The name comes from the Latin verb “occludere”, which means to restrict passage. It was first isolated by Furuse et al[2] in 1993 in cultures of chicken liver cells, based on the results of intensive screening of monoclonal antibodies. It is a component of TJs, highly conserved among species and without tissue specific isoforms (recently only one occludin isoforms resulting from alternative splicing has been reported)[3]. Occludin is only expressed under normal conditions in cells that form TJs. When cells that do not form TJs are led via transfection to express occludin, it is concentrated at cell contact sites, forming only a small number of short strands identified in freeze-fracture replicas[4]. However, when occludin was cotransfected with claudin-1, tight junctional fibrils were formed[4]. In addition, immunoreplica electron microscopy has documented the presence of occludin in tight junctional strands[5].

Occludin (Figure 1) consists of 504 amino acids. The amino terminal portion is intracellular and contains 57 amino acids. This portion contains a WW domain (PPYP), which is motif participating in the interaction with other signaling and regulatory proteins[6]. The four transmembrane domains consist of 21-24 amino acid residues, mostly hydrophobic. The two extracellular loops are 43 amino acids long each and the intravellular loop is 10 amino acids long. Both extracellular loops are rich in tyrosine and glycine (in the range of 20%-30%); however, the functional significance of this is not known yet[2]. The very long intracellular carboxyl terminal part (50% of the sequence) has several sites of potential interaction with other macromolecules and phosphorylation sites[2,7]. It is a highly hydrophilic sequence, containing stretches of charged amino acids, like EEEEE (aa 347-351) and RRGRRRRR (aa 363-370).

The electrophoretic mobility of occludin in polyacrylamide gels has revealed a wide range of bands from 62kDa to 82 kDa. This difference in mobility has been assigned to post-translational modifications and more precisely to differential phosphorylation on both tyrosine and serine/threonine residues. It has been proposed that the site and the degree of phosphorylation of occludin are important parameters in defining the localization and the function of the molecule. For example, highly phosphorylated occludin, especially in serine/threonine residues, is detected mainly in the junctional area and correlates with restricted transcellular permeability. On the contrary, non-phosphorylated occludin is detected intracellularly, in the vesicular compartment, and is considered to contribute less to transcellular permeability, being a pool of macromolecules easily subjected to regulatory stimuli. As mentioned above, the site of phosphorylation is also important. It has been suggested that high degree of phosphorylation at tyrosine residues correlates with loss of function and higher transcellular permeability.

Recently, a novel alternatively spliced form of occludin has been described[3], in which 56 amino acids are present close to the amino terminal of the molecule. This new form termed occludin 1B has been detected in TJs and widely co-expressed with the known occludin.

The name claudin comes from the Latin verb “claudere”, which means “to close”. Claudins, identified after occludin, are transmembrane proteins extremely crucial for TJ formation. The existence of another TJ component, in addition to occludin, was suspected when Saitou et al[8] created an occludin deficient cell line that was surprisingly able to form TJ strands. These TJs appeared structurally normal, but this finding should not underestimate the role of occludin since a later studied model of occludin deficient mice[9] presented abnormalities from several tissues e.g., chronic inflammation and hyperplasia of the gastric epithelium, calcification in the brain, testicular atrophy, thinning of the compact bone and loss of cytoplasmic granules in striated duct cells of the salivary glands.

Furuse et al[10] identified the first two members of the claudin superfamily, claudins 1 and 2. Comparing their sequence to already known proteins, they found similarity to at least three membrane-associated proteins: rat ventral prostate-1, clostridium perfringens enterotoxin receptor and oligodendrocyte specific protein[11]. Up until now, the claudin family consists of 24 members in humans[12,13] (Table 1).

Claudins are transmembrane proteins with MW ranging from 20 kDa to 27 kDa. They possess an intracellular amino-terminus, four transmembrane segments that form two extracellular and an intracellular loop, and a short carboxyterminal intramembrane tail (Figure 2)[14]. The first and fourth transmembrane part, as well as the first and second extracellular loops are highly conserved, whereas the COOH tail has the greatest variability among members of the claudin family[11]. The first extracellular loop is longer and more hydrophobic than the second one[11,14]; therefore it is possibly responsible for cell-to-cell contact. The second extracellular loop contains many charged residues, which regulate the affinity to ions[14] and possibly forms ion-selective paracellular channels. The carboxyterminal tail contains many phosphorylation sites, related to protein kinase C (PKC), casein kinase 2 (CK2) and a cAMP dependent kinase. Almost every claudin has a carboxyterminal Tyr-Val sequence, which acts as a PDZ-domain interacting motif. Exceptions are claudins 16, 11, 12 and 13. When claudins 1 and 2 were transfected in TJs, negative cells[4] induced cell-to-cell adhesion by forming TJ strands with no obvious structural and functional difference from their occludin positive homologues. At present, claudins are supposed to be the “backbone” of the TJ barrier[14-16].

