Multifunctional strands in tight junctions

Shoichiro Tsukita, Mikio Furuse and Masahiko Itoh

Department of Cell Biology, Kyoto University Faculty of Medicine, Yoshida-Konoe, Sakyo-ku,
Kyoto 606-850, Japan.

Tight junctions are one mode of cell-to-cell adhesion in epithelial and endothelial cellular sheets,　and are indispensable for multicellular organisms. They act as a primary barrier to the diffusion of solutes through the intercellular space, create a boundary between the apical and the basolateral plasma membrane domains, and recruit various cytoskeletal as well as signaling molecules at their cytoplasmic surface. New insights into the molecular architecture of tight junctions allow us now to discuss the structure and functions of this unique cell-cell adhesion apparatus in molecular terms.

The existence of separate fluid compartments with different molecular compositions is of particular importance for the development and maintenance of multicellular organisms. These compartments are delineated by various cellular sheets, which function as barriers to maintain the distinct internal environment of each compartment. For example, renal tubules, blood vessels and the peritoneal cavity are lined with epithelial, endothelial and mesothelial cellular sheets, respectively. Within these sheets, individual cells are mechanically linked with each other to maintain the structural integrity of the sheet, and the intercellular space between adjacent cells is sealed to prevent the diffusion of solutes through the intercellular space.
　　The junctional complex of simple epithelial cells is located at the most-apical part of the lateral membrane and consists of three distinct components: tight junctions (TJ), adherens junctions and desmosomes (1 )(Fig.1). On ultra-thin section electron micrographs, TJs appear as a series of apparent fusions ("kissing points"), involving the outer leaflets of the plasma membranes of adjacent cells (Fig.1, Fig.2b). At kissing points of TJs, the intercellular space is completely obliterated, whereas in adherens junctions and desmosomes, the apposing membranes are 15-20 nm apart (Fig.1b). In simple epithelial cellular sheets, adherens junctions and desmosomes mechanically link adjacent cells, whereas TJs are responsible for intercellular sealing (2,3).
　　But many physiological situations require that various materials are selectively transported across cellular sheets, and this occurs either by transcellular transport through the cell or by paracellular flux through TJs (4) (Box 1). So TJs are not simply impermeable barriers: they show ion as well as size selectivity, and vary in tightness depending on cell type (3,5).
　　In addition to the "barrier function", TJs are thought to be involved in a "fence function" (2,3). For the vectorial transport of materials across cellular sheets, plasma membranes are functionally divided into apical and basolateral domains that face the luminal and serosal compartments, respectively. Apical and basolateral membrane domains differ in the compositions of integral membrane proteins as well as lipids. However, since integral membrane proteins and lipids can diffuse laterally within the plane of the lipid bilayer of plasma membranes, some diffusion barrier is required at the border between apical and basolateral membrane domains. Since TJs look like a fence within plasma membranes at the most apical part of lateral membranes as shown below, it has been suggested that TJs are the morphological counterpart of a localized diffusion barrier.
　　In recent years, information on the molecular components of TJs, and in particular their cell adhesion molecules, has accumulated. Here, we will present an overview of our current understanding of the structure and functions of TJs in molecular terms.

Ultrastructure and components of TJ strands
The morphology of TJs has been intensively analyzed by freeze-fracture replica electron microscopy. On freeze-fracture replica electron micrographs, TJs appear as a set of continuous, anastomosing intramembranous particle strands or fibrils (TJ strands) on the P-face with complementary vacant grooves on the E-face (6) (Fig.2a). The number of TJ strands as well as the frequency of their ramification vary significantly depending on cell type, producing marked variation in the morphology of TJ strand networks. These observations led to our understanding of the three-dimensional structure of TJs (Fig.2c). Each TJ strand within the plasma membrane associates laterally with another TJ strand in the apposing membrane of adjacent cells to form "paired" TJ strands, where the intercellular space is obliterated.
　　Two types of models have been proposed to explain the chemical nature of TJ strands (Fig.3). In the 'protein model', TJ strands represent units of integral membrane proteins polymerized linearly within lipid bilayers, whereas in the 'lipid model' lipids organized in inverted cylindrical micelles are proposed to constitute TJ strands (7). Recent identification of TJ-specific integral membrane proteins strongly supports the "protein" model, although we cannot exclude the possibility that specific lipids may also be important for the formation of TJ strands.
　　Occludin (~60 kDa) was identified as the first integral membrane protein localized at TJs in chicken (8), and then also in mammals (9). Occludin has four transmembrane domains, a long carboxyl-terminal cytoplasmic domain and a short amino-terminal cytoplasmic domain (Fig.4a). No occludin-related genes have been identified yet, but two isoforms of occludin were found to be generated by alternative splicing (10).
　　On immuno-replica electron microscopy, anti-occludin antibodies exclusively labeled TJ strands (11), suggesting that occludin is directly incorporated into TJ strands. Furthermore, as the intensity of immunostaining with anti-occludin antibodies in various tissues correlates well with the number of TJ strands, the density of occludin molecules in TJ strands seems almost constant (11). But TJ strands can also be formed without occludin, as in some cell types such as endothelial cells in non-neuronal tissue and in Sertoli cells (in the human testis), occludin was not detected in TJ strands (12,13). More importantly, visceral endoderm cells differentiated from occludin-deficient embryonic stem cells still have well-developed networks of TJ strands (14). At present, the physiological functions of occludin are not well understood. Its possible functions will be discussed in detail below.
　　More recently, two other transmembrane protein "claudin-1 and claudin-2 " were identified as integral components of TJ strands (15). 　These proteins also possess four transmembrane domains, but do not show any sequence similarity to occludin (Fig.4b). To date, 24 members of the claudin family have been identified, mainly through database searches, in mouse and human (16,17 )(Table 1).
