Cell Junctions Overview

A cell junction is a specialized region of the plasma membrane, usually shared between two adjacent cells (or between a cell and the extracellular matrix), where transmembrane proteins cluster to perform a specific structural or communicative task. Despite the molecular diversity within each category, virtually every animal cell junction can be sorted into one of three functional classes.

Occluding junctions seal the gap between adjacent cells, converting a sheet of individual cells into a continuous barrier that restricts the free diffusion of ions and molecules through the space between cells (the paracellular route). The tight junction (zonula occludens) is the defining example in vertebrates; it also partitions the plasma membrane itself into distinct apical and basolateral domains, which is essential for the vectorial (directional) transport that epithelial cells perform.

Anchoring junctions mechanically couple the cytoskeleton of one cell to that of its neighbor, or to the extracellular matrix, distributing tension and shear stress across a tissue rather than concentrating it on any single cell. Anchoring junctions subdivide further by which cytoskeletal filament they attach to (actin or intermediate filaments) and by what they attach to (another cell or the matrix), giving four canonical types: adherens junctions and desmosomes (cell–cell), and focal adhesions and hemidesmosomes (cell–matrix).

Communicating junctions create direct cytoplasmic or symplastic continuity between adjacent cells, allowing ions, second messengers, and small metabolites to pass without crossing two plasma membranes and the extracellular space. Gap junctions perform this role in animal cells; the analogous, structurally unrelated structure in plants is the plasmodesma, which additionally allows RNA and even transcription factors to move between cells through a channel that traverses the shared cell wall. This three-way classification — seal, anchor, communicate — is not arbitrary bookkeeping. It reflects three genuinely separate biophysical problems that multicellular tissues had to solve independently, and each problem was solved with its own distinct molecular machine.

Occluding junctions: building a selective wall

The tight junction sits at the most apical point of the lateral membrane in epithelial cells, forming a continuous belt (the zonula occludens) that encircles each cell like a molecular gasket. It is built from claudins (a family of roughly two dozen tetraspan proteins in mammals) and occludin, both of which span the membrane four times and project extracellular loops that bind directly to the corresponding loops on the neighboring cell, forming rows of "kissing points" that seal the paracellular space. Intracellularly, these transmembrane proteins are tethered to the actin cytoskeleton through scaffolding proteins called ZO-1, ZO-2, and ZO-3 (zonula occludens proteins), which cluster the claudin/occludin strands and couple them mechanically to the cell's contractile machinery. Crucially, tight junctions are not simply impermeable walls — they are selective, tunable pores. Different claudin family members form paracellular channels with different ion selectivity and different pore sizes; claudin-2, for example, forms a cation-selective, water-permeable pore, which is why the proximal kidney tubule (rich in claudin-2) is a "leaky" epithelium optimized for bulk fluid reabsorption, while claudin-1 and claudin-4 dominate in "tight" epithelia like the distal nephron and skin, where the barrier must be nearly impermeable. This claudin code lets a single junctional architecture be retuned for wildly different physiological jobs simply by changing which claudin genes a given epithelium expresses. Tight junctions also enforce epithelial cell polarity. By fusing the outer leaflets of adjacent membranes at a defined line, the tight junction acts as a fence that blocks lateral diffusion of membrane lipids and proteins between the apical and basolateral domains. This is how, for example, glucose transporters (SGLT1) stay confined to the apical membrane of an intestinal cell while Na⁺/K⁺-ATPase stays confined to the basolateral membrane, a segregation that is the entire physical basis of directional nutrient absorption.

Anchoring junctions: distributing mechanical load

Anchoring junctions come in two cytoskeletal flavors and two partnership types, giving four combinations.

Adherens junctions (cell–cell, actin-linked) use cadherins — single-pass transmembrane glycoproteins whose extracellular cadherin (EC) repeat domains bind Ca²⁺ ions between repeats, rigidifying the ectodomain into a rod that projects from the membrane and binds, in a strictly homophilic and Ca²⁺-dependent manner, to cadherins on the neighboring cell. Classical cadherins like E-cadherin cluster at the zonula adherens, just basal to the tight junction, forming a continuous adhesion belt. Intracellularly, the cadherin tail binds β-catenin, which in turn binds α-catenin, which links the whole complex to filamentous actin (either directly or via vinculin and other actin-binding adaptors under tension). This actin belt can contract, and coordinated contraction of adherens junctions around many cells drives large-scale tissue movements such as epithelial folding during gastrulation and neural tube closure.

