Search

Summary Background The epidermis, which is a typical stratified epithelium, has tight junctions (TJs) in the granular layer, as do simple epithelia. So far, abnormalities of TJs and involvement of claudin-1 have been reported in tumours of simple epithelia. Objectives To examine the expression of TJ-associated proteins (occludin, ZO-1, claudin-1 and claudin-4) in normal human epidermis and in malignant disorders of keratinization. Methods Expression of the proteins in normal human epidermis, five cases of squamous cell carcinoma (SCC) of the skin and five cases of Bowen's disease (BD) was examined by immunofluorescence staining. Results In normal human epidermis, occludin, ZO-1 and claudin-4 were expressed at the cell–cell borders in the granular layer specifically or dominantly, whereas claudin-1 was expressed in the whole epithelium. In SCC, occludin, ZO-1, claudin-1 and claudin-4 were strongly expressed in tumour cells with keratinization such as cancer pearls. Claudin-1 was heterogeneously expressed in unkeratinized tumour cells, whereas expression of occludin, ZO-1 and claudin-4 was decreased or absent. In BD, aberrant expression of occludin, ZO-1, claudin-1 and claudin-4 was observed at the cell–cell borders in addition to their expression patterns observed in normal epidermis. Conclusions Expression of occludin, ZO-1 and claudin-4 is associated with keratinization in SCC and BD. However, the heterogeneous expression of claudin-1 in SCC is not determined only by keratinization.

The ezrin/radixin/moesin (ERM) proteins are involved in actin filament/plasma membrane interaction that is regulated by Rho. We examined whether ERM proteins are directly phosphorylated by Rho-associated kinase (Rho-kinase), a direct target of Rho. Recombinant full-length and COOH-terminal half radixin were incubated with constitutively active catalytic domain of Rho-kinase, and approximately 30 and approximately 100% of these molecules, respectively, were phosphorylated mainly at the COOH-terminal threonine (T564). Next, to detect Rho-kinase-dependent phosphorylation of ERM proteins in vivo, we raised a mAb that recognized the T564-phosphorylated radixin as well as ezrin and moesin phosphorylated at the corresponding threonine residue (T567 and T558, respectively). Immunoblotting of serum-starved Swiss 3T3 cells with this mAb revealed that after LPA stimulation ERM proteins were rapidly phosphorylated at T567 (ezrin), T564 (radixin), and T558 (moesin) in a Rho-dependent manner and then dephosphorylated within 2 min. Furthermore, the T564 phosphorylation of recombinant COOH-terminal half radixin did not affect its ability to bind to actin filaments in vitro but significantly suppressed its direct interaction with the NH2-terminal half of radixin. These observations indicate that the Rho-kinase-dependent phosphorylation interferes with the intramolecular and/ or intermolecular head-to-tail association of ERM proteins, which is an important mechanism of regulation of their activity as actin filament/plasma membrane cross-linkers.

Occludin is an integral membrane protein localizing at tight junctions (TJ) with four transmembrane domains and a long COOH-terminal cytoplasmic domain (domain E) consisting of 255 amino acids. Immunofluorescence and laser scan microscopy revealed that chick full-length occludin introduced into human and bovine epithelial cells was correctly delivered to and incorporated into preexisting TJ. Further transfection studies with various deletion mutants showed that the domain E, especially its COOH-terminal approximately 150 amino acids (domain E358/504), was necessary for the localization of occludin at TJ. Secondly, domain E was expressed in Escherichia coli as a fusion protein with glutathione-S-transferase, and this fusion protein was shown to be specifically bound to a complex of ZO-1 (220 kD) and ZO-2 (160 kD) among various membrane peripheral proteins. In vitro binding analyses using glutathione-S-transferase fusion proteins of various deletion mutants of domain E narrowed down the sequence necessary for the ZO-1/ZO-2 association into the domain E358/504. Furthermore, this region directly associated with the recombinant ZO-1 produced in E. coli. We concluded that occludin itself can localize at TJ and directly associate with ZO-1. The coincidence of the sequence necessary for the ZO-1 association with that for the TJ localization suggests that the association with underlying cytoskeletons through ZO-1 is required for occludin to be localized at TJ.

The ERM family members, ezrin, radixin, and moesin, localizing just beneath the plasma membranes, are thought to be involved in the actin filament/plasma membrane association. To identify the integral membrane protein directly associated with ERM family members, we performed immunoprecipitation studies using antimoesin mAb and cultured baby hamster kidney (BHK) cells metabolically labeled with [35S]methionine or surface-labeled with biotin. The results indicated that moesin is directly associated with a 140-kD integral membrane protein. Using BHK cells as antigens, we obtained a mAb that recognized the 140-kD membrane protein. We next cloned a cDNA encoding the 140-kD membrane protein and identified it as CD44, a broadly distributed cell surface glycoprotein. Immunoprecipitation with various anti-CD44 mAbs showed that ezrin and radixin, as well as moesin, are associated with CD44, not only in BHK cells, but also in mouse L fibroblasts. Furthermore, immunofluorescence microscopy revealed that in both BHK and L cells, the Triton X-100-insoluble CD44 is precisely colocalized with ERM family members. We concluded that ERM family members work as molecular linkers between the cytoplasmic domain of CD44 and actin-based cytoskeletons.

Several proteins contain the domain homologous to the N-terminal half of band 4.1 protein, indicating the existence of a superfamily. The members of this ‘band 4.1’ superfamily are thought to play crucial roles in the regulation of cytoskeleton-plasma membrane interaction just beneath plasma membranes. We examined the structural diversity of this superfamily by means of the polymerase chain reaction using synthesized mixed primers. We thus identified many members of the band 4.1 superfamily that were expressed in mouse teratocarcinoma F9 cells and mouse brain tissue. In total, 15 cDNA clones were obtained; 8 were identical to the corresponding parts of cDNAs for the known members, while 7 appeared to encode novel proteins (NBL1-7: novel band 4.1-like proteins). Sequence analyses of these clones revealed that the band 4.1 superfamily can be subdivided into 5 gene families; band 4.1 protein, ERM (ezrin/radixin/moesin/merlin/NBL6/NBL7+ ++), talin, PTPH1 (PTPH1/PTPMEG/NBL1-3), and NBL4 (NBL4/NBL5) families. The NBL4 family was first identified here, and the full-length cDNA encoding NBL4 was cloned. The deduced amino acid sequence revealed a myristoylation site, as well as phosphorylation sites for A-kinase and tyrosine kinases in its N-terminal half, suggesting its involvement in the phosphorylation-dependent regulation of cellular events just beneath the plasma membrane. In this study, we describe the initial characterization of these new members and discuss the evolution of the band 4.1 superfamily.