TJs in vivo usually express 2 claudins[15] with two notable exceptions, the oligodendrocytes and the Sertoli cells, both expressing a single claudin, claudin-11[11]. Claudins are also characterized by a differential tissue expression pattern[15,17] and a developmental selectivity. Interestingly, nephron expresses different claudins in different segments and this variability may be the basis for differential epithelial permeability in these segments[18-20]: claudins 5 and 15 are expressed at endothelial cells, 2, 10 and 11 at the proximal tubule segment, 1, 3 and 8 at the distal tubule and 1, 3, 4 and 8 at the collecting duct segment. As proved by Turksen et al[21,22] and Reyes et al[18], the expression of claudins in murine nephron changes during development. Hellani et al[23], who studied the expression of claudin 11 on Sertoli cells, also verified the above observations.

The complexity at the expression of different claudins in different tissues, in different stages of their development and to more than one combination, suggests that these molecules participate in both homophilic and heterophilic interactions within the same cell or between opposing cell membranes[24]. The possible combination models that claudins follow to their polymerization are presented in Figure 2. At the level of electron microscopy, claudins form different strand patterns, related to the P face or the E face of the cytoplasmic membrane. Claudins 1, 3 and 11 form P face strands, whereas claudins 2 and 5 are related to the E face strand pattern[15,24-26].

Claudins form a transmembrane and transcellular net that serves paracellular permeability, epithelial polarization and conservation of the transepithelial resistance (TER)[11,14,16], as well as selective permeation of charged molecules and ions[27,28]. These parameters vary among different tissues according to their claudin expression, as long as claudins are the only known variable elements in TJs[11].

Finally, some members of the claudin family are receptors for extracellular ligands, such as Clostridium perfringens enterotoxin, that bind directly to claudins-4 (high affinity) and -3 (low affinity)[29-31].

JAM was first described in 1998 by Martìn-Padura et al[32]. It is a type I transmembrane protein of MW 43 kDa, belongs to the immunoglobulin superfamily (Figure 3) and is localized in close proximity to the TJs of epithelial and endothelial sheets.

When JAM is expressed in cells that do not form TJs under physiological conditions, tight junctional strands are not observed, as is the case with occludin and claudins[33]. In this case, JAM molecules are accumulated only when cell-cell contact is established and both cells express JAM[32], suggesting a specific polymerization pattern by homophilic interactions. It appears that JAM plays a role in the process of formation of TJs, since antibodies against JAM do not disrupt already formed TJs[34].

Structurally, JAM consists of an extracellular amino terminal segment, a transmembrane domain and an intracellular carboxy terminal segment[32]. The extracellular part consists of 215 amino acids and contains two variable (V) type IgG domains. The V-V arrangement is novel between Ig domains and differentiates JAM-subfamily from other subfamilies of the Ig superfamily[32]. The intracellular part consists of 45 amino acids, where motifs appropriate for interactions with occludin[32] as well as other TJ associated macromolecules exist. The recently revealed X-ray structure of JAM suggests that first a U-shaped homodimer is formed and several homodimers interact and form an extensive network[35].

JAMs are subdivided based on the expression of type I or II PDZ-binding motifs in the intracellular C-terminus, which suggests that the 2 types interact with unique scaffolding and cytoplasmic proteins. JAM-A, JAM-B, and JAM-C (or JAM-1, JAM-2, and JAM-3) have type II binding motifs, whereas the atypical JAMs, including JAM-4, Coxsackie and adenovirus receptor and endothelial selective adhesion molecule, contain type I PDZ-binding domains[36].

It has been proposed that JAM may contribute to free diffusion of proteins within the lipid bilayer but more importantly to the restriction of cellular passage (leukocytes, monocytes, lymphocytes, etc.) through TJs between endothelial cells[14,32,35,37].

A recent addition to the list of TJ proteins is tricellulin. Structurally it is quite homologous to occludin, with 555 amino acids forming 4 transmembrane domains; both the amino and the carboxy termini are intracellular, forming two extracellular loops. It is localized at the tight junctional strands of tricellular contacts of epithelial cells[38]. It participates in epithelial barrier organization. It was recently found that recessive mutations in the tricellulin gene cause nonsyndromic deafness and that tricellulin participates in the junctions in cochlear and vestibular epithelial cells[39]. Recent studies indicate that tricellulin is also localized in special TJs of myelinating Schwann cells[40]. Tricellulin has also been proposed to play an inhibitory role in the process of epithelial to mesenchymal transition of gastric carcinoma cells[41].

