Tight Junctions and the Intestinal Barrier


The primary function of the gastrointestinal tract is to digest and absorb nutrients. To accomplish this, it must maintain a barrier between the luminal environment, technically a space outside the body, and the internal environment of the body; and it must selectively absorb and secrete nutrients, solutes, and water across the barrier. Separation of tissue spaces throughout the gastrointestinal tract is accomplished by continuous sheets of polarized columnar epithelial cells. An exception exists in the upper two-thirds of the esophagus which is covered by a nonkeratinizing squamous epithelium. Epithelial barriers are selective and capable of excluding potentially noxious luminal contents, such as gastric acid, colonic bacteria, and bacterial antigens, while at the same time capable of directional absorption and secretion of large volumes of solutes and water. Material can pass from one side of the epithelium to the other side along one of the two routes, either through the cell membranes or the space between them, referred to as the transcellular and paracellular pathways, respectively. The connection between individual epithelial cells is created by a series of intercellular junctions, the tight junction (TJ) being the most important for defining the characteristics of the paracellular barrier and its selectivity.


Intestinal barrier, Tight junction, Occludin, Claudin, Myosin light chain kinase, Pore pathway, Leak pathway, Ion transport, Cytokine, Lipopolysaccharide, Crohn’s disease, Celiac disease, Permeability index




adenomatous polyposis coli protein


blood vessel epicardial substance


coxsackievirus and adenovirus receptor


Crohn’s disease


cystic fibrosis transmembrane conductance regulator


crumbs protein homolog 3


enhanced green fluorescent protein


epithelial sodium channel


extracellular signal-regulated kinase

H + K + ATPase

hydrogen potassium ATPase


inflammatory bowel disease






junctional adhesion molecule


c-Jun N-terminal kinase




membrane-associated guanylate kinases


mitogen-activated protein kinase


MAL and related proteins for vesicle trafficking and membrane link


myosin light chain kinase

Na + K + ATPase

sodium potassium ATPase


nuclear factor kappa B


sodium hydrogen exchanger


nonsteroidal anti-inflammatory drug


domain shared by postsynaptic density protein (PSD95), Drosophila disc large tumor suppressor (Dlg1), and zonula occludens-1 protein


permeability index


sodium-glucose cotransporter


tight junction-associated MARVEL proteins


transepithelial resistance


tight junctions


tumor necrosis factor


tumor necrosis factor receptor


ulcerative colitis


zona occludin


In this chapter, we discuss the role of tight junctions (TJs) as an integral component of the intestinal epithelial barrier. The TJs, located at the boundary between apical and basolateral membrane of enterocytes, act to provide the barrier or “gate” function to paracellular permeation of water-soluble molecules. Herein, we discuss the components of the TJ complex and the major physiological regulators of intestinal TJ barrier. The TJ permeation occurs through at least two distinct permeation pathways; the pore pathway, which allows the flux of small-sized molecules (regulated in part by claudin-2), and the leak pathway, which permits the flux of large macromolecules (regulated in part by occludin and myosin light chain kinase). The mechanisms that mediate physiological (e.g., via Na + -nutrient cotransport) and pathological [e.g., inflammatory mediators such as tumor necrosis factor (TNF)-α or pathogenic bacteria] modulation of intestinal TJ barrier are detailed as it relates to intestinal epithelial homeostasis and alteration in intestinal TJ permeability. Additionally, the role of the defective intestinal TJ barrier in the pathogenesis of clinical diseases such as celiac disease and inflammatory bowel disease (IBD) are discussed. Thus, the basic components of the TJ complex, the specific TJ pathways, the physiological and pathological regulators of TJ permeability, and the clinical importance of defective intestinal TJ barrier are deliberated in this chapter.


The primary function of the gastrointestinal (GI) tract is to digest and absorb nutrients. To accomplish this, it must maintain a barrier between the luminal environment, technically a space outside the body, and the internal environment of the body; and it must selectively absorb and secrete nutrients, solutes, and water across the barrier. Separation of tissue spaces throughout the GI tract is accomplished by continuous sheets of polarized columnar epithelial cells. An exception exists in the upper two-thirds of the esophagus, which is covered by a nonkeratinizing squamous epithelium. Epithelial barriers are selective and capable of excluding potentially noxious luminal contents, such as gastric acid, colonic bacteria, and bacterial antigens, while at the same time capable of directional absorption and secretion of large volumes of solutes and water. Material can pass from one side of the epithelium to the other along one of two routes, either through the cell membranes or the space between them, referred to as the transcellular and paracellular pathways, respectively. The connection between individual epithelial cells is created by a series of intercellular junctions, the tight junction (TJ) being the most important for defining the characteristics of the paracellular barrier and its selectivity.

The specific characteristics of epithelial barriers vary widely throughout the GI tract, matched to each organ’s transport functions. However, in all cases, disruption of the barrier leads to a loss of normal transport and inflammation due to tissue damage or antigen exposure. In this chapter, we focus primarily on the role of the TJ in the intestinal barrier. We begin with the role of the TJ and paracellular pathway in normal transport. In recent years, a large number of proteins have been identified as components of the TJ and the functions of many of these proteins are being unraveled. This allows interpretation of the barrier’s physiologic properties on a stronger cellular and molecular foundation. We review the latest advances in this area. The intestinal TJ barrier is highly regulated and we review mechanisms and physiologic relevance for the GI tract. Finally, we review some of the intestinal disorders that have an associated defect in intestinal TJ barrier and the implications of TJ barrier defect in the disease pathogenesis. Since our previous version of this chapter, there have been number of important advances that better define the molecular and cellular processes that affect the TJ barrier function under both normal conditions and during pathologic states. In this latest version, we have taken a comprehensive approach to cover wide variety of topics but not all in equal depth. Although we tried to cover as much of the relevant historical advancements, core concepts, and the latest advances in the field, due to the overwhelming amount of high quality original publications in this area, it was not possible to cite all major advancements. Where appropriate, the reader is referred elsewhere for a more complete presentation, particularly of current controversies and unresolved issues.

Intrinsic and Extrinsic Elements of the Barrier

The term “epithelial barrier function” is often used to describe all the mechanisms contributing to homeostasis of the epithelial barrier. The single layer of continuous epithelial cells and their intercellular junctions constitute the intrinsic elements of the barrier. The magnitude of this barrier is most often measured as the transepithelial electrical resistance (TER) and the permeability to paracellular markers, such as mannitol and inulin. TER correlates with the ability to separate ionic charge across the epithelia, reflected in either a transepithelial electrical potential difference or the current that creates the potential, measured experimentally as the short circuit current ( I sc ). Extrinsic elements include the innate and acquired mucosal immune system, protective secretion of mucus, bicarbonate, IgA and antimicrobial peptides as well as mechanism of epithelial repair or restitution. The contribution of each element varies along the GI tract, with mucus secretion being the most constant along the entire length from mouth to anus. Our goal in this chapter is to focus predominantly on the intrinsic barrier of epithelial cells and TJs in health and disease.

The Intestinal Epithelial Barrier and Transcellular and Paracellular Transport

Transport Pathways

Throughout the GI tract, transport of electrolytes, solutes, and water across epithelia occurs across both a transcellular and paracellular pathways ( Fig. 25.1 ). The transcellular route for hydrophilic molecules, for example, Na + , Cl , and glucose, is governed by the profile of membrane pumps, carrier, and channels expressed in a particular cell type. The passive movement across the lipid component of the membrane is very limited for charged and hydrophilic molecules. For example, the electrical resistance across model lipid membrane bilayers is in the range of 10 6 –10 9 ohm × cm 2 whereas the resistance across real membranes in the GI tract is 3–4 orders of magnitude less, reflecting facilitated conductance through protein-based channels ( Table 25.1 ). The profile of conductance proteins differs among epithelia explaining their unique functions. Individual transporters also show a polarized distribution to either the apical or basolateral membrane surface as the basis for directional transport. For example, the apical H + -K + ATPase of gastric parietal cells is responsible for secreting hydrochloric acid within the stomach. Na + -dependent bile acid transporters are positioned on the hepatocyte’s sinusoidal surface and the apical-luminal surface in the ileum to produce the enterohepatic circulation of bile salts. The cystic fibrosis transmembrane regulator (CFTR), a chloride channel, is positioned on the apical surface of biliary, pancreatic, and intestinal surfaces to bring about luminal Cl secretion which is followed by Na + and water secretion.

Fig. 25.1

Equivalent electrical circuit model of the intestinal epithelial cell layer. Only resistive elements are shown. Series resistance across the transcellular pathway is the sum of individual resistance across the apical ( R ap ) and basolateral membranes ( R bl ). These are in parallel with resistances of the TJ ( R TJ ) plus the lateral intercellular space ( R LIS ). The R LIS is small, the membrane resistances are usually high and the epithelial resistance is governed by resistance of the TJ.

Table 25.1

Electrical Characteristics of Some Epithelia

Epithelium a Species R cell b R paracellular P Na /P Cl c
Proximal tubule Dog 6–7 1.4
Gallbladder Rabbit 229 21 3.3
Duodenum Rat 98
Jejunum Rat 67 51 10.0
Ileum Rabbit 115 100 2.5
Distal colon Rabbit 730 385 0.6
Mouse surface 132 3,200
Crypt 429
Gastric fundus Necturus 2,826 10,573
Urinary bladder Rabbit 160,000 300,000
Cell lines d
Caco-2 Human colon 125–250 3.0
LLC-PK 1 Pig prox. tubule 100 0.6
MDCK Dog 60–4000 10.0

a All values can be found in Powell.

b Electrical resistance values in ohms × cm 2 .

c Permeability ratio of Na + versus Cl . P Na /P Cl in free solute is 0.66. Paracellular pathways with ratios above this value are more permeable for Na + than Cl , i.e., cation-selective.

d Values for cell lines are the personal observations of Dr. C. Van Itallie.

