Regulation and roles for claudin-family tight junction proteins


  • Mary K. Findley,

    1. Division of Pulmonary, Allergy and Critical Care Medicine, Emory University School of Medicine, Atlanta, Georgia
    Search for more papers by this author
  • Michael Koval

    Corresponding author
    1. Division of Pulmonary, Allergy and Critical Care Medicine, Emory University School of Medicine, Atlanta, Georgia
    2. Department of Cell Biology, Emory University School of Medicine, Atlanta, Georgia
    • Division of Pulmonary, Allergy, and Critical Care Medicine, Emory University School of Medicine, Whitehead Biomedical Research Building, 615 Michael St., Suite 205, Atlanta, GA 30322
    Search for more papers by this author
    • Tel: +404-712-2976. Fax:+404-712-2974


Transmembrane proteins known as claudins play a critical role in tight junctions by regulating paracellular barrier permeability. The control of claudin assembly into tight junctions requires a complex interplay between several classes of claudins, other transmembrane proteins and scaffold proteins. Claudins are also subject to regulation by post-translational modifications including phosphorylation and palmitoylation. Several human diseases have been linked to claudin mutations, underscoring the physiologic function of these proteins. Roles for claudins in regulating cell phenotype and growth control also are beginning to emerge, suggesting a multifaceted role for claudins in regulation of cells beyond serving as a simple structural element of tight junctions. © 2009 IUBMB IUBMB Life, 61(4):431–437, 2009


Epithelial and endothelial cell monolayers form barriers to separate organs into functional subcompartments. This distinct compartmentalization and isolation from the external environment is crucial for the function of organ systems in multicellular organisms. Cells control these selective barriers by regulating the movement of water, ions and proteins across the monolayer, thus generating polarity of cellular structure and function1. The movement of ions and molecules between cells is known as paracellular permeability and is regulated by sites of cell-cell contact known as tight junctions, a complex of transmembrane and peripheral proteins that is tethered to the cytoskeleton. While tight junctions require the coordinated activity of several different proteins, the specificity of tight junction permeability is regulated by transmembrane proteins known as claudins2, a name derived from the Latin claudere, which means “to close.” The primary role of claudins is the regulation of paracellular selectivity to small ions. Claudins form what are functionally equivalent to charge selective pores which promote specific ion permeability, although there are also examples of claudins which more generally increase rather than restrict paracellular permeability as well3. The mechanisms which regulate claudin assembly and function are just beginning to be elucidated. Here, we summarize current progress in understanding how claudins control paracellular permeability.


To date, 23 distinct human claudins have been identified4. Claudins are 20–27 kDa transmembrane proteins which span the bilayer four times, where the N- and C- termini are oriented towards the cytoplasm and there are two extracellular loop domains5. For at least two claudins, claudin-10 and claudin-18, alternative splicing can also generate claudin isoforms with different properties6, 7.

Little is known about function of the relatively short, 7-amino acid, N-terminus. The cytoplasmic C-terminus sequence varies considerably in length (from 21 to 63 residues) and sequence between isoforms. All claudins have C-terminal PDZ binding motifs that enables direct interaction with tight junction cytoplasmic proteins such as ZO-1,-2, and -3, multi-PDZ domain protein (MUPP)-1 and PALS-1 associated TJ protein (PATJ). In particular, interactions with the cytoplasmic scaffolding proteins ZO-1 and ZO-2 indirectly link claudins to the actin cytoskeleton which stabilizes the tight junction and is required to maintain their permeability characteristics8.

The first extracellular loop of claudins is ∼52 residues and influences paracellular charge selectivity9. The first extracellular loop domain has a signature motif [Gly-Leu-Trp-x-x-Cys-(8–10aa)-Cys] which is highly conserved and is found in closely related proteins, such as epithelial membrane proteins (EMPs, PMP22, MP20)4. The cysteines in this signature motif may form an intramolecular disulfide bond to stabilize protein conformation; however, this remains to be determined. The second extracellular loop of claudins is shorter (16–33 residues) and less well characterized, although molecular modeling suggests the second extracellular loop is folded in a helix-turn-helix motif which participates in claudin-claudin interactions10.


Freeze fracture immunogold analysis has shown that tight junctions are composed of a network of strands of which claudins are the major transmembrane constituent11. These strands are roughly the diameter of a gap junction channel, which has led to the suggestion that claudins are organized into a basic hexameric unit, similar to connexins in gap junctions. Consistent with this, claudins have been shown to form ladders of stable oligomers in native gel electrophoresis increasing in molecular mass up to a hexamer12, 13. However, to date little is known about the mechanisms which regulate claudin oligomerization and whether this is a process which occurs prior to assembly into tight junction strands.


Signature di-cysteine palmitoylation motifs are conserved throughout the claudin protein family and were found to enhance incorporation of claudins into tight junctions14. Palmitoylated claudins more efficiently portion into detergent resistant membranes as compared with nonpalmitoylated mutants, suggesting lipid rafts play a role in tight junction assembly15. Tetraspanins, such as CD9 and CD81, are also palmitoylated which is critical to their function as membrane chaperones and organizers of higher order membrane domains16. A link between tetraspanins and claudins is suggested by the demonstration that CD9 directly interacts with claudin-1 in a cholesterol sensitive manner17. Whether CD81 plays a role in regulating claudin assembly is not yet known, however, CD81 acts as a co-receptor with claudin-1, -6, or -9 to facilitate Hepatitis C binding and entry into cells18. Also, it is unknown where claudins oligomerize however, as palmitoylation and tetraspanin oligomerization occurs predominantly in the Golgi apparatus, this suggests the Golgi apparatus as a potential location for claudin oligomerization. This would be analogous to the connexin family of gap junction proteins which oligomerize in the Trans Golgi Network4.

