Tight and Anchoring Junctions

An epithelial or endothelial sheet is a single layer (or stack) of cells that must control traffic between two different body compartments—gut lumen versus blood, for instance. The tight junction (zonula occludens) is the structure that makes this possible: it is not a single protein but a beltlike network of transmembrane strands, built chiefly from claudins (a family of ~27 related proteins in mammals) and occludin, that runs around the apical-most perimeter of each cell and fuses with the corresponding strands on the neighboring cell. This fusion narrows the intercellular (paracellular) space to near-zero at discrete points, forcing most solutes that cross the epithelium to go through cells (transcellular route) rather than between them, unless the specific claudins present create a regulated paracellular pore. Tight junctions also act as a fence, restricting the diffusion of lipids and proteins between the apical and basolateral plasma membrane domains, which is essential for maintaining epithelial cell polarity.

Anchoring junctions solve a different problem: mechanical continuity. The adherens junction (zonula adherens) sits just basal to the tight junction in a typical epithelial junctional complex and uses classical cadherins—E-cadherin in most epithelia, N-cadherin in neural and mesenchymal tissue, VE-cadherin in endothelium—as single-pass transmembrane proteins whose extracellular cadherin (EC) repeat domains bind Ca²⁺ ions between repeats, rigidifying the ectodomain into a rod that makes homophilic, Ca²⁺-dependent contacts with cadherins on the adjacent cell. Intracellularly, the cadherin tail binds β-catenin, which binds α-catenin, which links (directly and indirectly, via vinculin and other actin-binding proteins) to the actin cytoskeleton, creating a continuous belt of adhesion coupled to a contractile actomyosin ring that can generate and transmit tension across the tissue.

Desmosomes are spot-like (macula adherens) junctions that use a different cadherin subfamily—desmogleins and desmocollins—whose cytoplasmic tails bind plakoglobin and plakophilin, which in turn recruit desmoplakin, and desmoplakin anchors intermediate filaments (keratins in epithelia, desmin in cardiac muscle) rather than actin. Because intermediate filaments form a tension-resistant, rope-like network throughout the cytoplasm, desmosomes distribute mechanical stress across an entire tissue, which is why they are especially dense in tissues under high mechanical load: epidermis, cardiac muscle, and bladder urothelium.

Assembling the tight junction. Claudin monomers span the membrane four times, with two extracellular loops (ECL1 and ECL2) projecting into the paracellular space. Adjacent claudin molecules first oligomerize in the plane of one membrane (cis-interactions), then ECL1 of a claudin on one cell binds ECL2-type surfaces of a claudin on the facing cell (trans-interactions), polymerizing into continuous strands that appear as a branching network of intramembranous particles by freeze-fracture electron microscopy. Because ECL1 contains a variable, charged region, different claudins create paracellular channels with different ion selectivity: claudin-2, for example, forms a cation-permeable, water-permeable pore (important in kidney proximal tubule and gut crypts), while claudin-1 and claudin-4 tighten the barrier and restrict paracellular flux. Occludin and the tricellulin protein (at three-cell junctional points) stabilize strands and contribute to barrier integrity but cannot form strands alone. The cytoplasmic tails of claudins and occludin bind ZO-1, ZO-2, and ZO-3 (zonula occludens proteins), PDZ-domain scaffolds that cluster the transmembrane strands, link them to the underlying actin cytoskeleton, and recruit signaling proteins that regulate junction assembly, so the tight junction is simultaneously a diffusion barrier and a signaling platform.

Adherens junction mechanics. Cadherin ectodomains require Ca²⁺ to adopt their rigid, adhesion-competent conformation; each linker between successive EC repeats binds three Ca²⁺ ions, and chelating calcium (e.g., with EGTA) causes the ectodomain to flex and lose adhesive strength almost immediately—the classic manipulation used to disrupt cadherin junctions experimentally. Trans-dimers form initially as weak "strand-swap" contacts between the outermost EC1 domains of opposing cadherins, and lateral clustering (cis-dimerization) on the membrane surface converts these into a zipper-like, high-avidity adhesive interface. Intracellularly, p120-catenin binds the juxtamembrane domain and stabilizes cadherin at the membrane (preventing its endocytosis), while β-catenin/α-catenin link to actin—not as a rigid static tether but as a force-sensitive connection: α-catenin unfolds under tension to expose a vinculin-binding site, recruiting more actin-binding machinery precisely where mechanical load is highest, reinforcing junctions under strain.