Desmosomes (cell–cell, intermediate-filament-linked) use desmosomal cadherins (desmogleins and desmocollins) instead of classical cadherins. Rather than an actin belt, desmosomes are spot-welds — button-like plaques distributed across the lateral membrane — whose cytoplasmic plaque proteins desmoplakin, plakoglobin, and plakophilin anchor bundles of keratin intermediate filaments. Because intermediate filaments are far more tensile-resistant than actin filaments, desmosomes are the junction of choice in tissues under high mechanical stress: epidermis, cardiac muscle, and bladder epithelium are dense with desmosomes.

Focal adhesions (cell–matrix, actin-linked) use integrins, heterodimeric transmembrane receptors (one α and one β subunit) that bind extracellular matrix ligands such as fibronectin and, intracellularly, nucleate a large multiprotein plaque (talin, vinculin, paxillin, α-actinin) that links to actin stress fibers. Integrin binding is a two-way, allosteric conversation: matrix engagement changes integrin conformation, which changes intracellular signaling ("outside-in" signaling), while intracellular signals can activate integrins to increase matrix affinity ("inside-out" signaling) — a property essential to platelet aggregation and leukocyte adhesion during inflammation.

Hemidesmosomes (cell–matrix, intermediate-filament-linked) resemble focal adhesions functionally but use a different integrin (α6β4) and anchor keratin filaments rather than actin, via the plaque protein plectin, providing the mechanical rivet that fastens basal epidermal cells to the underlying basal lamina.

Communicating junctions: shared cytoplasm

A gap junction is built from twelve copies of a four-pass transmembrane protein called a connexin: six connexins oligomerize into a hemichannel called a connexon in one cell's membrane, and this connexon docks head-to-head with a connexon in the adjacent cell's membrane, forming a continuous aqueous channel that directly connects the two cytoplasms. The channel has a functional pore diameter of roughly

1.0–1.5 nm, allowing free passage of ions, small metabolites (ATP, cyclic AMP, IP₃, glucose) and molecules up to about

1 kDa, while excluding proteins and nucleic acids. Because connexons can be built from different connexin isoforms (21 genes in humans) and different combinations of connexins in the two docking hemichannels (homotypic vs. heterotypic channels), gap junctions have tunable permeability and gating properties, and can be closed rapidly by a drop in cytoplasmic pH or a rise in Ca²⁺ — a protective mechanism that seals off a damaged cell from its healthy neighbors. Gap junctions cluster into plaques containing hundreds to thousands of connexons, visible in freeze-fracture replicas as a crystalline array of intramembranous particles. Plants achieve an analogous but structurally unrelated outcome with plasmodesmata: channels that traverse the shared cell wall, lined by fused plasma membrane and typically containing a central tubule of appressed endoplasmic reticulum called the desmotubule. The cytoplasmic sleeve between desmotubule and plasma membrane allows symplastic transport of ions, sugars, and small proteins, and — remarkably — its pore size can be actively dilated by the plant to permit passage of transcription factors, small RNAs, and viral movement proteins, giving plants a regulated, macromolecular cell-to-cell communication system with no true animal equivalent.

The three-way classification of junctions did not emerge from theory; it emerged from decades of electron microscopy, physiology, and molecular genetics converging on the same conclusion from different directions.

Freeze-fracture electron microscopy revealed the structural signatures of each junction type. In the 1960s and 1970s, researchers including Daniel Friend, Nicholas Gilula, and colleagues developed freeze-fracture, a technique in which frozen membranes are cleaved along their internal hydrophobic midline, exposing the interior of the lipid bilayer, then replicated in platinum for viewing. This technique showed that tight junctions consist of continuous, anastomosing strands of intramembranous particles that fuse the two membrane leaflets at discrete points — the physical basis of the "kissing points" model, later confirmed to correspond to claudin polymers when claudin genes were cloned by Shoichiro Tsukita and Mikio Furuse in the mid-1990s. The same technique showed gap junctions as crystalline, hexagonally packed arrays of particles, quite distinct from tight junction strands, giving early, purely morphological evidence that these were fundamentally different machines — one sealing a gap, the other spanning it.