There is accumulating evidence that claudins constitute the backbone of TJ strands. Immuno-replica electron microscopy revealed that claudins are exclusively localized on TJ strands (16,18,19). Exogenously expressed claudins conferred cell-aggregation activity on L fibroblasts with their concomitant concentration at cell-cell contact planes (20) and led to the formation of a large network of TJ strand-like structures (21) (Fig.4d). Occludin itself has no ability to reconstitute such well-organized strands, but when occludin was introduced into claudin-expressing L transfectants, it was incorporated into reconstituted claudin-based strands (21).
　　The expression pattern of claudins varies considerably among tissues (15,16) (Table 1). Some claudins are known to be expressed in specific cell types; for example claudin-5/TMVCF is expressed only in endothelial cells of blood vessels (19), and claudin-11/OSP is only in oligodendrocytes and Sertoli cells (18). Most cell types, however, express more than two claudin species in various combinations to constitute TJ strands: Within individual single strands, distinct species of claudins are co-polymerized to constitute "heteropolymers", and between adjacent strands within "paired" strands, claudins adhere with each other in a homotypic as well as heterotypic manner (22,23).
　　The last transmembrane component of TJs is JAM (junctional adhesion molecule; ~40 kDa) (24). There are three JAM-related proteins (25,26), which belong to the immunoglobulin superfamily: they have a single transmembrane domain and their extracellular portion is thought to be folded into two immunoglobulin-like domains (Fig.4c). Preliminary freeze-fracture replica electron microscopy revealed that exogenously expressed JAM does not reconstitute TJ strands in L transfectants and that it associates laterally with the claudin-based backbone of TJ strands in epithelial cells. JAM was shown to be involved in cell-cell adhesion/junctional assembly of epithelial/endothelial cells (24,25,27,28) as well as in the extravasation of monocytes through endothelial cells (24), but our knowledge on its function is still fragmentary.

A ziplock with diversified permeability
TJs vary in tightness in a tissue-dependent manner (2,3). The tightness of TJs can be directly measured as transepithelial electric resistance (TER). 　The number of TJ strands was found to correlate well with the TER values of TJs in various tissues (29,30). For example, in the kidney, epithelial cells of the proximal and distal tubules bear 1-2 and 4-7 TJ strands, respectively, and the epithelial cells of the distal tubules exhibit much higher TER than those of the proximal tubules. However, exceptions to this correlation have also been reported (31,32). For example, the two existing strains of Madin-Darby canine kidney (MDCK) epithelial cells -MDCK I and MDCK II - show marked disparity in their TER. Stevenson et al. reported that MDCK I cells have a 30-60 fold higher TER than MDCK II cells, but the number of TJ strands in these strains is very similar (33). These observations indicate that individual paired TJ strands also vary in quality, i.e. tightness, not only in number.

The number of strands. The number of TJ strands is an important factor in determining the barrier properties of TJs, but the molecular mechanism underlying regulation of the strand number remains unknown. When MDCK I cells, which express claudin-1 and claudin-4, were specifically depleted of claudin-4, a marked decrease was observed in the number of TJ strands and in their barrier function (34). Mice lacking claudin-11/OSP, which is expressed specifically in oligodendrocytes and Sertoli cells in wild-type mice, were recently generated, and in these mice TJ strands were absent in myelin sheaths as well as in Sertoli cells (35). Furthermore, when claudins were overexpressed in L fibroblasts, a large network of TJ strands was formed (21). In addition, overexpression of occludin in MDCK cells was also shown to cause an increase in the number of TJ strands to some extent (36). These findings would suggest that the number of TJ strands is determined by the total amount of expressed claudins (and occludin) in individual cells. However, the regulation of the number of TJ strands is probably more complicated. In epithelial cells that already express claudins, overexpression of claudins did not lead to a significant increase in the number of TJ strands (37) , suggesting that an upper limit exists. Interestingly, when a claudin-1 mutant lacking its binding ability to underlying cytoskeletons were overexpressed in MDCK cells, aberrant TJ strands were formed (37). This finding suggests the possible involvement of the underlying proteins in the regulation of the TJ strand number, but how the upper limit is set remains a mystery.
　　In addition to the strand number, the complexity of the network pattern may also be an important factor determining the barrier properties of TJs (30). The network patterns of the reconstituted TJs in L fibroblasts varied markedly among claudin species. For example, claudin-1-induced TJ strands form a large network through frequent ramifications (21), whereas claudin-11/OSP-induced strands scarcely branched and ran parallel to each other (18). This observation is likely to be relevant in vivo, as claudin-11/OSP-based TJ strands in myelin sheaths and Sertoli cells are mostly parallel with little branching (18,35). It is tempting to speculate that the complexity of the TJ strand network is determined by the combination and the mixing ratio of the expressed claudin species.

Extracellular aqueous pores. The extracellular portion of TJ strands probably functions as a ziplock to create a primary barrier against paracellular diffusion (Fig.5). By comparing the TER and the morphology of TJ strands in various epithelia, Claude found that as TJ strands increase in number, the TER value increases logarithmically (30), and then to explain this relationship the existence of aqueous pores was postulated within the paired TJ strands, which take both open and closed states (5,30,38) (Box 2). As mentioned above, however, some exceptions to this relationship between the number of TJ strands and the TER value have been found (31,32). The difference in tightness of individual TJ strands could be explained by the heterogeneity of aqueous pores in terms of their probability being open or closed (33).
　　But what is the chemical nature of these aqueous pores? Recent studies on hereditary hypomagnesemia provided a clue to answer this question (39). Most Mg++ is resorbed from the urine through the paracellular pathway in the thick ascending limb of Henle, but in these patients this resorption is reduced, resulting in severe hypomagnesemia. Positional cloning identified claudin-16/paracellin-1 as the gene responsible for this disease. In good agreement, claudin-16/paracellin-1 is exclusively expressed in the thick ascending limb of Henle. This finding suggested that claudin-16/paracellin-1 forms aqueous pores that function as Mg++ "paracellular" channels.