To examine the functions of ERM family members (ezrin, radixin, and moesin), mouse epithelial cells (MTD-1A cells) and thymoma cells (L5178Y), which coexpress all of them, were cultured in the presence of antisense phosphorothioate oligonucleotides (PONs) complementary to ERM sequences. Immunoblotting revealed that the antisense PONs selectively suppressed the expression of each member. Immunofluorescence microscopy of these ezrin, radixin, or moesin "single-suppressed" MTD-1A cells revealed that the ERM family members are colocalized at cell-cell adhesion sites, microvilli, and cleavage furrows, where actin filaments are densely associated with plasma membranes. The ezrin/radixin/moesin antisense PONs mixture induced the destruction of both cell-cell and cell-substrate adhesion, as well as the disappearance of microvilli. Ezrin or radixin antisense PONs individually affected the initial step of the formation of both cell-cell and cell-substrate adhesion, but did not affect the microvilli structures. In sharp contrast, moesin antisense PONs did not singly affect cell-cell and cell-substrate adhesion, whereas it partly affected the microvilli structures. These data indicate that ezrin and radixin can be functionally substituted, that moesin has some synergetic functional interaction with ezrin and radixin, and that these ERM family members are involved in cell-cell and cell-substrate adhesion, as well as microvilli formation.

The expression of cell-cell adhesion molecule, E-cadherin and its associated proteins, alpha- and beta-catenins in human gastric carcinomas was examined by Western blotting. All the seven gastric carcinoma cell lines expressed E-cadherin except KATOIII, which was derived from pleural effusion of a scirrhous type stomach cancer or Borrmann's type-4 carcinoma. The expression of alpha-catenin was not detected in HSC43 derived from scirrhous carcinoma, while HSC39 expressed abnormal beta-catenin caused by genetic alteration. In gastric carcinoma cases, the levels of E-cadherin and alpha-catenin were significantly lower in poorly differentiated adenocarcinomas and scirrhous carcinomas when compared to other types of gastric carcinomas. Deeply invasive carcinomas expressed E-cadherin and alpha-catenin at lower levels. However, the expression level of alpha-catenin was not necessarily consistent with that of E-cadherin. One of 10 gastric carcinomas examined showed complete deletion of alpha-catenin gene in Southern blotting. beta-catenin was expressed at lower level in poorly differentiated adenocarcinomas than in well-differentiated adenocarcinomas. These findings suggest that reduction in the expression of E-cadherin and its associated molecules, catenins, is involved in the development and infiltrative growth of scirrhous type gastric carcinomas.

E-cadherin is a transmembrane glycoprotein that is regulated by its associated proteins, including alpha-catenin. To understand intercellular adhesion of cancerous tissues more precisely, we investigated E-cadherin and alpha-catenin expression in 87 human breast cancers immunohistochemically. All of the normal mammary glands consistently showed strong alpha-catenin expression on cell-cell boundaries. In contrast, various expression patterns were observed in cancerous tissues; 22 (25%) evaluated as uniformly positive, 32 (37%) as heterogeneous and 33 (38%) as uniformly negative. We found significant correlations between reduction in alpha-catenin expression and clinicopathological features, especially tumor invasion and metastasis. Furthermore, a significant correlation was found between E-cadherin and alpha-catenin expressions. These results suggest that E-cadherin mediated adhesion system may be affected by reduction in alpha-catenin expression and that alpha-catenin may play an important role in cancer invasion and metastasis.

Recently, we found that ZO-1, a tight junction-associated protein, was concentrated in the so called isolated adherens junction fraction from the liver (Itoh, M., A. Nagafuchi, S. Yonemura, T. Kitani-Yasuda, Sa. Tsukita, and Sh. Tsukita. 1993. J. Cell Biol. 121:491-502). Using this fraction derived from chick liver as an antigen, we obtained three monoclonal antibodies specific for a approximately 65-kD protein in rats. This antigen was not extractable from plasma membranes without detergent, suggesting that it is an integral membrane protein. Immunofluorescence and immunoelectron microscopy with these mAbs showed that this approximately 65-kD membrane protein was exclusively localized at tight junctions of both epithelial and endothelial cells: at the electron microscopic level, the labels were detected directly over the points of membrane contact in tight junctions. To further clarify the nature and structure of this membrane protein, we cloned and sequenced its cDNA. We found that the cDNA encoded a 504-amino acid polypeptide with 55.9 kDa. A search of the data base identified no proteins with significant homology to this membrane protein. A most striking feature of its primary structure was revealed by a hydrophilicity plot: four putative membrane-spanning segments were included in the NH2-terminal half. This hydrophilicity plot was very similar to that of connexin, an integral membrane protein in gap junctions. These findings revealed that an integral membrane protein localizing at tight junctions is now identified, which we designated as "occludin."

We have obtained a monoclonal antibody (33A-3D) that specifically recognize desmoyokin, a 680 kDa desmosomal plaque protein that is well characterized in bovine muzzle epidermis. A cDNA clone (DY6, 3693 bp) was isolated by immunoscreening a mouse keratinocyte expression library with 33A-3D, and it was confirmed that DY6 has a partial coding sequence for desmoyokin. DY6 consists of highly homologous repeats about 128 residues long. Furthermore, the 128-residue repeats exhibit a quasi seven-residue substructure, which we believe will adopt an antiparallel beta-sheet structure. Surprisingly, the amino acid sequence showed a significant homology with AHNAK, a newly identified human gene encoding a 700 kDa protein, which was suggested to be down-regulated in neuroblastoma. From its extensive homology, the similarity in both size and structure, and the identical patterns on Southern blot analysis of genomic DNAs, desmoyokin and AHNAK protein are thought to be identical. Although the desmoyokin/AHNAK protein is detected in a variety of cell types at both protein and mRNA levels, its distribution in keratinocytes (associated closely with cell membrane) is quite different from that in cells other than keratinocytes (distributed diffusely in the cytoplasm). These findings suggest that the desmoyokin/AHNAK protein is a ubiquitous molecule with a unique structure and appears to have different distributions (and probably different functions) among different cells.