Werner Loewenstein's electrophysiological experiments (1960s) proved that gap junctions create direct cytoplasmic continuity. Loewenstein and colleagues impaled adjacent cells in insect salivary gland epithelium and liver with two microelectrodes each and injected current into one cell while recording voltage in the neighbor. They found that current injected into one cell produced a voltage change in the adjacent cell far larger than could be explained by current flowing through the extracellular space and two separate plasma membranes — the cells were electrically coupled, implying a low-resistance pathway between their cytoplasms. Loewenstein went further and injected fluorescent dyes of defined molecular weight (fluorescein, ~330 Da) into single cells, showing the dye spread rapidly to neighboring cells, while larger labeled dextrans above roughly 1–1.5 kDa did not — establishing both the existence of a direct cytoplasmic channel and its size cutoff, decades before the connexin genes were cloned.

Tight junction "leakiness" was measured directly using extracellular tracers, proving occluding junctions are the physical barrier to paracellular diffusion. In a classic experiment, Marilyn Farquhar and George Palade (1963) used electron-dense tracers infused into the vasculature or gut lumen of animals and examined, by electron microscopy, how far the tracer penetrated into the intercellular space between epithelial cells. They found the tracer (for example, lanthanum salts or peroxidase) penetrated the lateral intercellular space from the basal side but was stopped abruptly at the apical-most point of cell-cell contact — precisely where the zonula occludens is located — directly demonstrating that the tight junction, not the whole lateral membrane, is the diffusion barrier, and cementing the "occluding junction" concept as a distinct functional class from adherens junctions and desmosomes, which the tracer passed freely.

The molecular identification of claudins (Furuse, Tsukita, and colleagues, 1998) proved that a specific protein family, not merely membrane apposition, creates the paracellular seal. When claudin genes were transfected into cultured fibroblasts — cells that normally have no tight junctions at all — the fibroblasts formed strand-like structures at cell-cell contacts resembling tight junction strands in freeze-fracture images, and neighboring transfected cells showed reduced paracellular tracer permeability. This gain-of-function experiment is the direct molecular proof that claudins are necessary and sufficient to build the occluding strand, closing the loop opened by Farquhar and Palade's tracer studies thirty-five years earlier.

Cadherin-mediated adhesion was proven by classic cell-mixing and blocking-antibody experiments. Masatoshi Takeichi and colleagues (1977 onward) showed that cells expressing different cadherin subtypes, when mixed together in suspension, sort out into separate homogeneous clusters rather than mixing randomly — direct evidence that cadherins mediate homophilic, subtype-specific adhesion, a finding foundational to understanding how tissues maintain sharp boundaries during development (and foreshadows the deeper mechanistic treatment in Cell Adhesion Molecules). Adding function-blocking antibodies against E-cadherin to cultured epithelial cells caused the cells to lose their adherens junctions and disperse, while restoring cadherin function (by transfecting cadherin-negative cells with a cadherin cDNA) restored aggregation — a clean necessary-and-sufficient demonstration.

Example 1 — Diagnosing a barrier defect from a permeability assay. A gastroenterologist studies two segments of cultured intestinal epithelium, A and B, grown as confluent monolayers on porous filters (a standard transepithelial transport assay). Segment A has a transepithelial electrical resistance (TEER) of 1,000 Ω·cm² and shows almost no passage of a 4 kDa fluorescent dextran added to the apical chamber after 2 hours. Segment B has a TEER of only 50 Ω·cm² and shows substantial dextran leakage into the basal chamber over the same period. Immunostaining shows segment B has markedly reduced claudin-4 and occludin expression at cell-cell borders, while E-cadherin and desmoplakin staining are normal in both segments. Reasoning: TEER measures the resistance of the paracellular (and, to a lesser extent, transcellular) route to ion flow; a high TEER means ions cannot easily leak between cells, implying an intact tight junction seal. The dextran assay independently measures paracellular macromolecular leak. Segment B's combination of low TEER, high macromolecular leakage, and specifically reduced claudin/occludin (with normal adherens junction and desmosome markers) localizes the defect precisely to the occluding junction — the anchoring junctions holding the cells together mechanically are intact (the monolayer doesn't fall apart), but the paracellular seal has failed. This is the pattern seen in inflammatory conditions such as active inflammatory bowel disease, where cytokines (notably TNF-α and IL-13) downregulate specific claudins and internalize occludin, producing a "leaky gut" without necessarily destroying the epithelium's mechanical integrity.