Primary transcellular transport is “active,” powered by ATP hydrolysis to move ions against an electrical or concentration gradient. The prime example is the ubiquitously expressed Na + -K + -ATPase, which moves three Na + ions out the basolateral surface in exchange for two K + ions, with the net effect being to generate an inwardly directed Na + and outwardly directed K + gradient and negative intracellular electrical potential. The high membrane conductance for K + and its exit from the cell further enhances the intracellular negative electrical potential. These electrical and chemical gradients are then used in “secondary” active transport to couple energetically unfavorable uphill movement of nutrients, such as glucose or amino acids, to the downhill movement of Na + through, for example, the Na + -coupled glucose cotransporter (SGLT-1) of the jejunum. As a final generalization, the characteristics of transcellular transport are highly regulated by short-term signals, for example, hormone-stimulated bicarbonate secretion from pancreatic ducts, and long-term transcriptional control, for example, aldosterone-stimulated expression of the Na + -K + -ATPase or electrogenic sodium channel ENAC. We outline these features of transcellular transport before proceeding with a detailed discussion of the paracellular pathway, to highlight the sharp distinction to paracellular transport, which is passive, nonrectifying and does not appear to be as highly regulated at least by physiologic stimuli. The transcellular and paracellular transport can be coordinated via intracellular signaling and actin cytoskeleton to enhance the total transport.

The Apical Junction Complex

The paracellular barrier to material movement coincides with continuous cell-cell contacts located at the apical end of their lateral surfaces ( Figs. 25.2 and 25.3 ). The earliest histological description of what we now refer to as the “apical junction complex” comes from the late 19th century. When sections of small intestine were stained with vital dyes a distinct intercellular density was observed between cells at the apical end of the lateral space. The English literature referred to this as the terminal bar ; other names reveal an assumed role in intercellular adhesion, e.g., “Schlussleiten,” and “bandelettes de fermeture.” The first speculation about a barrier function is attributed to Bonnet. After examining several different GI tissues obtained from an executed man, he concluded that the terminal bar was a general feature of all epithelia and might play a role in segregating the distinct fluid compositions found in different regions of the GI tract.

Fig. 25.2

Junction types within the apical junction complex between intestinal epithelial cells. Left: Two columnar epithelial cells with apical brush border typical of the small intestine. A thick band of peri-junctional actin and myosin filaments connected to the tight and adherens junctions are typical of intestinal epithelial cells. Center: The “terminal bar” or apical junction complex is amplified revealing a series of intercellular contacts including the tight and adherens junctions, desmosomes, and gap junctions. Tight and adherens junctions are linked to the actin cytoskeleton and desmosomes to intermediate filaments. Right: TJ contacts are further magnified revealing rows of claudin strands adhering between adjacent cells to seal the paracellular space.

Fig. 25.3

(A) Freeze-fracture electron microscopic replica of the TJ region of mouse jejunum, showing the interconnected network on claudin-based strands crossing the membrane. Continuous rows of claudins from adjacent cells adhere and seal the paracellular space. Above the barrier contact zone a few apical microvilli are visible. (B) Transmission electron micrograph of the apical junction complex region of two adjacent mouse mammary epithelial cells, rotated at 90° around a vertical axis to the image in (A). Lanthanum hydroxide (black) was added to the basolateral side; it freely diffuses through the intercellular space until it is partially blocked at the TJ from reaching the apical side. The TJ is recognized as a region of very close cell-cell apposition. Microvilli are seen on the apical surfaces. (C) Immunofluorescent microscopic localization of the TJ protein ZO-1 in mouse distal colon. The epithelial surface is at the top and two crypts are visible descending toward the bottom of the image. ZO-1 at the TJ is visible at the apical end of the lateral cell contacts, from crypt to surface. Tangential sectioning in the crypt reveals the continuous circumferential location of TJs around each cell. Bar, 10 μm.

(Image C is courtesy of J. Holmes.)

With the first ultrastructural images of intestinal epithelia in 1963, the apical junction complex was revealed as a set of morphologically distinct junction types ( Fig. 25.2 ). Each of these functions in cell-cell adhesion and signal transduction and provide links to the cytoskeleton. The TJ is invariably the most apical. It appears along the apical to basal axis, in transverse sections, as a series of close cell-to-cell contacts, or “kisses.” In freeze fracture images ( Fig. 25.3 A), the contacts are revealed as continuous rows of transmembrane protein particles. Actin filaments terminate on the plasma membrane directly at the contacts and participate in the regulation of the TJ barrier, , and are known to bind the peripheral membrane scaffolding proteins ZO-1 and cingulin ( Table 25.2 ). Below this is the adherens junction, location of the intercellular adhesion molecule cadherin and its cytoplasmic binding partner beta-catenin and extensive attachments to a ring of peri-junctional actin filaments. The importance of cadherin in adhesion and maintaining the differentiated cell phenotype are underscored by its frequent mutation as a final step facilitating metastasis of colon cancer. Beta-catenin has a second role in the nucleus where it signals cell growth and adenoma formation unless degraded by interacting with the adenomatous polyposis coli protein (APC). Human mutations in APC leave beta-catenin free to signal cell growth and transformation into adenomatous polyps. Below the adherens junctions are desmosomes, whose transmembrane proteins, while homologous to cadherin, are linked to intermediate filaments not actin. Desmosomes serve to protect the alimentary epithelia from shear-induced damage. Gap junctions that are made up of intercellular channels formed by six subunits of connexins that are bundled together in homomeric or heteromeric fashion and allow transfer of small metabolites and second messengers like Ca ++ and IP3, between adjacent cells for coordination of epithelial functions like secretion and exocytosis.

Table 25.2

Proteins Located at the Tight Junction

Category Protein Function
Transmembrane Claudin(s) Barrier and pore selectivity
TAMPs Signaling scaffold, adhesion, barrier regulation
JAM(s) Many
CAR Coxsackie virus receptor
PDZ-Scaffolding ZO-1 MAGUK, binds occludin, claudin, ZAK, JAM, ZAK, actin, ASIP, ZONAB
ZO-2 Binds ZO-1, actin, claudins, fos, jun, CEBP
ZO-3 Binds ZO-1, actin, claudins
MUPP-1 13 PDZs and binds claudins
Polarity ASIP/PAR-3 Atypical PKC binding protein
PAR-6 Cdc42–Par6–Par3–aPKC interaction required for polarity and junction formation
PAT-J AKA Discs-lost
aka Crumbs polarity protein
Cell polarity complex component
Regulates assembly of tight junctions
Kinases ZAK Binds and phosphorylates zo-1
aPKC Binds polarity proteins par-3, par-6, interaction required for junction assembly
src Occludin phosphorylation blocks zo-1 binding
yes Binds occludin
Phosphatases PTEN Tumor suppressor binds MAGI-2 and 3
PP2A binds aPKC, disassembles junction
Transcription factors ZONAB ErbB-2 activator
HuASH1 Drosophila ash1 homolog
Fos, Jun
GTP-binding proteins Rab 3B Mutants inhibit ldlr delivery
Rab 13 Mutants inhibit claudin-1 delivery
Gαi2 Binds sh3 of zo-1
AF6 Binds to Ras, ZO-1 and actin
GEF-H1 Guanine nucleotide exchange factor influences permeability
Vesicle targeting Sec6/8 Exocyst complex
VAP33 Binds occludin and v-snares
Other Cingulin
Blood vessel Epicardial Substance (Bves)
Binds ZO-MAGUKs, JAM-1, actin
Regulates cell adhesion

Viewed by freeze-fracture election microscopy ( Fig. 25.3 A), the TJ barrier coincides with a network of transmembrane strands. In unfixed tissue, the strands often appear as rows of individual particles, now known to be a family of transmembrane adhesion molecules claudins. The Latin root, claudere , means to close. Rows of claudins from each cell meet in the intercellular space forming adhesive contacts and a semipermeable seal. The complexity (number and cross-linking) of strands differs among various tissues. It was long thought that the number of strands correlated with the resistance of the barrier. Consistent with this, in the small intestine, the complexity of strands increases at the crypt-to-villus transition. Since discovery of the barrier forming proteins, this structure-function correlation has been called into question; the molecular species of claudin in a particular junction appear be an important determinant. However, composition of other transmembrane TJ proteins is also likely to be important in the regulation and development of TJ barrier function. The specific proteins that affect paracellular flux of solutes of varying size need continuing clarification.

Barrier Properties: Resistance, Flux, and Permselectivity

Early electron microscopic studies ( Fig. 25.3 B) interpreted the close membrane apposition of adjacent cells at TJ contact points as membrane fusion, and even suggested convergence of the outer leaflets of the lipid bilayer. Supported by studies showing the inability of electron-dense proteins, such as hemoglobin and colloidal lanthanum, to pass through the TJs, these analyses led to the popular view of the TJ as an absolute barrier to paracellular flux. Although commonly thought of as an impermeant seal, it might be more appropriate to compare TJs to sieves. However, the characteristics of the selectively permeable TJ barrier vary widely among different tissues, within different cell types of a single tissue, and in response to physiological and pathophysiological stimuli ( Table 25.1 ). Thus, while the paracellular barrier is most often assessed by electrical conductance, TER, or transepithelial flux of small fluid phase markers, such as mannitol or polyethylene glycol, measurement of only one or two of these parameters provides an incomplete picture of overall barrier function.