The difference between MDCK I and MDCK II cells is probably also due to their different expression of claudins (40): MDCK I express primarily claudin-1 and claudin-4, whereas MDCK II cells also express large amounts of claudin-2 in addition to claudin-1 and -4. When claudin-2 was introduced into MDCK I cells, the TER value of these MDCK I transfectants fell to the level of MDCK II cells without any changes in the number of TJ strands. In contrast, exogenously expressed claudin-3 did not affect the TER value of MDCK I cells. Therefore, it is likely that claudin-2 constitutes aqueous pores with high conductance within paired TJ strands of MDCK II cells.
These findings led to the conclusion that claudins not only form the backbone of TJ strands but also extracellular aqueous pores, and that the combination and the mixing ratios of claudin species determine the tightness of individual TJ strands (23). But it is also possible that TJ strands are simply repeatedly broken and annealed, and that this contributes to the tightness of individual strands. To date, the information is not available regarding the stability of strands, and this issue should be examined in detail in molecular terms in future studies.
　　Occludin has also been shown to be involved in the barrier function of TJs, but at present, how occludin is involved remains unclear. Occludin-deficient mice were born normal, but as they grew up, they began to show complex phenotypes including significant growth retardation, chronic inflammation and hyperplasia of the gastric epithelium and mineral deposition in the brain (41). TJs in most organs of occludin-deficient mice such as intestinal epithelial cells seem normal in terms of their morphology and TER. Consistently, transfection of carboxyl-terminally truncated occludin into MDCK cells induced redistribution of endogenous occludin, leaving occludin-deficient TJ strands with normal appearance, but did not decrease their TER (42). These findings are inconsistent with previous observation that addition of synthetic peptides corresponding to the second extracellular loop of occludin into culture medium removed endogenous occludin from TJs, resulting in marked decrease in their TER value (43). Interestingly, overexpression of full-length as well as carboxyl-terminally truncated occludin in MDCK cells raised TER, and, paradoxically, increased mannitol flux (36,37,42). These observations suggested that occludin contributes to the electrical barrier function of TJ to some extent and possibly to the formation of aqueous pores within TJ strands through which non-charged solute flux occurs. Although the paracellular flux of non-charged solute has not yet been examined in occludin-deficient mice, it is possible that the multiple defects found in these mice are attributed to disappearance of putative "occludin-based" aqueous pores from TJ strands. Of course, it is also possible that occludin plays some important role in other TJ-related functions such as fence function and/or signaling events rather than barrier function, which could explain the cause of multiple defects in occludin-deficient mice.

A magnetic bar for PDZ-containing proteins
The thickness of TJ strands (6) (~10 nm; see Fig.2a) is similar to the diameter of the gap junctional channel (connexon) consisting of 6 connexin molecules that also bear four transmembrane domains. Therefore, it is not likely that claudins are aligned in a single line to constitute TJ strands, but that they are packed more densely in the strands. It is then expected that the cytoplasmic surface of individual TJ strands appears as a tooth-brush consisting of densely packed, numerous short carboxyl-terminal cytoplasmic tails of claudins. In addition to these claudin tails, relatively long carboxyl-terminal tails of occludin are probably intermingled.
　　Many cytosolic proteins have been reported to associate with cytoplasmic surface of TJs. As the first component of TJs, a peripheral membrane protein with a molecular mass of 220 kDa was identified through monoclonal antibody production, and was named ZO-1 (Zonula Occludens-1) (44). When ZO-1 was immunoprecipitated from cell lysates of MDCK cells, two proteins with molecular masses of 160 kDa and 130 kDa were coprecipitated (45,46). As these proteins were also localized at TJs, they were designated as ZO-2 and ZO-3, respectively. ZO-1, ZO-2 and ZO-3 have sequence similarity with each other (47-50): they contain three PDZ domains (PDZ1, PDZ2 and PDZ3), one SH3 domain, and one guanylate kinase-like (GUK) domain (Fig.6).
　　The PDZ domain was initially reported to specifically bind to the carboxyl-terminal Glu-Ser/Thr-Asp-Val motif, but it is now known that it recognizes more diverse 4-amino acid sequences most ofthem ending in Val. Interestingly, most claudin tails have a Val at their carboxyl termini (15,16); the only known exception is claudin-12. This suggests that these carboxyl termini directly bind to PDZ domains. If so, the cytoplasmic surface of TJ strands may function as a magnetic bar that strongly attracts and recruits many PDZ-containing proteins (Fig.5). Indeed, the PDZ1 domains of ZO-1, ZO-2 and ZO-3 were recently shown to bind directly to the carboxyl termini of claudins (51). No notable differences were detected in the affinity of different claudins to PDZ1 domains. ZO-1, ZO-2 and ZO-3 also directly bind to the carboxyl-terminal tail of occludin, but their GUK domains, not PDZ domains, are involved in this interaction (50,52-54). As ZO-1, ZO-2 and ZO-3 are localized to TJs in occludin-deficient mice, this interaction may not be essential for their recruitment to TJs (41,51). Furthermore, JAM, which is concentrated around TJ strands, also ends in Val, and was recently shown to directly bind to ZO-1 (55,56) and other PDZ-containing proteins (57). Thus, it is possible that JAM is also involved in the recruitment of various PDZ-containing proteins to TJs.
　　In addition to ZO-1, ZO-2 and ZO-3, several PDZ-containing proteins are recruited to the cytoplasmic surface of TJs, but it remains unknown whether these proteins directly bind to the carboxyl termini of claudins (Fig.6): MAGI-1 (Membrane Associated Guanylate Kinase Inverted-1) (58,59), MAGI-2 (60) and MAGI-3 (61), each of which contains 6 PDZ domains, and mammalian homologues of C.elegans PAR (partitioning) gene products, PAR-3 (ASIP, atypical PKC isotype-specific interacting protein)62 and PAR-663-65, which contain three and one PDZ domain(s), respectively. The list of PDZ-containing proteins localized at TJs will probably continue to increase. These proteins are multi-domain proteins, and may function as adapters at the cytoplasmic surface of TJ strands, which recruit various proteins including cytoskeletal as well as signaling molecules (Fig.6, Table 2).
