INTRODUCTION

Top Abstract Introduction References

POLARIZED EPITHELIAL CELLS function, in part, to provide a permeability barrier between two very different environments and to enable vectorial transport across the cellular layer. Highly specialized cellular components have evolved that allow epithelia to form an impermeant barrier and segregate cell surface membrane proteins and lipids into the apical and basolateral membrane domains. The junctional complex of polarized epithelia is a highly developed structure that was first appreciated in detail by electron microscopy in the 1960s. The junctional complex includes several well-defined structures including gap junctions, desmosomes, adherens junctions, and the tight junction (TJ) (see Fig. 1). Gap junctions mediate communication between cells by allowing small molecules to pass from cytoplasm to cytoplasm of neighboring cells, thereby metabolically and electrically coupling them together (reviewed in Ref. 33). Desmosomes are "buttonlike" points of intercellular contact that rivet cells together and provide anchoring sites for intermediate filaments (Fig. 1) (reviewed in Ref. 23). Adherens junctions form a continuous belt (the adhesion belt) and function to hold neighboring cells together through a family of Ca²⁺-dependent cell-cell adhesion molecules (cadherins) that are linked to actin and myosin filaments (Fig. 1) (reviewed in Refs. 26, 59). The TJ is the most apical component of the junctional complex and functions as the "fence" separating apical from basolateral domains, and is the major paracellular barrier. The protein components of the TJ and the basis for regulation of paracellular permeability have recently been reviewed (2). This review will focus upon recent advances in our understanding of how TJs are formed. The assembly of TJs in polarized epithelia is a critical event during tubular and ductal development and during recovery from ischemic or toxic injury (e.g., intestine and kidney). Over the past several years, work from a number of laboratories has helped to identify important proteins within the TJ and to define key events in TJ formation.

View larger version (24K): [in this window] [in a new window]   Fig. 1.   Epithelial cell junctional complex. Tripartite junctional complex contains the tight junction (TJ), adherens junctions, and desmosomes. The most apical component of the junctional complex is the TJ, which functions as a permeability barrier and separates apical from basolateral membrane domains. Occludin is the only known protein of the TJ with domains in the paracellular space. The adherens junction anchors cells together through the Ca2+-dependent cell adhesion molecules, E-cadherins. The adherens junction forms a continuous belt linked to actin and myosin filaments. Desmosomes are buttonlike structures that hold cells together and provide anchoring sites for intermediate filaments. An additional type of cellular junction, gap junctions, allows small molecules to pass from one cell into another (not shown).

	PROTEINS OF THE TIGHT JUNCTION

Before reviewing events important for the assembly of TJs, it will be useful to summarize the current understanding of proteins known to be associated with the TJ. Some of the major proteins of the TJ are schematized in Fig. 2, and the known TJ proteins are summarized and listed in Table 1.

View larger version (23K): [in this window] [in a new window]   Fig. 2.   Schematic representation of occludin and proteins that coprecipitate with ZO-1. Occludin has two extracellular loops that may interact with other molecules in the paracellular space. The COOH terminus (C) of occludin interacts with ZO-1, and immunoprecipitates of ZO-1 coprecipitate ZO-2 and several unidentified phosphoproteins (see text for details) including p130. In the ATP-depletion model, fodrin also tightly associates with ZO-1. There are many other molecules that have been described in the TJ, and these are summarized in Table 1 and in the text. N, NH2 terminus.