Radixin is a barbed end-capping actin-modulating protein which was previously reported to be concentrated at cell-to-cell adherens junctions (AJ) and cleavage furrows. Recently, cDNA encoding mouse radixin was isolated, showing that radixin is highly homologous to but distinct from ezrin. From mouse teratocarcinoma cells we isolated and analyzed cDNA encoding another radixin-related protein. Sequence analysis has demonstrated that this protein is a mouse homologue of human moesin (98.3% identity) and that it shares 71.7% and 80.1% identity with ezrin and radixin, respectively. Translation experiments in vitro combined with immunoblot analyses led us to conclude that there is a gene family consisting of ezrin, radixin and moesin. These members are coexpressed in various types of cells. Then, by immunofluorescence microscopy, we closely analyzed their distribution using polyclonal and monoclonal antibodies, which could recognize all three members. In addition to cell-to-cell AJ and cleavage furrows, it was shown that they were concentrated at microvilli and ruffling membranes in various types of cells. Furthermore, the cell-to-substrate AJ (focal contacts) were clearly stained by anti-radixin pAb only after the apical/lateral membranes and cytoplasm were removed by the zinc method. We conclude that at least one of the members of the ezrin-radixin-moesin family is concentrated at specific regions where actin filaments are densely associated with plasma membranes.

Recently we developed an isolation procedure for the cell-to-cell adherens junctions (AJ; cadherin-based junctions) from rat liver (Tsukita, Sh. and Sa. Tsukita. 1989. J. Cell Biol. 108:31-41). In this study, using the isolated AJ, we have obtained two mAbs specific to the 220-kD undercoat-constitutive protein. Immunofluorescence and immunoelectron microscopy with these mAbs showed that this 220-kD protein was highly concentrated at the undercoat of cell-to-cell AJ in various types of tissues and that this protein was located in the immediate vicinity of the plasma membrane in the undercoat of AJ. In the cells lacking typical cell-to-cell AJ, such as fibroblasts, the 220-kD protein was immunofluorescently shown to be coconcentrated with cadherin molecules at cell-cell adhesion sites. These localization analyses appeared to indicate the possible direct or indirect association of the 220-kD protein with cadherin molecules. Furthermore, it was revealed that the 220-kD protein and alpha-spectrin were coimmunoprecipitated with the above mAbs in both the isolated AJ and the brain. The affinity-purified 220-kD protein molecule looked like a spherical particle, and its binding site on the spectrin molecule was shown to be in the position approximately 10-20 nm from the midpoint of spectrin tetramer by low-angle rotary-shadowing electron microscopy. Taking all these results together with biochemical and immunological comparisons, we are persuaded to speculate that the 220-kD protein is a novel member of the ankyrin family. However, the possibility cannot be excluded that the 220-kD protein is an isoform of beta-spectrin. The possible roles of this 220-kD protein in the association of cadherin molecules with the spectrin-based membrane skeletons at the cadherin-based cell-cell adhesion sites are discussed.

Radixin is an actin barbed-end capping protein which is highly concentrated in the undercoat of the cell-to-cell adherens junction and the cleavage furrow in the interphase and mitotic phase, respectively (Tsukita, Sa., Y. Hieda, and Sh. Tsukita. 1989 a.J. Cell Biol. 108:2369-2382; Sato, N., S. Yonemura, T. Obinata, Sa. Tsukita, and Sh. Tsukita. 1991. J. Cell Biol. 113:321-330). To further understand the structure and functions of the radixin molecule, we isolated and sequenced the cDNA clones encoding mouse radixin. Direct peptide sequencing of radixin and immunological analysis with antiserum to a fusion protein were performed to confirm that the protein encoded by these clones is identical to radixin. The composite cDNA is 4,241 nucleotides long and codes for a 583-amino acid polypeptide with a calculated molecular mass of 68.5 kD. Sequence analysis has demonstrated that mouse radixin shares 75.3% identity with human ezrin, which was reported to be a member of the band 4.1 family. We then isolated the cDNA encoding mouse ezrin. Sequence analysis and Northern blot analysis revealed that radixin and ezrin are similar but distinct (74.9% identity), leading us to conclude that radixin is a novel member of the band 4.1 family. In erythrocytes the band 4.1 protein acts as a key protein in the association of short actin filaments with a plasma membrane protein (glycophorin), together with spectrin. Therefore, the sequence similarity between radixin and band 4.1 protein described in this study favors the idea that radixin plays a crucial role in the association of the barbed ends of actin filaments with the plasma membrane in the cell-to-cell adherens junction and the cleavage furrow.

The in situ organization of the inner dynein arms of Chlamydomonas axonemes was investigated by analyzing a number of tilt-series thin-section electron micrographs taken with improved fixation and staining protocols. The use of mutant axonemes lacking the outer dynein arms and the radial spokes enabled us to detect a novel feature: inner-arm projections arranged in two rows. The arrangement of projections within each row was deduced by making a model that fitted both tilt-series longitudinal thin-section images and cross-section images. The model suggested that, within a repeating unit of 96 nm, one of the two rows, positioned more centrifugally, consists of four discrete domains and the other, positioned more centripetally, consists of three pairs of globular structures. Whereas the centripetal row resembles the inner-arm arrangement reported previously, the centrifugal row appears to have been identified for the first time in this study. These findings make necessary the re-examination of the arrangement of various inner-arm subspecies within the axoneme.