Desmosome mechanics. Desmoglein and desmocollin isoforms (Dsg1–4, Dsc1–3, expressed in tissue-specific and layer-specific patterns in skin) form a similar Ca²⁺-dependent extracellular adhesive interface, but their cytoplasmic domains recruit plakoglobin (a close relative of β-catenin) and plakophilins, which cluster desmoplakin at the plaque. Desmoplakin's rod domain directly binds keratin intermediate filaments, so each desmosome nucleates a local convergence point where keratin networks from two adjacent cells are functionally—though not physically continuous, since the plasma membrane still separates them—coupled into one tissue-wide mechanical syncytium. Because intermediate filaments resist stretching far better than actin, this arrangement is what lets skin withstand shearing forces without cell-cell separation.

Order of assembly. In a classic epithelial cell reforming junctions (e.g., after low-calcium disruption and calcium switch), adherens junctions typically nucleate first via cadherin clustering, recruiting actin and generating the initial adhesive "spot welds"; tight junction components are then recruited to a position just apical to these nascent adherens contacts, and the whole junctional complex matures and moves to its final apical position as the cell polarizes, showing that anchoring junctions provide a structural and positional cue for tight junction formation.

Farquhar and Palade (1963): defining the junctional complex. Before the molecular players were known, Marilyn Farquhar and George Palade used thin-section transmission electron microscopy on several epithelia (intestine, gallbladder, kidney tubule) and systematically described a conserved, ordered sequence of three junctions near the apical surface: a zonula occludens (tight junction, where adjacent membranes appeared to touch and obliterate the intercellular space), a zonula adherens (a belt-like junction just below it with a visible intercellular gap and dense cytoplasmic plaque with associated filaments), and desmosomes (spot-like structures further basal, also with dense plaques). This paper established both the names still used today and the crucial insight that the tight junction, not a continuous "cement," was the actual diffusion barrier—demonstrated using electron-dense tracers (such as lanthanum) infused into the lumen, which penetrated the intercellular space up to, but not through, the tight junction, directly visualizing where the paracellular seal occurs.

Freeze-fracture EM: strands, not sheets. Tight junctions initially seemed to be a fused membrane sheet, but freeze-fracture electron microscopy (splitting the membrane along its hydrophobic midline and viewing the exposed leaflet) revealed instead a branching network of intramembranous particle strands on the P-face (protoplasmic face) of the membrane, matching complementary grooves on the E-face. This directly showed that the tight junction is built from discrete, polymerizing transmembrane protein strands—later identified as claudins—rather than continuous membrane fusion, and that the number of strands correlates with barrier "tightness" (leaky epithelia like proximal renal tubule have few strands; tight epithelia like urinary bladder have many).

Furuse, Tsukita and colleagues: identifying claudins and proving they form the strands (1998). Occludin was identified first (1993) as a component of tight junction strands, but occludin-knockout mice, remarkably, still formed morphologically normal tight junction strands, proving occludin was not the essential strand-forming protein. Mikio Furuse and Shoichiro Tsukita's group then purified tight-junction-enriched membrane fractions from chicken liver and identified two novel ~22-kDa proteins, claudin-1 and claudin-2. The decisive experiment: when claudin cDNA was transfected into cultured mouse L fibroblasts—cells that normally have no tight junctions and no strands at all—the cells spontaneously formed freeze-fracture-visible strands at contacts with neighboring transfected cells, and these strands showed adhesive, junction-like properties. This was the direct proof that claudins are both necessary and sufficient to polymerize into tight junction strands; expressing different claudin family members in these fibroblasts also showed that mixing claudin types changes strand morphology and paracellular permeability, establishing that claudin identity dictates barrier selectivity.