Occludin. The points of contact between cells visualized by electron microscopy reveal rows of intramembrane particles that form long branching fibrils circumscribing the cell (16). These fibrils are thought to represent linear polymers of transmembrane proteins that associate with similar particles in neighboring cells. The paracellular permeability between epithelial cells behaves as if the barrier contains pores or channels with a distinct preference for cations and molecules between 8 and 18 Å (9, 16). Recently, a transmembrane protein that may provide the basis of these structures has been identified, and it is called occludin. Monoclonal antibodies to junctional membranes from chicken liver were used to identify a 65-kDa protein that is exclusively localized by immunoelectron microscopy at the TJ of many different cell types (21). Mammalian homologs of chicken occludin have now been identified and are ~90% identical to each other but only 50% identical to chicken (3). The topology of occludin predicts the NH₂ and COOH termini to be in the cytoplasm with two extracellular loops projecting into the paracellular space (Fig. 2). The loops within the paracellular space may interact with loops originating from occludin in the neighboring cell or unidentified molecules to promote interaction and "sealing" of the paracellular space. Occludin migrates on sodium dodecyl sulfate-polyacrylamide gel electrophoresis as a series of bands between 62 and 82 kDa, depending upon the degree of phosphorylation. Phosphorylation occurs on serine or threonine residues, and the degree of phosphorylation may affect localization in the cell. The less phosphorylated forms (smaller sizes) are found in the basolateral membrane, cytosol, and TJ, whereas the more heavily phosphorylated form (larger size) is concentrated exclusively in the TJ (44). Recent evidence confirms the notion that occludin is a functional component of the paracellular pathway. Expressing chicken occludin in baculovirus caused multilamellar structures to accumulate in the cytoplasm that resembled TJs (20). Induction of chick occludin in MDCK cells caused an increase in transepithelial resistance (TER) and an increase in the number of TJ strands (35). Interestingly, as the duration of exposure to the inducing agent [isopropyl--D-thiogalactopyranoside (IPTG)] was increased, there was not only an increase in TER but also in mannitol flux. This "paradox" has been seen in other studies (7) and is thought to reflect different characteristics of the paracellular pathway. Treatment of A6 cells with peptides to the two extracellular loops indicates that loop two has a direct role in forming the paracellular permeability barrier. Cells treated with a peptide to the occludin extracellular loop 2, but not loop 1, reduced transepithelial resistance and increased paracellular flux of molecules up to 40 kDa. This occurred by reducing the amount of occludin at the TJ without obviously affecting cell viability, morphology, or other TJ protein levels (58). These results raise the possibility that occludin can be independently regulated without affecting other TJ proteins and that at least one of the extracellular loops is important to paracellular permeability. The COOH-terminal cytoplasmic domain of occludin is important for interaction with another TJ protein, ZO-1, as COOH-terminal deletions of ~150 amino acids prevented association with ZO-1 (22). Association with ZO-1 may also be important for localization of occludin to the TJ: in one study, COOH-terminal deletions of occludin failed to localize in the TJ (22), but in another study, a large COOH-terminal deletion (~250 amino acid) of occludin was found in the TJ, although in a discontinuous pattern (7). Similar to cells overexpressing occludin (35), cells transfected with COOH-terminally deleted occludin paradoxically increased both TER and paracellular flux while disturbing the separation of lipids into apical and basolateral membrane domains (7). The finding of increased TER and paracellular flux in cells expressing COOH-terminally deleted occludin has been proposed to occur through different effects of the mutant occludin on the sealing (electrical) features of the TJ and the flux characteristics of potential pores or channels in the TJ (7). The reasons for the different localization of COOH-terminally truncated occludin in the two studies are not clear but may result from differences in the extent of the COOH-terminal deletions and different systems of study.

ZO-1/ZO-2. Of several cytoplasmic proteins associated with the TJ, ZO-1 and ZO-2 (zona occludens 1 and 2) are the best characterized (Fig. 2). ZO-1 was identified in 1986 using specific monoclonal antibodies raised to a mouse liver-TJ fraction (48, 49). ZO-1 is found in TJs and in filtration slits of glomerular epithelial cells and in some cadherins junctions such as the intercalated discs of cardiac myocytes (31). ZO-2 appears to more restricted to TJs. TJs among various cell types can differ significantly, and multiple isoforms of ZO-1 and ZO-2 may contribute to these differences (2, 8). ZO-1 and ZO-2 interact with each other (25), and ZO-1 binds to the COOH-terminal tail of occludin (Fig. 2) (22).