Radixin is a barbed end-capping actin-modulating protein which was first identified in isolated cell-to-cell adherens junctions from rat liver (Tsukita, Sa., Y. Hieda, and Sh. Tsukita, 1989. J. Cell Biol. 108:2369-2382). In the present study, we have analyzed the distribution of radixin in dividing cells. For this purpose, an mAb specific for radixin was obtained using chicken gizzard radixin as an antigen. By immunofluorescence microscopy with this mAb and a polyclonal antibody obtained previously, it was clearly shown in rat fibroblastic cells (3Y1 cells) that radixin was highly concentrated at the cleavage furrow during cytokinesis. Radixin appeared to accumulate rapidly at the cleavage furrow at the onset of furrowing, continued to be concentrated at the furrow during anaphase and telophase, and was finally enriched at the midbody. This concentration of radixin at the cleavage furrow was detected in all other cultured cells we examined: bovine epithelial cells (MDBK cells), mouse myeloma cells (P3 cells), rat kangaroo Ptk2 cells, mouse teratocarcinoma cells, and chicken fibroblasts. Furthermore, it became clear that the epitope for the mAb was immunofluorescently masked in the cell-to-cell adherens junctions. Together, these results lead us to conclude that radixin is present in the undercoat of the cell-to-cell adherens junctions and that of the cleavage furrow, although their respective molecular architectures are distinct. The possible roles of radixin at the cleavage furrow are discussed with special reference to the molecular mechanism of the actin filament-plasma membrane interaction at the furrow.

The function of ERM (ezrin/radixin/moesin) proteins as general cross-linkers between actin filaments and plasma membranes is regulated downstream of Rho, through the transition between active and inactive forms. To directly examine the conformational change between the active and inactive forms of ERM proteins, we applied low-angle rotary-shadowing electron microscopy to the radixin molecules, wild-type, T564A-non-phosphorylated-type, and T564E-phosphorylated-type, since most of the active forms are reportedly stabilized in cells by the C-terminal threonine phosphorylation. As a result, the T564A- and wild-type radixin molecules yielded the globular closed forms, approximately 8-14 nm in diameter, with some striations on their surfaces. In contrast, the T564E-radixin molecules tended to take elongated open forms, in which two globular structures measuring approximately 8 nm and approximately 5 nm in diameter were associated with both ends of the filamentous structures. The filamentous structure took either a approximately 20-25 nm-long straight course or a folded course. Taken together with the biochemical and the crystal structural results obtained to date, the closed and open forms represent the inactive and active forms of radixin as cross-linkers between actin filaments and plasma membranes.

At tight junctions (TJs), claudins with four transmembrane domains are incorporated into TJ strands. Junctional adhesion molecule (JAM), which belongs to the immunoglobulin superfamily, is also localized at TJs, but it remains unclear how JAM is integrated into TJs. Immunoreplica electron microscopy revealed that JAM showed an intimate spatial relationship with TJ strands in epithelial cells. In L fibroblasts expressing exogenous JAM, JAM was concentrated at cell–cell adhesion sites, where there were no strand-like structures, but rather characteristic membrane domains free of intramembranous particles were detected. These domains were specifically labeled with anti-JAM polyclonal antibody, suggesting that JAM forms planar aggregates through their lateral self-association. Immunofluorescence microscopy and in vitro binding assays revealed that ZO-1 directly binds to the COOH termini of claudins and JAM at its PDZ1 and PDZ3 domains, respectively. Furthermore, another PDZ-containing polarity-related protein, PAR-3, was directly bound to the COOH terminus of JAM, but not to that of claudins. These findings led to a molecular architectural model for TJs: small aggregates of JAM are tethered to claudin-based strands through ZO-1, and these JAM aggregates recruit PAR-3 to TJs. We also discuss the importance of this model from the perspective of the general molecular mechanisms behind the recruitment of PAR proteins to plasma membranes.

We identified Xenopus pericentriolar material-1 (PCM-1), which had been reported to constitute pericentriolar material, cloned its cDNA, and generated a specific pAb against this molecule. Immunolabeling revealed that PCM-1 was not a pericentriolar material protein, but a specific component of centriolar satellites, morphologically characterized as electron-dense granules, approximately 70-100 nm in diameter, scattered around centrosomes. Using a GFP fusion protein with PCM-1, we found that PCM-1-containing centriolar satellites moved along microtubules toward their minus ends, i.e., toward centrosomes, in live cells, as well as in vitro reconstituted asters. These findings defined centriolar satellites at the molecular level, and explained their pericentriolar localization. Next, to understand the relationship between centriolar satellites and centriolar replication, we examined the expression and subcellular localization of PCM-1 in ciliated epithelial cells during ciliogenesis. When ciliogenesis was induced in mouse nasal respiratory epithelial cells, PCM-1 immunofluorescence was markedly elevated at the apical cytoplasm. At the electron microscopic level, anti-PCM-1 pAb exclusively labeled fibrous granules, but not deuterosomes, both of which have been suggested to play central roles in centriolar replication in ciliogenesis. These findings suggested that centriolar satellites and fibrous granules are identical novel nonmembranous organelles containing PCM-1, which may play some important role(s) in centriolar replication.

Tight junctions (TJs) in endothelial cells are thought to determine vascular permeability. Recently, claudin-1 to -15 were identified as major components of TJ strands. Among these, claudin-5 (also called transmembrane protein deleted in velo-cardio-facial syndrome [TMVCF]) was expressed ubiquitously, even in organs lacking epithelial tissues, suggesting the possible involvement of this claudin species in endothelial TJs. We then obtained a claudin-6–specific polyclonal antibody and a polyclonal antibody that recognized both claudin-5/TMVCF and claudin-6. In the brain and lung, immunofluorescence microscopy with these polyclonal antibodies showed that claudin-5/TMVCF was exclusively concentrated at cell–cell borders of endothelial cells of all segments of blood vessels, but not at those of epithelial cells. Immunoreplica electron microscopy revealed that claudin-5/TMVCF was a component of TJ strands. In contrast, in the kidney, the claudin-5/TMVCF signal was restricted to endothelial cells of arteries, but was undetectable in those of veins and capillaries. In addition, in all other tissues we examined, claudin-5/TMVCF was specifically detected in endothelial cells of some segments of blood vessels, but not in epithelial cells. Furthermore, when claudin-5/TMVCF cDNA was introduced into mouse L fibroblasts, TJ strands were reconstituted that resembled those in endothelial cells in vivo, i.e., the extracellular face–associated TJs. These findings indicated that claudin-5/TMVCF is an endothelial cell–specific component of TJ strands.