　　Through the recruitment of various types of proteins to TJs via PDZ-containing proteins, a huge macromolecular complex is expected to be formed at the cytoplasmic surface of TJ strands (Fig.5). What is the physiological functions of this complex? Firstly, as actin filaments bind to the carboxyl-terminal portions of ZO-1/ZO-2 (53,54,66), this complex cross-links TJ strands to actomyosin cytoskeletons. This interaction has been thought to play a role in the regulation of TJ functions. Interestingly, similar accumulation of PDZ-containing proteins occurs in the postsynaptic density in neurons, where these PDZ-containing proteins are directly involved in the synaptic signal transduction and its regulation (67). In cell-matrix adhesion, a huge macromolecular complex is known to be formed at integrin-based adhesion sites, and to play crucial roles in the extracellular matrix-dependent signaling, although this complex formation is not based on PDZ-containing proteins (68). However, in cell-cell adhesion, such macromolecular complex is not well developed at the cadherin-based adhesion site, i.e. adherens junctions. Therefore, it is tempting to speculate that the huge macromolecular complex formed at the cytoplasmic surface of TJs plays central roles in the intercellular signaling of epithelial/endothelial cells, being involved in the regulation of their proliferation, differentiation and polarization. In this connection, it is interesting to point out that TJs recruit a tumor suppressor gene product (PTEN) (60,61), a quaternary complex of cell polarity-related gene products (PAR-3, PAR-6, atypical PKC, cdc42) (63-65) and vesicular transport-related proteins (Rab3B, Rab13, Sec6/8 products) (69-71).

A fence within plasma membranes
TJ strands are heteropolymers of integral membrane proteins, occludin and claudins, which are embedded within plasma membranes, and continuously encircle the top of individual epithelial/endothelial cells to delineate the border between the apical and basolateral membrane domains. Therefore, it is likely that TJ strands act as a 'fence', limiting the lateral diffusion of lipids and proteins between the apical and basolateral membrane domains (Fig.5).
　　The apical membrane of epithelial cells is enriched in glycosphingolipids and sphingomyelin (72,73). Interestingly, this membrane displays a striking asymmetric organization of lipids across the lipid bilayer, and glycosphingolipids as well as sphingomyelin are concentrated in its outer leaflet (74,75). Such polarized localization of lipids suggest the existence of the diffusion barrier, especially for the lipids in the outer leaflet. This was confirmed experimentally. When fluorescently labeled lipids were inserted into the outer leaflet of the apical membrane of cultured epithelial cells, they retained on the apical surface. In contrast, fluorescently labeled lipids inserted into the inner leaflet of the apical membrane quickly redistributed to the basolateral surface (76,77).
　　It is reasonable to speculate that TJs restrict the lateral diffusion of not only lipids but also integral membrane proteins. As the intercellular space is completely obliterated at TJs, integral membrane proteins with extended extracellular portions could not cross TJs. However, it is also clear that, in addition to TJs, there are other mechanisms behind the asymmetric distribution of certain integral membrane proteins within plasma membranes (78): The cytoskeletal proteins underlying the plasma membranes can restrict the lateral diffusion of proteins within membrane domains and the targeted vesicle fusion is also important. Early studies showed that the disruption of intercellular junctions (for instance by incubation with low calcium medium) resulted in the intermixture of membrane proteins from the apical and basolateral domains (79,80). However, in these experiments, not only TJs but also adherens junctions/desmosomes disappeared, and the cytoskeletal architecture was also affected. Thus, it is difficult from these experiments to conclude that TJs act as a diffusion barrier for membrane proteins. This issue, i.e. the importance of TJs in the asymmetric distribution of integral membrane proteins, remains controversial (81). In this connection, expression of constitutively-active RhoA and Rac1 small GTPases in MDCK cells was recently reported to result in disorganization of TJ strand networks as well as the disruption of the junctional fence for lipids but not for integral membrane proteins (82).
　　When carboxyl-terminally truncated occludin was overexpressed in cultured MDCK cells, fluorescently-labeled sphingomyelin added to the apical membrane domain was redistributed to the basolateral surface (42). The polarized distribution of integral membrane proteins was not apparently affected. These findings suggest the possible involvement of occludin in the diffusion barrier, especially for lipids. The relationship between claudins and the diffusion barrier in epithelial cells has not yet been examined. Considering that occludin-deficient mice were born with normal epithelial cells in the intestine and the kidney (41), it is still premature to further discuss the relationship between the "fence" function of TJs and occludin/claudins.

Future directions
TJs have attracted a great deal of interest from investigators in various fields, but lack of information concerning the TJ-specific integral membrane proteins has hampered studies of the molecular biology of TJs. The recent identification of major components of TJ strands have facilitated the molecular assessment of the morphological and physiological observations of TJs that have been accumulated over years. Based on the accumulated information on occludin and claudins, we have discussed here multiple functions of TJ strands at the molecular level; the barrier, signaling, and fence functions (Fig.5). In addition to the issues described above, the challenge of understanding many intriguing open questions on TJs lies ahead of us.
　　The identification of occludin/claudins leads immediately to many basic questions on the TJ strand itself. How are occludin and heterogeneous claudins arranged in individual TJ strands? To what extent are TJ strands dynamically polymerized and depolymerized? How is it regulated? Are some lipids required for the polymerization of occludin/claudins? How can TJ strands restrict the lateral diffusion of lipids only in outer leaflets of the membranes? As the polymerization of integral membrane proteins in a linear fashion is unique, the elucidation of the basic physico-chemical properties of TJ strands may constitute one of the big challenges for years to come.
