Tight junctions: from molecules to gastrointestinal diseases

ABSTRACT

Intestinal epithelium functions as a tissue barrier to prevent interaction between the internal compartment and the external milieu. Intestinal barrier function also determines epithelial polarity for the absorption of nutrients and the secretion of waste products. These vital functions require strong integrity of tight junction proteins. In fact, intestinal tight junctions that seal the paracellular space can restrict mucosal-to-serosal transport of hostile luminal contents. Tight junctions can form both an absolute barrier and a paracellular ion channel. Although defective tight junctions potentially lead to compromised intestinal barrier and the development and progression of gastrointestinal (GI) diseases, no FDA-approved therapies that recover the epithelial tight junction barrier are currently available in clinical practice. Here, we discuss the impacts and regulatory mechanisms of tight junction disruption in the gut and related diseases. We also provide an overview of potential therapeutic targets to restore the epithelial tight junction barrier in the GI tract.

KEYWORDS:

• Tight junction
• paracellular permeability
• barrier function
• intestinal epithelium
• gastrointestinal disease

Introductory note

Tight junction is the key determinant of barrier function in various epithelial cell types. In the intestine, tight junction disruption can cause the leaky gut that is known to be associated with pathogenesis and progression of the gastrointestinal diseases. Therefore, targeting tight junction integrity can be the therapeutic approach for the treatment of related diseases in the gastrointestinal tract. Unfortunately, there is no tight junction-targeted drug in the clinical setting. To discover and develop a such drug that can recover tight junction integrity, insights into the physiological and pathophysiological mechanisms of tight junction regulation are required. In this review, we provide the comprehensive views of evidence regarding the mechanisms of tight junction assembly and integrity as well as related gastrointestinal diseases. In addition, we also discuss about the small molecules and known drugs that can be used for targeting tight junction integrity.

1. A brief history of tight junction discovery

The tight junction (TJ) was first visualized by transmission electron microscopy (TEM) in regions between adjacent epithelial cells in 1963¹ and was shown to be related to paracellular transport.^2–4 Freeze-fracture electron microscopy demonstrated an anastomosing network of TJ strands that were superficial and deep in “leaky” and “tight” epithelia,^5,6 respectively. The morphology of TJ strands was found to be strongly correlated with epithelial barrier function, as indicated by the transepithelial electrical resistance (TER) in the intestinal epithelial cells.⁷ The lipid micelle model was proposed as the first description of the molecular identity of TJs, in which the exoplasmic leaflets of neighboring cell membranes are partly fused together as lipidic hexagonal cylinders,^8–10 establishing a membrane fence. Later, it was reported that fluorescent-tag lipids were unable to diffuse in paracellular space,¹¹ thus refuting this hypothesis.

At present, TJs are known as a group of proteins that seal paracellular space between adjacent cells, near the apical membrane. Zonula occludens-1 (ZO-1), the first TJ protein, was discovered in 1986,¹² while cingulin was identified as another peripheral element of TJ in 1989.¹³ In 1993, occludin was identified as the first transmembrane protein contributing to TJ-related barrier function.¹⁴ Other transmembrane proteins acting on TJs, such as claudins, junctional adhesion molecules (JAMs), and MAL and related proteins for vesicle trafficking and membrane link domain-3 (MarvelD3), were later discovered.^15–18 In multicellular organisms, there are bicellular and tricellular points of cell–cell contacts. Recently, tricellulin and angulin-1 have been proposed as functional tricellular TJs that can determine the paracellular passaging of macromolecules.^19,20

In addition to being a legitimate barrier, some TJ proteins, for example, claudin-2 and claudin-15, can form paracellular pores for the permeability of water and ions in enterocytes.^21,22 Enhanced TJ permeability can result from defective integrity of either barrier-forming or activating pore-forming TJs, also known as leak and pore pathways, respectively.²³ Importantly, compromised epithelial barrier function can amplify a positive-feedback cycle of immune-associated barrier dysfunction, which is closely postulated with both the pathogenesis and the progression of gastrointestinal diseases. Unfortunately, no drug that can directly modify TJ integrity is currently available. To discover drugs targeting TJ recovery, it is important to obtain mechanistic insights into the regulation of epithelial TJs.

2. Structure and molecular architecture of tight junctions

Tight junctions (TJs) are dynamic structures, which are formed by the aggregation of an assortment of specific proteins at the apical membrane. TJ components can be distinguished by their localization into transmembrane proteins and cytoplasmic scaffolding proteins.

2.1 Transmembrane proteins

The transmembrane proteins of TJs are structurally classified into three groups: 1) single-spanning proteins, including Junctional adhesion molecule (JAM), Crumbs protein homolog 3 (Crb3), and Coxsackievirus and adenovirus receptor (CAR); 2) a tri-spanning protein, Blood vessel epicardial substance (BVES); and 3) tetra-spanning proteins of the claudin and tight junction-associated MARVEL protein (TAMP) families, which include occludin, tricellulin, and MarvelD3 (Figure 1). JAMs, claudin, occludin, and tricellulin have been extensively studied and shown to be involved in TJ permeability in intestinal epithelial cells.

Figure 1. Cellular architecture of tight junction. Tight junctions can be divided into bi- and tricellular tight junctions based on their localization. For example, tricellulin and angulin are considered as tricellular tight junctions. Most of tight junctions are transmembrane proteins located in bi-cellular point of cell-cell contact including occludin, claudins, and JAMs. Indeed, occludin and claudins consist of four membrane-spanning domains whereas JAM is a single membrane-spanning protein with large extracellular Ig-like domain. On the other hand, ZO-1, afadin, and cingulin are cytosolic scaffolding tight junction proteins that play an important role in arranging tight junction strands. The scaffolding tight junctions consist of a number of binding sites responsible for protein interaction.

JAMs belong to the immunoglobulin superfamily and are involved in cell–cell adhesion and permeability in the epithelium and endothelium.²⁴ There are four types of JAM, namely, JAM-A, JAM-B, JAM-C, and JAM-4, but the role of JAM-4 in TJs is still unclear.^25,26 Two pairs of JAM-A dimers from apposing cell surfaces ensure intercellular interaction through the immunoglobulin-like domain, while their PDZ-binding motif at the C-terminus allows TJ regulation by myriad scaffolding proteins.^25,27 JAM-A plays significant roles in apical membrane apposition junctions, regulation of macromolecule permeability, and formation of epithelial permeability.²⁸

laudins are known as fundamental components of TJs involved in modulating the paracellular permeability of ions and molecules.²² At least 26 members of the mammalian claudin family have been identified.²⁹ In TJs, claudins assort themselves into membrane strands via homo- or heteromeric interaction, and two anti-parallel strands connect to form the pore-like structure regulating the paracellular movement of small molecules.^22,28,29 This gated permeability and integrity of the claudin strands is largely reported to be sophisticatedly determined by the dynamic expression of pore-forming and siege-forming claudins (i.e., claudin-2 and claudin-4, respectively).^22,28,29 Indeed, pore-forming claudins are defined as paracellular ion channels. On the other hand, siege-forming claudins are non-ion channel type of claudins that can form paracellular barriers. The assortment of claudin substitutes in strand and modifications of TJ proteins by cellular signaling have been proposed as factors regulating TJ permeability, which has discussed later.²²

Occludin exhibits the MARVEL motif at a bicellular region, which allows the trans-homodimeric apposition of apical membranes.³⁰ The cytosolic C-terminus of occludin is composed of various phosphorylated residues for signaling kinases and a docking site for scaffolds.^30,31 Although occludin is considered an important TJ component, depletion of occludin in epithelial cells resulted in intact barrier, unaltered permeability, and increased localization of other TAMPs to the junctional area.³¹ Occludin may involve in TJ assembly with assistance of scaffolding proteins, i.e., zonula occludens (ZO)-1, but the physiological functions of each single protein are not fully understood.³² Moreover, recent work indicated that ZO-1 was not important for establishment of barrier function, but it played vital roles in intestinal mucosal repair.³³

In contrast to occludin, tricellulin, or MARVELD2 is predominantly found and colocalized with MARVELD3 at tricellular contacts where the edges of three cell surfaces are conjoined.¹⁵ Recent studies suggested that tricellulin is not only important for controlling macromolecule permeability, but also has prominent roles in the mechanical regulation at bicellular and tricellular regions.³⁴

2.2 Scaffolding proteins

A plaque of scaffolding proteins is incorporated into the TJ area with various domains that bind specifically to motifs of the integral part of TJs, other scaffold proteins, and the cytoskeleton to regulate TJ assembly, junction permeability, and related signaling.