Electrical Resistance

Epithelia are classified as “tight” or “leaky” based on their overall electrical resistance. The small intestine, colon, and renal proximal tubule are typical examples of leaky epithelia, while gastric fundus, renal collecting duct, and urinary bladder are “tight” epithelia. In either case, paracellular ionic conductance can be measured by mounting tissue with the mucosal and serosal surfaces facing electrically isolated fluid-filled chambers with current and voltage electrodes on both sides. This is the standard Ussing chamber configuration. An equivalent circuit diagram of the epithelium can be developed as one that includes the transcellular pathway, represented by apical and basolateral membrane resistances series, in parallel with resistance of the paracellular pathway ( Fig. 25.1 ). Because membrane resistances are generally very high, conductance by the transcellular pathway can generally be discounted and the overall resistance interpreted as representing the paracellular pathway. However, this is not the case when transcellular and paracellular resistances are similar. This may occur when plasma membrane conductance is enhanced, as can occur when large numbers of ion channels, such as CFTR, are opened, or, alternatively, when paracellular resistance is very high, as in urinary bladder. Nevertheless, in most cases, the effects on overall transepithelial resistance reflect changes in paracellular resistance. Although the TJ and lateral intercellular space are arranged in series such that both contribute to paracellular resistance, the contribution of the lateral intercellular space is small and usually ignored. Since membrane resistances are generally high, it is the variation in TJ resistance that determined whether an epithelium is leaky or tight.

Electrophysiologic studies prior to about 1960 primarily used tight epithelia, such as frog skin and urinary bladder, where the paracellular resistance is very high (more than 1500 Ω cm 2 ). This reinforced the misconception that the TJ is an impermeant barrier. Such studies also provided evidence that barrier function requires viable tissue, as the progressive tissue injury following devascularization correlated with decreases in overall electrical resistance. Because leaky epithelia, such as gallbladder and small intestine, always demonstrated low resistance, it was assumed that this was due to tissue fragility and damage during mounting. Despite this interpretation, the high-conductance pathway across leaky epithelia displayed charge and size selectivity, which would not be expected if the “shunt” pathway was simply a result of tissue damage. The controversy was settled when the shunt pathway was localized to the paracellular space using conductance-scanning methods. By passing a microelectrode over the gallbladder epithelial surface a high conductance shunt could be demonstrated at the intercellular junctions. Use of smaller electron microscopic tracers, such as ionic lanthanum, also allowed ultrastructural visualization of permeation through the TJ. These findings led to a paradigm shift in which the idea of a selectively permeable TJ was accepted. Work over the past 40 years has characterized these permeability properties in detail and resulted in our current understanding of TJs as barriers that contain several classes of transport channels that can be distinguished in both functional and molecular terms.

Charge Selectivity

Experimentally, ion selectivity is often measured as a ratio of cation to anion permeability. As sodium and chloride are the most common cation and anion, respectively, in physiological solutions, ion selectivity is most commonly reported as (PNa + /PCl ) ( Table 25.1 ). However, care must be taken in interpreting PNa + /PCl values, as the ratio in solution is 0.66 due to the smaller size of hydrated sodium, relative to chloride, ions. Thus, PNa + /PCl values above 0.66 should be considered cation selective. This is true of most epithelia, although the actual ratio can vary 30-fold between anatomical sites. While this is of great physiological significance, it should be recalled that, relative to transmembrane ion channels, this level of discrimination between anions and cations is very low. Further, such transmembrane ion channels easily distinguish ions with similar size and charge densities, such as Na + and K + , while the ability of TJs to separate these is very poor. However, monovalent and divalent cations are treated differently by some TJs, and, as discussed below, we now know that this and overall ion selectivity reflect the composition of TJ in terms of claudin family members.

Charge selectivity is a characteristic feature of paracellular barriers and is essential for creating transepithelial gradients that direct passive paracellular transport. For example, many active transcellular absorptive processes within the intestine take advantage of the electrochemical gradient created by the Na + /K + ATPase. However, while the Na + /K + ATPase pumps Na + to the basolateral space, Na + -nutrient cotransport occurs at the luminal, or apical, plasma membrane. This requires that free luminal Na + be sufficient to drive these processes. Thus, if there was no effective means of serosal to mucosal Na + flux, Na + poor diets would result in malabsorption and osmotic diarrhea. Given that there are no defined mechanisms for transcellular Na + secretion (into the lumen), it stands to reason that this transport must occur by the cation-selective paracellular route that favors Na + over Cl , which is the other most abundant ion in physiological settings. Defects in paracellular Na + transport may, in part, explain the abnormalities that occur when claudin-15, which enhances paracellular Na + flux, is knocked out in mice. Although inherited claudin mutations have not been associated with GI disease in humans, mutations that disrupt trafficking or expression of claudins 16 and 19 in the thick ascending limb of the renal tubule result in failure of paracellular Ca 2 + and Mg 2 + absorption and cause the autosomal recessive disease familial hypomagnesemia with hypercalciuria and nephrocalcinosis. Nevertheless, alterations in expression and trafficking of claudin proteins are present in intestinal diseases and can be induced by specific inflammatory mediators. Whether these changes contribute to disease pathogenesis or, alternatively, represent an adaptive response, is an area in need of further study. In this respect, a recent study demonstrated that claudin-2-mediated enhanced water efflux offered relief from mucosal pathogen colonization, prolonged pathogen shedding, excessive cytokine responses, and major tissue injury in experimental Citrobacter rodentium colitis model.

Size Selectivity

In addition to charge selectivity, the TJ discriminates between solutes on the basis of size. This property has been recognized for over 30 years, but, for the most part, detailed analysis has not been possible due to the limited number of probes available. This obstacle has been overcome using mixtures of polyethylene glycol oligomers with hydrodynamic radii ranging from ~ 3 to ~ 7 Å. The relationship between size and apparent permeability of these probes changed sharply at radii of ~ 4 Å, indicating a size-restricted population of paracellular pores. This is similar to the ~ 6 Å pores reported within small intestinal villous epithelium and may reflect the same structure. Notably, recent data have shown that claudin-2 expression, which enhances paracellular flux of monovalent cations, also increases flux of polyethylene glycol oligomers with radii less than 4 Å. Thus, it appears that a claudin-based class of paracellular channels defines overall charge selectivity as well as size selectivity of this route, which has been termed the pore pathway.

Despite the claudin-2-based cutoff at ~ 4 Å radii, larger polyethylene glycol oligomers are still able to traverse epithelial monolayers, albeit at much lower rates than smaller solutes. The magnitude of this component of paracellular flux does not vary substantially with solute radius, suggesting that these molecules traverse a size nonrestricted route that is distinct from the claudin-based pore pathway. The route of large solute flux has been referred to as the leak pathway and may correlate with the 50–60 Å channels described within crypt epithelium. Enhanced leak pathway flux is the means by which IFN-γ and TNF-α increase paracellular permeability, and it has, therefore, been proposed that leak pathway regulation is primarily associated with pathologic processes. While this may be true, the converse is not, as some inflammatory mediators, for example, IL-13, which enhances claudin-2 expression, increases flux across the pore pathway without affecting the leak pathway. Based on recent studies, occludin-dependent regulation of leak pathway is now clearly recognized. Transient occludin depletion in mature Caco-2 cell monolayers was shown to cause a progressive size-dependent increase in paracellular flux rate (three- to fourfold increase for urea (molecular radius 2.9 Å), five- to sixfold increase for mannitol (molecular radius 4.1 Å), ∼ 25-fold increase for inulin (molecular radius 15 Å), ~ 25-fold increase for 10-kDa dextran (molecular radius 23 Å), and ~ 45-fold increase for 70-kDa dextran (molecular radius 36 Å) ( Fig. 25.4 ). These findings were supported by another study showing charge-nonselective reduction in macromolecular flux with stable knockdown of occludin in Caco-2BBe cells. In the same study, C-terminal occludin/ELL domain was shown to be critical for anchoring and exchange of occludin at the TJ. Although occludin knockdown has been shown to have differential effect on the expression of claudins, possible due to different approaches and heterogenicity of cell lines, the contribution of claudins to the noncharge selective, large-size leak pathway in the absence of occludin is unclear.

Fig. 25.4

Graph of molecular radius of paracellular markers vs. relative increase in flux rate following occludin siRNA transfection (relative correlation coefficient, r = 0.98).

(From Al-Sadi R, Khatib K, Guo S, Ye D, Youssef M, Ma T. Occludin regulates macromolecule flux across the intestinal epithelial tight junction barrier. Am J Physiol Gastrointest Liver Physiol 2011; 300 :G1054–G1064, with permission.)

Myosin light chain kinase (MLCK), which causes contraction of peri-junctional actomyosin, is also a key regulator of TJ permeability. Close association of TJ with the cytoskeleton described by Madara was later shown to be the basis for cytoskeletal control of tight-junction permeability. Under physiological conditions, MLCK-induced increase in pore pathway permeability during Na + glucose cotransport is associated with condensation of peri-junction actomyosin. Under physiopathological conditions, such as TNF-α-mediated increase in TJ permeability, the enhancement in leak pathway permeability is associated with MLCK-dependent occludin endocytosis. While MLCK activation do not affect pore pathway directly, in vivo secondary immune activation in constitutively active MLCK expressing mice has been shown to increase claudin-2 expression and alter TJ Na + /Cl selectivity in IL-13-dependent way. Also, proinflammatory cytokines are known to increase leak pathway permeability in MLCK transcription and activation-dependent way, as discussed in detail, later in this chapter.