Claudins, comprising a multigene family, constitute tight junction (TJ) strands. Clostridium perfringens enterotoxin (CPE), a single approximately 35-kD polypeptide, was reported to specifically bind to claudin-3/RVP1 and claudin-4/CPE-R at its COOH-terminal half. We examined the effects of the COOH-terminal half fragment of CPE (C-CPE) on TJs in L transfectants expressing claudin-1 to -4 (C1L to C4L, respectively), and in MDCK I cells expressing claudin-1 and -4. C-CPE bound to claudin-3 and -4 with high affinity, but not to claudin-1 or -2. In the presence of C-CPE, reconstituted TJ strands in C3L cells gradually disintegrated and disappeared from their cell surface. In MDCK I cells incubated with C-CPE, claudin-4 was selectively removed from TJs with its concomitant degradation. At 4 h after incubation with C-CPE, TJ strands were disintegrated, and the number of TJ strands and the complexity of their network were markedly decreased. In good agreement with the time course of these morphological changes, the TJ barrier (TER and paracellular flux) of MDCK I cells was downregulated by C-CPE in a dose-dependent manner. These findings provided evidence for the direct involvement of claudins in the barrier functions of TJs.

Tight junctions (TJs) are essential structures for the physiological functions of epithelial and endothelial cells, and have been suggested to have both barrier and fence functions. Tight junctions create a primary barrier to the diffusion of solutes through the paracellular pathway, and also function as a fence between apical and basolateral membrane domains, to create and maintain cell polarity of epithelial and endothelial cells. Several peripheral membrane proteins have been shown to be concentrated at the cytoplasmic surface of TJs. However, TJ-specific integral membrane proteins had not been identified until recently, and the lack of information concerning TJ-specific integral membrane proteins has hampered a more direct assessment of the function of TJs at the molecular level. Here, we present an overview of current progress in the identification of TJ-specific integral membrane proteins.

Occludin has been identified from chick liver as a novel integral membrane protein localizing at tight junctions, and the cDNA encoding its mammalian homologue was identified very recently (Ando-Akatsuka, Y., M. Saitou, T. Hirase, M. Kishi, A. Sakakibara, M. Itoh, S. Yonemura, M. Furuse, Sh. Tsukita, J. Cell Biol. 133, 43-47 (1996)). Here we describe the basic properties of mammalian occludin in epithelial cells at the DNA, RNA, and protein levels. The human occludin gene was mapped to chromosome band 5q13.1 by fluorescent in situ hybridization. Northern blotting identified several occludin mRNA bands, suggesting the possible expression of several alternatively spliced products. Occludin mRNA was detected in cultured epithelial cells, but not in cultured fibroblasts. The mRNA level was high in the testis, kidney, liver, lung, and brain, which reportedly bear well-developed tight junctions. We then produced monoclonal and polyclonal antibodies using recombinant mouse occludin as the antigen, which reacted not only with mouse, but also human, dog and pig occludin. These antibodies recognized several bands around 60 kDa in epithelial cells but not in fibroblasts. Immunofluorescence microscopy of various tissues revealed that the staining intensity of occludin correlated well with the number of tight junction strands in epithelial cells. By contrast, the staining of ZO-1, a well-characterized tight junction-associated protein, was not specific for tight junctions. Furthermore, the exclusive concentration of occludin at tight junctions in epithelial cells was confirmed by immunoreplica electron microscopy.

The tight junction seals cells together at a subapical location and functionally separates the plasma membrane into an apical and a basolateral domain. This junction is one of the most characteristic structural markers of the polarized epithelial cell. Recently, occludin has been identified as an integral transmembrane protein localizing at the tight junction and directly associated with ZO-1, an undercoat-constitutive cytoplasmic protein. We have investigated occludin expression in conjunction with ZO-1 in normal epithelia and cancers of human digestive tract by immunostaining with a new antibody raised against human occludin. In the normal simple columnar epithelium, occludin was expressed together with ZO-1 as a single line at the apical cell border. However, in the esophagus, which has a stratified squamous epithelium, no occludin expression could be detected, but ZO-1 was expressed in the spinous layer. As for cancers, both occludin and ZO-1 showed the same expression in differentiated adenocarcinoma cells as in normal epithelium, but in poorly differentiated adenocarcinomas, the expression of these two proteins was reduced. There was significant correlation between tumor differentiation and expression of these proteins. These results suggest that occludin, together with ZO-1, is involved in the formation of gland-like structures. In addition, occludin expression can serve as a histopathological indicator for differentiation in gastrointestinal adenocarcinomas.

A fraction enriched in cell-to-cell adherens and tight junctions was isolated from the chick liver, and the undercoat-constitutive proteins were extracted from this isolated junctional fraction. Monoclonal antibodies (mAbs) were then obtained by injecting this extract into rats, and five antigens were identified to be concentrated in the isolated junctional fraction. We have characterized one mAb (E14 mAb) and its antigen (E14). By immunoblotting of the isolated junctional fraction the E14 mAb reacted strongly with a single band of approximately 155 kDa, and E14 was highly concentrated in the isolated junctional fraction. Immunofluorescence microscopy revealed that the E14 mAb exclusively stained the junctional complex region of the liver, renal epithelial cells, and the cell-cell border of endothelial cells in various tissues. The intercalated disc of the heart was also significantly stained. However, the E14 signal was hardly detected from intestinal epithelial cells. By immunoelectron microscopy using renal epithelial cells, E14 was mainly detected in the fibrous structures associated with the cell-to-cell adherens junction. We conclude that E14 is a novel undercoat-constitutive protein found in certain types of cell-to-cell adherens junctions.