　　Several intriguing questions stand out at the cellular level. One of the most pressing questions concerns the molecular mechanism underlying the polarized formation of TJs at the most apical region of the lateral membranes in epithelial cells. What is the relationship between tight junctions, adherens junctions and desmosomes during epithelial polarization? How are occludin, claudins, cadherins and their underlying molecules integrated into the polarized junctional complex during epithelial polarization? In this connection, it should be pointed out that there is an alternative model for the function of ZO-1/ZO-2/ZO-3 which was not discussed above. In that model, ZO-1/ZO-2/ZO-3 recruit TJ proteins such as claudins and occludin to their final destination at the interface between the apical and basolateral membrane domains. Another outstanding issue concerns the regulation of the TJ barrier. As indicated above, TJs vary in tightness in a cell type-dependent manner. The tightness of TJs is also known to be dynamically and finely regulated in individual cells depending on various physiological and pathological requirements (2,3) (see Box 1). The information of the molecular mechanism underlying these regulations is still fragmentary, but several signaling pathways such as serine/threonine phosphorylation, tyrosine phosphorylation, heterotrimeric G proteins and small G proteins are thought to be involved in their regulation (83). The transcription of occludin was reported to be down-regulated by TNF alpha/interferon gamma (84) and/or by activation of the MAP kinase cascade (85,86), but there is no information available regarding the transcriptional regulation of claudins by these or other signaling pathways. The cytoplasmic tail of occludin was shown to be heavily phosphorylated on serine and threonine residues (87), whereas the phosphorylation of claudins has not yet been examined.
　　Finally, another important challenge for future studies of TJs is to examine their possible involvement in various diseases. As mentioned above, mutations in claudin-16/paracellin-1 were shown to cause hereditary hypomagnesemia (39). Furthermore, recent positional cloning identified claudin-14 as the gene responsible for hereditary deafness (88). This claudin species is expressed in hair cells in the cochlea of the inner ear. As TJs in these cells play crucial roles in establishment of two compositionally distinct compartments in the inner ear, mutations in the claudin-14 gene would cause deafness. In addition to hereditary diseases, claudins appear to have something to do with various pathological conditions including inflammation (89). Furthermore, the involvement of occludin86 as well as claudins90 in tumorigenesis has been suggested in recent years.
　　We are only just beginning to understand the functions of TJs in molecular terms. Our picture of the molecular architecture of TJs remains incomplete, and other important constituents need to be identified. Further development of the molecular biology of TJs will lead to a better understanding of the roles of TJs not only in normal physiology but also in disease.

References
1. Farquhar, M. G. & Palade, G. E. Junctional complexes in various epithelia. J. Cell Biol. 17, 375-412 (1963).
2. Schneeberger, E. E. & Lynch, R. D. Structure, function, and regulation of cellular tight junctions. Am.J. Physiol. 262, L647-L661 (1992).
3. Gumbiner, B. Breaking through the tight junction barrier. J.Cell Biol. 123, 1631-1633 (1993).
4. Spring, K. Routes and mechanism of fluid transport by epithelia. Annu. Rev. Physiol. 60, 105-119 (1998).
5. Reuss, L. Tight junction permeability to ions and water. In ﾒTight Junctionsﾓ (Cereijido,M. ed.) p.49-66, CRC press, London. (1992).
6. Staehelin, L. A. Further observations on the fine structure of freeze-cleaved tight junctions. J.Cell Sci. 13, 763-786 (1973).
7. Kachar, B. & Reese, T. S. Evidence for the lipidic nature of tight junction strands. Nature 296, 464-466 (1982).
8. Furuse, M. et al. Occludin: a novel integral membrane protein localizing at tight junctions. J.Cell Biol. 123, 1777-1788 (1993).
9. Ando-Akatsuka,Y. et al. Interspecies diversity of the occludin sequence: cDNA cloning of human, mouse, dog, and rat-kangaroo homologues. J.Cell Biol. 133, 43-47 (1996).
10. Muresan, Z., Paul, D. L. & Goodenough, D. A. Occludin1B, a variant of the tight junction protein. Mol.Biol.Cell 11, 627-634 (2000).
11. Saitou, M. et al. Mammalian occludin in epithelial cells: its expression and subcellular distribution. Eur.J.Cell Biol. 73, 222-231 (1997).
12. Hirase, T. et al. Occludin as a possible determinant of tight junction permeability in endothelial cells. J.Cell Sci. 110, 1603-1613 (1997).
13. Moroi, S. et al. Occludin is concentrated at tight junctions of mouse/rat but not human/guinea pig Sertoli cells in testes. Am.J.Physiol. 274, C1708-C1717 (1998).
14. Saitou, M. et al. Occludin-deficient embryonic stem cells can differentiate into polarized epithelial cells bearing tight junctions. J.Cell Biol. 141, 397-408 (1998).
15. Furuse, M., Fujita, K., Hiiragi, T., Fujimoto, K. & Tsukita, Sh. Claudin-1 and -2: Novel integral membrane proteins localizing at tight junctions with no sequence similarity to occludin. J.Cell Biol. 141, 1539-1550 (1998).
16. Morita, K., Furuse, M., Fujimoto, K. & Tsukita, Sh. Claudin multigene family encoding four-transmembrane domain protein components of tight junction strands. Proc.Natl.Acad.Sci.USA 96, 511-516 (1999).
17. Tsukita, Sh. & Furuse, M. Occludin and claudins in tight junction strands: Leading or supporting players? Trend.Cell Biol. 9, 268-273 (1999).
18. Morita, K., Sasaki, H., Fujimoto, K., Furuse, M., & Tsukita, Sh.. Claudin-11/OSP-based tight junctions in myelinated sheaths of oligodendrocytes and Sertoli cells in testis. J.Cell Biol. 145, 579-588 (1999).
19. Morita, K., Sasaki, H., Furuse, M. & Tsukita, Sh. Endothelial claudin: Claudin-5/TMVCF constitutes tight junction strands in endothelial cells. J.Cell Biol. 147, 185-194 (1999).
20. Kubota, K. et al. Ca++-independent cell adhesion activity of claudins, integral membrane proteins of tight junctions. Curr.Biol. 9, 1035-1038 (1999).
21. Furuse, M., Sasaki, H. Fujimoto, K. & Tsukita, Sh. A single gene product, claudin-1 or -2, reconstitutes tight junction strands and recruits occludin in fibroblasts. J.Cell Biol. 143, 391-401 (1998).
22. Furuse, M., Sasaki, H. & Tsukita, Sh. Manner of interaction of heterogeneous claudin species within and between tight junction strands. J.Cell Biol. 147, 891-903 (1999).