Example 2 — Predicting the tissue consequence of a connexin mutation. Connexin 26 (encoded by the gene GJB2) is the dominant connexin isoform in the cochlear supporting cells of the inner ear, where gap junctions recycle K⁺ ions that flow into hair cells during sound transduction back to the endolymph, a recycling loop essential for maintaining the endocochlear potential (~+80 mV) that powers hair cell mechanotransduction. Mutations in GJB2 are the single most common cause of hereditary non-syndromic deafness, accounting for a substantial fraction of congenital hearing loss in many populations. Reasoning: Given the physiology, a loss-of-function GJB2 mutation should be predicted (before consulting any clinical data) to disrupt K⁺ recycling between supporting cells, causing K⁺ to accumulate in the space around hair cells rather than being shunted back to the endolymph. This would collapse the ionic gradient that hair cell mechanotransduction channels depend on, predicting sensorineural hearing loss without necessarily any accompanying skin, cardiac, or intestinal defect if the mutation only affects the specific connexin isoform. This matches the actual clinical genetics: most GJB2-associated deafness is non-syndromic (isolated to hearing) precisely because connexin 26 has a fairly restricted, though not exclusive, tissue distribution compared to more widely expressed connexins such as connexin 43 (whose loss-of-function mutations, by contrast, cause the multi-organ disorder oculodentodigital dysplasia, affecting eyes, teeth, digits, and heart conduction). This illustrates a general principle: the tissue-specific phenotype of a junction-gene mutation is predictable from which cell types express that specific isoform, not just from the general class (communicating junction) the gene belongs to.

Werner R. Loewenstein's electrophysiological dye-coupling and current-injection experiments in the 1960s were the decisive functional proof that some cell junctions create direct cytoplasmic continuity — years before anyone had cloned a connexin gene or resolved a gap junction structure at atomic resolution. Working first at Columbia and then at the University of Miami, Loewenstein built on the earlier morphological observations of "nexus" or gap-like contacts seen by electron microscopists, and asked a sharper, quantitative question: does current actually flow between these cells, and if so, through what kind of pathway? His double-microelectrode current-injection experiments and his dye-injection size-cutoff experiments together established both the existence and the approximate size limit of the intercellular channel, giving the field a functional definition of the gap junction that pre-dated, and was later beautifully confirmed by, the molecular cloning of connexins by Gilula, Beyer, Paul, and others in the 1980s.

Cell junction biology is not an abstract taxonomy — it is directly load-bearing in medicine, biotechnology, and agriculture. In dermatology, autoantibodies against desmoglein-3 (an anchoring-junction cadherin) cause pemphigus vulgaris, a blistering autoimmune disease in which the epidermis loses cell-cell cohesion and separates within the living cell layers; understanding the specific desmosomal target has enabled targeted immunosuppressive and B-cell-depleting therapies (rituximab) rather than blunt, broad immunosuppression. In gastroenterology, drugs and dietary interventions aimed at restoring claudin expression and tight-junction integrity are being explored for inflammatory bowel disease and for reducing gut permeability ("leaky gut") implicated in some autoimmune and metabolic conditions. In oncology, loss of E-cadherin-based adherens junctions is a defining molecular event in the epithelial-to-mesenchymal transition that allows carcinoma cells to detach from a primary tumor and invade — E-cadherin is functionally a tumor suppressor, and its transcriptional silencing (often via the Snail/Slug repressors) is used clinically as a prognostic marker of invasive potential. In cardiology, connexin 43 gap junctions coordinate the wave of electrical excitation across cardiac muscle; their disruption or mislocalization after myocardial infarction is a major contributor to life-threatening arrhythmias, and connexin-targeted antiarrhythmic peptides are an active area of drug development. In audiology, genetic screening for GJB2 (connexin 26) mutations is now a standard part of the workup for congenital hearing loss, directly informing genetic counseling. In agriculture and plant virology, because plasmodesmata regulate the spread of viruses, small RNAs, and signaling proteins between plant cells, engineering crops with altered plasmodesmal gating is an active strategy for breeding virus-resistant and stress-tolerant plant varieties. And in drug delivery, pharmaceutical scientists deliberately and transiently open tight junctions using claudin-binding peptides or specific claudin-modulating agents to enhance absorption of orally or nasally delivered drugs that would otherwise be blocked by the intestinal or nasal epithelial barrier — a direct engineering application of the same claudin biology that Farquhar, Palade, Furuse, and Tsukita first characterized.

Ask the tutor: How do claudin isoform combinations actually determine whether a tight junction is ion-selective, water-permeable, or nearly impermeable?