The ERM proteins, ezrin, radixin, and moesin, are involved in the actin filament/plasma membrane interaction as cross-linkers. CD44 has been identified as one of the major membrane binding partners for ERM proteins. To examine the CD44/ERM protein interaction in vitro, we produced mouse ezrin, radixin, moesin, and the glutathione-S-transferase (GST)/CD44 cytoplasmic domain fusion protein (GST-CD44cyt) by means of recombinant baculovirus infection, and constructed an in vitro assay for the binding between ERM proteins and the cytoplasmic domain of CD44. In this system, ERM proteins bound to GST-CD44cyt with high affinity (Kd of moesin was 9.3 +/- 1.6nM) at a low ionic strength, but with low affinity at a physiological ionic strength. However, in the presence of phosphoinositides (phosphatidylinositol [PI], phosphatidylinositol 4-monophosphate [4-PIP], and phosphatidylinositol 4.5-bisphosphate [4,5-PIP2]), ERM proteins bound with a relatively high affinity to GST-CD44cyt even at a physiological ionic strength: 4,5-PIP2 showed a marked effect (Kd of moesin in the presence of 4,5-PIP2 was 9.3 +/- 4.8 nM). Next, to examine the regulation mechanism of CD44/ERM interaction in vivo, we reexamined the immunoprecipitated CD44/ERM complex from BHK cells and found that it contains Rho-GDP dissociation inhibitor (GDI), a regulator of Rho GTPase. We then evaluated the involvement of Rho in the regulation of the CD44/ERM complex formation. When recombinant ERM proteins were added and incubated with lysates of cultured BHK cells followed by centrifugation, a portion of the recombinant ERM proteins was recovered in the insoluble fraction. This binding was enhanced by GTP gamma S and markedly suppressed by C3 toxin, a specific inhibitor of Rho, indicating that the GTP form of Rho in the lysate is required for this binding. A mAb specific for the cytoplasmic domain of CD44 also markedly suppressed this binding, identifying most of the binding partners for exogenous ERM proteins in the insoluble fraction as CD44. Consistent with this binding analysis, in living BHK cells treated with C3 toxin, most insoluble ERM proteins moved to soluble compartments in the cytoplasm, leaving CD44 free from ERM. These findings indicate that Rho regulates the CD44/ERM complex formation in vivo and that the phosphatidylinositol turnover may be involved in this regulation mechanism.

Rho small GTP-binding protein regulates various cell functions, such as formation of stress fibers and focal adhesions, cell motility, membrane ruffling, cytokinesis and smooth muscle contraction in mammalian cells and bud formation in the yeast Saccharomyces cerevisiae. As to the functioning sites of Rho in Saccharomyces cerevisiae, we have recently shown that RHO1 protein, a homologue of mammalian RhoA, is concentrated to the growth region of the cells where cortical actin patches are clustered. However, in mammalian cells, the functioning sites of Rho have not yet been studied. In the present study, MDCK cell lines stably expressing myc-tagged RhoA (myc-RhoA) were prepared and localization of myc-RhoA was first immunohistochemically examined using an anti-myc antibody. In the resting cells, almost all of myc-RhoA was observed in the cytosol. When the cells were stimulated with phorbol ester or hepatocyte growth factor, membrane rufflings were induced and myc-RhoA was translocated to the membrane ruffling area. Moreover, myc-RhoA was translocated from the cytosol to the cell-cell adhesion sites when the cells were transferred from a low to normal Ca2+ medium. RhoA was also concentrated to the cleavage furrows during cytokinesis in Swiss 3T3 cells. Translocation of myc-RhoA to the membrane ruffling area was inhibited by prior microinjection into the cells of Rho GDI, a negative regulator of Rho which inhibits activation of Rho, or of C3, an exoenzyme of Clostridium botulinum which ADP-ribosylates Rho and inhibits its functions, indicating that both activation and functioning of Rho are essential for the translocation of Rho. The ERM (Ezrin, Radixin, Moesin) family members were colocalized with RhoA at all of these sites. However, RhoA was not apparently observed at the focal adhesion plaque where vinculin was localized. These results suggest that at least one of the functioning sites of Rho is the ERM family-controlled actin filament/plasma membrane association sites.

Cadherin cell adhesion molecules play an essential role in creating tight intercellular association and are considered to work as an invasion suppressor system of cancer cells. They form a molecular complex with catenins, a group of cytoplasmic proteins including alpha- and beta-catenins. While alpha-catenin has been demonstrated to be crucial for cadherin function, the role of beta-catenin is not yet fully understood. In this study, we analyzed the cadherin-catenin system in two human cell lines, HSC-39 and its putative subline HSC-40A, derived from a signet ring cell carcinoma of stomach. These cells grow as loose aggregates or single cells, suggesting that their cadherin system is not functional. In these cell lines, an identical 321-base pair in-frame mRNA deletion of beta-catenin was identified; this led to a 107-amino-acid deletion in the NH2-terminal region of the protein. Southern blot analysis disclosed a homozygous deletion in part of the beta-catenin gene. On the other hand, these cells expressed E-cadherin, alpha-catenin, and plakoglobin of normal size. Immunoprecipitation analyses showed that E-cadherin was coprecipitated with the mutated beta-catenin but not with alpha-catenin, and antibodies against beta-catenin did not copurify alpha-catenin. However, the recombinant fusion protein containing wild-type beta-catenin precipitated alpha-catenin from these cells. These results suggest that the dysfunction of E-cadherin in these cell lines is due primarily to its failure to interact with alpha-catenin, and that this defect results from the mutation in beta-catenin. Thus, it is most likely that the association between E-cadherin and alpha-catenin is mediated by beta-catenin, and that this process is blocked by NH2-terminal deletion in beta-catenin. These findings indicate that genetic abnormality of beta-catenin is one of the mechanisms responsible for loosening of cell-cell contact, and may be involved in enhancement of tumor invasion in human cancers.

We have isolated the human α-catenin gene (CTNNAl), which encodes a cadherin-associated protein, and have determined its primary structure and chromosomal localization. The transcript of CTNNAl is 3.4 kb long and consists of 16 coding exons encoding 906 amino acids and at least one 5’ noncoding exon. The 102-kDa predicted protein is the same size as the murine homolog, and the amino acid sequences of the two proteins are 99.2% homologous. Analysis by reverse transcription-PCR revealed that this gene is expressed ubiquitously in normal tissues. It was mapped to chromosome band 5q31 by fluorescent in situ hybridization.

The cell-to-cell adherens junction is a site for cadherin-mediated adhesion where actin filaments are densely associated with the plasma membrane through its well developed plasmalemmal undercoat, a "plaque" structure. Recently, we succeeded in isolating the cell-to-cell adherens junctions from rat liver, and in identifying some novel AJ plaque constituents, including src-like tyrosine kinases, radixin, alpha-catenin, and a 220kD protein. The application of genetic engineering techniques to AJ plaque proteins has recently generated a wealth of novel observations, leading to the speculation that these proteins are involved in tumorigenesis and metastasis. This paper reviews these findings and discusses some functions of the AJ plaque proteins in normal cells, as well as in tumorigenesis and metastasis.