2.2.1 PDZ-containing scaffolding proteins

PDZ-containing proteins form a junctional plaque through binding with TJ integral proteins.³⁵ The evolutionarily conserved, ~90-residue-long PDZ sequence allows the scaffolding proteins to perform two major functions: 1) anchoring the intracellular C-terminus of transmembrane proteins (TJs, ion channels, receptors, etc.) and 2) establishing scaffolding networks via interactions with other PDZ proteins through PDZ–PDZ dimerization.^36,37

Zonula occludens (ZO), including ZO-1 and its homologs ZO-2 and ZO-3, is a group of PDZ-related scaffolding proteins that belong to the superfamily of membrane-associated guanylate kinase-like proteins, possessing sites of three tandem PDZ domains at the N-terminus^38,39 (Figure 1). The first PDZ domain interacts with claudins, the binding affinity of which is regulated by phosphorylation at the PDZ-binding motifs of claudins.⁴⁰ The second PDZ domain establishes dimerization with another PDZ2, while PDZ3 is bound by JAM-A. In addition to PDZ regions, ZO-1 engages with other scaffolding proteins and cytoskeletal components, such as afadin, occludin, and α-catenin, through Src homology 3 (SH3), U5, guanylate kinase-like (GUK) regions, and actin-binding region (ABR).^41,42 Moreover, the interactive zone of PDZ3 together with SH3-GUK-U6 facilitates the phase separation of ZO-1 and polymerization of ZO proteins during ZO assembly.^43,44

ZO-1 is a fundamental component of TJs that interacts with other scaffolds and F-actin to regulate epithelial barrier permeability.^45–47 In mouse epithelia, ZO-1 depletion was shown to cause delayed TJ formation and reassembly with integrity being unaffected, but TJ disruption was not detected in ZO-2-depleted cells.⁴⁸ However, double deletion of ZO-1 and ZO-2 established leaky monolayers with the complete loss of TJ strands,⁴⁵ suggesting that ZO-1 and ZO-2 have overlapping functions in anchoring the TJ strand. ZO-1 also plays a key role in conveying the signaling between the TJ structure and the actin cytoskeleton.⁴⁹ The ABR of ZO-1 normally forms a weak and intermittent interaction with F-actin.⁵⁰ Either strong binding or no interaction among ZO-1, ABR, and F-actin causes elevated TJ permeability.^49,50 Notably, ZO-1 deficiency in intestinal epithelium was demonstrated to lead to the concentration of F-actin and contraction of actomyosin fibers, but did not depend on ABR loss.⁴⁴ It is suggested that the reorganization of F-actin after change in ZO-1 may indirectly engage ZO-1-binding scaffolds, namely, ZO-2, cingulin, paracingulin, and afadin.⁴¹

Other PDZ domain-containing scaffolding proteins, in addition to ZOs, including partitioning defective 3 (Par3), Par6, protein associated with Lin‑7 1 (PALS1), PALS1‑associated tight junction (PATJ), multi-PDZ domain protein 1 (MUPP), and afadin, have been implicated in TJ assembly as well as apico-basal polarity.^{32,41,51–54} Interestingly, afadin has been identified to bind to ZO-1, JAM-A, actin, and nectin (adherens junction protein).^46,55 Afadin plays profound roles in the initiation of TJ assembly and polarization by gathering ZO-1 and JAM-A to the adherens junction area.^41,51 Deficiency of afadin in mouse intestine was reported to cause impairments of the TJs and adherens junctions of intestinal crypt cells. In addition, reduction in afadin expression was found to be associated with the progression of colon and pancreatic cancers.^56–58

2.2.2 Non-PDZ scaffolding proteins: cingulin and paracingulin

Cingulin and paracingulin homodimers consist of a coiled rod and globular head, which have a shared ZO-1-Interaction-Motif (ZIM) at the N-terminus.^59,60 Cingulin was found to bind to TJ components, namely, JAM-A, ZO-1, and ZO-2. Moreover, cingulin exhibits high affinity to actin⁶¹ and microtubules,^62,63 with this affinity being regulated by AMP-activated protein kinase (AMPK) phosphorylation.^62,63 Furthermore, cingulin as well as paracingulin downregulates the activity of Rho family proteins, resulting in the inhibition of claudin-2 expression and cell proliferation.^64–66

3. Intestinal epithelium and permeability pathways

3.1 Intestinal epithelial barriers

Mucosal surfaces that are composed by distinct cell types and separate the internal environment from the external milieu are necessities of maintenance of first-line of defense and body homeostasis. There are multiple sites that face the external environment, including mucosal surfaces of the respiratory, urinary, and GI tracts, as well as the skin. Although the skin is the most visible interface, the entire surface of mucosal layers is actually much larger than the whole surface area of the skin.^67,68 In the GI tract, epithelium-lined mucosal surfaces form a selective barrier to limit the passage of dietary antigens and virulence factors from the intestinal lumen.^23,67–70 The intestinal epithelium also regulates vectorial transport for nutrient and mineral absorptions and secretion of unwanted compounds.^71–77 In fact, vectorial transport across epithelial layers and epithelial tissue barriers requires the integrity of tight junction. Increased intestinal permeability can ultimately lead to activation of the mucosal immune system and its sequelae to cause the pathogenesis and progression of inflammation-associated mucosal diseases.^{21,70,78–85}

Increased intestinal permeability can be due to tight junction-dependent and tight junction-independent mechanisms. Two tight junction-dependent permeability pathways have been proposed, namely, the leak and pore pathways.²³ The pathways can be defined as involving defects of barrier-forming and activation of pore-forming tight junctions, respectively. Meanwhile, increased tight junction-independent unrestricted pathway permeability occurs at sites of apoptosis-induced epithelial damage and tissue erosion.^{23,68–70,84,86} A comprehensive model of the barrier permeability pathways is illustrated in Figure 2.

Figure 2. Tight junction-dependent paracellular permeability pathways. There are the well-known paracellular permeability pathways including leak, pore, and unrestricted pathways. Leak and pore are tight junction-dependent paracellular permeability pathways. Indeed, leak pathway is a size-selective permeability whereas pore pathway is size- and charge-selective permeability. Of note, pore-forming tight junction is sometimes called as paracellular ion channel. Conversely, unrestricted pathway occurs at the site of apoptotic cells or tissue damage.

3.2 Leak and unrestricted permeability pathways

The leak pathway is considered to be a low-conductance, size-selective paracellular pathway, which allows the permeability of macromolecules with a hydrodynamic radius of ~8–100 Å.^{23,68–70,84,86} Defects of barrier-forming tight junctions, for example, ZO-1, occludin, claudin-1, claudin-4, and tricellulin, have been implicated in the leak pathway.^{32,70,87–96} Among the several proposed signaling pathways, the activation of myosin light-chain kinase (MLCK)-induced, caveolin-1-dependent occludin internalization has been widely accepted as one of the major mechanisms of tumor necrosis factor (TNF)-mediated leak pathway permeability.^97–101 Notably, the C-terminal coiled-coil occludin/ELL domain (OCEL) of occludin and its K433 amino sequence are required for TNF-induced leak pathway permeability.¹⁰² Three differently sized routes of macromolecular flux have been proposed to contribute to the leak pathway, which are permeable to molecules with hydrodynamic radii of 8, 10, and 60 Å.^103–108 In isolated rat small intestine, the capacity for permeability of the upper part of the villi is generally lower than that of the lower part and the crypt is highly permeable to molecules with radii of up to 60 Å.¹⁰⁶ In response to activation of Na⁺/glucose cotransport-1 (SGLT1) in Caco-2 cell monolayers, tight junction permeability was found to be significantly increased, as demonstrated by a decreased TER value.¹⁰⁰ Consistent with this, paracellular flux of [³H]mannitol (hydrodynamic radius of 3.6 Å), but not [¹⁴C]inulin (hydrodynamic radius of 11.5 Å), was also increased, indicating that a decrease in TER resulted from tight junction-dependent permeability, not epithelial damage.¹⁰⁰ This study also provided evidence of the smallest route (hydrodynamic radius of 3.6 Å) of an uncharged, size-selective permeability pathway of macromolecules. Notably, the existence of differently sized routes of leak pathway that are regulated by neither different tight junction components nor the degree of decrease of a similar barrier-forming tight junction has been demonstrated. Upon severe tissue damage, the unrestricted pathway, which is a high-conductance and size/charge-independent pathway, occurs in ulcerated epithelia.⁷⁰ This nonselective permeability event allows the transmucosal transport of large particles (hydrodynamic radius of >100 Å) such as proteins and bacteria into circulation; it is found in patients with necrotizing enterocolitis and in dextran sulfate sodium (DSS)-treated mice.^84,109,110 The 4-kDa fluorescein isothiocyanate-dextran (4-kDa FITC-dextran) is generally used as a probe for investigating leak pathway permeability.¹¹¹ However, using only one probe, it is not possible to distinguish leak and unrestricted pathways experimentally.¹¹² Both in vitro and in vivo multiplex analyses of size-selective macromolecular permeability assays have recently been developed to enable spontaneous measurements of leak and unrestricted permeability pathways.^112,113 These methods are based on the capacity for permeability of a mixture of multi-colored and differently sized fluorescent probes.

