for 1.5 ml microfuge tubes. The method for immunoprecipitation was essentially the same as described above. Anti-Mesh (995-1 and -2) and anti-Ssk (6981-1 and -2) antibodies were used for the immunoprecipitation and the immunocomplexes were separated by SDS-PAGE, transferred to polyvinylidene difluoride membranes and probed with an anti-Ssk (6981-1; 1∶1000) and an anti-Mesh (995-1; 1∶1000) antibodies.

For the ectopic expression of Mesh in S2 cells, Mesh–EGFP and mCherry (Clontech) cDNA were subcloned into pMT-V5His (Invitrogen) and EGFP cDNA was subcloned into pAC-V5His (Invitrogen). S2 cells were cultured at 25°C in Schneider medium containing 10% fetal bovine serum and antibiotics. DNAs (pAC-EGFP, pMT-mCherry, and pMT-Mesh-EGFP) were transfected into cells using the Effectene kit (Qiagen), and cells were cultured for 2 days before immunostaining or aggregation assay. To induce the expression from pMT vectors, copper sulfate (final concentration: 500 mM) was added to the culture medium at 24 h after transfection. For immunostaining, the cells were transferred onto concanavalin A-coated coverslips and incubated for 2 h before fixation with methanol-acetone (1∶1) for 10 min at −20°C. After washing with PBS with 0.05% Tween 20 (PBST), the fixed cells were blocked with 10% calf serum in PBST. Samples were then incubated with anti-GFP antibody (Roche) for 30 min at room temperature, followed by incubation with Alexa Fluor 488-conjugated secondary antibody (Invitrogen) for 30 minutes. After washing with PBST, cells were embedded in Fluorsave (Calbiochem). For aggregation assay, the cells were gently dissociated by repeated pipetting and the cell concentrations were readjusted with cell culture medium to 1×10⁶ cells/ml. The cells were shaken at 100 rpm on a rotation platform at room temperature. Aggregation of the cells was analyzed after 2 h. Images were captured with a camera (ORCA-AG; Hamamatsu Photonics) mounted to a microscope (IX71; Olympus) with UPlanSApo 20×NA 0.75 objective lens (Olympus) using IP Lab (ver. 3.9.5r3) acquisition software (BD Biosciences).

Embryos (1–15 h after laying) were put on yeast paste containing Alexa Fluor® 555-labeled dextran (MW 10,000 Invitrogen) to feed newly hatched larvae. After 10–15 h, first-instar larvae were washed with water. Images were acquired with a confocal microscope (model TCS-SPE; Leica) and its accompanying software using an HC PLAN Apochromat 20×NA 0.7 objective lens (Leica). Images were processed with Adobe Photoshop®.

We are grateful to S. Yonemura, A. Nagafuchi, and all the members of Furuse laboratories for helpful discussions. We also thank F. Matsuzaki for the antibody and the fly stocks, and R. G. Fehon, P. J. Bryant, the Bloomington Stock Center, the Drosophila Genetic Resource Center at Kyoto Institute of Technology and the fly stocks of National Institute of Genetics (NIG-Fly) for fly stock.