E-cadherin (E-cad) plays a major role in the maintenance of cell-cell adhesion in epithelial tissues, and impaired E-cad expression correlates with tumour invasion and metastasis. Alpha-catenin (alpha-cat), an undercoat protein of adherens junctions, binds to the cytoplasmic domain of E-cad and is essential for linking E-cad to actin-based cytoskeleton. We investigated E-cad and alpha-cat expression in 60 human gastric cancers immunohistochemically. The 60 gastric cancers were classified into 18 (30%) in which alpha-cat expression was preserved, and 42 (70%) reduced cases. The reduction of alpha-cat expression was significantly related to dedifferentiation, depth of invasion, infiltrative growth and lymph node metastasis. We also examined the co-expression of alpha-cat and E-cad. Seventeen (28%) tumours preserved both molecules [alpha-cat(+)/E-cad(+)] and 33 (55%) tumours reduced both [alpha-cat(-)/E-cad(-)], whereas 9 (15%) tumours exhibited alpha-cat(-)/E-cad(+). The frequency of lymph node metastasis in alpha-cat(-)/E-cad(+) tumour (67%) was significantly higher than that in alpha-cat(+)/E-cad(+) tumours (24%) and was close to that in alpha-cat(-)/E-cad(-) tumours (82%). The frequency of haematogenous liver metastasis in alpha-cat(-)/E-cad(+) tumours (44%) was significantly higher than that in alpha-cat(+)/E-cad(+) tumours (6%) or alpha-cat(-)/E-cad(-) tumours (9%). Thus, in all E-cad(+) tumours, the frequency of lymph node and liver metastasis was higher in alpha-cat(-) tumours than in alpha-cat(+) tumours. alpha-Cat expression is apparently better at predicting tumour invasion and metastasis than E-cad expression.

The cadherins are a family of transmembrane glycoproteins responsible for calcium-dependent cell-cell adhesion. This adhesion is mediated by a group of cytoplasmic proteins, the catenins, which act inside the cell to couple the cadherin molecule to the microfilament cytoskeleton. Dysfunction of E-cadherin-dependent cell-cell adhesion has been demonstrated to contribute to the acquisition of invasive potential of malignant adenocarcinoma cells. The potential role of alterations of catenin expression in tumor cell invasion is largely unexplored. We have previously found that E-cadherin is frequently down-regulated in clinical samples of prostate cancer (Umbas, R., Schalken, J. A., Aalders, T. W., Carter, B. S., Karthaus, H. F. M., Schaafsma, H. E., Debruyne, F. M. J., and Isaacs, W. B. Cancer Res., 52: 5104-5109, 1992). In this study, we further investigate this adhesion system in both benign and malignant human prostate cells in culture. Using antibodies to E-cadherin and its cytoplasmic accessory protein, alpha-catenin, we find that 5 of 6 human prostate cancer cell lines have reduced or absent levels of one or the other or both of these molecules when compared to normal prostatic epithelial cells. Only the LNCaP prostate cancer cell line is indistinguishable from normal prostate epithelium with respect to its E-cadherin-alpha-catenin complement. Interestingly, the PC-3 line is characterized by the presence of E-cadherin, but the complete lack of alpha-catenin found at both the RNA and protein level. This lack of alpha-catenin gene expression is explained by Southern analysis, which reveals a homozygous deletion of a large portion of the alpha-catenin gene in PC-3 cells. This loss of alpha-catenin is functionally manifested by negligible Ca(2+)-dependent aggregation of these cells in vitro, when compared to LNCaP cells. These results confirm that E-cadherin-dependent cell-cell adhesion is frequently aberrant in prostate cancer cells, and suggest that in a subset of prostate cancers, this adhesion may be inactivated by loss of alpha-catenin rather than E-cadherin itself. Furthermore, these results demonstrate that mutational inactivation of the alpha-catenin gene is one mechanism responsible for the loss of normal cell-cell adhesion in prostate cancer.

Primary pulmonary adenocarcinoma rarely shows cavitation with a fluid level on chest roentgenograms. Herein we describe such a case misdiagnosed as pulmonary tuberculosis. The patient was a 63-year-old, female who had never smoked. Chest roentgenograms revealed a cavitary lesion in the left lower lobe, possessing a prominent fluid level. Fiberoptic bronchoscopic aspirate was positive for acid-fast bacilli on stains. Since there was no improvement with antituberculous chemotherapy, a left lower lobectomy was performed. The present case is of interest in the light of cavity formation in pulmonary carcinoma. The diagnosis and roentgenographic features are discussed.

We previously identified a 220-kD constitutive protein of the plasma membrane undercoat which colocalizes at the immunofluorescence microscopic level with cadherins and occurs not only in epithelial M., S. Yonemura, A. Nagafuchi, Sa. Tsukita, and Sh. Tsukita. 1991. J. Cell Biol. 115:1449-1462). To clarify the nature and possible functions of this protein, we cloned its full-length cDNA and sequenced it. Unexpectedly, we found mouse 220-kD protein to be highly homologous to rat protein ZO-1, only a part of which had been already sequenced. This relationship was confirmed by immunoblotting with anti-ZO-1 antibody. As protein ZO-1 was originally identified as a component exclusively underlying tight junctions in epithelial cells, where cadherins are not believed to be localized, we analyzed the distribution of cadherins and the 220-kD protein by ultrathin cryosection immunoelectron microscopy. We found that in non-epithelial cells lacking tight junctions cadherins and the 220-kD protein colocalize, whereas in epithelial cells (e.g., intestinal epithelial cells) bearing well-developed tight junctions cadherins and the 220-kD protein are clearly segregated into adherens and tight junctions, respectively. Interestingly, in epithelial cells such as hepatocytes, which tight junctions are not so well developed, the 220-kD protein is detected not only in the tight junction zone but also at adherens junctions. Furthermore, we show in mouse L cells transfected with cDNAs encoding N-, P-, E-cadherins that cadherins interact directly or indirectly with the 220-kD protein. Possible functions of the 220-kD protein (ZO-1) are discussed with special reference to the molecular mechanism for adherens and tight junction formation.