3.3 Activation of paracellular ion channel and pore pathway permeability

The pore pathway permeability is defined as a high-conductance, and size- and charge-selective paracellular route that primarily relies on expression of the pore-forming claudins.^23,70,114 Notably, the pore pathway accommodates the permeability of small, charged ions with a pore size of ~4–8 Å.^{21,23,69,80,104,115} Polyethylene glycol (7 Å) and creatinine (6 Å) have been used as probes for the in vivo investigation of pore pathway permeability.^80,104 Indeed, 4-kDa FITC-dextran or fluorescein, a smaller uncharged, paracellular fluorescent probe (9 Å diameter), cannot pass through the pore pathway.⁸⁶ There are many in vitro models that can be used for demonstrating pore pathway permeability, including cultured intestinal epithelial cells (Caco-2 and T84 cell lines) and Madin–Darby canine kidney (MDCK) cells.^104,116 Notably, two distinct clones of MDCK, namely, MDCK1 (without expression of pore-forming claudins) and MDCK2 (endogenous expression of pore-forming claudins), can be used for experiments on pore pathway permeability. In enterocytes, claudin-2 and claudin-15 are the dominant paracellular monovalent cation and water channels, with different degrees of expression.^78,117–121 In fact, claudin-2 is highly expressed in intestinal tissue at birth, but its expression gradually declines during post-neonatal development until being almost experimentally undetectable in adulthood. In contrast, claudin-15 has a low level of expression in children, but relatively high expression in adults. The expression of claudin-2 and claudin-15 is primarily restricted to crypt epithelia in both rodents and humans.^21,78,120 Nevertheless, little is known about the contribution of claudin-15 to pore pathway permeability in GI disease. Conversely, the impact of claudin-2 expression on increased activity of the pore pathway has been extensively investigated and shown to be related to disease progression in experimental colitis and active inflammatory bowel disease (IBD) patients.^{21,84,86,122,123} Generally, claudin-2-mediated paracellular Na⁺ conductance can be indirectly measured by dilution and bi-ionic potential assays.^23,124 Recently, the trans-tight junction patch clamp approach has been developed to elucidate the gating characteristics of an independent, single claudin-2 channel in Tet-off regulated claudin-2-inducible MDCKI monolayers, which revealed the claudin-2-dependent conductance of ~9 pA.¹²⁵ This method also identified three main states of non-rectifying claudin-2 gating: open, stable, and transient closed states.¹²⁵

4. Tight junction dynamics

4.1 Evidence of tight junction mobilization

As proteins related to the structure of the cell, the integrity of tight junctions was thought to provide a static, impermeable barrier at the site of apical intercellular space complexes. However, this hypothesis was proven to be almost entirely incorrect. In fact, the concept of tight junction dynamics has a long history of over 30 years.¹²⁶ In the last decade, fluorescence recovery after photobleaching (FRAP) revealed different mobile characteristics of fluorescent protein-tagged tight junctions during a steady state.^127–129 Although claudin-1 stably localized at the tight junction region, ZO-1 and occludin are differentially exchangeable at subcellular levels. Notably, tight junction-to-cytoplasm switching of ZO-1 was found to occur in an energy-dependent manner, which required activity of the actin-binding region (ABR) and MLCK. Indeed, ZO-1 switching could occur via either MLCK-dependent or MLCK-independent mechanisms, which are slow and fast kinetics, respectively.²³ Meanwhile, occludin passively diffuses between apical and lateral membrane.²³ Therefore, it is widely accepted that tight junction remodeling can spontaneously occur.¹²⁷ These kinetic behaviors of three representative tight junction proteins refute the theory of tight junctions having a static architecture (Figure 3a).

Figure 3. Tight junction dynamics. A) Movement of tight junction at steady state. Tight junction is motile as seen from the kinetic active movement of ZO-1 between tight junction area and cytoplasmic pool that requires energy and MLCK in fast-moving state. Occludin always moves along lateral membrane that is passive, no energy required. In contrast, claudin-1 is stable in the tight junction region. B) CK2 is the regulator of tight junction-dependent pore pathway permeability. After being activated by IL-13, claudin-2 is overexpressed and move toward to tight junction region, resulting in trans-homodimerization of claudin-2 to form paracellular cation channel. This formation is partly due to CK2-mediated phosphorylation of S408 of occludin that promotes cis-homodimerization of occludin. Inhibition of CK2 can lead to abolishment of cis-homodimerization of occludin. Therefore, this event can promote ZO-1/occludin/claudin-2 interaction via U5-GUK and PDZ1, leading to prevention of trans-homodimerization of claudin-2. Indeed, CK2 inhibitor is considered as an indirect inhibitor of claudin-2-dependent pore pathway permeability.

4.2 Roles of casein kinase-2 (CK2) on tight junction interactions

Interaction between each tight junction protein has been reported to regulate barrier function. As a scaffolding protein, ZO-1 has multiple interacting domains, for example, U5-GuK and PDZ1.^{47,48,130–133} In general, casein kinase-2 (CK2) phosphorylates the S408 amino acid of occludin, resulting in occludin–occludin homodimerization. This in turn allows trans-homodimerization of claudin-2 to form a paracellular cation channel that promotes pore pathway permeability.¹³⁴ Meanwhile, the inhibition of CK2-mediated phosphorylation of occludin S408 promotes the cis-heterotrimeric complex of ZO-1/occludin/claudin-2. Indeed, ZO-1 can bind to non-phosphorylated occludin and claudin-2 via U5-GuK and PDZ1 domains, respectively. This interaction prevents claudin-2 trans-homodimerization, leading to the attenuation of claudin-2-dependent pore pathway permeability.¹³⁴ Therefore, CK2 inhibitor has been considered as an indirect inhibitor of claudin-2 (Figure 3b).²¹

4.3 Integrative functions of bi- and tri-cellular tight junctions

In addition, coordinated functions of bicellular and tricellular tight junctions have been reported. The expression of claudin-1 and claudin-5 enhanced the enrichment of tricellulin and occludin at the intercellular point of cell–cell contacts.¹³⁵ Moreover, MarvelD3 was reported to interact with occludin and tricellulin at tight junction regions, but no interaction between occludin and tricellulin was observed.¹³⁵ Tricellulin was found to be capable of localizing on a bi-cellular tight junction in occludin-knockdown cells and occludin-knockout mice to compensate for deficient bi-cellular tight junction-dependent barrier function of occludin.¹³⁶ Furthermore, recent evidence indicated that double knockout of occludin and tricellulin in MDCKII cells led to decreasing numbers of anastomosing networks and branches compared with those in controls.¹³⁷ Notably, in occludin/tricellulin double-knockout cells, the rate of paracellular permeability was significantly elevated. It was noted that this effect was not found upon single knockout of either occludin or tricellulin.¹³⁷ This suggested that occludin and tricellulin play crucial roles in forming barriers to macromolecular passage.