23. Tsukita, Sh. & Furuse, M. Pores in the wall: Claudins constitute tight junction strands containing aqueous pores. J.Cell Biol. 149, 13-16 (2000).
24. Martin-Padura, I. et al. Junctional adhesion molecule, a novel member of the immunoglobulin superfamily that distributes at intercellular junctions and modulates monocyte transmigration. J.Cell Biol. 142, 117-127 (1998).
25. Palmeri, D., van Zante, A., Huang, C. C., Hemmerich, S. & Rosen, S.D. Vascular endothelial junction-associated molecule, a novel member of the immunoglobulin superfamily, is localized to intercellular boundaries of endothelial cells. J.Biol Chem. 275, 19139-19145 (2000).
26. Aurrand-Lons, M. A., Duncn, L., Du Pasquire, L. & Imhof, B. A. Cloning of JAM-2 and JAM-3; an emerging junctional adhesion molecular family? Curr.Top.Microbiol.Immunol. 251, 91-98 (2000).
27. Bazzoni, G. et al. Homophilic interaction of junctional adhesion molecule. J.Biol.Chem. 275, 30970-30976 (2000).
28. Liu, Y., Nustrat, A., Schnell, F. J., Reaves, T. A., Walsh, S., Pochet, M., and Parkos, C. A. Human junction adhesion molecule regulates tight junction resealing in epithelia. J.Cell Sci. 113, 236 3-23764 (2000).
29. Claude, P. & Goodenough, D. A. Fracture faces of zonulae occludentes from ﾒtightﾓ and ﾒleakyﾓ epithelia. J.Cell Biol. 58, 390-400 (1973).
30. Claude, P. Morphological factors influencing transepithelial permeability: a model for the resistance of the zonula occludens. J.Membr.Biol. 10, 219-232 (1978).
31. Martinez-Palomo, A. & Erlij, D. Structure of tight junctions in epithelia with different permeability. Proc.Natl.Acd.Sci.USA. 72, 4487-4491 (1975).
32. Mollgard, K., Malinowski, D. N. & Saunders, N. R. Lack of correlation between tight junction morphology and permeability properties in developing choroid plexus. Nature 264, 293-294 (1976).
33. Stevenson, B. R., Anderson, J. M., Goodenough, D. A. & Mooseker, M. S. Tight junction structure and ZO-1 content are identical in two strains of Madin-Darby Canine Kidney cells which differ in transepithelial resistance. J.Cell Biol. 107, 2401-2408 (1988).
34. Sonoda, N. et al. Clostridium perfringens enterotoxin fragment removes specific claudins from tight junction strands: Evidence for direct involvement of claudins in tight junction barrier. J.Cell Biol. 147, 195-204 (1999).
35. Gow, A. et al. CNS myelin and Sertoli cell tight junction strands are absent in OSP/claudin-11 null mice. Cell 99, 649-659 (1999).
36. McCarthy, K. M. et al. Occludin is a functional component of the tight junction. J.Cell Sci. 109, 2287-2298 (1996).
37. McCarthy, K. M. et al. Inducible expression of claudin-1-myc but not occludin-VSV-G results in aberrant tight junction strand formation in MDCK cells. J.Cell Sci. 113, 3387-3398 (2000).
38. Cereijido, M., Gonzales-Mariscal, L. & Contreras, G. Tight junction: barrier between higher organisms and environment. NIPS 4, 72 (1989).
39. Simon, D. B. et al. Paracellin-1, a renal tight junction protein required for paracellular Mg2+ resorption. Science 285, 103-106 (1999).
40. Furuse, M., Furuse, K., Sasaki, H. & Tsukita, Sh. Conversion of Zonula Occludentes from tight to leaky strand type by introducing claudin-2 into MDCK I cells. J.Cell Biol. in press (2001).
41. Saitou, M. et al. Complex phenotype of mice lacking occludin, a component of tight junction strands. Mol.Biol.Cell 11, 4131-4142 (2000).
42. Balda, M. S. et al. Functional dissociation of paracellular permeability and transepithelial electrical resistance and disruption of the apical-basolateral intramembrane diffusion barrier by expression of a mutant tight junction membrane protein. J.Cell Biol. 134, 1031-1049 (1996).
43. Wong, V. & Gumbiner, B. M.. A synthetic peptide corresponding to the extracellular domain of occludin perturbs the tight junction permeability barrier. J.Cell Biol. 136, 399-409 (1997).
44. Stevenson, B. R., Siliciano, J. D., Mooseker, M. S. & Goodenough, D. A. Identification of ZO-1: a high molecular weight polypeptide associated with the tight junction (zonula occludens) in a variety of epithelia. J.Cell Biol. 103, 755-766 (1986).
45. Gumbiner, B., Lowenkopf, T. & Apatira, D. Identification of a 160-kDa polypeptide that binds to the tight junction protein ZO-1. Proc.Natl.Acad.Sci.USA. 88, 3460-3464 (1991).
46. Balda, M. S., Gonz�lez-Mariscal, L., Matter, K., Cereijido, M. & Anderson, J. M. Assembly of the tight junction: The role of diacylglycerol. J.Cell Biol. 123, 293-302 (1993).
47. Itoh, M. et al. The 220-kD protein colocalizing with cadherins in non-epithelial cells is identical to ZO-1, a tight junction-associated protein in epithelial cells: cDNA cloning and immunoelectron microscopy. J.Cell Biol. 121, 491-502 (1993).
48. Willott, E., Balda, M. S., Fanning, A. S., Jameson, B., van Itallie, C. & Anderson, J.M. The tight junction protein ZO-1 is homologous to the Drosophila discs-large tumor suppresser protein of septate junctions. Proc.Natl.Acad.Sci.USA. 90, 7834-7831 (1993).
49. Jesaitis, L. A. & Goodenough, D. A. Molecular characterization and tissue distribution of ZO-2, a tight junction protein homologous to ZO-1 and the Drosophila discs-large tumor suppresser protein. J. Cell Biol. 124, 949-961 (1994).