Protein Components of the TJ

Beginning in 1986 with identification of ZO-1 as the first TJ-associated protein, there has been continuous growth to almost 40 distinct proteins or protein families ( Table 25.2 ). Surprisingly, this number far exceeds the known components of other intercellular junctions, such as adherens, desmosomes, and gap junctions ( Fig. 25.2 ), and it is likely that many identified TJ proteins serve regulatory, rather than structural functions. This is certainly the case of enzymes, such as atypical protein kinase C isoforms and MLCK, which localize to the TJ and have been shown to regulate epithelial polarity, junction assembly, and barrier function. These will be discussed later in the context of specific regulatory events. In contrast, it is now clear that claudin proteins define the charge, and size selectivity of the TJ barrier. The specific roles played by other TJ-associated proteins, such as ZO-1 and occludin are less certain, but recent data have shed light on their functions in TJ stabilization and barrier regulation. Through an interesting evolutionary convergence, many tight and adherens junction proteins also serve as receptors or coreceptors for pathogen entry. Finally, polarity complexes are concentrated at the TJ.

Transmembrane Proteins and TJ Stability

The transmembrane proteins of TJs fall into three main groups: the single transmembrane domain proteins [e.g., JAM (junctional adhesion molecule), Crb3 (Crumbs protein homolog 3), CAR (coxsackievirus and adenovirus receptor), etc.]; the triple transmembrane domain protein, Bves (blood vessel epicardial substance); and the four-transmembrane domain proteins of the claudin and TAMP (tight junction-associated MARVEL proteins) families, which include occludin, tricellulin, and MarvelD3. A large number of transmembrane proteins serving diverse functions are found at TJs. Given the restricted, though, presumably, friendly confines of the TJ, it has been assumed that these proteins are embedded in a densely packed array that restricts their mobility. Together with the identification of multiple sites of interaction between TJ proteins, this led to the presumption that these proteins are tightly anchored and immobile. While possible, this conclusion would seem at odds with data demonstrating rapid regulation of TJ structure and function and, potentially, with the discovery that TJ membrane microdomains are enriched in glycolipids and cholesterol, i.e., membrane rafts, in which membrane fluidity is greater than more abundant phospholipid membrane domains.

The stability or mobility of proteins within the TJ has been studied using fluorescent-tagged TJ proteins. Importantly, these fluorescent fusion proteins were carefully validated and demonstrated to (1) colocalize and cofractionate with their endogenous, unlabeled counterparts ; (2) not interfere with TJ function ; and (3) complement TJ protein deficiencies in mice and in cell lines. Thus, one can conclude that their behaviors with regard to dynamic behavior are related to, and likely the same as, those of endogenous TJ proteins. Consistent with the conclusion that TJ-associated proteins are stably anchored, EGFP-claudin-1 was found to have a mobile fraction of only ~ 25%. While the hypothesis that such anchoring depends on the interaction of claudin-1 with ZO-1 (discussed below), anchoring of a truncated EGFP-claudin construct lacking the ZO-1 binding domain was similar to that of full-length EGFP-claudin although, as discussed below, trafficking to the TJ was severely impaired. Such data in epithelia were also consistent with the observation that, in fibroblasts, claudin-1-EGFP, which forms TJ strand-like fibrils despite blockade of ZO-1 binding by the C-terminal EGFP, demonstrated little fluorescent recovery after photobleaching. In contrast to claudin-1, occludin displayed a mobile fraction of ~ 75%. Occludin mobility occurred by diffusion within the plasma membrane and was dependent on membrane fluidity, as it could be inhibited by cholesterol extraction, with methyl-β-cyclodextrin, or reduced temperature, but was not an energy-dependent process. In contrast, while the mobile fraction of the peripheral membrane or plaque protein ZO-1 was similar to that of occludin, it was recovered by energy-dependent exchange with a cytosolic ZO-1 pool and was unaffected by cholesterol extraction or reduced temperature. While the functional significance of these observations is poorly understood at present, the data do show that protein interactions at the TJ are labile and that, despite the densely packed nature of the TJ, the raft-like biophysical properties of TJ membranes support free diffusion of TJ proteins.

Data including the observation that claudin-1-EGFP forms TJ-like strands in fibroblasts, which lack endogenous TJs, suggest that claudins form the core of TJ strands in epithelia. Claudins are the only junction-associated proteins that have been found to form such strands, although occludin can incorporate into claudin-based strands. Other transmembrane proteins, such as JAM-1, CAR, and connexin, are located around, but not incorporated into the strands. Beside varying stability of various TJ proteins within the plasma membrane, various modes of intracellular transport such as clathrin-mediated transport, caveolar transport, and micropinocytosis, as well as signaling molecules such as small GTPaseRabs and their adaptors have been shown to regulate biogenesis and function of TJs.

PDZ Containing Scaffolding Proteins

The interface between the transmembrane proteins and most cytoplasmic components is formed by a set of scaffolding proteins with multiple PDZ domains. PDZ domains, named for the proteins PSD-95, discs large, and ZO-1, are protein-binding modules that recognize target sequences at the extreme C-termini of transmembrane proteins. They are well characterized for their ability to cluster signal transduction complexes at specialized membrane contacts, such as synapses and TJs. PDZ-proteins at the TJ include the ZO-MAGUK proteins ZO-1, ZO-2, and ZO-3; the MAGUK relatives MAGI-1, MAGI-2, and MAGI-3; a protein with 13 PDZ called MUPP-1, and several of the polarity components cited below including PAR-3, PAR-6, PALS1, PATJ, scribble, and Dlg. Most of these proteins contain several PDZ domains and bind the tails of claudins and JAMs. Some, like the ZO-MAGUKs have other protein-binding modules, including SH3 and enzymatically inactive guanylate kinase (GuK) domains, which are known to bind additional targets including α-catenin and occludin. In addition, ZO-1 has a well-defined actin-binding domain as well as less well-characterized U5, U6, and ZU5 domains.

Although much work remains, several functions of ZO-1 in TJ assembly and regulation are recognized. ZO-1 recruitment to nascent adherens junctions requires the U5 motif, likely as a result of U5-dependent interactions with α-catenin, AVRCF, and AF-6/afadin. However, it is equally possible that ZO-1 and ZO-2 are involved in recruitment of other proteins to the nascent adherens junction. For example, mammary epithelial (Eph4) cells deficient in both ZO-1 and ZO-2 have a marked delay in adherens junction assembly, as do MDCK cells lacking only ZO-1. ZO-1 is able to rescue this defect. Data regarding ZO-2 are less clear, as one study showed that knockdown in MDCK cells delayed E-cadherin, ZO-1, and occludin recruitment to the plasma membrane, while a separate study found no effect of ZO-2 knockdown. Thus, while differences between labs make it difficult to precisely compare the results, these data do suggest that ZO-1 and ZO-2 have overlapping, but distinct functions. This is consistent with the observation that knockout of either ZO-1 or ZO-2 results in embryonic lethality in mice as well as the association of PDZ1 domain ZO-2, but not ZO-1, mutations with familial hypercholanemia in human patients.

Following adherens junction assembly, ZO-1 and ZO-2 direct polymerization of claudin proteins into the TJ strands. The interaction between claudin proteins and the first PDZ domain of ZO-1 and ZO-2 is important in the first phase of this process, as claudins are not recruited to the junctional complex in EPh4 cells that lack both ZO proteins and, as noted above, claudin proteins in which the PDZ binding motif are either mutated or blocked are trafficked to TJs inefficiently. However, the SH3-U5-GuK-U6 region is also critical to regulation of claudin polymerization. Although the means by which claudin polymerization is controlled requires further study, the U6 motif does appear to negatively regulate this process, as U6 deletion results in aberrant strand assembly.

In addition to the accumulation of data defining a role for ZO proteins in TJ assembly, recent work has defined a role of these proteins, particularly ZO-1 in barrier regulation. ZO-1, but not ZO-2, knockdown in MDCK layers increased paracellular permeability of large solutes, i.e., the leak pathway. Although charge selectivity was not examined, ZO-1 knockdown did not affect paracellular flux of small polyethylene glycols, suggesting that pore pathway barrier function was intact. ZO-1-dependent leak pathway maintenance did not require protein regions beyond the U6 domain, including the actin-binding region (ABR), as ZO-1 mutants truncated just after the U6 domain or in which the ABR was deleted internally were both able to restore barrier function. A following study has found that while ZO-1 PDZ1 is important for the normal organization of both the TJ and the AJC cytoskeleton, the SH3 domain and the U5 motif are required to recruit ZO-1 to the AJC, which is crucial for normal TJ and cytoskeletal organization. The PDZ2 domain is necessary for establishment of the ZO-1, occludin, and claudin-2 association and resultant normal permeability. This study suggests that the coordination of multiple ZO-1 domains is required for normal TJ structure and function. In addition to linking occludin and claudins, ZO-1 is able to connect to F-actin via an ABR located within the C-terminal half of the protein. Recent data indicate that the ABR is required for physiological, MLC-dependent barrier regulation. In this respect, MLCK activity has been shown to regulate the leak pathway in part by energy-dependent mechanical tension induced pulling apart of the TJ barrier. In vitro and in vivo analyses have shown that MLCK inhibition enhances TJ barrier function and is associated with stabilization of ZO-1, as assessed by fluorescence recovery of ZO-1. This ZO-1 stabilization required the ABR. Moreover, either ZO-1 knockdown or dominant negative expression of the free ABR prevented barrier function increases after MLCK inhibition. Together with other data, these results suggest that MLCK-dependent ZO-1 exchange and ABR-dependent ZO-1 anchoring are critical determinants of TJ barrier function and regulate the leak pathway. More recent studies have found that transmembrane TJ proteins were enriched among those binding the N terminus of ZO-1 and cytoskeletal proteins were enriched in among those binding to the C terminus of ZO-1. One such ZO-1-binding protein, F-BAR-domain protein transducer of cdc42-dependent actin activity (TOCA-1; also called formin-binding protein 1-like) has been shown to regulate actin assembly and TJ barrier function via recruiting actin nucleation-promoting factor N-WASP to TJs. Thus, ZO-1 mediates interaction between cytoskeletal elements (regulated by MLCK) and the transmembrane TJ proteins, including occludin and claudins. Also, association between ZO-2, JAM-A, actin filament-binding protein afadin activates small GTPase Rap2c and control apical cytoskeleton contraction.