4.4 Claudin–claudin interactions

There are at least two types of claudin–claudin interaction: trans- and cis-interactions.¹³⁸ Trans-interaction involves the head-to-head binding of claudins between adjacent cells, while cis-interaction involves the lateral interaction between claudins within a single membrane. Tight junction assembly is related to health and diseases as well as being involved in dynamic rearrangements.^87,90,139 Recently, using cells transfected with claudin-15 as a model, a spatiotemporally elastic remodeling process of tight junctions has been reported.¹⁴⁰ Heterotypic claudin interaction, which can be in trans and cis forms, has been found to contribute to the maintenance of barrier function in many cell types.^140–142 Indeed, claudins were originally known as the backbone of the anastomosing network of tight junction strands.^{141,143–148} Several lines of evidence have indicated that extracellular loops (EL1 and EL2) of claudins are important for both trans- and cis-claudin–claudin interactions.¹⁴⁹ Charge of K65 (positive charge) in Claudin-1, E158 (negative charge) in Claudin-3, and the polarity of Q56 and Q62 in Claudin-3 and Q57 in Cldn5 are required for establishing tight junction strands.¹⁴¹ Notably, it has been reported that the first extracellular domains of claudin-2 and claudin-4 are a key determinant of paracellular charge selectivity and barrier function, but these domains are not involved in the formation of tight junction strands.¹⁵⁰ Several lines of evidence have indicated homo- and heterodimeric interactions of claudins to form tight junction strands and exert pore-forming activity. In the case of interaction between two barrier-forming claudins, this interaction leads to the establishment of barrier function. However, heterodimeric interaction of barrier-forming and pore-forming claudins has never been fully elucidated. It would be interesting to obtain more insight into this type of interaction since it may act synergistically or antagonistically, which would deepen our understanding of the regulation and restoration of tight junction permeability.

5. Cellular signaling regulators in TJ structures and function

5.1 Myosin light chain kinase

The myosin light-chain kinase (MLCK) is a Ca²⁺/calmodulin-dependent serine/threonine kinase.¹⁵¹ MLCK orchestrates a series of cytoskeleton-related regulatory proteins and controls actomyosin arrangement and contraction of cardiac, smooth, and skeletal muscle.^70,151 In addition, MLCK was previously identified as a diagnostic and drug target for cancer.¹⁵² There are at least three different MLCK-encoding genes: MYLK1, MYLK2, and MYLK3. MYLK2 and MYLK3 are skeletal- and cardiac-specific MLCK-encoding genes,^153,154 respectively, while MYLK1 encodes long non-muscle isoforms (~210 kDa) including MLCK1 (full-length “long MLCK”) and MLCK2.¹⁵⁵ The alternative splicing variants of long MLCK in intestinal epithelia determine the differential functions and localization of MLCK. In the GI tract, MLCK1 localizes mainly in villous epithelium and is highly abundant near the peri-junctional actomyosin ring (PAMR), whereas MLCK2 is expressed ubiquitously along the crypt-villus axis.¹⁵⁶ The intimate association between the apical cytoskeleton and intestinal epithelial tight junctions has long been investigated.^157–160 MLCK1 expression and recruitment to PAMR have been proposed to contribute to tight junction permeability in physiological and pathological circumstances. Indeed, physiological increase in TJ permeability by MLCK was observed in paracellular nutrient absorption. The activation of Na⁺-glucose cotransport requires MLCK upregulation to increase paracellular permeability.¹⁰⁰ This physiological event could compensate the saturated glucose transporters. Conversely, synergistic activation of IFN-γ and TNF promotes MLCK1 recruitment to PAMR by upregulating TNF receptor 2 to enhance tight junction permeability, contributing to colitis and immune-mediated diarrhea.^{84,85,161–163} When MLCK1 is recruited to PAMR, it generally phosphorylates the myosin light chain (MLC), resulting in contraction of the cytoskeleton and occludin internalization.^{100,161,164–166} Therefore, MLCK is widely accepted as a therapeutic target for diseases related to tight junction disruption. Nevertheless, direct enzymatic inhibition of MLCK is not feasible for patients due to its association with severe adverse effects and death.

5.2 Rho-ROCK

The small Rho GTPase family regulates various cellular functions, at least in part via three main G-protein types: Rho, cell division control protein 42 (CDC42), and Rac.¹⁶⁷ These signal molecules and the partners with which they interact intracellularly are involved in the regulation of early embryonic development, cell proliferation, adhesion, polarity, cytokinesis, and survival.¹⁶⁸ Rho-associated coiled-coil-containing kinase (ROCK), also called Rho kinase, is a downstream effector of RhoA GTPase that appears to regulate several pathways involved in paracellular permeability, for example, extracellular signal-regulated kinase (ERK), NF-κB signaling, myocardin-related transcription factor-A (MRTFA), and MLCK pathways.^169–172 Notably, ROCK acts as a key regulator in cytoskeletal rearrangement and actin organization via the suppression of Thr696 and Thr853 phosphorylation of the myosin-binding subunit (MBS).^173–175 Several in vitro studies have revealed that ROCK inhibition could prevent neuronal apoptosis,¹⁷⁶ decrease intestinal fibrosis and vascular stiffening,¹⁷⁷ and enhance E-cadherin stability in necrotizing enterocolitis.¹⁷⁷ A previous study also demonstrated that rotavirus disrupted tight junction integrity in polarized MDCK cells via a RhoA/ROCK/MLC-dependent mechanism.¹⁷⁸ Simvastatin, a lipid-lowering drug, abrogated the Rho/ROCK signaling pathway, resulting in the amelioration of sepsis-induced intestinal barrier dysfunction, and increasing the expression of tight junction proteins in rats with cecal ligation and puncture.¹⁷⁹ Interestingly, ROCK inhibitor was also shown to inhibit lipopolysaccharide (LPS)-induced MLC phosphorylation and occludin downregulation. Furthermore, ROCK was also able to reverse LPS-induced TER decrease and attenuate the paracellular permeability of 4-kDa FITC-dextran in Caco-2 cell monolayers.¹⁸⁰ Finally, ROCK inhibitor was reported to suppress the expression of claudin-2 and attenuate disease severity in experimental necrotizing enterocolitis (NEC).¹⁸¹ These findings strongly suggest that the Rho/ROCK pathway is a therapeutic target for restoring disrupted intestinal tight junctions.

5.3 PI3K–PTEN and AKT–mTOR pathway

Phosphoinositide 3-kinase (PI3K) is a signaling molecule involved in cell growth, proliferation, and subcellular trafficking of vesicles. The phosphatase and tensin homolog (PTEN) is known as a tumor suppressor that functionally antagonizes the oncogenic effect of PI3K and vice versa.¹⁸² Generally, PI3K-induced generation of phosphatidylinositol (3,4,5)-trisphosphate (PIP₃) results in stimulation of the canonical AKT pathway, which promotes the upregulation of mammalian target of rapamycin complex 1 (mTORC1) signaling.¹⁸² In the intestinal epithelial cell line Caco-2, LPS were reported to induce intestinal barrier dysfunction by decreasing both mRNA and protein expression of ZO-1 and occludin, with concomitant expression of PTEN.¹⁸³ Notably, stimulation of the PI3K/AKT pathway was found to antagonize PTEN function to ameliorate LPS-mediated intestinal epithelial barrier disruption.¹⁸³ Despite the mechanism behind this not being definitively clarified, it is possible that activation of the PI3K pathway indirectly restores tight junctions by promoting proliferative phenotypes and PTEN inhibition may be a plausible target for the treatment of colitis. However, conflicting evidence on this possibility has been reported. For example, the inhibition of PTEN was shown to exacerbate colitis in IL10^{–/ –} mice. Notably, in vivo model is complex, in which increases in the abundance of colitogenic bacteria Akkermansia and Bacteroides in the gut of IL10^{–/ –} mice have been reported in parallel with the administration of PTEN inhibitors.¹⁸⁴ These findings revealed the importance of communication between gut microbiota and mucosal intracellular signaling in determining tight junction integrity, although this still requires further investigation.

5.4 ERK1/2

The Raf–MEK–ERK1/2 pathway integrates external stimuli to influence a broad range of cellular processes, including cell survival, proliferation, transcription, and metabolism.¹⁸⁵ The association of ERK1/2 on the epithelial barrier and the homeostasis of TJ has been a subject of ongoing investigation. Recent models of inflammation-induced barrier breach demonstrated that ERK1/2 activation underlay the reduction of monolayer integrity, increase in epithelial permeability, and TJ disassembly in several epithelial tissues.^186–189 In intestinal epithelial monolayers (Table 1), IL-6 or IL-17 exhibited ERK1/2 stimulation, resulting in decreased monolayer integrity as well as increased pore-forming claudin-2 expression.¹⁹⁷ The upregulation of claudin-2 upon ERK1/2 activity was also reported in mouse rectal CMT93-II cells.¹⁹⁸ Moreover, ERK1/2 enhancement of claudin-2 expression is likely mediated by its downstream transcription factor AP-1, which could act as a transcriptional enhancer of claudin-2.¹⁹⁹

Table 1. Roles of ERK pathway in the regulation of tight junction-dependent permeability.