Claudin knockouts and human disease reveal ion-selective pores. Claudin-11 knockout mice lack tight junctions in myelin and Sertoli cells and are infertile with CNS conduction defects; claudin-14 mutations cause human hereditary deafness by disrupting the endocochlear potential; and most tellingly, claudin-16 (paracellin-1) loss-of-function mutations cause a human disease of renal Mg²⁺ and Ca²⁺ wasting (familial hypomagnesemia with hypercalciuria and nephrocalcinosis), because claudin-16 normally forms a paracellular pore in the thick ascending limb that specifically permits Mg²⁺/Ca²⁺ reabsorption between cells. Losing one claudin gene causes a highly specific ion-transport disease, direct genetic proof that individual claudins set the ionic selectivity, not just the general "tightness," of the paracellular pathway.

Cadherin calcium-dependence and homophilic specificity. Classic experiments by Masatoshi Takeichi's group in the 1980s showed that antibodies against a cell-surface glycoprotein (which he named cadherin) blocked calcium-dependent aggregation of dissociated embryonic cells, and that removing extracellular Ca²⁺ (with EGTA) instantly caused epithelial sheets to round up and separate, proving both the identity of the adhesion molecule and its absolute Ca²⁺ dependence. Transfection experiments were equally decisive: L cells (which lack cadherins and do not aggregate) transfected with E-cadherin cDNA formed compact, calcium-dependent aggregates, while cells expressing different cadherin types (E- vs. P-cadherin) preferentially sorted with cells expressing the same cadherin type in mixed aggregates—direct evidence for homophilic, type-specific adhesion, foreshadowing how differential cadherin expression drives tissue boundary formation.

Desmosome identity established by autoimmune disease. The proof that desmogleins are the adhesive core of desmosomes came partly from human autoimmune disease: patients with pemphigus vulgaris and pemphigus foliaceus produce autoantibodies against desmoglein-3 and desmoglein-1 respectively, and passive transfer of purified patient IgG into neonatal mice reproduces the human blistering phenotype—epidermal cells separating from each other (acantholysis) exactly at the desmosome. This Ig-transfer experiment, performed by Amagai and colleagues in the 1990s, causally confirmed that desmoglein is both necessary for keratinocyte adhesion and the actual antigenic target whose loss of function causes clinical blistering.

Example 1 — a dominant-negative claudin experiment. A researcher expresses a truncated claudin-4 construct lacking its second extracellular loop (ECL2), tagged with GFP, in a Madin-Darby canine kidney (MDCK) epithelial monolayer that normally has a transepithelial electrical resistance (TER) of 1000 Ω·cm². Wild-type claudin-4 overexpression is known to increase TER (tighten the barrier) because claudin-4 restricts cation flux. This truncated construct still localizes to the tight junction (its transmembrane domains and cytoplasmic tail, which binds ZO-1, are intact) but cannot make proper trans-interactions because ECL2 is missing. Reasoning through the outcome: because claudins first oligomerize in cis within one membrane, the mutant claudin-4 will co-assemble with endogenous wild-type claudins and other claudin family members already in the strand. Because it cannot complete a normal trans-bond, its presence disrupts strand continuity at every point it is incorporated, acting as a dominant-negative—not merely inactive but actively poisoning the polymer, the way one broken link degrades a whole chain. Predicted result: TER falls well below the baseline 1000 Ω·cm² (say, to 150–250 Ω·cm²) even though total claudin protein and ZO-1 recruitment look normal by immunofluorescence—illustrating that tight junction function depends on ectodomain trans-interaction integrity, not merely on correct protein localization. This is exactly the logic used in real studies dissecting which claudin domains are required for strand formation versus mere membrane targeting.