Cadherins are a family of transmembrane glycoproteins which are responsible for calcium-dependent cell-cell adhesion. At least two types of proteins called alpha- and beta-catenin are known to be closely associated with the cytoplasmic domain of cadherin molecules and to play a crucial role in the regulation of cadherin cell adhesion function. Sequence analyses of cDNAs encoding these catenins have revealed that alpha- and beta-catenins have a similarity to vinculin and Drosophila armadillo protein, respectively. The possible involvement of these catenin molecules in the molecular mechanism of human cancer invasion and metastasis is discussed.

Tumor metastasis is initiated by disaggregation of invasive cells from the primary tumor - a step that requires a breakdown of intercellular adhesion. To investigate the mechanism of dysfunction in cell-cell adhesion of cancerous tissues, we evaluated E-cadherin (E-cad) and α-catenin (α-cat) expression of human esophageal cancer immunohistochemically. Eighty percent of the tumors showed reduced expression of both of the molecules. The frequency of α-cat negative tumors was observed in 47%, but none of them were E-cad negative. These results suggest that esophageal cancers lose α-cat expression more frequently than E-cad, and the loss of α-cat mights cause dysfunction of E-cad-mediated intercellular adhesion.

A human lung cancer cell line, PC 9, was analyzed to elucidate the molecular mechanisms of dysfunction of cadherin-mediated cell-cell adhesion in cancer. Although PC 9 cells strongly expressed E-cadherin at the cell membrane, which was indistinguishable immunochemically from functional E-cadherin, they did not show tight cell-cell adhesion and had reduced E-cadherin-mediated aggregation activity. Immunoprecipitation with E-cadherin and Western blot analysis revealed that PC 9 cells did not express alpha-catenin, a cadherin-associated protein, suggesting that this was the cause of the cadherin dysfunction in the cell line. In addition, Northern and Southern blot analyses disclosed homozygous deletion of part of the alpha-catenin gene, which might have resulted in the loss of alpha-catenin expression in PC 9 cells.

An F-actin binding protein was purified from bovine liver by means of DNase I affinity, hydroxylapatite and DEAE-cellulose column chromatographies. It consisted of a single polypeptide chain having an apparent molecular weight of 68,000 with a Stokes radius of 35A. Electron microscopy of rotary shadowed specimens showed that the 68kD protein is a globular protein. This protein showed a higher affinity for F-actin in the presence of Ca2+ than in its absence, which is opposite to the actin-binding property shown by nonmuscle alpha-actinin or fimbrin. The 68kD protein had no F-actin severing and capping activity. Interestingly, the 68kD protein was found to aggregate liposomes at micromolar Ca2+ concentrations. Immunoblot analysis and partial protein sequence data identified the 68kD protein as an annexin VI (p68) homologue. Immunocytochemical studies showed that the 68kD protein was localized along stress fibers as well as membrane ruffles, microspikes and focal contacts, raising the possibility that annexin VI may contribute to control membrane-microfilament interaction in the cell. © 1992 Wiley-Liss, Inc.

To approach the transmembrane signaling pathway in the cell-to-cell adherens junctions (AJ), AJ-specific tyrosine phosphorylation was analyzed. When various types of rat adult tissues were pretreated with sodium orthovanadate, a potent inhibitor of tyrosine phosphatase, immunofluorescence microscopy showed that anti-phosphotyrosine polyclonal antibody specifically stained the undercoat of the cell-to-cell AJ. This indicates that the tyrosine kinase activity is elevated at the undercoat of the cell-to-cell AJ of adult tissues. To identify tyrosine kinases responsible for the high level of tyrosine phosphorylation at AJ, we have performed in vitro phosphorylation experiments with cell-to-cell AJ isolated from rat liver (Tsukita, Sh. and Sa. Tsukita. 1989. J. Cell Biol. 108:31-41) and immunoblotting analyses with specific antibodies for tyrosine kinases. As a result, three proto-oncogenic tyrosine kinases of src family, c-yes, c-src, and lyn kinases, were identified as major tyrosine kinases in the cell-to-cell AJ of hepatocytes. Furthermore, it was immunofluorescently shown that at least two of these kinases, c-yes and c-src kinases, were enriched at the cell-to-cell AJ of various types of cells including hepatocytes. Based on these findings, it is concluded that, in various types of cells, specific proto-oncogenic tyrosine kinases of src-family (c-yes and c-src) are enriched to work as signal mediators in the cell-to-cell AJ where the level of tyrosine phosphorylation is elevated.

Radixin is a barbed end-capping actin-modulating protein which was first identified in isolated cell-to-cell adherens junctions from rat liver (Tsukita, Sa., Y. Hieda, and Sh. Tsukita, 1989. J. Cell Biol. 108:2369-2382). In the present study, we have analyzed the distribution of radixin in dividing cells. For this purpose, an mAb specific for radixin was obtained using chicken gizzard radixin as an antigen. By immunofluorescence microscopy with this mAb and a polyclonal antibody obtained previously, it was clearly shown in rat fibroblastic cells (3Y1 cells) that radixin was highly concentrated at the cleavage furrow during cytokinesis. Radixin appeared to accumulate rapidly at the cleavage furrow at the onset of furrowing, continued to be concentrated at the furrow during anaphase and telophase, and was finally enriched at the midbody. This concentration of radixin at the cleavage furrow was detected in all other cultured cells we examined: bovine epithelial cells (MDBK cells), mouse myeloma cells (P3 cells), rat kangaroo Ptk2 cells, mouse teratocarcinoma cells, and chicken fibroblasts. Furthermore, it became clear that the epitope for the mAb was immunofluorescently masked in the cell-to-cell adherens junctions. Together, these results lead us to conclude that radixin is present in the undercoat of the cell-to-cell adherens junctions and that of the cleavage furrow, although their respective molecular architectures are distinct. The possible roles of radixin at the cleavage furrow are discussed with special reference to the molecular mechanism of the actin filament-plasma membrane interaction at the furrow.

Localization of junctions between inner enamel-secretory ameloblasts was examined by immunofluorescence microscopy using antibodies against adherens junction proteins, radixin, vinculin, and A-CAM. All antibodies used stained the boundary between the ameloblasts exclusively in the plane where F-actin was abundant. This suggests that the adherens junctions in the ameloblasts are involved in cell-to-cell movement with actin-based microfilament bundles.