Models	ERK1/2 stimulator	Effects on barrier and tight junction	ERK1/2 inhibition	Mechanism	Ref
Caco-2 monolayers and mouse intestine	TNF-alpha and i.p. TNF-alpha	Reduced TER, Increased inulin flux, Increased mucosal-to-serosal dextran flux	PD-98059, siERK1/2	Phosphorylated transcription factor, ElK-1, Increased MLCK promoter activity	^Citation190
Caco-2 monolayers	TNF-alpha	Reduced TER, Reduced expression of occludin and ZO-1	Polyphenol, Ellagic acid	Induced MLCK gene expression and MLC phosphorylation	^Citation191
Caco-2 monolayers	IL-1beta	Reduced TER, increase inulin flux	PD-98059	Phosphorylated transcription factor, ElK-1, Increased MLCK promoter activity	^Citation192
Caco-2 BBe1	Leptin	Increased FITC-dextran flux	U0126, LY	-	^Citation193
Caco-2 monolayers	Secondary bile acid-deoxycholic acid	Increased FITC-dextran flux, Reduce expression of occludin and ZO-1	Flavonoid derivative -(-) epicatechin, U0126	Phosphorylation of MLC	^Citation194
Mice and Caco-2 cell	Intragastric-LPS, LPS and membrane ephrinA1/ephA2 pathway stimulation	Reduced TER, Increased FITC-dextran flux, Reduced expression of occludin and claudin-1	Attenuation of ephrinA1/ephA2 pathway and PD98059	-	^Citation195
Caco-2 monolayers	EGF	Prevented acetaldehyde-induced tight junction disruption	-	-	^Citation196

In addition to the elevation of claudin-2, studies have revealed that ERK1/2 induced barrier disruption and increased TJ permeability through the ETS-like-1 protein (Elk)-1–MLCK pathway (Table 1). Indeed, activation of ERK1/2 by the hallmark cytokines of IBD, TNFα, or IL-1β, stimulated phosphorylation of the transcription factor Elk-1.^190,192 As an MLCK gene enhancer, Elk-1 ultimately augmented the paracellular permeability of Caco-2 monolayers.^190,192 In contrast, some findings paradoxically showed that ERK1/2 activation led to protection of the epithelial barrier in distinct tissues and experimental models.^196,200 However, the effects of ERK1/2 stimulation on the intestinal epithelial barrier integrity still require further clarification, particularly to distinguish the roles of ERK1/2 in gut epithelial tight junctions under physiological and pathological conditions. ERK1/2 as revealed in intestinal epithelial cells is now known to serve as a mediator of inflammation-induced tight junction disassembly.

5.5 AMP-activated protein kinase

Accumulating evidence of AMP-activated protein kinase (AMPK), the master regulator of cellular energy, has been extensively reported in modulating TJs. Various models of AMPK activation have been established and widely used, including energy deprivation, induction of intracellular calcium, and use of a pharmacological activator to induce phosphorylation of the AMPKα subunit at the Thr-172 residue, leading to enhanced epithelial barrier functions and TJ assembly.^91,201–204 Since then, AMPK stimulation has been widely suggested as a therapeutic target for TJ-associated GI diseases, for example, IBD²⁰⁵ and colorectal cancer.^206,207 Although definitive findings on the mechanisms by which AMPK promotes TJ assembly and controls permeability have not been obtained, this review classifies the notable modalities by which AMPK regulates TJs into two different features.

First, AMPK activation was shown to increase phosphorylation of claudin, for example, claudin-1 and claudin-4, in mammary gland epithelial cells²⁰⁸ and saliva gland cells,²⁰⁹ respectively. Indeed, the phosphorylation of claudin-1 protected itself against ubiquitin-mediated destruction, preserving the TJ complex under osmotic stress,²⁰⁸ while claudin-4 phosphorylation enhanced its interaction with occludin, augmenting TJ assembly.²⁰⁹

Interaction with TJ scaffolding proteins appeared to be another mechanism by which AMPK regulates TJ assembly. For instance, Wu et al. found that the phosphorylation of Ser1102 on AF-6 occurs by AMPK’s promotion of afadin-mediated ZO-1 assembly.⁵¹ Moreover, AMPK was shown to be required for the homeostasis of TJ–actin–microtubule arrangement and led to cingulin phosphorylation, which improved cingulin–microtubule binding but reduced actin affinity to mitigate TJ permeability.^62,63 Furthermore, upon energy deprivation which the Par3–Par6–aPKC polarity complex is inactive,²¹⁰ activated AMPK tried to maintains the apical polarity complex via the direct phosphorylation of actin-binding protein girdin, maintaining apical junction polarity and integrity during energy stress.²¹¹ Taking these findings together, AMPK promotes TJ assembly and preserves TJ integrity in response to cellular stress via the phosphorylation of claudin and TJ scaffolds.

5.6 Atypical PKCs

Atypical PKCs, PKCι/λ, and PKCζ, identified as being functionally insensitive to intracellular Ca²⁺ or diacylglycerol, have been found to regulate TJ assembly and apico-basal polarization, as reviewed extensively elsewhere.^53,212 In short, aPKC undergoes tripartite interaction with Par3–Par6 complex at the apical junction area mediated through PKC-interacting domains on Par3 and Par6, while Par3 is anchored to JAM-A and nectin.^41,213,214 Upon embryonic development, Par6 and its colocalized Cdc42 were shown to be initially required for PKCζ activation to induce epithelial polarization.²¹⁵ Active PKCζ-mediatedPar3 phosphorylation promotes the apical membrane localization of Par3–aPKC complex. However, this complex induces the exclusion of basal-specific proteins, Lethal giant larvae, from the apical complex to promote compartmentalization.²¹⁶ In addition, PKCζ also facilitated occludin and ZO-1 assembly via the phosphorylation of Thr residues at the C-terminus of occludin.²¹⁷ The inhibition of aPKCs by either a pseudosubstrate of PKCζ or ATP depletion revealed the dephosphorylation and redistribution of occludin, ZO-1, and Par3 in epithelial cell lines.^210,217 The dephosphorylation of occludin might partly occur as a result of protein phosphatase (PP)2A activity. However, the interaction between aPKC and PP2A requires further investigation.²¹⁸ Taking these findings together, aPKC along with Par3–Par6 has been shown to play a crucial role in promoting epithelial polarization and regulating TJ assembly.

5.7 Noncoding RNAs

Noncoding RNAs (ncRNAs), RNA transcripts without the capacity to produce proteins, have recently emerged as regulators of epigenetics and gene expression.^219,220 MicroRNAs (miRNAs) are small ncRNAs of approximately 22 nucleotides in length. Owing to their versatility, they can repress or activate messenger RNAs (mRNAs) to subsequently inhibit or stimulate translation,^221,222 respectively. In addition to miRNAs, long noncoding RNAs (lncRNAs) are ncRNAs of at least 200 nucleotides in length. They are also involved in mRNA stability and post-transcriptional regulation, similarly to miRNAs.^223,224 Intriguingly, the dysregulation of ncRNAs is associated with gut barrier disruption and inflammation, which are the pathophysiological features of several gastrointestinal diseases. Therefore, the functions and applications of these ncRNAs have received growing attention in research on such diseases.^225–228

miRNAs can modulate TJ expression, which is involved in diseases associated with a defective intestinal epithelial barrier. One of the most interesting miRNAs in this context is the evolutionarily conserved miR-21, which was shown to be overexpressed in colonic biopsies of ulcerative colitis patients.^229,230 In a model of barrier dysfunction induced by intestinal ischemia–reperfusion injury, the expression of microRNA-21 (miR-21-5p and miR-21-3p) was increased, while that of TJ proteins, including occludin and CLDN1, was decreased.²³¹ Similarly, miR-21-5p was suggested to potentially induce the epithelial permeability of Caco-2 cells due to the downregulation of ADP ribosylation factor 4, which modulates occludin and CLDN4 abundance.²³² Profound increases in miR-122a and miR-223 were also identified in IBD models. miR-122a was shown to mediate TNFα-induced barrier leakage, while miR-223 upregulation in enterocytes resulted in the reduction of CLDN8.^233,234 Additional miRNAs have recently been found to be involved in other TJ-associated disorders. For example, miR-29a expression was predominantly elevated in intrauterine growth-restricted (IUGR) porcine neonates.²³⁵ miR-29a was shown to hamper epithelial integrity and diminish the translation of claudin-1 through binding with the 3, untranslated region (UTR) at the CLDN1 transcript in an intestinal porcine epithelial cell line (IPEC-1).²³⁵ In addition, in irritable bowel syndrome (IBS), decreases in miR-16 and miR-125b-5p were identified, leading to increased expression of the CGN and CLDN2 genes, respectively.²³⁶