50. Haskins, J., Gu, L., Wittchen, E. S., Hibbard, J. & Stevenson, B. R. ZO-3, a novel member of the MAGUK protein family found at the tight junction, interacts with ZO-1 and occludin. J.Cell Biol. 141, 199-208 (1998).
51. Itoh, M., Furuse, M., Morita, K., Kubota, K,, Saitou. M. & Tsukita Sh. Direct binding of three tight junction-associated MAGUKs, ZO-1, ZO-2, and ZO-3, with the COOH termini of claudins. J.Cell Biol. 147, 1351-1367 (1999)
52. Furuse, M. et al. Direct association of occludin with ZO-1 and its possible involvement in the localization of occludin at tight junctions. J.Cell Biol. 127, 1617-1626 (1994).
53. Itoh. M, Morita, K. & Tsukita, Sh. Characterization of ZO-2 as a MAGUK family member associated with tight and adherens junctions with a binding affinity to occludin and a catenin. J.Biol.Chem. 274, 5981-5986 (1999).
54. Fanning, A. S., Jameson, B. J., Jesaitis, L. A. & Anderson, J. M. The tight junction protein ZO-1 establishes a link between the transmembrane protein occludin and the actin cytoskeleton. J.Biol.Chem. 273, 29745-29753 (1998).
55. Bazzoni, G., Martinez-Estrada, O.M., Orsenigo, F., Cordenonsi, M., Citi, S. & Dejana, E. Interaction of Junctional Adhesion Molecule with the tight junction components ZO-1, cingulin, and occludin. J. Biol. Chem. 275, 20520-20526 (2000).
56. Ebnet, K., Schulz, C. U., Meyer Zu Brickwedde, M. K., Pendle, G. G. & Vestweber, D. Junctional adhesion molecule interacts with the PDZ domain-containing proteins AF-6 and ZO-1. J.Biol.Chem. 275, 27979-27988 (2000).
57. Martinez-Estrada, O. M., Villa, A., Breviario, F., Orsenigo, F., Dejana, E. & Bazzoni, G. Association of junctional adhesion molecule with calcium/calmodulin-dependent serine protein kinase (CASK/LIN-2) in human epithelial Caco-2 cells. J.Biol.Chem. in press (2000).
58. Dobrosotskaya, I., Guy, R. K. & James, G. L. MAGI-1, a membrane-associated guanylate kinase with a unique arrangement of protein-protein interaction domains. J.Biol.Chem. 272, 31589-31597 (1997).
59. Ide, N. et al. Localization of membrane-associated guanylate kinase (MAGI)-1/BAI-associated protein (BAP) 1 at tight junctions of epithelial cells. Oncogene 18, 7810-7815 (1999).
60. Wu, X. et al. Evidence for regulation of the PTEN tumor suppressor by a membrane-localized multi-PDZ domain containing scaffold protein MAGI-2. Proc.Natl.Acad.Sci.USA. 97, 4233-4238 (2000).
61. Wu, X. et al. Interaction of the tumor suppressor PTEN/MMAC with a PDZ domain of MAGI3, a novel membrane-associated guanylate kinase. J.Biol.Chem. 275, 21477-21485 (2000).
62. Izumi, Y. et al. An atypical PKC directly associates and colocalizes at the epithelial tight junction with ASIP, a mammalian homologue of Caenorhabditis elegans polarity protein PAR-3. J.Cell Biol. 143, 95-106 (1998).
63. Joberty, G., Petersen, C., Gao, L. & Macara, I. G. The cell-polarity protein Par6 links Par3 and atypical protein kinase C to Cdc42. Nature Cell Biol. 2, 531-539 (2000)
64. Lin, D., Edwards, A. S., Fawcett, J. P., Mbamalu, G., Scott, J. D. & Pawson, T. A mammalian PAR-3-PAR-6 complex implicated in Cdc42/Rac1 and aPKC signalling and cell polarity. Nature Cell Biol. 2, 540-547 (2000)
65. Qiu, R.G., Abo, A. & Steven Martin, G. A human homolog of the C. elegans polarity determinant par-6 links rac and cdc42 to PKCzeta signaling and cell transformation. Curr. Biol. 10, 697-707 (2000)
66. Itoh, M., Nagafuchi, A., Moroi, S. & Tsukita, Sh. Involvement of ZO-1 in cadherin-based cell adhesion through its direct binding to a catenin and actin filaments. J.Cell Biol. 138, 181-192 (1997).
67. Hata, Y., Nakanishi, H. & Takai, Y. Synaptic PDZ domain-containing proteins. Neurosci.Res. 32, 1-7 (1998).
68. Giancotti, F.G. & Ruoslahti, E. Integrin signaling. Science 285, 1028-1032 (1999).
69. Weber, E. et al. Expression and polarized targeting of a rab3 isoform in epithelial cells. J.Cell Biol. 125, 583-594 (1994).
70. Zahraoui, A. et al. A small rab GTPase is distributed in cytoplasmic vesicles in non polarized cells but colocalizes with the tight junction marker ZO-1 in polarized epithelial cells. J.Cell Biol. 124, 101-115 (1994).
71. Grindstaff, K. K. et al. Sec6/8 complex is recruited to cell-cell contacts and specifies transport vesicle delivery to the basal-lateral membrane in epithelial cells. Cell 93, 731-740 (1998).
72. Forstner, G.G. & Wherrett, J.R. Plasma membrane and mucosal glycosphingolipids in the rat intestine. Biochim.Biophys.Acta 306, 446-459 (1973).
73. Chapelle, S. & Gilles-Baillien, M. Phospholipids and cholesterol in brush border and basolateral membranes from rat intestinal mucosa. Biochim.Biophys.Acta 753, 269-271 (1983).
74. Barsukov, L.I., Bergelson, L.D., Spiess, M., Hauser, H. & Semenza, G. Phospholipid topology and flip-flop in intestinal brush-border membrane. Biochim.Biophys.Acta 862, 87-99 (1986).