Zonula occludens-1 (ZO-1) is a 210-225 kDa phosphoprotein that interacts with occludin[44], claudins[45] and JAM[46] and also ZO-2[47], ZO-3[47], AF-6[46], cingulin[48] and the actin cytoskeleton[44]. Therefore, it plays a key role in bringing several components together and in connecting tight junctional proteins to the cytoskeleton. Cells that do not form TJs contain ZO-1 either dispersed in the cytoplasm or concentrated in cadherin-enriched adherens junctions[49] (Table 2).

ZO-2 is a 160 kDa protein present in two different isoforms. Table 3 shows the relationships of ZO-2 to other TJ molecules and cytosceleton fibers[14]. ZO-2 reacts with the splicing factor SC35[50] and with the transcription factors Fos, Jun and C/EBP[51].

ZO-3 is a 130 kDa that differs from the other two members of ZO group in that it lacks the long carboxy-tail (Table 4). According to the model of Wittchen et al[47], ZO-1 binds to ZO-2 and ZO-3; however, there is no direct interaction between ZO-2 and ZO-3 (Figure 4).

Open in New Tab   Full Size Figure   Download Figure

Figure 4 Zonula occludens. JAM: Junctional adhesion molecule; ZO: Zonula occludens.

AF-6/Afadin is a 205 kDa protein that seems to be more important for TJ formation and development than for TJ stabilization[14]. It is expressed in both TJ and AJ, and it seems to correlate with: (1) ZO-1 and activated Ras protein have antagonistic effect for the same binding site[52]; (2) JAM; and (3) F actin. AF-6 also interacts with cingulin[48]. Afadin expresses a 190 kDa splicing variant[53], found at the postsynaptic densities of neuronal tissues.

Cingulin is named after the Latin word “cingere”, to encircle. It is a 140-160 kDa protein. It is characterized by a globular N terminal head, a central α-helical coiled-coil domain and a C terminal tail. Through its central rods, dimmers are formed that interact with ZO-1, ZO-2, ZO-3, JAM, F actin and myosin. The majority of these interactions are supported by its globular head[14]. Cingulin is a serine phosphorylated protein, independent from PKC[54].

One hundred and fifty five kilodaltons protein that belongs to a family of proteins with alpha helical coiled coil domain and an ATPase domain. It sustains phosphorylation and its phosphorylated form is detected at TJ areas, whereas the non-phosphorylated form dissociates form TJs[14].

TJs are expressed by the intestinal columnar epithelium. As the TJs of every epithelial tissue, there are occludin and special claudins expressed at intestinal TJs that mainly regulate the permeability of the epithelial layer under normal conditions. TJ strands are copolymers of heterogeneous claudin species and occludin, and heterogeneous claudin species constitute the backbone of TJ strands in situ[24].

Occludin associated with ZO-1 in a linear strand is detected among the intestinal epithelia and is strictly related to the differentiation status[83]. In highly differentiated adenocarcinomas, occludin and ZO-1 are normally expressed and form TJ structures. On the contrary, low differentiation carcinomas are characterized by low or absent occludin expression, whereas ZO-1, although normally expressed, appears to concentrate at the inner membrane area in a dotted pattern.

The intestinal epithelium is in continuous contact with the microbial ecosystem of the gut, with which it establishes a dynamic relationship. Based mainly on culturing techniques it, had initially been estimated that more than 500 bacterial species inhabit the human gut[84]. With the advancement of metagenomic technology, the magnitude of bacterial species diversity was raised to 15 000 to 36 000 species based on rRNA sequence analysis[85]. A recent release of the data from the Metagenomics of the Human Intestinal Tract project revealed a total of 3.3 million non-redundant microbial genes in human fecal specimens[86]. Intestinal epithelial cells and bacteria interact in a continuous so-called “cross-talk”, for mutual benefit. TJs participate in this interaction in a primary or a secondary way. They possess proteins that act as pathogen receptors and directly bind to the bacterial wall. Moreover, TJs can be secondarily affected by cytokine expression and by rearrangement of the actin cytoskeleton induced by other bacterial mechanisms, such as intimin-Tir mechanism of enteropathogenic Escherichia coli[87]. Some known interactions between bacterial strains and tight junctional macromolecules are described in (Table 5)[29,31,61,62,87-93]. Probiotics have a positive effect on TJ barrier, thus enhancing the epithelial resistance to pathogens, also reducing the paracellular permeability of antigens that may cause inflammation[94].