Moreover, accumulating evidence has shown the relationships of lncRNAs to gastrointestinal diseases. lncRNAs are generally known to play crucial roles in the regulation of gene expression, mRNA stabilization, and translational and post-translational modifications,^223,224,237 while some serve as competing endogenous RNAs by interacting with certain miRNAs to mitigate mRNA suppression.²³⁸ For example, the lncRNA uc.173 was shown to abolish the interaction of miR-29b with CLDN1 mRNA, thus promoting CLDN1 translation and eventually augmenting gut barrier function.²³⁹ In IBD colonic biopsies, the expression of the lncRNA colon cancer-associated transcript-1 (CCAT1) was shown to correlate with IBD-specific cytokines.²⁴⁰ CCAT1 disrupted ZO-1 and occludin architecture by concealing the function of miR-185-3p, which normally represses MLCK translation in Caco-2 cells, so it was suggested as a novel MLCK regulator.²⁴⁰ Additionally, upregulated lncRNA H19 in IRI and IBD models promoted permeability by encoding miR-675-5p, which abolished ZO-1 expression.^241,242 In contrast, some lncRNAs were found to enhance barrier function. For instance, PlncRNA1 preserved the expression of ZO-1 and occludin by sponging miR-34c in Caco-2 cells.²⁴³ In addition, the lncRNA SPRY4 intronic transcript 1 (SPRY4-IT1) similarly enhanced the intestinal epithelial barrier function through augmenting the expression of mRNAs encoding CLDN1, CLDN3, and JAM-1.²⁴⁴ In summary, these studies have revealed the involvement of ncRNAs in gastrointestinal disorders and TJ biology. However, it remains unclear yet whether alteration of ncRNAs essentially causes barrier-associated diseases or it change is a result of pathogenesis and compensatory mechanism. The conclusive models of tight junction regulatory signaling are shown in Figure 4.

Figure 4. Intracellular regulatory signaling of tight junction-dependent paracellular permeability.

6. The modulators enhancing TJ function in the GI tract

As TJs play key roles in GI barrier disruption and the aforementioned GI diseases, the emerging TJ-modulating agents have attracted substantial attention as enhancers of GI epithelial barrier integrity. Some diet-derived and small-molecule compounds acting on TJ physiology are reviewed here.

6.1 Polyphenols

Polyphenols, the ubiquitous secondary metabolites found in plant-based foods, are primarily classified as flavonoids and non-flavonoids (i.e., phenolic acids, stilbenes, coumarins, and lignans), based on their structures.^245,246 They have beneficial activities, such as antioxidant and anti-inflammatory properties.²⁴⁷ Moreover, increasing evidence has shown the additional effect of polyphenols on GI epithelial mucosal homeostasis. Several flavonoids, such as quercetin, epigallocatechin gallate (EGCG), kaempferol, and rutin, have been reported to enhance GI barrier function, characterized by an increase in TER, in in vitro human intestinal epithelial models, including Caco-2 and T84 cell lines. Distinct expression patterns of TJ proteins (CLDN1/3/4, ZO-1/2, and occludin) and protein assembly were reported to be differentially promoted by divergent types of flavonoid.^248–251 Besides, polyphenol-rich bee pollen extracts exhibited a protective effect against intestinal barrier dysfunction in Caco-2 cells.²⁵² Moreover, propolis (bee glue) promoted a healthy GI mucosal epithelial barrier through AMPK activation and the ERK 1/2 signaling cascade in Caco-2 cells and rats.²⁵³ Grape polyphenols were also revealed to promote the expression of TJ proteins, as well as counteract the production of reactive oxygen species (ROS), mediated by LPS, and exert anti-inflammatory effects in Caco-2 cells.²⁵⁴ Furthermore, catechin and procyanidin B2, the dietary compounds rich in polyphenols, were found to augment the GI epithelial barrier function in co-cultured Caco-2 and HT29-MTX cells.²⁵⁵ Finally, resveratrol, a stilbene compound, also improved gut barrier impairment in mice.²⁵⁶

6.2 Oligosaccharides

Many well-known types of dietary oligosaccharide, including chitosan oligosaccharide, fructooligosaccharide, and galactomannan pentasaccharide, have attracted attention in the nutrition therapy industry. Chitosan, a positively charged linear heteropolysaccharide derived from chitin, which is the second most common natural polysaccharide in the world, is extracted from the shells of shrimp, lobster, and crab.^257,258 Chitosan is utilized as an enhancer of oral-based drug delivery.²⁵⁹ However, chitosan was also revealed to upregulate the expression of TJ proteins (e.g., CLDN1, occludin, ZO-1) and strengthen GI barrier function in a mouse model of ulcerative colitis.²⁶⁰ Indeed, chitooligosaccharide (COS), a degraded product of chitosan, also augmented GI barrier integrity in a mouse closed-loop model and mucus-secreting human colonic HT-29 cells.^206,261 In addition, fructooligosaccharide (FOS) relieved 5-fluorouracil (5-FU)-induced GI mucosal damage and gut inflammation in BALB/c mice.²⁶² FOS also stimulated TJ assembly in an AMPK-dependent manner in T84 and intestinal epithelial cells.²⁶³ Furthermore, pentasaccharide of mannan (MOS5) was recently reported to possess the capacity to enhance tight junction assembly in intestinal epithelial cell monolayers, via a mechanism involving AMPK activation.²⁶⁴

6.3 Short-chain fatty acids

Many short-chain fatty acids (e.g., butyrate, acetate, and propionate) have been revealed to promote the alteration of TJ permeability by activating lipoxygenase (LOX) in the small intestinal epithelium.^265,266 Butyrate is commonly found in bacterial fermentation, which is a by-product of dietary fiber in the large intestine.²⁶⁷ Several benefits of butyrate have been reported, including enhancing the permeability of the intestinal barrier through upregulating claudin-1 transcription, which is related to the association between SP1 and claudin-1 promoter.²⁶⁸ In addition, butyrate was shown to protect against Campylobacter jejuni invasion and translocation across Caco-2 cell monolayers.²⁶⁹ Indeed, the mechanism of action of butyrate in facilitating TJ assembly and maintaining the intestinal barrier in Caco-2 cell monolayers was shown to be dependent on AMPK.²⁷⁰ In addition, a recent study showed that short-chain fatty acids can act as ligands of G-protein-coupled receptors (GPCRs), such as GPR41, GPR43, and GPR109A, to activate PKCβ that can in turn phosphorylate ZO-1 in order to increase intestinal barrier function.²⁷¹

6.4 Small-molecule compounds

As mentioned above, MLCK is widely recognized as a therapeutic target for tight junction disruption-related GI diseases including IBD.⁷⁰ In addition, MLCK-mediated TJ dysfunction was reported to be associated with IBS.^272,273 Consistent with this, a membrane-permeating peptide that inhibits MLC kinase (PIK) was reported to prevent enteropathogenic Escherichia coli (EPEC)-, TNF-, and IFN-γ-induced TER defects and suppress MLC phosphorylation in Caco-2 cell monolayers.²⁷⁴ PIK was also found to be effective for reducing fluid loss in a mouse model of diarrhea induced by T-cell activation.⁸⁵

However, the inhibition of MLCK is not clinically feasible due to the associated severe toxicity. No approach is currently available that directly targets tight junction integrity without modulating immune function, which always compromises immunity.⁷⁰ To safely restore the functions of the GI epithelial barrier and protect against cytokine-dependent, MLCK-mediated TJ disruption, a recent study has disclosed the non-enzymatic interruption of MLCK recruitment to perijunctional actomyosin rings as a novel therapeutic.¹⁶¹ A small molecule called divertin, which can bind specifically to the immunoglobulin-like occludin cell adhesion molecule (IgCAM) of epithelial long MLCK1, was shown to prevent cytokine-mediated MLCK1 recruitment to PAMR without interfering with other facilitative mechanisms (e.g., epithelial migration and wound repair).¹⁶¹