75. Rothman, J.E., Tsai, D.K., Dawidowicz, E.A. & Lenard, J. Transbilayer phospholipid asymmetry and its maintenance in the membrane of influenza virus. Biochemistry 15, 2361-2370 (1976).
76. Dragsten, P.R., Blumenthal, R. & Handler, J.S. Membrane asymmetry in epithelia: is the tight junction a barrier to diffusion in the plasma membrane? Nature 294,718-722 (1981).
77. van Meer, G. & Simon, K. The function of tight junctions in maintaining differences in lipid composition between the apical and basolateral cell surface domains of MDCK cells. EMBO J. 5, 1455-1464.
78. Nelson, W.J. Regulation of cell surface polarity from bacteria to mammals. Science 258, 948-954 (1992).
79. Pisam, M. & Ripoche, P. Redistribution of surface macromolecules in dissociated epithelial cells. J.Cell Biol. 71, 909-920 (1976).
80. Ziomek, C.A., Shulman, S. & Edidin, M. Redistribution of membrane proteins in isolated mouse intestinal epithelial cell. J.Cell Biol. 86, 849-857 (1980).
81. Vega-Salas, D.E., Salas, P.J.I., Gundersen, D. & Rodriguez-Boulan, E. Formation of the apical pole of epithelial (Madin-Darby canine kidney) cells: polarity of an apical protein is independent of tight junctions while segregation of a basolateral marker requires cell-cell interaction. J.Cell Biol. 104, 905-916 (1987).
82. Jou, T-S., Schneeberger, E. E. & Nelson, W.J. Structural and functional regulation of tight junctions by RhoA and Rac1 small GTPases. J.Cell Biol. 142, 101-115 (1998).
83. Tsukita, Sh., Furuse, M. & Itoh, M. Structural and signaling molecules come together at tight junctions. Curr.Opin.Cell Biol. 11, 628-633 (1999).
84. Mankertz, J. et al. Expression from the human occludin promoter is affected by tumor necrosis factor a and interferon g. J.Cell Sci. 113, 2085-2090 (2000).
85. Chen, Yh., Lu, Q., Schneeberger, E. E. & Goodenough, D. A. Restoration of tight junction structure and barrier function by down-regulation of the mitogen-activated protein kinase pathway in ras-transformed Madin-Darby canine kidney cells. Mol.Biol.Cell 11, 849-862 (2000).
86. Li, D. & Mrsny, R. J. Oncogenic Raf-1 disrupts epithelial tight junctions via downregulation of occludin. J.Cell Biol. 148, 791-800 (2000).
87. Sakakibara, A., Furuse, M., Saitou, M., Ando-Akatsuka, Y. & Tsukita, Sh.. Possible involvement of phosphorylation of occludin in tight junction formation. J.Cell Biol. 137, 1393-1401 (1997).
88. Wilcox, E. R. et al. Mutations in the gene encoding tight junction claudin-14 cause autosomal recessive deafness DFNB29. Cell 104, 165-172 (2001).
89. Kinugasa, T., Sakaguchi, T. & Reinecker, H. C. Claudins regulate the intestinal barrier in response to immune mediators. Gastroenterology 118, 1001-1011 (2000).
90. Hough, C. D. et al. Large-scale serial analysis of gene expression reveals genes differentially expressed in ovarian cancer. Cancer Res. 60, 6281-6267 (2000).
91. Madara, J. L. Tight junction dynamics: Is paracellular transport regulated? Cell 53, 497-498 (1988).
92. Powell, D. W. Barrier function of epithelia. Am.J.Physiol. 241, G275-G288 (1981).
93. Madara, J. L. Regulation of the movement of solutes across tight junctions. Annu.Rev.Physiol. 60, 143-159 (1998).
94. Briehl, M.M. & Miesfeld, R.L. Isolation and characterization of transcripts induced by androgen withdrawal and apoptotic cell death in the rat ventral prostate. Mol.Endocrinol. 5, 1381-1388 (1991).
95. Katahira, J., Inoue, N., Horiguchi, Y., Matsuda, M. & Sugimoto, N. Molecular cloning and functional characterization of the receptor for Clostridium perfringens enterotoxin. J.Cell Biol. 136, 1239-1247 (1997).
96. Sirotkin, H. et al. Identification, characterization, and precise mapping of a human gene encoding a novel membrane-spanning protein from the 22q11 region deleted in Velo-Cardio-Facial syndrome. Genomics 42, 245-251 (1997).
97. Bronstein, J.M., Popper, P., Micevych, P.E. & Farber, D.B. Isolation and characterization of a novel oligodendrocyte-specific protein. Neurology 47, 772-778 (1996).
98. Citi, S., Sabanay, H., Jakes, R., Geiger, B. & Kendrick-Jones, J. Cingulin, a new peripheral component of tight junctions. Nature 333, 272-276 (1988).
99. Cordenonsi, M. et al. Cingulin contains globular and coiled-coil domains and interacts with ZO-1, ZO-2, ZO-3, and myosin. J.Cell Biol. 147, 1569-1582 (1999).
100. Zhong, Y. et al. Monoclonal antibody 7H6 reacts with a novel tight junction-associated protein distinct from ZO-1, cingulin, and ZO-2. J.Cell Biol. 120, 477-483 (1993).
101. Keon, B. H., Sch廓er, S., Kuhn, C., Grund, C. & Franke, W. W. Symplekin, a novel type of tight junction plaque protein. J.Cell Biol. 134, 1003-1018 (1996).
102. Saha, C., Nigam, S. K. & Denker, B. M. Involvement of Gai2 in the maintenance and biogenesis of epithelial cell tight junctions. J.Biol.Chem. 273, 21629-21633 (1998).
103. Balda, M. S. & Matter, K. The tight junction protein ZO-1 and an interacting transcription factor regulate ErbB-2 expression. EMBO J. 19, 2024-2033 (2000).
104. Nakamura, T. et al. huASH1 protein, a putative transcription factor encoded by a human homologue of the Drosophila ash1 gene, localizes to both nuclei and cell-cell tight junctions. Proc.Natl.Acad.Sci.USA 97, 7284-7289 (2000).