Both ZO-1 and ZO-2 belong to the membrane-associated guanylate kinase (MAGUK) family of proteins (reviewed in Ref. 1). Members of this recently described family are often found at sites of cell-cell contact and may function to couple extracellular signaling pathways with the cytoskeleton. MAGUK family members share several conserved motifs including an SH3 domain, guanylate kinase domain, and PDZ domain(s). The PDZ domains are named for the three proteins in which the domain was first recognized (i.e., the initial letters of PSD-95, Dlg, and ZO-1). PSD-95 is a protein of the postsynaptic density, and Dlg is the product of the Drosophila lethal(1)discs-large-1 tumor suppressor gene that is located on the cytoplasmic surface of septate junctions. The SH3 domain is likely to be important for interaction with other signaling molecules or the cytoskeleton. The guanylate kinase domains are similar to the ATP-dependent enzyme that converts GMP to GDP, but there is significant variability of this domain among family members. PDZ domains appear to interact with the COOH-terminal cytoplasmic tail of transmembrane proteins, and this may be the mechanism of ZO-1-occludin interactions. Recently, the binding specificity for several PDZ containing proteins was determined (46), and it is likely that internal amino acid sequences can also bind to PDZ domains. Some PDZ proteins such as p55 in red blood cells and hDlg (human homolog of Drosophila dlg) bind directly to the actin binding protein 4.1. In contrast, other PDZ proteins, such as Lin-2 in Caenorhabditis elegans, are not associated with the membrane but appear to be important for signaling vulval cell induction (reviewed in Refs. 1 and 34).

Cingulin, 7H6, and others. Several other proteins have been localized to the TJ including cingulin (15), 7H6 (62), rab 13 (57), G_i-2 (17, 18), and protein kinase C (PKC) (19, 50). Cingulin is a 140-kDa protein described in 1988 (14) and is located in the junctional regions of epithelial cells from a variety of epithelial tissues (reviewed in Ref. 11). Cingulin appears to be two peptides intertwined as a "coiled coil" and is localized in close proximity to the vinculin-rich cytoskeletal belt associated with adherens junctions of chick embryonic kidney cells and in the TJ of chicken intestine (15). 7H6 (155 kDa) is a protein within the TJ of hepatocytes and epithelial cells that is recognized by a specific monoclonal antibody that was generated to a bile canaliculus-rich membrane fraction from liver (62). 7H6 has been described in endothelial cells, and in both endothelial and epithelial cells it may function to regulate paracellular permeability (45, 61). In addition, immunoprecipitations from ³²P-labeled MDCK cells have identified a series of phosphoproteins. ZO-1 coprecipitates ZO-2 and labeled proteins of 330, 130, and 65 kDa (Fig. 2; Table 1) (6, 50, 61). Immunoprecipitations of cingulin coprecipitate a band of ~200 kDa that is not ZO-1 (13). The ~130-kDa proteins appears to be homologous to the ZO-1/ZO-2 family (B. Stevenson, personal communication). The small G proteins, Rab 13 and Rab3B (57, 60) in addition to the heterotrimeric G proteins, G_i-2, and the recently described G family member, G₁₂ (17-19), have been identified in this region and may participate in the maintenance and/or regulation of TJ assembly. The tyrosine protooncogene c-Yes and the Src substrate p120 are also found near the TJ (41, 54). Recently, a novel protein (Symplekin) has been described on the cytoplasmic side of the TJ (32). This 127-kDa protein is not found in the junctions of endothelial cells but is widely expressed among many other cell types. Interestingly, Symplekin appears to be predominantly located in the nucleoplasm and is recruited to the TJ in those cells forming TJs. A nuclear localization has also been described for ZO-1 in subconfluent cells, but the implications for the nuclear localization of ZO-1 in TJ assembly are not clear (24).