6.5 Amino acids and peptides

Some amino acids and peptides also exert protective effects on the GI mucosal barrier. Glutamine (Gln), the most common essential amino acid in the human body, was found to improve GI barrier function and attenuate intestinal permeability, as detected by an increase in TER value in Caco-2 cell monolayers. The depletion of Gln led to the reduction of protein expression levels of CLDN1, occludin, and ZO-1.^275–277 Tryptophan (Trp) was also demonstrated to cause a dose-dependent increase in TER in Caco-2 cells.²⁷⁸ Additionally, a casein-derived peptide enhanced the TER value, indicating an increase in GI barrier function, and increased the levels of occludin gene and protein expression in in vitro human intestinal epithelial Caco-2 cells, as well as in ex vivo mouse jejunal and ileal loops.²⁷⁹

6.6 Vitamins and minerals

Other studies of TJ-modulating agents including vitamins and minerals used as food supplements have been performed. Vitamin A is a fat-soluble vitamin applied for maintaining good visual health, antioxidant capacity, and mucosal immune response.^280,281 Vitamin A deficiency was reported to worsen diarrhea and GI mucosal damage, which was reversed by vitamin A supplementation.²⁸² Mechanistically, this might occur via the downregulation of TJ-related mRNA expression (e.g., occludin, CLDN1/3/4, and ZO-1).^280,283,284 In addition, vitamin A exerted a protective effect on intestinal barrier integrity, as it ameliorated the reduction of TER and restored Clostridium difficile toxin A and LPS-mediated GI mucosal damage in Caco-2 cells and IPEC-J2 cells,^285,286 respectively. Vitamin D is essential for bone health and GI barrier integrity. Its deficiency is associated with the risk of GI epithelial defect and IBD.²⁸⁷ In a model of colitis induced by dextran sulfate sodium (DSS), it was demonstrated that vitamin D receptor (VDR)-knockout mice showed more severe diarrhea and ulceration than a control group.²⁸⁷ Consistent with this, upon VDR siRNA knockdown, the downregulation of TJ proteins, including CLDN1, ZO-1/2, and E-cadherin, and the reduction of TER were revealed, which were relieved by 1,25-dihydroxyvitamin D3 [1,25(OH)₂D₃], suggesting the role of VDR-mediated vitamin D signaling in the regulation of TJs.²⁸⁷ Zinc is an essential mineral found in some foods using as dietary supplements. Zinc also plays a pivotal role in regulating TJ proteins, as zinc supplementation can relieve the GI mucosal damage induced by malnutrition, inflammation, and infection.^288,289 GPR39, a zinc-sensing receptor, serves as a sensor for extracellular zinc, and Gpr39 knockout in mice was shown to lead to decreases in ZO-1 and occludin expression and severe symptoms of colitis.^290,291 Moreover, in Caco-2 cells, zinc depletion was found to cause a decline in TER and an increase in intestinal epithelial permeability.²⁹² Fortunately, zinc supplementation and GPR39-specific agonist helped reverse the decline in ZO-1 and occludin expression and enabled the reassembly of TJs in Caco-2 cells and T84 cells, at least in part via an AMPK-dependent mechanism.^91,293

6.7 Other known drugs

Metformin is known as a drug used in the treatment of type 2 diabetes and cardiovascular complications by attenuating inflammatory responses via the suppression of nuclear factor κB (NF-κB) and the activation of AMPK.^294–296 In mice orally administered metformin, the metformin delayed the onset of fructose-induced liver stenosis through reduced fructose function, in which alterations of MMPs and TIMP-1 concentrations in the intestine occurred.²⁹⁷ Apart from being an antidiabetic drug, metformin was found to effectively benefit gut epithelium by suppressing inflammation and improving ileal epithelial barrier function in interleukin-10-deficient mice, probably through AMPK activation.²⁹⁷ AMPK is clearly a target of metformin, which can improve mucosal integrity as a treatment of colitis.²⁰⁵ A recent study showed that metformin can also regulate TJ-related protein expression by reducing the effects of TNF-α on Caco-2 cell permeability, which results in inhibiting the MLCK–MLC signaling pathway.²⁹⁸

7. Tight junction and GI diseases

7.1 Role of the tight junction in the esophagus and related diseases

Food travels through the esophagus from the pharynx to the stomach. The barrier functions of esophageal squamous stratified epithelium involve providing mucosal and submucosal defense against noxious intraluminal substances and regulating paracellular ion transport, which is mediated by claudins, the tetra-spanning transmembrane proteins of tight junctions, located at the apical and lateral membranes of esophageal epithelial cells.^299–301 Claudins (CLDN)1, 4, and 7 are prominently expressed in the esophagus, which has basal pH of ~5–6.^300,302 The esophagus is not normally exposed to a strongly acidic environment. Therefore, a decrease in esophageal pH leads to impaired esophageal barrier functions.³⁰³ The changes of claudin expression patterns and the acidic disruption of the esophagus are associated with several esophageal diseases.³⁰⁴

An intra-esophageal pH of <4 caused by acid reflux from the stomach can disrupt the esophageal barrier integrity and can be detected by 24-h ambulatory esophageal impedance pH monitoring.^302,305,306 The early morphological lesions caused by unanticipated acidic changes are dilated intercellular spaces related to the expression of tight junctions, which are found in the non-erosive form of gastroesophageal reflux disease (GERD).³⁰⁷ CLDN 1 and 4 were found to be decreased in patients with GERD.³⁰⁸ Consistent with this, the expression of CLDN 4 was downregulated and the esophageal barrier function was attenuated in acid-disrupted esophageal epithelial models.^309,310 The damaged esophageal squamous mucosal layers can be replaced with metaplastic columnar epithelium, which is more acid-resistant and contributed by CLDN 18, as found in patients with Barrett’s esophagus.^307,311 Moreover, the overexpression of CLDN 2 was found to be related to the reflux of bile acid and the development of BE.³¹² In eosinophilic esophagitis, CLDN 7 expression was suppressed by the presence of transforming growth factor-β1 (TGF-β1), whereas the upregulation of CLDN 7 relieved TGF-β1-induced esophageal barrier dysfunction.³¹³ Moreover, in esophageal squamous cell carcinoma, reduced expression of CLDN 1, 4, and 7 was identified, which might be candidate prognostic biomarkers for esophageal cancer.^314–316 CLDN 18, meanwhile, was found to be expressed in esophageal adenocarcinoma.³¹⁷

7.2 Role of the tight junction in the stomach and related diseases

In the stomach, the tight junctions formed by adjacent gastric epithelial cells play a pivotal role in regulating the paracellular permeability of ions, nutrients, and other solutes and preventing digestive enzymes and toxins for entering the systemic circulation.³¹⁸ CLDN 3, 4, 5, 12, 18, and 23 are widely expressed in the stomach.²² The tight gastric epithelium can tolerate local acidity, in contrast to esophageal epithelium. However, the elevation of pH by the urease activity of Helicobacter pylori was found to attenuate TER and interfere with gastric epithelial integrity, relating to occludin internalization and MLC phosphorylation.^318–320 H. pylori infection can markedly disrupt both pore and leak pathways of tight junctions. Indeed, the alterations of claudin protein expression can be affected in gastric cancer.³¹⁸ Notably, the dysregulation of claudins can result in a number of gastric diseases.

The gastric epithelium is considered to have greater tightness, depending on the number and depth of tight junction strands, than the esophageal epithelium.³¹⁸ Therefore, the gastric mucosal layers have low water and solute permeability and are resistant to stomach-secreted acidic juice.³¹⁸ The altered expression patterns of CLDNs can give rise to the development of gastric cancers. CLDN 3, 4, and 7 were revealed to be upregulated in gastric cancers, while other CLDNs showed decreased expression.³²¹ In particular, the reduction of stomach-specific CLDN 18 expression was reported in atrophic gastritis and CLDN 18 deficiency was found to be related to the poor prognosis of patients with gastric cancer.^322,323 CLDN 18 was also found to help protect gastric epithelium from gastric acid and pepsin and to modulate the paracellular pathway.³²⁴ Moreover, CLDN 18 plays a crucial role in regulating cell lineage differentiation in the stomach.³²² H. pylori infection is a causal risk factor of chronic gastritis and gastric tumor. In H. pylori-infected mice, the loss of CLDN 18 expression was found to occur first in gastric glands, while subsequently being found in the gastric neck and in the surface epithelial cells. The development of gastric neoplasia was also revealed in gastric tissues from CLDN 18-knockout mice, as the intestinalization of gastric mucosa was addressed.³²² Moreover, H. pylori-induced gastritis and gastric cancer were found to be associated with the change of the expression patterns of CLDNs, from gastric-type to intestinal-type CLDNs.^324–327 In addition to H. pylori-dependent gastric tumor, the long-term absence of CLDN 18 could initiate the formation of gastric tumors in a chronic gastritis mouse model without H. pylori infection through several signaling cascades related to certain chemokines and Wnt signaling pathways.³²⁸

7.3 Role of the tight junction in intestinal epithelium and related diseases

Intestinal epithelium plays important roles in supporting the vectorial transport of nutrients and fluid absorption that requires cell polarity.^70,72,75,77 Indeed, epithelial cell polarity in intestinal epithelial cells depends on the integrity of tight junctions.⁷⁰ Intestinal epithelium also forms physical barriers to protect against noxious substances from the gut lumen, which also relies on the integrity of tight junctions. Therefore, disruption of tight junctions in the intestinal epithelia can disturb homeostasis.