TJ Associated MARVEL Proteins

Occludin, an ~ 65 kD atetraspan protein, was the first TJ-associated transmembrane protein identified. Although this discovery was greeted with excitement, subsequent understanding of occludin function has been elusive. The role of occludin in TJ assembly was first called into question by the observation that occludin-deficient embryonic cells can differentiate into polarized epithelial cells with structurally and functionally intact TJs. As discussed below, this prompted further studies that resulted in the discovery of claudins. Controversy over the contributions of occludin to TJs was enhanced further by the observation that occludin knockout mice are viable, and did not appear to have GI or renal disease, and as assessed using Ussing chambers, have intact intestinal barrier function. Since intestinal barrier assessment was only performed at 6-weeks of age, it remained unclear as to whether the TJ barrier defects were present at an earlier age but were subsequently compensated by other TJ proteins or associated proteins. Nevertheless, occludin knockout mice do have severe health defects. Males have testicular atrophy and are sterile, while females do not effectively suckle their young. In addition, occludin knockout mice of either gender are small, develop chronic inflammation and hyperplasia of the gastric epithelium with loss of parietal and chief cells, brain calcifications, osteopenia, and salivary gland defects. Subsequent reports have shown that siRNA or microRNA-induced knockdown of occludin caused and increase in intestinal TJ permeability in vitro and in vivo, in Caco-2 monolayers, and in live mice undergoing intestinal perfusion, respectively. As discussed earlier, occludin depletion resulted in a selective and preferential increase in macromolecular flux, suggesting that occludin plays a role in modulation of the leak pathway. Similarly, Buschmann et al. also found that siRNA-induced knockdown of occludin caused an increase in macromolecular flux across the Caco-2 BBe monolayers. They showed that a 107-amino acid occludin/EEL domain at the C-terminal, which contains binding sites for ZO-1 and actin, was required for the occludin-dependent enhancement of TJ barrier. Occludin deficiency has also been shown to promote ethanol-induced disruption of mouse colonic epithelial barrier function and occludin overexpressing transgenic mice are protected to certain extent against TNF-α-induced increase in leak pathway permeability. Overall, the studies with transient knockdown in mature cell monolayers, stable knockdown in cell monolayers, and exogenous mice intestinal expression have clearly demonstrated a critical role of occludin in intestinal epithelial TJ barrier.

Tricellulin, so named because it is concentrated at tricellular TJs, was the next protein identified as a member of MARVEL family. Occludin and tricellulin both contain tetraspanning MARVEL (MAL and related proteins for vesicle trafficking and membrane link) domains and bind ZO-1 via their carboxy terminal cytoplasmic tail, although the tail is not part of the MARVEL domain. However, in contrast to occludin knockdown in MDCK cells, tricellulin knockdown severely compromises barrier function of Eph4 monolayers. Interestingly, the effects of tricellulin knockdown in Caco-2 cells are less severe.

Tricellulin mutations have been reported in humans and linked to autosomal recessive deafness, likely as a result of defects in function of junctions between supporting and hair cells within the vestibular and cochlear epithelia. Remarkably, these tricellulin mutations, which were most often truncations that disrupted ZO-1 binding, caused deafness without creating a syndrome involving other organs. This limited distribution of disease does not reflect absence of tricellulin expression in other tissues, suggesting that another protein may compensate for tricellulin loss.

Because both occludin and tricellulin contain MARVEL domains, some suggested an evolutionary relationship between these proteins and a third MARVEL domain-containing protein, marvelD3. However, other analyses suggested that this was not the case. The issue was resolved by an evolutionary analysis demonstrating that human MARVEL domain-containing proteins can be segregated into four groups, one of which contains occludin, tricellulin, marvelD3, and their splice variants. This family has, therefore, been referred to as TJ-associated MARVEL proteins (TAMPs). MarvelD3 was found at both bicellular and tricellular TJs of Caco-2 intestinal epithelial monolayers as well as murine jejunum, hepatocytes, and renal tubules. However, while one study found that marvelD3 knockdown enhanced steady-state TER, another found that marvelD3 knockdown delayed TER development but did not affect final TER. This difference may relate to experimental conditions, or siRNA sequences and cell lines; although both studies used Caco-2 cells, the clones employed were clearly different, as the TERs achieved varied significantly. Further, analysis showed that combined knockdown of marvelD3 and either occludin or tricellulin caused a greater delay in TER development than knockdown of either protein individually.

Despite the analyses above, defining specific functions of occludin, tricellulin, and marvelD3 has been challenging. Further, data from occludin knockout animals, patients with tricellulin mutations, and siRNA knockdown cells suggest that a simple knockout/knockdown deletional approach will not be sufficient to dissect functions of the TAMPs. However, the association of occludin internalization with barrier loss induced by actin disruption or TNF core family cytokines show a strong link between acute occludin endocytosis and barrier loss. This was recently explored using a transgenic mouse that expresses functional EGFP-occludin, as well as endogenous occludin, within the intestinal epithelium. While TNF-induced occludin endocytosis depletes the protein from large regions of the intestinal epithelial TJ of wild-type mice, occludin was present throughout the intestinal epithelial TJs of transgenic mice. This maintenance of occludin at TJs of transgenic mice was not due to failure of occludin endocytosis but likely a result of occludin overexpression. The effect of TNF-α on jejunal barrier function was also attenuated, and diarrhea was entirely prevented in EGFP-occludin transgenic mice. In parallel with this in vitro work, an in vitro study in MDCK cells also found that occludin was required for TNF-induced barrier loss. Also, occludin stabilization by hypoxia responsive prolyl hydroxylase domain protein (PHD3) via prevention of the interaction between the E3 ligase Itch and occludin was found to be important for intestinal TJ barrier function. On the other hand, partitioning-defective protein 3 (Par-3) silencing blocked the trafficking of occludin from or through the Golgi complex to the cell surface and delayed TJ barrier development.

The above data suggest an important TJ barrier regulatory role for occludin. However, much work remains, including analysis of protein interactions mediated by occludin and functional significance of the occludin tail hyperphosphorylation that is associated with localization at the TJ and development of barrier function. For example, occludin is a key TJ protein in the claudin-2-independent TJ barrier enhancement in response to serine proteases. Recent studies have demonstrated that the C-terminal occludin tail associates with specific kinases and phosphatases, and that these interactions are able to regulate TJ assembly and barrier function. Moreover, it is now clear that phosphorylation of specific residues within the occludin tail modifies interactions with ZO-1. However, the sites identified are not within the region of occludin responsible for ZO-1 binding, suggesting that phosphorylation induces conformational changes in other regions of the tail. Moreover such a modification of interaction between transmembrane proteins and ZO-1 can also help understand the overlap between leak and pore pathway. For instance, CK2-mediated phosphorylation of occludin phosphorylation at S408 by CK2 enhances its diffusion within the membrane, thereby reducing occludin binding to ZO-1 and claudin-2 and allowing flux across claudin-2 pores. In contrast, CK2 inhibition and occludin dephosphorylation promote occludin- ZO-1-claudin-2 complex that reduces claudin-2-mediated pore function. Definition of these structural changes as well as the means by which they impact protein interactions and barrier function remains an important area of further studies.


The discovery of claudins in 1998 significantly advanced our understanding of the TJ barrier. Claudins were first shown to have the ability to form strands and confer cell-to-cell adhesion. Subsequent works clearly link claudins to the selective barrier properties of the TJ ( Table 25.3 ).

Table 25.3

Heritable Human Diseases of Tight Junction Proteins

Gene Disease Pathology/Mechanism Ref.
Cldn-1 Ichthyosis and sclerosing cholangitis Affects skin and bile ducts
Cldn-8 Barrter’s with hypomagnesemia S151P mutation, unproven pathogenesis
Cldn-14 Nonsyndromic deafness, Cochlear hair cell degeneration
Cldn-14 Nephrolithiasis GWAS found intronic SNPs
Human FHHNC a Defective renal Mg ++ reabsorption
Bovine interstitial nephritis
Cldn-19 FHHNC with ocular disease Defective renal Mg ++ reabsorption
Retinal development defect
PMP22 Peripheral polyneuropathies Demyelination
HNPP b Gene deletion
Charcot-Marie-Tooth Type 1A Gene duplication
Dejerine-Sottas syndrome Mutations
ZO-2 Familial cholanemia Defective PDZ-claudin binding
Tricellulin Nonsyndromic deafness Defective ZO-1 binding
Mutations with possible indirect effects in TJ proteins
P63 AEC c Decrease transcription of claudin-1 gene
WNK Kinase pseudohypoaldosteronism type II Abnl claudin-4 phosphorylation Increased Cl- permeability

Cln3 and 4 are hemizygous in patients with Williams-Beuren Syndrome. Reviewed in NEJM 2010 vol 362(3) pp. 239–252 in CV and craniofacial dysmorphisms.

a Familial hypomagnesemia hypercalciuria with nephrocalcinosis.

b Hereditary neuropathy with liability to pressure palsies.

c Ankylobleparon-ectodermal dysplasia-clefting.