	MECHANISMS OF TIGHT JUNCTION ASSEMBLY

Studies with cultured cell monolayers and intact tissues have been able to identify some of the second messenger and signaling pathways important for the assembly of TJs. To date, multiple signal transduction pathways have been implicated in TJ biogenesis including kinases (6, 12, 36, 38), Ca²⁺ (37, 51, 52), G proteins (5, 18), calmodulin, adenosine 3',5'-cyclic monophosphate (cAMP), and phospholipase C (5). Some of these signaling molecules are schematized in Fig. 4. Unraveling these pathways will be a major challenge in the next several years and is complicated by the observation that different epithelial cells may assemble and regulate TJs differently. There are likely to be some parallels between epithelial and endothelial cell TJ assembly, but epithelial cells are more diverse and will likely utilize some unique mechanisms. MDCK cells (epithelial cells derived from dog kidney) have been extensively utilized for the study of junction formation (reviewed in Ref. 10). Two different models of TJ assembly in epithelial cells have been used: the Ca²⁺ switch model and the ATP depletion-repletion model. The mechanisms of junctional assembly appear to be different in each of these models (53). The Ca²⁺ switch model has been widely utilized for studies of TJ assembly and is based the observation that MDCK cells plated at confluence establish TJs in 12-15 h through a process requiring protein synthesis, formation of an actin filament ring in close contact with the lateral membrane, calmodulin, and a Ca²⁺-dependent exocytic fusion of TJ proteins. Monolayers incubated in the absence of Ca²⁺ (low calcium, LC) lack cell-cell contact, intercellular junctions, and apical-basolateral polarization of lipids and protein. The formation of junctions can be followed by measurement of TER, an indicator of monolayer integrity (Fig. 3). Raising extracellular Ca²⁺ (normal calcium, NC; Ca²⁺ switch) triggers a series of molecular events that restores TJs, polarity, and TER (Fig. 3).

View larger version (51K): [in this window] [in a new window]   Fig. 3.   Comparison of the Ca2+ switch and ATP depletion-repletion models of TJ biogenesis. In both models, MDCK cells are plated at confluence on filters. Cells maintain an epithelial phenotype with apical and basolateral membranes domains that are defined by the TJ (hatched circles). Adherens belt is depicted as broken lines through the middle of the cell. When placed in low Ca2+ (µM), the cells become rounded, lose apical and basolateral domains, and allow unregulated flux across the monolayer (followed by transepithelial resistance, a reliable indicator of TJ integrity). In low Ca2+ (LC, µM), transepithelial resistance (TER) is near zero, and ZO-1 is found in intracellular granules (hatched circles) that also contain catenins. Upon restoring normal Ca2+ (NC, mM; Ca2+ switch), ZO-1 translocates to the lateral membrane and becomes more tightly associated with actin cytoskeleton. In contrast, during ATP depletion, TER also falls to near zero, but ZO-1 remains in the subapical lateral membrane and becomes more tightly associated with fodrin (depicted as 3 connected ovals along the lateral membrane) and other cytoskeletal proteins. With ATP repletion, ZO-1 interactions with cytoskeletal proteins normalize, and TER returns toward baseline.

Reestablishing the architecture of the actin cytoskeleton appears to be critical for the biogenesis of TJs. Most of the actin is positioned under the apical junctional complex where myosin II and several actin binding proteins including -actinin, vinculin, and radixin have been identified (Fig. 4). ATP-dependent contraction of the apical actin cytoskeleton of enterocytes was recognized in the mid 1970s (42), and it is well established that actin-disrupting drugs, such as cytochalasin, also perturb the paracellular barrier (47). Drugs that perturb the actin cytoskeleton are likely to disrupt the TJ through effects on actin originating on the perijunctional ring that projects onto the cytoplasmic surface of the TJ (29). Myosin movement along actin filaments is regulated by ATP and phosphorylation of the regulatory light chain by Ca²⁺-calmodulin-activated myosin light-chain kinase. In several systems, increases in intracellular Ca²⁺ can affect phosphorylation of myosin regulatory light chain contraction of perijunctional actin and cause increased paracellular permeability (55). Recent advances in understanding the signaling pathways important for cytoskeletal rearrangements during cell migration may have relevance for the biogenesis and/or maintenance of TJs. For example, the small GTP binding protein Rho regulates actin filament organization, and in polarized epithelial cells Rho also regulates organization and permeability of the TJ (Fig. 4) (39). Other proteins such as Rac and focal adhesion kinase (FAK) play a role in membrane ruffling and establishment of focal adhesion (reviewed in Ref. 28), although it is unknown whether these molecules also participate in regulation of TJs.