Dendritic cells, antigen presenting cells that belong to the subepithelial immune system, mediate a novel mechanism of bacterial uptake[95]. DCs send dendrites that express TJ proteins (occludin, claudin 1, ZO1) through the gaps between epithelial cells. These dendrites disassociate transcellular interactions, creating new ones between their surface proteins and previous cells. In this way, they send their dendrites to the epithelial surface and sample bacteria, without disrupting the integrity of the epithelial barrier.

Several lines of evidence suggest that TJs are involved in the pathogenesis of inflammatory and non-inflammatory intestinal diseases. TJ disruption leads to an inadequate epithelial barrier and, subsequently, to an incontrollable water and electrolyte loss. Secondly, intestinal TJ disruption is implicated in the pathogenesis of diverse extraintestinal autoimmune and inflammatory diseases. Thirdly, intestinal TJs may be secondarily disrupted in the course of several intestinal or extraintestinal diseases, leading to their further aggravation through promotion of systemic responses to endotoxin escape from the gut lumen.

Pathogenetic role in intestinal diseases: Changes at the level of TJs are related to the inflammatory bowel diseases (IBD), Crohn’s disease and ulcerative colitis. These are diseases that display a course of recurrent exacerbation and subsidence periods. At the peak of the IBDs, polymorphonuclear neutrophils transmigrate through the epithelial layer and induce inflammation at the intestinal surface. Transmigration is accompanied by an increase of the epithelial permeability to ions and macromolecules, as a result of downregulation of TJ. More specifically, occludin seems to be downregulated in a mechanism different from that of other TJ proteins[96]. The observation that, in clinically asymptomatic Crohn’s disease patients, increased intestinal epithelial permeability precedes clinical relapse by as much as 1 year, raised the assumption that permeability defect may be an early event in disease reactivation. Further evidence supporting that abnormal intestinal permeability occurs early in the pathogenesis of Crohn’s disease is provided by a study demonstrating intestinal barrier disruption, even in the non-inflamed parts of ileum from patients with Crohn’s disease[97]. The primary and potentially genetically determined association of intestinal TJs disruption with the evolution of IBD came from studies demonstrating increased gut permeability in otherwise healthy relatives of people with IBD[98]. However, the genetic factors implicated in these phenomena have not been revealed yet and the opposite opinion, that TJ disruption is a secondary event to the inflammatory process in IBD, is still under consideration[98,99]. Notably, disease exacerbations and the risk of developing IBD have been associated with emotional stress[100]. Experimental and clinical data provided evidence that psychological stress exerts injurious effects on intestinal TJs and increases gut permeability in IBD and irritable bowel syndrome[101,102]. Therefore, the pathogenetic role of intestinal TJs disruption in IBD beyond its potential genetic basis has an environmental stress-related component as well.

Celiac disease is an immune-mediated enteropathy triggered by an inappropriate T cell-mediated response to ingested gluten and its component gliadin. Clinical and experimental studies suggest that altered intestinal barrier function might play an inciting role in the development of celiac disease by allowing gliadin to cross the intestinal barrier and activate the immune system. Zonulin is the 47 kDa intestinal epithelial protein analogue of zonula occludens toxin (Zot) of Vibrio cholerae. Zonulin is normally expressed and secreted to the surface of intestinal and other epithelia (heart, brain). Zonulin and Zot have the same receptor on the cell surface and both trigger actin polymerization by PLC and PKCa pathways. Zonulin has been proposed to be the initial factor for the pathogenesis of celiac disease[87]. The pathogenetic role of intestinal TJs disruption in celiac disease is supported by studies demonstrating that increased intestinal permeability exists prior to disease onset, persists in asymptomatic patients who were on a gluten-free diet and is also present in a significant proportion of healthy first-degree relatives of patients with celiac disease who also have increased intestinal permeability[103-106].