Claudins are a family of tetraspan proteins, ranging from 20 to 25 kDa characterized by a conserved amino acid motif in the first extracellular loop (W-GLW-C-C). The human genome contains at least 26 claudin genes. The puffer fish Takifugu contains 56 claudin genes, although this animal has other examples of gene expansions that lack obvious biologic significance. The barrier-forming junctions of invertebrates, septate junctions, are structurally quite different from TJ, thus it was surprising to find claudin-like proteins in both flies and worms and to find that they are required to form epithelial barriers. Drosophila has six claudin sequences. Several of them are expressed at the barrier-forming septate junctions and, in addition to affecting barrier function, appear to regulate size of epithelial tubes. Interestingly, claudin-15 knockout results in mice with mega-intestine and zebrafish with multiple gut lumens, suggesting that claudins regulate epithelial tube development in vertebrates as well as flies.

The first extracellular loops of claudins are highly conserved, ranging from 41 to 55 residues, contain a specific ion-binding site, and appear to make a major contribution to TJ ion selectivity and barrier function. Site-directed mutagenesis of specific residues within the first extracellular loop of claudins can reverse the charge selectivity of the TJ pore, while inclusion of both acidic and basic residues in that region can reduce permeability to both cations and anions. The second loops, which are also highly conserved, range from 10 to 21 residues and appear to be more involved in homotypic and heterotypic adhesion and are also the site of Clostridium perfringins enterotoxin binding in claudins 3 and 4. The claudin cytoplasmic tails range from 21 to 44 residues and are the least well-conserved regions. With the probable exception of claudin- 12, -19a, -21, and -24 to -27, all claudin cytoplasmic tails end in PDZ-binding motifs, which bind PDZ domains in the cytoplasmic scaffolding proteins ZO-1 and MUPP1 and other proteins. PDZ-mediated interactions of claudins with ZO-1 and ZO-2 are required for efficient delivery to the TJ, although this may not be universally true.

The expression of several claudins has been documented within the mouse intestine. Claudins 2, 3, 7, and 15 are the most highly expressed claudins in gastrointestinal tract Many claudins show unique expression patterns along both longitudinal (duodenum-colon) and vertical (crypt-villus) axes. For instance, while claudin-13 was detectable only in the colon, claudin-18 was detectable only in the duodenum and jejunum. Claudin-8 expression increases progressively from small intestine to colon, while expression of claudin-15 decreases from small intestine to colon. Under normal physiological conditions, the expression of claudin-2 is restricted to the deep crypts. Furthermore, expression of at least few claudins is distributed differentially between the apical TJ and the basolateral membrane (see Ref. for a review). As true for many membrane proteins, the expression of claudins is also regulated differentially during development. These spatiotemporal expressions of claudins are thought to confer regional gradients of size and charge selectivity, but the mechanisms and physiological implications are incompletely defined.

It is now clear that claudins are largely responsible for paracellular permeability of small solutes, including ions and water. Although intrinsically, all claudins are barrier forming, some claudins form pores that are similar in size selectivity but are either cationic or anionic charge selective. The outcome of such claudin-based paracellular permeability is probably defined by temporal composition of various claudins at the TJ, their association with cytoplasmic plaque proteins, and their biochemical modification by processes such as sumolylation, phosphorylation, palmitoylation, etc. Based on ultrastructure, electrophysiology, and genetic studies, it is now known that claudin-2 pores are permeable to small cations and water and are exquisitely size-selective. Native or exogeneous claudin-2 expression can explain differences in barrier function in leaky or tight monolayers of several cell lines, under different experimental conditions. In addition to claudin-2, claudin-10b, and claudin-15 are known to form cation pores, while claudin-10a and -17 anion pores. Barrier forming group of claudins include claudin-1, -3, -4, -5, -6, -8, -12, -18, and -19. Further details of claudin function in and beyond epithelial TJ barrier have been discussed in recent reviews.

Consistent with the view that claudins regulate TJ permeability, mutations in claudin-16, also known as paracellin-1, prevent paracellular absorption of divalent cations in the renal tubule. This results in the disease familial hypomagnesemia with hypercalciuria and nephrocalcinosis, which is also associated with claudin-19 mutations. Claudin-11 and -14 mutations have also been associated with deafness, claudin-1 mutations have been linked to neonatal sclerosing cholangitis and ichthyosis. Similarly, mice lacking claudin-1 die shortly after birth from rapid evaporative water loss from the skin. They do not live long enough to investigate whether they recreate the biliary pathology. However, a point mutation in the claudin-binding PDZ domain of ZO-2 is the basis of Amish familial hypercholanemia. Claudin-18 knockout mice develop atrophic gastritis shortly after birth due to the paracellular H + leak in their stomach epithelia and also osteoporosis due to increased bone resorption by osteoclasts. Conditional deletion of claudin-7 in mouse has been shown to increase the paracellular flux of small organic solutes and to induce colonic inflammation. While specific claudin mutations have not been linked to GI disease, inflammatory, and infectious diseases are strongly associated with changes in claudin protein expression and trafficking. For instance, claudin-2 is consistently upregulated in human IBD as well as in animal models of IBD, while claudin-3 and -4 are usually upregulated in gastrointestinal cancers.

JAM and Other Single Transmembrane Domain TJ Proteins

Junctional adhesion molecule-A (JAM-A), a cell adhesion molecule localizing at intercellular junctions is known to play a role in immune cell trafficking across the epithelium and endothelium, cell cycle regulation, and barrier function. In the intestine, epithelial expression of JAM-1, which is also referred to as JAM-A, is reduced in IBD. Moreover, JAM-A knockout animals have intestinal barrier defects and enhanced cell turnover in the absence of exogenous stress as well as greater sensitivity to dextran-sulfate sodium-induced colitis and high saturated fat, fructose, and cholesterol diet-induced steatohepatitis. Increased expression of JAM-A is also associated with the enhancement of TJ barrier function induced by gut acting antibiotics as well as probiotic Bacillus subtilis. Other single transmembrane domain TJ proteins localized to the TJ include the coxsackie virus-adenovirus receptor CAR and Crumbs protein homolog 3 (Crb3). CAR, a cell adhesion molecule and a receptor for type B coxsackievirus and subgroup C adenovirus, regulates E-cadherin stability via promoting its endocytosis. In a study on human preimplantation embryos and embryonic stem cells, expression of soluble CAR was detected in undifferentiated blastomeres while transmembrane CAR was present in epithelial-like cell types, such as the trophectoderm (TE) and the outer layer of hESC colonies, suggesting a role of CAR in early embryogenesis and differentiation. Crumbs protein homolog 3 (Crb3/PATJ) is an apical organizer crucial for the maintenance of epithelial polarity. Crb3 is essential for postnatal viability, and Crb3 knockout E18.5 mouse embryos show impaired differentiation of intestinal apical membrane and fusion of villi apart from cystic kidneys and accumulation of mucosubstances in airways.

Additional Tight Junctional Proteins and Functions

Besides controlling paracellular permeability, TJs are also considered to be a signaling platform for cellular processes including morphogenesis, cell polarity, and differentiation via interactions with actin cytoskeletal elements, kinases and phosphatases, and the intracellular trafficking apparatus. TJ is home to a large number of kinases, including c-src and c-yes, both of which induce junction disassembly. Atypical protein kinase C (PKC) isoforms, including those involved in development of epithelial polarity, are also concentrated within the tight and adherens junctions. Finally, several transcription factors are located at the junction or bind to junction proteins.

TJ proteins also function as receptors for pathogens, interact with pathogen effector molecules, and form an essential step in their pathogenesis. Interactions between CAR and adenoviral capsid protein fiber have been demonstrated to promote adenoviral viral entry and escape, while CAR-type B coxsackievirus interaction leads to occludin internalization dependent coxsackie virus entry. Similarly, claudin-1 and occludin collaborate with at least two other cell surface proteins to facilitate hepatitis C virus entry. Diversity of occludin sequences confer species specificity to hepatitis C virus infection. Further, occludin splice variants and claudin-1 polymorphisms may impact susceptibility of some human hosts to hepatitis C virus. This theme of pathogen receptors being sequestered within the tight and adherens junctions extends beyond coxsackie virus, adenovirus, and hepatitis C. For example, Listeria monocytogenes use E-cadherin as a receptor for infection. Interestingly, E-cadherin is normally inaccessible, and is only exposed and available for L. monocytogenes binding at sites of cell extrusion.

Regulation of Intestinal Epithelial TJ Barrier

TJs are formed by many specific proteins and are connected with the cytoskeleton. The intestinal TJs are highly dynamic and their regulations can change in response to both external and intracellular stimuli. A number of signaling molecules have been implicated in the regulation of TJ function, including MLCK, Ca ++ , protein kinase C, G proteins, phospholipase A2 and C. In many intestinal and systemic diseases, alteration of TJ proteins has been implicated as a critical component to barrier dysfunction. Moreover, permeability of the TJs can be modified by bacterial toxins, cytokines, growth factors, hormones, and drugs.