View larger version (28K): [in this window] [in a new window]   Fig. 4.   Potential interaction of signaling molecules in TJ biogenesis. Fodrin and the subapical cytoskeleton are shown under the apical membrane as a circle and in the midst of cross-hatched lines. TJ complex is below actin/fodrin and is depicted as a complex of occludin (transmembrane protein) with phosphoproteins (ZO-1, ZO-2, and P-130), which can potentially be phosphorylated on serine and tyrosine residues. Other possible signaling molecules including G subunits, small GTPases (Rab/Rho), tyrosine kinases (Tyr-kinases) and potential tyrosine phosphates (?TP) are depicted in the vicinity of the TJ complex. Stimulation of phospholipase C and perhaps other effectors leads to formation of inositol trisphosphate (IP3) and diacylglycerol (DAG). IP3 releases intracellular calcium from the endoplasmic reticulum (ER) which, in conjunction with DAG, activates protein kinase C (PKC). PKC translocates to the TJ, but it is not known whether the phosphorylation of TJ proteins occurs via PKC or unidentified kinase(s). There may also be calcium-dependent and -independent isoforms in the TJ. The receptor(s) that initiates these events is not known, but may include receptor tyrosine kinases, G protein coupled receptors or cell adhesion molecules (CAM).

The ATP depletion-repletion model of TJ biogenesis may have relevance to recovery from ischemia reperfusion or hypoxia reoxygenation injuries. Although this model has not been as extensively characterized as the Ca²⁺ switch, a major difference appears to be the nature of the interactions of TJ proteins with the cytoskeleton (Fig. 3). ZO-1 has been proposed to interact with actin cytoskeleton through spectrin (31), and recent work with the ATP depletion-repletion model of junction assembly supports this finding (53). Transepithelial resistance drops rapidly and reversibly in parallel with declining ATP levels, and ZO-1, ZO-2 and cingulin shift into a Triton X-100-insoluble pool, consistent with increased cytoskeletal interaction during ATP depletion. Analysis of immunoprecipitations from ATP-depleted cells identified ZO-1 and fodrin (a spectrin analog that links proteins to the cytoskeleton) within a large molecular weight complex. This is in contrast to the findings with the Ca²⁺ switch model; as the TJ is disassembled in LC, TJ proteins become more soluble, suggesting that they are less tightly associated with the actin cytoskeleton (Fig. 3). The differences in interactions between the cytoskeleton and TJ proteins in these two models suggests that the mechanisms of TJ assembly are likely to significantly vary depending upon the cause of junctional disassembly.

The essential role of Ca²⁺ in the formation of intracellular junctions is well established. Extracellular Ca²⁺ is required for homotypic interactions of E-cadherin and is likely to be the initial event of junctional complex formation (26). However, the source of intracellular Ca²⁺ that is critical for TJ biogenesis has important implications for understanding the signaling mechanisms involved in TJ formation (Fig. 4). Using fura 2-loaded MDCK cells and continuous spectrofluorometric measurements, it was demonstrated that there were large increases in intracellular Ca²⁺ that were particularly evident at points of cell-cell contact (37), the site of junction formation and apical membrane biogenesis. These observations were extended by chelating intracellular Ca²⁺ and demonstrating a marked attenuation of TER development (52). Furthermore, chelation of intracellular Ca²⁺ retarded the movement of ZO-1 from intracellular sites to the plasma membrane during the switch. A fraction of ZO-1 was redistributed from the Triton-soluble to Triton-insoluble pool during the Ca²⁺ switch, which could be inhibited by prior chelation of Ca²⁺. Although these studies demonstrate the importance of intracellular Ca²⁺, it was only recently that a role for potentially regulated intracellular Ca²⁺ stores was demonstrated (51). Thapsigargin (TG) inhibits endoplasmic reticulum (ER) Ca²⁺-adenosinetriphosphatase (Ca²⁺-ATPase) and depletes intracellular ER stores of Ca²⁺; this renders the cell insensitive to further stimulation of ER Ca²⁺ release. Selective depletion of calcium stores prior to initiation of cell-cell contact disrupts the biogenesis of desmosomes and TJs without obviously affecting cell-cell contact or E-cadherin (51). The sorting of ZO-1 and the desmosomal protein desmoplakin I are disturbed in TG-treated cells despite the presence of a normal intracellular Ca²⁺ concentration. The dependence of TJ formation on intracellular Ca²⁺ stores is consistent with a role for classic signaling pathways that utilize heterotrimeric G proteins and PKC (schematized in Fig. 4). Although the sorting of TJ proteins appears to be dependent upon intact internal Ca²⁺ stores and PKC, PKC does not appear to be required for biochemical stabilization of ZO-1 into the cytoskeletal fraction. Internal Ca²⁺ stores are required for maintaining ZO-1 in the TJ, but TG and PKC inhibitors have minimal effects on already established tight monolayers. These findings emphasize that different mechanisms are required for the maintenance and assembly of TJs.