Over the past half-century, several lines of evidence have indicated the association between leaky gut syndrome and IBD.^329–332 Indeed, increased gut permeability was found in patients with Crohn’s disease and their relatives.³³² Of particular importance is that MLCK activation, claudin-2 overexpression, and occludin internalization have been found in intestinal tissues from both forms of IBD, namely, Crohn’s disease and ulcerative colitis.^333–335 In mice with TNFR2 and MLCK knockout specifically in intestinal epithelium, CD4⁺CD45RB^hi adoptive transfer failed to induce gut permeability, MLC phosphorylation, occludin internalization, and claudin-2 expression.⁸⁴ Furthermore, knockout of intestinal TNFR2 and MLCK also delayed the onset of T-cell transfer-induced colitis.^84,116 Consistent with this, the inhibition of MLCK recruitment to PAMR and CK2 inhibitor was found to attenuate immune-mediated leak and pore permeability pathways,^21,161 respectively. These findings suggest that the prevention of intestinal tight junction disruption may be a therapeutic option for IBD. A previous study reported that the upregulation of claudin-2 was associated with increased gut permeability.¹⁸¹ In addition, MLCK activation and tight junction disruption were also reported as key etiological factors of T-cell activation-induced diarrhea.⁸⁵ Furthermore, EPEC infection induced MLCK activation and occludin internalization, resulting in diarrhea.^274,336 Clostridium difficile‑induced colitis was also found to be related to loss of ZO-1 and ZO-2.³³⁷ Recent studies have additionally identified that enteric infection with V. cholerae El Tor variant caused intestinal barrier dysfunction with an unknown molecular defect and contributed to the pathogenesis of secretory diarrhea.^202,338 Notably, enhancement of claudin-2-induced diarrhea in response to Citrobacter rodentium infection-mediated IL-22 upregulation has been considered as a mechanism of innate immunity responsible for bacterial clearance.⁸⁰ Apart from colitis and diarrhea, intestinal barrier dysfunction was also reported to be found in both adult and pediatric IBS.²⁷² In support of this, MLCK expression was shown to be increased in parallel with a reduction of ZO-1 expression in intestinal tissues from an IBS mouse model.²⁷³ Notably, increased intestinal epithelial permeability is IBS could lead to the promotion of low-grade chronic systemic inflammation, resulting in visceral hypersensitivity.²⁷³ Indeed, the relationship between increased tight junction-dependent paracellular permeability and gastrointestinal diseases is shown in Figure 5.

Figure 5. The relationship between increased tight junction-dependent paracellular permeability and the pathogenesis and progression of gastrointestinal diseases.

8. Future perspective

In this review, we provide information regarding the physiological roles of intestinal tight junction integrity. We also give examples with experimental evidence supporting the impact of intestinal tight junction dysfunction on the pathogenesis of gastrointestinal diseases. For instance, intestinal tight junction disruption and gut epithelial barrier leakage have been considered as key pathogenic factors in IBD. It has been reported that the upregulation of TNF and IL-13 in intestinal tissues of IBD can promote both leak and pore permeability pathways, as feedforward and positive feedback controls of cytokine-induced intestinal barrier loss and vice versa as a vicious cycle. Increased intestinal barrier permeability was also identified as a factor determining disease severity and progression of IBD. Apart from IBD, TNF-mediated occludin downregulation was also proposed as a drug target for T-cell activation-induced diarrhea. Moreover, tight junction disruption-mediated intestinal barrier loss has been implicated in infectious diarrhea with unknown contributions to the diarrheal pathogenesis. Nevertheless, no FDA-approved therapeutics that can directly restore tight junction integrity are currently available.

At present, accumulating lines of evidence have indicated the existence of many well-recognized drug targets that have been proposed for the recovery of tight junctions, such as MLCK1. Divertin, a small molecule that can specifically bind to IgCAM3 of epithelial long MLCK1, is the first drug candidate for this purpose. Indeed, divertin non-enzymatically suppressed cytokine-induced MLCK1 recruitment to PAMR, which promotes leak pathway-dependent barrier loss. In addition, many AMPK activators, for example, metformin, polyphenols, short-chain fatty acids, oligosaccharides, and some nutrient-sensing GPCR agonists, were also shown to enhance tight junction integrity. Hence, further investigation should focus on the effectiveness and safety profiles of divertin and AMPK activators in human subjects for potential future use.

Claudin-2-mediated paracellular cation permeability is one of the drug targets for IBD. Based on a previous study, although no claudin-2 inhibitor is currently available, the inhibition of CK2 has recently been proposed as an effective strategy for treating experimental colitis. Therefore, CK2 inhibitor may be clinically useful for patients with diseases related to the overactivation of claudin-2, including IBD. However, there are some concerns since claudin-2-mediated intestinal fluid secretion causing diarrhea in response to infection is a component of the innate immunity to promote bacterial clearance. The use of a claudin-2 inhibitor may attenuate diarrhea in the early phase of enteric infection but may also delay the process of bacterial clearance and prolong disease. Furthermore, claudin-2 suppression may result in luminal fecal dehydration, which can promote intestinal obstruction.

Several lines of evidence have also supported the relationship between tight junction disruption and NEC, as well as IBS. In addition to intestinal diseases, tight junction disruption was also reported to be associated with other diseases that arise along the GI tubes, including GERD, Barrett’s esophagus, esophageal adenocarcinoma, and H. pylori-dependent gastric tumor. Despite the pathogenic impacts of tight junction disassembly in these diseases currently being unclear, phenotypic abnormalities of tight junctions may be of interest as diagnostic and prognostic markers for these conditions.

Concluding remarks

In this review, we discuss about the cellular architecture of epithelial tight junction in the gastrointestinal tract. Indeed, based on localizations tight junction at subcellular level, tight junction can be differentially divided into bi- and tri-cellular tight junctions. According to the physical properties of tight junction, there are, at least, pore-forming and barrier-forming tight junctions. In fact, tight junction proteins can be either transmembrane or intracellular scaffolding proteins. In addition, we also furnish the comprehensive views of the relationship between intestinal tight junction disruption and gastrointestinal diseases including Barrett’s esophagus, GERD, H. pylori-induced gastritis, IBD, IBS, cholera, and various types of cancers along the gastrointestinal tract. Nevertheless, there is currently no FDA-approved therapy for tight junction disruption-related diseases. Of particular importance, activation of AMPK and inhibition of MLCK recruitment to PAMR have been markedly proposed as the therapeutic approaches for recovery of intestinal tight junction disruption and related diseases, especially IBD. Furthermore, we also show evidence regarding other possible therapeutic targets, small molecules, and known drugs that have been of interest for the development of new drug for targeting intestinal tight junction recovery.

Author contributions

AM: Writing - original draft, lead; funding acquisition; figure preparation; writing - review and editing

NP: Writing - original draft; writing - review and editing

PRS: Writing - review and editing

PC: Writing - original draft

WS: Writing - original draft, funding acquisition.

PP: Conceptualization, lead; funding acquisition, supporting; lead; writing - original draft, lead; writing - review and editing, lead

Acknowledgments

This project is funded by National Research Council of Thailand (NRCT) (Grant No. N42A650225) (to PP). This work is supported by Chulabhorn Royal Academy (Project code ISFO2-004/2565) (to PP), the Mahidol Medical Scholars Program (to WS), Mahidol University (Basic Research Fund: fiscal year 2021) (to AM), and Office of the Permanent Secretary, Ministry of Higher Education, Science, Research and Innovation (Grant No. RGNS 64-177) (to AM).

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was supported by the National Research Council of Thailand [N42A650225]; Office of the Permanent Secretary, Ministry of Higher Education, Science, Research and Innovation [RGNS 64-177]; Chulabhorn Royal Academy [ISFO2-004/2565].