The intestinal TJ barrier is rapidly regulated in response to extracellular factors that activate various intracellular signaling pathways. As the fluid composition of the luminal and serosal compartments undergo continual change during the various digestive and inter-digestive phases and vary depending on the types and amounts of food consumed, types of medication ingested, luminal bacterial composition and the bacterial load, the type and amount of digestive and proteolytic enzymes secreted into the lumen, the inflammatory state of the intestinal mucosa and the amount of pro-inflammatory cytokines and inflammatory mediators present, and the ionic and solute content of the luminal and serosal fluid, intestinal TJ barrier undergoes continual change in response to the changes in the extracellular fluid composition. The modification of TJ barrier function and paracellular permeability is dynamically regulated by various extracellular factors and stimuli and is closely associated with normal health and susceptibility to diseases. Therefore, in this section, we discuss those factors that have historical importance (such as the cytochalasins and the luminal osmolarity) and those that have been more extensively studied and for which the mechanisms have been supported by experimental studies. The central role of MLCK and actin-myosin contraction in the regulation of intestinal TJ barrier has been firmly established. For many of the intestinal TJ barrier modulating factors, MLCK-induced peri-junctional actin-myosin contraction and the subsequent cytoskeletal-TJ protein interaction (possibly mediated via the actin-ZO-1 and ZO-1-occludin/claudin interactions) appears to be a common mechanism leading to the functional and morphological opening of the intestinal TJ barrier. In addition, many studies have also suggested that the alterations in TJ protein expression, including occludin and claudins, and endocytosis of transmembrane TJ proteins are important mechanisms leading to the increase in TJ permeability. More recently, investigations have focused on the role of miRNAs in the regulation of intestinal barrier function. Major gaps in the understanding of the intracellular processes that regulate intestinal TJ barrier remain and provide an important opportunity for future investigations in this area.

Cytochalasins and the Intestinal TJ Barrier

Cytochalasins have been widely used as pharmacologic agents to examine actin-dependent cellular functions and were among the first agents to be used to study the role of cytoskeletal elements in epithelial TJ barrier modulation. Cytochalasins disrupt actin filaments by several mechanisms including direct severing of actin filaments, inhibition of actin subunit polymerization, and by inducing reactive cellular responses. The cytochalasin disruption of actin microfilaments produces a morphologic alteration in intestinal TJs and an increase in intestinal TJ permeability in both the ex vivo and in vitro intestinal epithelial systems. The cytochalasin D treatment of Ussing chamber mounted guinea pig small intestinal tissue produced a concentration dependent (0.1–20 μg/mL) drop in TER. The drop in small intestinal tissue TER directly correlated with an increase in Na + and mannitol flux, confirming an increase in intestinal epithelial TJ permeability. The dual-flux studies indicated that the increased permeation through the paracellular pathways fully accounted for the increase in mannitol and Na + flux. The cytochalasin D increase in intestinal TJ permeability was also accompanied by condensation or aggregation of microfilaments in the peri-junctional acto-myosin ring, especially in the regions of multiple cellular contacts. The condensation of the peri-junctional microfilaments produced a “pulse-string” type contraction of the brush borders of surface intestinal epithelial cells with a bulging or convex appearance of the epithelial surface ( Fig. 25.5 ), a decrease in number of microvilli at the junctional contact areas, and a disturbance in distribution, decrease in number, and loss of mesh-work like organization of TJ strands ( Fig. 25.6 ). These findings suggested that the cytochalasin D-induced increase in intestinal TJ permeability was mediated in part by a contractile-tension generated by contraction of the actin/myosin ring at the level of the TJs. Cytochalasin D appears to affect epithelial TJ actin filaments in two distinct stages: a rapid energy-independent severing of actin filaments into smaller fragments followed by reorganization of actin fragments to form cytoskeleton clumps by an energy-dependent process. The cytochalasin D-induced aggregation of peri-junctional microfilaments, contraction of the peri-junctional actin/myosin ring, alterations in TJ strands, and increase in TJ permeability were inhibited by energy depletion induced by 2,4-dinitrophenol (DNP), indicating the energy requirement in these processes. Similarly, cytochalasin (B and D) treatment of filter-grown intestinal epithelial monolayers Caco-2 and T84 cells also produced an acute drop in TER and an increase in epithelial permeability to paracellular markers mannitol and inulin ( Fig. 25.7 ). In contrast, the disintegration of Caco-2 microtubules with tublin depolymerizing agent colchicine did not have any effect on Caco-2 TJ barrier function or junctional localization of TJ proteins, suggesting that intact microtubules are not required for the acute maintenance of the intestinal epithelial TJ barrier function. The cytochalasin B-induced increase in Caco-2 TJ permeability required metabolic energy and was rapidly reversible (within hours) following cytochalasin B removal, indicating that the cytochalasin effect was not due to a permanent cell damage or cell death but by a rapidly reversible process.

Fig. 25.5

Scanning electron micrographs of vehicle control (A, B) and CD-exposed (C, D) (10 ng/mL, 60 min) mucosal sheets. The three villus ridges display smooth surfaces with intermittent linear folds. As seen in B, control villi are covered by polygonal absorptive cells with flat apical surfaces. In contrast, CD-exposed tissues display a cobblestone-like appearance of the villus surfaces (C). Higher magnification (D) shows this cobblestone effect is due to pulse-string contraction of the brush borders of individual absorptive cells resulting in a convex apical absorptive cell surface and flaring of microvilli. Bars, 20 μm.

(From Madara JL, Barenberg D, Carlson S. Effects of cytochalasin D on occluding junctions of intestinal absorptive cells: further evidence that the cytoskeleton may influence paracellular permeability and junctional charge selectivity. J Cell Biol 1986; 102 :2125–36, with permission.)

Fig. 25.6

Freeze-fracture replicas of villus absorptive cell occluding junctions. (A) Control junctions are composed of a net-like mesh of cross-linked strands or grooves. Peri-junctional microvilli are densely aligned above the junction. Bar. 0.1 μm. (Shadow angle, approximately left to right.) (B) Junction exposed to 10 μm/mL CD for 40 min. Junction is composed of an irregular array of strands that underlie occasional broad protrusions of the apical membrane (arrowheads) . Geometric irregularities produced by such protrusions result in a fracture plane which only focally includes the apical-most strand (straight arrows) . Many peri-junctional microvilli are lost and intramembrane particles penetrate into the incompletely isolated intrajunctional compartments (curved arrow) . Bars, 0.1 μm. (Shadow angle, approximately left to right.)

(From Madara JL, Barenberg D, Carlson S. Effects of cytochalasin D on occluding junctions of intestinal absorptive cells: further evidence that the cytoskeleton may influence paracellular permeability and junctional charge selectivity. J Cell Biol 1986; 102 :2125–2136, with permission.)

Fig. 25.7

Effect of Cyto B (5 μg/mL) on Caco-2 epithelial resistance and paracellular permeability. (A) Cyto B (5 μg/mL) effect on Caco-2 epithelial resistance expressed as Ω cm 2 . Inset: magnified view of the early time course. (B) Cyto B (5 μg/mL) effect on mucosal-to-serosal flux of paracellular marker mannitol expressed in nmol/cm 2 . Values are means ± SE; n = 4.

(From Ma TY, Hoa NT, Tran DD, Bui V, Pedram A, Mills S, Merryfield M. Cytochalasin B modulation of Caco-2 tight junction barrier: role of myosin light chain kinase. Am J Physiol Gastrointest Liver Physiol 2000; 279 :G875–G885, with permission.)

The cytochalasin B-induced increase in Caco-2 TJ permeability also correlated with sequential changes in peri-junctional actin and myosin filaments ( Fig. 25.8 ). In Caco-2 monolayers, actin and myosin filaments are localized in a belt-like manner surrounding the apical junctional area. Cytochalasin treatment produces a rapid energy-independent severing of actin filaments into small fragments (early phase response). Within seconds of cytochalasin B treatment, the peri-junctional actin filaments are severed and become fragmented and are present diffusely in the cytoplasm at the level of the TJs ( Fig. 25.8 B). This early phase fragmentation is then followed by an energy-dependent process in which the severed actin fragments reorganize to form large cytoskeletal aggregates containing actin and myosins filaments (late phase response). By 15–30 min of cytochalasin B treatment, the fragmented actin and myosin filaments coalesce to form large cytoskeletal clumps or “foci” near the peri-junctional areas ( Fig. 25.8 D). Similar effects of cytochalasin D on actin filament breakage and cytoskeletal clump formation (containing actin, myosins, and tropomyosins) were also demonstrated in the African green monkey kidney cells (BSC1cells).

Fig. 25.8

Effect of cytochalasin B (Cyto B) (5 μg/mL) on perijunctional Caco-2 actin microfilaments. Caco-2 F-actin filaments were labeled with fluorescein-conjugated phalloidin. The sequential effect of Cyto B (5 μg/mL) on Caco-2 actin microfilaments at time 0 (A) and 1 (B), 15 (C), and 30 min (D) is shown in the photomicrographs (original magnification, × 80). By 1 min of Cyto B exposure, peri-junctional actin filaments were fragmented and present diffusely throughout the cytoplasm. By 15–30 min of Cyto B exposure, actin fragments coalesced to form large actin clumps or “foci” near the perijunctional areas.

(From Ma TY, Hoa NT, Tran DD, Bui V, Pedram A, Mills S, Merryfield M. Cytochalasin B modulation of Caco-2 tight junction barrier: role of myosin light chain kinase. Am J Physiol Gastrointest Liver Physiol 2000; 279 :G875–G885, with permission.)