As mentioned above, the assembly of TJs appears to be regulated, in part, by protein phosphorylation events. Over the past several years a variety of nonspecific and specific PKC agonists and antagonists have been used to elucidate some of the mechanisms relevant to TJ formation. The PKC inhibitor 1-(5-isoquinolinylsulfonyl)-2-methylpiperazine (H-7) markedly inhibits the development of transepithelial resistance of MDCK cells after switching from LC to NC media and inhibits the sorting of ZO-1 and cingulin from an intracellular site to the lateral membrane (36). Immunoprecipitates of ZO-1/ZO-2 complex in cells labeled with [³²P]orthophosphate identify unknown phosphorylated proteins at 130 kDa (6, 50) as well as 330 and 65 kDa (50). This complex can be detected in cells cultured in LC or NC, suggesting they exist as a preformed complex. Consistent with the effect of protein kinase inhibitors on TER and ZO-1 translocation, the PKC agonist 1,2-dioctanoylglycerol (DiC₈; Ref. 6) has the opposite effect. DiC₈ stimulates translocation of ZO-1 from the cytoplasm to the membrane and promotes actin cytoskeletal reorganization. Treatment of cells with PKC agonists also increases the number of TJ fibrils and is associated with decreased paracellular flux of [³H]mannitol, although there was only a small increase in transepithelial resistance. As predicted from the studies with PKC agonists, the PKC inhibitor calphostin blocks development of TER in the Ca²⁺ switch and prevents translocation of ZO-1 to the membrane (50). PKC activity increases shortly after switching to NC and there is a significant increase in total PKC activity of the membrane fraction. By immunofluorescence, PKC- translocates to the lateral membrane colocalizing with ZO-1 (50), and others have also found PKC- in the TJ of epithelial cells (19). However, there was no change in extractability of ZO-1 in the presence of calphostin C, suggesting that PKC is not required for cytoskeletal association, even though intracellular Ca²⁺ is clearly required (52). PKC- may also regulate TJ assembly as cells expressing dominant negative mutants were resistant to 12-O-tetradecanoylphorbol-13-acetate (TPA)-induced increases in paracellular flux and TER (43). Several studies have shown that ZO-1 can be directly phosphorylated (30, 50), but the implications for TJ maintenance and biogenesis are unknown. Although the kinase(s) responsible for phosphorylating ZO-1 have not been identified, a serine protein kinase has been partially characterized that selectively interacts with the SH3 domain of ZO-1 and is associated with the junctional complexes extracted from MDCK cells (4).