Pathogenetic role in extraintestinal diseases: A combination of predisposing genetics, dysregulated intestinal barrier function and aberrant immune responses play an inciting role in type 1 diabetes. Increased gut permeability is believed to facilitate increased exposure to antigens that can trigger autoimmune destruction of the insulin-producing pancreatic beta cells[107]. Several lines of experimental and clinical data suggest that intestinal TJs disruption and increased gut permeability is an early event with a pathogenetic association with disease onset. Specifically, experimental studies with the Biobreeding diabetes-prone rat (BBDP), an inbred line in which autoimmune diabetes spontaneously develops when weaned onto a normal diet, showed that increased intestinal permeability, associated with decreased expression of the TJ protein claudin-1, precede the onset of insulitis and clinical diabetes[108]. Also, increased intestinal permeability was found in diabetic patients at various stages of disease progression and their relatives; however, prediabetic subjects had the greatest increase[109,110]. Zonulin has been proposed to be the initial factor for the pathogenesis of diabetes mellitus type 1[110].

Food allergies are expressed as adverse multisystemic immune-mediated reactions to ingested food proteins/antigens. Current pathogenetic aspects highlight the pivotal role of increased intestinal permeability, which permits increased dietary antigen transport across the intestinal barrier and exposure of dietary antigens to the mucosal immune system, leading to the development of the dietary antigen-specific response. In support of this hypothesis, intestinal permeability in infants with food allergies was significantly increased compared with that seen in healthy children[111]. In subjects with food allergies, intestinal permeability remained increased even after 6 mo of an allergen-free diet, indicating that the increased permeability probably preexists and is independent of food antigen stimulation[112]. Additional data supporting a role for increased intestinal permeability in the development of food antigen sensitization and food allergies are provided by recent clinical studies that demonstrate an association between increased intestinal permeability and the development of new-onset food allergies in patients after liver and heart transplantation under immunosuppressant therapy. These allergies were not related to the passive transfer of food antigen-specific IgE or lymphocytes from donors to previously nonallergic recipients, as were developed even in cases of non-allergic donors[113,114]. It is considered that immunosuppressive agents disrupt intestinal TJs, thus increasing permeability and facilitating presentation of food allergens to the immune system[115].

Beyond diabetes type 1, a similar mechanism has been proposed to contribute to the pathogenesis of diverse autoimmune diseases like atopic dermatitis, ankylosing spondylitis, Hashimoto’s thyroiditis and autoimmune hepatitis. The classical paradigm of autoimmune pathogenesis involving specific gene makeup and exposure to environmental triggers has been recently challenged by the addition of a third element, the loss of intestinal barrier function[116]. Disruption of intestinal TJs in genetically predisposed subjects might trigger a pathological immune response. According to this theory, once the autoimmune process is activated, it is not auto-perpetuating, but rather can be modulated or even reversed by preventing the continuous interplay between genes and environment, or in other words, by preventing loss of gut barrier function[116]. A zonulin - induced intestinal TJs disruption has been proposed as the responsible mechanism of barrier dysfunction allowing antigens to invade subepithelial and cause autoimmune reactions[87].

Secondarily affected in intestinal and extraintestinal diseases: Diverse intestinal and extraintestinal diseases during their course exert injurious effects on the integrity of intestinal TJs. Disruption of intestinal barrier function further subsequently aggravates through promotion of a systemic inflammatory response and the structural and functional integrity of the diseased and other organs, leading to clinical deterioration. Thus, although secondarily affected, dysfunction of the gut barrier exerts a pivotal role for the clinical outcome of diverse diseases. There are two main theories of how this may occur. The first one supports the view that after an initial insult, the intestinal barrier is compromised, thus allowing the passage of intestinal bacteria and endotoxins in mesenteric lymph nodes, portal circulation and normally sterile extraintestinal tissues (bacterial translocation)[117]. This process causes systemic infections and promotes a systemic inflammatory response both associated with distant organ failure and development of a septic state[118]. The second theory supports that, after the initial injurious insult and disruption of gut barrier integrity, bacteria and endotoxins crossing the mucosal barrier activate an intestinal inflammatory response, even when these translocating factors are trapped within the gut wall or intestinal lymph nodes and do not reach the systemic circulation[119]. Thus, the gut becomes a proinflammatory organ and gut-derived inflammatory factors, carried mainly in mesenteric lymph, induce a systemic inflammatory response and multiple organ failure[119-121]. In both theories, the intestinal TJs are further disrupted under the influence of systemic cytokinemia, further aggravating intestinal barrier function and leading to a vicious cycle.

The above described mechanisms may occur in diverse clinical states such as in intestinal ischemia[122], hemorrhagic shock[123,124], in critically ill patients[125], total parenteral nutrition[126], radiation enteritis[127], burns[128], ileus[129], acute pancreatitis[130], sepsis[131], cardiac bypass surgery[132], chemotherapy[133], obstructive jaundice[134,135], alcoholic liver disease[136], liver resections[137,138] and liver cirrhosis[139-141].