Example 2 — a clinical vignette integrating both junction types. A newborn presents with widespread, fragile, easily denuded skin at birth (a "lethal acantholytic epidermolysis" phenotype), plus woolly, sparse hair, and evidence of a leaky epidermal water barrier (transepidermal water loss far above the normal ~4–8 g/m²/h for intact neonatal skin). Biopsy shows keratinocytes detaching from one another within the epidermis, with intact hemidesmosomes at the basement membrane (so the epidermis is not detaching from the dermis, only cell-from-cell within the epidermis). Reasoning: cell-cell (not cell-matrix) detachment implicates either desmosomes or adherens junctions, not Integrins and Cell-Matrix Adhesion-based hemidesmosomes, which the biopsy shows are intact. The combination of skin fragility and hair abnormality is the classic signature of a desmoplakin mutation, because desmoplakin anchors keratin intermediate filaments in both keratinocytes and hair follicle cells; loss of function collapses the keratin network's connection to desmosomal plaques, so cells hold together loosely enough to survive normal handling but shear apart under the mechanical stress of birth and early life. A separate, purely epidermal water-barrier defect (without hair involvement) would more likely point to a tight junction claudin defect (as in some ichthyoses), since only tight junctions—not desmosomes—establish the paracellular seal that limits transepidermal water loss. This vignette shows how junction anatomy (barrier vs. mechanical rivet) predicts distinct, distinguishable clinical phenotypes.

Shoichiro Tsukita, working at Kyoto University through the 1990s and 2000s, transformed tight junction biology from a purely descriptive, electron-microscopic field into a molecular one. Trained in cell biology and skilled in both freeze-fracture EM and biochemical purification, Tsukita and his close collaborator Mikio Furuse painstakingly fractionated liver plasma membranes to isolate the actual protein strands seen by EM, rather than relying on antibody screens or homology guesses—an approach that let them discover occludin and, crucially, the entire claudin family, which turned out to be far larger and more functionally diverse than anyone anticipated. Tsukita's laboratory then generated the claudin and occludin knockout mice and the fibroblast transfection systems that functionally proved claudins build tight junction strands, work that reframed the tight junction from a static "seal" into a dynamically tunable, genetically programmable barrier whose selectivity is written into which claudin genes a given epithelium expresses.

Gastroenterology and the "leaky gut" hypothesis. Inflammatory bowel disease, celiac disease, and some food allergies involve cytokine-driven (TNF-α, IL-13) down-regulation of tight-junction claudins and redistribution of claudin-2, increasing paracellular permeability to luminal antigens; drugs targeting these cytokine pathways partly work by restoring barrier integrity, and larazotide acetate, a peptide that blocks pathological tight junction opening, has been tested in celiac disease.

Nephrology. Claudin-16 and claudin-19 mutations cause familial hypomagnesemia with hypercalciuria and nephrocalcinosis, directly informing diagnosis of otherwise puzzling electrolyte-wasting syndromes, and claudin-2 biology explains segment-specific differences in renal solute handling exploited in understanding diuretic action.

Dermatology and autoimmunity. Pemphigus vulgaris and pemphigus foliaceus are diagnosed and monitored by testing for anti-desmoglein-1/3 autoantibodies, and understanding the desmosome mechanism directly guides treatment with immunosuppression and B-cell-depleting therapy (rituximab) targeting the autoantibody-producing clones.

Cardiology. Mutations in desmosomal genes (plakophilin-2, desmoplakin, desmoglein-2) cause arrhythmogenic right ventricular cardiomyopathy, in which weakened intercalated-disc desmosomes lead to myocyte detachment, fibrofatty replacement, and life-threatening arrhythmias—genetic testing of these genes is now standard in cardiomyopathy work-up.

Oncology. Loss of E-cadherin expression (via mutation, promoter methylation, or transcriptional repression by Snail/Slug/Twist) is a hallmark of epithelial-to-mesenchymal transition, enabling carcinoma cells to detach from primary tumors and invade—E-cadherin immunohistochemistry is used diagnostically, and hereditary diffuse gastric cancer is caused by germline CDH1 (E-cadherin) mutations.

Vaccine and toxin biology. The Clostridium perfringens enterotoxin binds claudin-3/4 directly and is used experimentally to acutely disrupt tight junctions, informing both toxin mechanism studies and drug-delivery strategies that transiently open epithelial barriers to enhance absorption of otherwise poorly permeable therapeutics.

Regenerative and developmental biology. Cadherin-based sorting behavior underlies tissue boundary formation and organoid self-organization, concepts that connect directly forward to how cells build and remodel their surrounding The Extracellular Matrix and adhere to it through Integrins and Cell-Matrix Adhesion.

Ask the tutor: How does phosphorylation of ZO-1 or claudin cytoplasmic tails dynamically regulate tight junction permeability during inflammation?