As actin-myosin contraction in smooth muscle and other cell types is mediated by MLCK activation, the possibility that the cytochalasin-induced increase in Caco-2 TJ permeability was also mediated by MLCK-activated actin-myosin interaction was considered. In brief, MLCK catalyzes myosin light chain (MLC) phosphorylation leading to the activation of Mg ++ -myosin ATPase, which hydrolyzes ATP to generate the mechanical energy needed for the actin-myosin contraction. The cytochalasin-induced alteration in peri-junctional actin and myosin filaments and increase in Caco-2 TJ permeability were accompanied by an increase in MLCK activity ( Fig. 25.9 ). The inhibition of cytochalasin- induced MLCK activation by MLCK inhibitors (ML-7, ML-9, KT-5926) prevented the late phase cytoskeletal aggregation and the increase in Caco-2 TJ permeability but not the early phase severing of actin filaments, suggesting that MLCK activation mediated the aggregation of the cytoskeletal filaments and the subsequent increase in Caco-2 TJ permeability. The cytochalasin severing of actin filaments has been proposed to be the triggering event for the MLCK activation. Consistent with this possibility, villin induced severing of actin filaments has been shown to induce Mg ++ -myosin ATPase activity and actin-myosin interaction. The requirement of metabolic energy and mechanical contraction of actin-myosin filaments in the TJ barrier opening was supported by the studies showing that Mg ++ -myosin ATPase inhibitor (2,3-butadione monoxime) and metabolic inhibitors (sodium azide, 2,4-dinitrophenol) inhibit the cytoskeletal clump formation (late-phase response), alteration in junctional localization of ZO-1 proteins and the increase in TJ permeability, but not the severing of the actin filaments (early-phase response). Together, these studies suggested that the cytochalasin-induced increase in intestinal TJ permeability was mediated by cytochalasin-induced activation of MLCK, which causes aggregation of peri-junctional microfilaments and the contraction of the peri-junctional actin-myosin ring, which in turn leads to a pulse-string-type contraction of the cell membrane and centripedal-tension generated opening of the intestinal TJ barrier. The MLCK-dependent retraction of the apical membrane and TJ barrier creates a large channel opening in-between cells and increases paracellular flux of both small and large-sized solutes, consistent with the generation of leak pathway.

Fig. 25.9

Effect of Cyto B on Caco-2 MLCK activity. Caco-2 monolayers were exposed to Cyto B for increasing time periods (0–30 min). Subsequently, Caco-2 monolayers were lysed, and Caco-2 MLCK was immunoprecipitated. The activity of the immunoprecipitated MLCK was determined by in vitro kinetic measurement of MLC phosphorylation. Phosphorylated MLC (P-MLC; ~ 19.5 kDa) was separated by 10% SDS-PAGE, stained with Coomassie blue solution, and autoradiographed. Cyto B produced a time-dependent activation of Caco-2 MLCK with the peak activation occurring between 5 and 10 min after Cyto B exposure.

(From Ma TY, Hoa NT, Tran DD, Bui V, Pedram A, Mills S, Merryfield M. Cytochalasin B modulation of Caco-2 tight junction barrier: role of myosin light chain kinase. Am J Physiol Gastrointest Liver Physiol 2000; 279 :G875–G885, with permission.)

The cytochalsasin D depolymerization of actin also causes a disturbance in junctional localization of TJ protein occludin, characterized by internalization and disassembly of occludin from junctional areas and loss of continuity in occludin junctional localization. Latrunculin A (LatA), an actin depolymerizing agent, has been shown to irreversibly disrupt the actin network and therefore disrupt the TJ barrier irreversibly in kidney derived cell line MDCK, and induced the formation of randomly dispersed actin aggregates. LatA also caused a disturbance in junctional localization of occludin and internalization of occludin in kidney derived cell line MDCK. Live cell imaging coupled with simultaneous recordings of barrier function showed that occludin internalization was temporally related to barrier loss. The latA-induced occludin internalization was mediated by caveolin-1 and dynamin II-dependent endocytosis, and inhibition of caveolae-mediated endocytosis prevented the latA-induced occludin internalization and drop in MDCK TER. These data suggested that latrunculin-induced actin depolymerization induces caveolae-mediated endocytosis of TJ proteins, and that internalization of transmembrane TJ proteins leads to the loss of TJ barrier function. The actin depolymerization-induced activation of MLCK is presumably also involved in the endocytosis of the TJ proteins.

Luminal Osmolarity and Solvent Drag Effect

The relationship between luminal osmolarity, intestinal water flux, and intestinal paracellular permeability has been extensively examined in vivo by recycling perfusion of isolated rat small intestinal segment with perfusate solutions having varying osmolarity. The rate of intestinal absorption of various sized paracellular markers including mannitol, PEG 400, PEG 900, and inulin was linearly related to the increasing concentration of permeability markers, consistent with a passive uptake mechanism. In the in vivo rat intestinal perfusion studies, changing the luminal perfusate pH (from 6.0 to 7.5), varying the unstirred water layer resistance by increasing the luminal flow rate (from 1 to 3 mL/min), or disruption of the mucus layer by treatment with mucolytic agent acetylcysteine did not affect the intestinal flux rates of paracellular markers.

In a leaky or low-resistance epithelia such as the small intestine, junctional or paracellular pathway is the major permeation pathway for the passive water and ionic flux. The passive water flux may be bidirectional, and the direction of the net water flux may be regulated by the differences in osmotic gradient or hydrostatic pressure across the intestinal epithelial barrier. A linear relationship exists between decreasing luminal osmolarity and increasing paracellular water flux. The contribution of solvent drag on the absorption of paracellular markers may be assessed by manipulating the water flux by changing the osmolarity of the luminal perfusate solution. Decreasing the luminal osmolarity from 600 to 225 mOsm results in an osmotic gradient-dependent mucosal-to-serosal water flux ( Fig. 25.10 ). As the water moves rapidly across the paracellular or junctional pathways, hydrophilic solutes in the luminal solution are also carried along via a solvent drag. There is a direct correlation between increasing intestinal water flux and increasing flux of the hydrophilic solutes via the solvent drag ( Fig. 25.11 ).

Fig. 25.10

Relationship between small intestinal luminal osmolarity and water flux.

(From Ma TY, Hollander D, Erickson RA, Truong H, Krugliak P. Is the small intestinal epithelium truly “tight” to inulin permeation? Am J Physiol 1991; 260 :G669–G676, with permission.)

Fig. 25.11

Relationship between jejunal water flux and inulin flux ( n = 3–9 rats). Jejunal water flux was varied by changing osmolarity of luminal perfusate. Plotted values are mean of water and inulin absorption.

(From Ma TY, Hollander D, Erickson RA, Truong H, Krugliak P. Is the small intestinal epithelium truly “tight” to inulin permeation? Am J Physiol 1991; 260 :G669–G676, with permission.)

Solvent drag allows the intestine to absorb all luminal glucose, even when concentrations exceed the capacity of the transcellular transport system; at low concentrations, the transcellular pathway absorbs luminal glucose, but as concentrations increase, a growing fraction of glucose absorption occurs via the paracellular pathway. The relative contribution of diffusive and convective (or solvent drag effect) component to the passive transport of permeability probes and solvent drag reflection coefficient σ f (an indicator of dependency on solvent drag effect) may be calculated using Fick’s law and modified Kedam-Katchalsky equation of solvent drag effect. Fick’s first law expresses the flux rate J D at which a given solute diffuses across a semipermeable membrane as diffusional coefficient, P D .

<SPAN role=presentation tabIndex=0 id=MathJax-Element-1-Frame class=MathJax style="POSITION: relative" data-mathml='PD=−JD/C2−C1,’>PD=JD/(C2C1),PD=−JD/C2−C1,
P D = − J D / C 2 − C 1 ,
where C 1 and C 2 are concentrations of solute on the luminal and serosal side, respectively. The solute flux across the semipermeable epithelium by solvent drag is expressed quantitatively by a modified Kedem-Katchalsky equation.
<SPAN role=presentation tabIndex=0 id=MathJax-Element-2-Frame class=MathJax style="POSITION: relative" data-mathml='JSD=JV1−σfC1+C2/2,’>JSD=JV(1σf)((C1+C2)/2),JSD=JV1−σfC1+C2/2,
J SD = J V 1 − σ f C 1 + C 2 / 2 ,
where J SD is the solute flux rate by solvent drag, (1 − σ f ) is the coefficient of solvent drag, J V is the solvent flow rate, and σ f is the solvent drag reflection coefficient. Combining the above two equations provides an equation describing the relative contribution of diffusion and solvent drag to the net passive solute flux, J S , across a porous epithelium.
<SPAN role=presentation tabIndex=0 id=MathJax-Element-3-Frame class=MathJax style="POSITION: relative" data-mathml='JS=JD+JSD’>JS=JD+JSDJS=JD+JSD
J S = J D + J SD

<SPAN role=presentation tabIndex=0 id=MathJax-Element-4-Frame class=MathJax style="POSITION: relative" data-mathml='JS=−PDΔC+JV1−σfC1+C2/2′>JS=PDΔC+JV(1σf)((C1+C2)/2)JS=−PDΔC+JV1−σfC1+C2/2
J S = − P D Δ C + J V 1 − σ f C 1 + C 2 / 2

Only gold members can continue reading. Log In or Register to continue

Apr 21, 2019 | Posted by in ABDOMINAL MEDICINE | Comments Off on Tight Junctions and the Intestinal Barrier
Premium Wordpress Themes by UFO Themes