Activation of PKC pathways appears to be downstream of E-cadherin-mediated events. There is no obvious effect of the PKC agonist (DiC₈) on the cellular distribution of E-cadherin, and stimulation of PKC with the agonist overcomes the block on ZO-1 translocation from cytosol to membrane that is seen with anti-E-cadherin antibodies. Nevertheless, this does not exclude more subtle regulation of the adherens junction by PKC. Catenins link E-cadherin to the actin cytoskeleton and are required for the normal epithelial cell phenotype. The -, -, and -catenins associate with ZO-1 in LC and 2 h after Ca²⁺ switch, but these associations were not detected in confluent monolayers (40). The -catenin appears to be important for the translocation of ZO-1 from the cytosol to the TJ. In invasive colon carcinoma cell lines lacking -catenin, treatment with a PKC activator (TPA) causes restoration of a normal phenotype by stimulating desmosomal and TJ proteins without inducing -catenin expression or affecting E-cadherin (56). Studies with both agonists and antagonists of PKC are consistent with a central role for PKC in TJ biogenesis. Taken together, these results highlight the importance of phosphorylation events, in part through PKC, in TJ formation. Although less clear, some evidence suggests that PKC may also have role in maintaining TJs. A protein kinase inhibitor can prevent disassembly of TJs induced by low extracellular Ca²⁺ (12), but treatment of confluent monolayers with a PKC inhibitor in NC had no effect on TER or ZO-1 staining (50). In other cell types, such as thyroid follicular cells, different kinases or alternate mechanisms may be important. Stimulation of cAMP/PKA promotes barrier function of thyroid cells probably via stabilization of Ca²⁺-dependent cell adhesion (38). It is likely that other protein kinases (PKA) also participate in the maintenance and assembly of TJs.

Since PKC is critical for mobilizing intracellular Ca²⁺ during TJ biogenesis, it is quite likely that heterotrimeric G proteins are involved in this process as well. Evidence from several sources support an important role for G proteins in TJ biogenesis. Several G subunits including G_i-2, G₁₂, and G_o have been localized in the vicinity of the TJ (17-19, 27), and G_s may present in this region as well (Denker, unpublished observations). The specific G proteins involved in this process have not been identified. Studies in MDCK cells using the Ca²⁺ switch model of TJ biogenesis demonstrate a variety of effects when cells are treated activators [AlF₃ or guanosine 5'-O-(3-thiotriphosphate)] and inactivators (pertussis toxin) of some G protein families (5). In these studies, direct stimulation of adenylyl cyclase with forskolin or the use of cAMP analogs significantly reduced TER development in the Ca²⁺ switch. Direct activation of G_s with cholera toxin should also lead to increased cAMP levels and result in lower TER, but cholera toxin caused only a small reduction in TER development (5). Another line of evidence implicating G subunits in TJ formation comes from Ca²⁺ switch experiments on MDCK cell lines stably transfected with G subunits. G_o (pertussis toxin family member 69% identical to G_i-2) expressed in MDCK cells is localized to the TJ and can be immunoprecipitated with ZO-1 (18). Stably expressed G_o or a constitutively activated mutant (impaired GTPase activity) of G_o (Q205L) had no effect on TER in confluent monolayers. However, during the Ca²⁺ switch, activated G_o (Q205L) MDCK cells achieved significantly higher peak TER values and accelerated the rate of TJ biogenesis (18). The complexity of signaling pathways and the multitude of G proteins within a cell make it difficult to clearly establish the relevant G subunits and their role in TJ biogenesis. Nevertheless, these findings are consistent with the concept of multiple G subunits affecting TJ biogenesis. Furthermore, treatment of confluent monolayers with AlF₃ also affected TER (5) suggesting that G proteins may also participate in the steady-state regulation of TJ integrity and paracellular permeability.

The complexity of the TJ in regulating the paracellular pathway and providing the fence that separates apical and basolateral membrane domains is now apparent. Although there has been significant progress in the last few years toward identifying the components of the TJ, our understanding of assembly and maintenance of this structure remains far from complete. In many cell types this is a highly dynamic structure important to many critical cellular functions. In this regard, it is not surprising that the TJ is composed of a complex array of regulated proteins that are in place to ensure its integrity, allow regulation, and permit reestablishment under a variety of biological circumstances. Of particular interest in the future will be understanding the signaling mechanisms underlying reassembly of the TJ after ischemic and other insults. This will provide the basic understanding that may then make it possible to intervene therapeutically with agents that modulate these signaling pathways in circumstances like acute renal failure and intestinal ischemia.