Claudin-Claudin Interactions

Epithelia and endothelia typically express multiple claudin isoforms in a tissue specific manner, suggesting the potential of claudin isoforms to intermix in tight junction strands. The interactions between different classes of claudins have the potential to control barrier permeability through the formation of tight junctions composed of multiple claudins. Claudins interact in two different ways: laterally in the plane of the membrane (heteromeric interactions) or head to head binding between adjacent cells (heterotypic interactions). Very little is known about claudin heteromeric binding. Claudin homomultimers composed of up to six monomers have been observed biochemically12, 13, 19 and lateral interactions of claudin-5 within cells has been detected using fluorescence resonance energy transfer (FRET)20.

Cells with claudin-null backgrounds have helped define specificity in claudin heterotypic interactions12, 19, 21. Using this approach, heterotypic and heteromeric compatibility have been found to be determined by distinct mechanisms. For example, claudin-3 and claudin-4 are heteromerically compatible when expressed in the same cell, however, they do not heterotypically interact despite having extracellular loop domains that are highly conserved at the amino acid level21. By contrast, claudin-3 heterotypically interacts with other claudins, including claudin-1 and claudin-512, 19, 21. However, a single point mutation in the first extracellular domain of claudin-3 to convert Asn(44) to the corresponding amino acid in claudin-4 (Thr) produced a claudin capable of heterotypic binding to claudin-4 while still retaining the ability to bind to claudin-1 and claudin-5. Thus far, the only claudin found to be capable of heterotypic binding appears to be claudin-3, suggesting that heterotypic interactions may be rare and that homotypic claudin interaction may be more common. Whether this is the case will require analysis of claudins beyond the handful already studied in detail.

Claudin-Occludin Interactions

Occludin is another tetraspan transmembrane protein associated with tight junctions22. For many years occludin was assumed to be the protein which formed the physical basis of the tight junction barrier; however, transgenic occludin deficient mice are viable and have normal barrier function23, which ultimately lead to the identification of claudins as the barrier forming component of tight junctions24. Although occludin-deficiency does not impair barrier function, peptides that mimic the occludin extracellular loop domains can disrupt epithelial barrier function by enhancing junctional disassembly suggesting that occludin associates with tight junctions in a regulatory capacity25. Regulation of barrier function is also suggested by studies where cells transfected with a C-terminal truncated occludin construct show increased paracellular flux of uncharged molecules26. In this case, transepithelial resistance was not altered, suggesting the increase in paracellular flux was due to increased turnover (or treadmilling) of the tight junction strands1, most likely due to a disruption of occludin-scaffold protein interactions.

Recently, a claudin-1 extracellular loop peptide was also found to disrupt tight junction assembly and function27. Interestingly, the claudin-1 peptide was found to bind to occludin, suggesting a direct interaction27. The structural basis for this interaction as well as whether this reflects a potential heterotypic or heteromeric interaction between claudins and occludin remains to be determined.

Claudin-Scaffold Protein Interactions

Claudin-ZO protein interactions are required for tight junction assembly8. This was demonstrated using cultured epithelial cells in where expression of both ZO-1 and ZO-2 was suppressed. In ZO-1/ZO-2 null cells, claudins failed to localize to tight junctions and the cells had poor barrier function. However, transfection with either ZO-1 or ZO-2 enabled proper claudin localization and function8. Thus, ZO-1 and ZO-2 have overlapping function, which suggests that linking claudins to the actin cytoskeleton through these scaffold proteins is essential for tight junction formation.

As most claudins have a conserved C-terminal YV domain, they are likely to interact with zona occludens (ZO) -1, -2, and -3, as demonstrated for claudin-1 through claudin-828. Claudin-1, -5, and -8 have also been shown to interact with MUPP1 and claudin-1 can bind PATJ, underscoring the multiplicity of scaffold proteins capable of interacting with claudins29, 30. Claudin-scaffold interactions are also specified by other motifs in the C-terminal domain. This was demonstrated using a series of tail swap constructs, where tight junction strand count and morphology correlate with the claudin tail motif, whereas ion permeability characteristics depended on the remainder of the protein, most notably, the extracellular loop domains31.


Claudin phosphorylation is another mode of regulating paracellular permeability. Phosphorylation has been linked to both increases and decreases in tight junction assembly and function. For instance, Protein Kinase A-mediated phosphorylation has been shown to decrease assembly of claudin-3 into tight junctions32, yet is necessary for claudin-16 assembly and function33. Claudin-1 and Claudin-4 phosphorylation by Protein Kinase C-theta is required for assembly into intestinal epithelial tight junctions34. Conversely, PP2A phosphatase activity has been shown to decrease claudin-1 phosphorylation and barrier function, although in this case, ZO-1 phosphorylation was also reduced, so this may be reflect altered interactions between claudin-1 and ZO-135. Consistent with the ability of a phosphatase to impair interactions with ZO-1 and the cytoskeleton, claudin-1 detergent solubility is enhanced by PP2A.

Conversely, phosphorylation via myosin light chain kinase (MLCK) and rho kinase is more frequently associated with tight junction disassembly and increased paracellular permeability associated with inflammation and in response to environmental stresses such as chronic alcohol abuse36, 37. Although claudin-5 phosphorylation correlates with increased MLCK activity37, the relative importance of direct claudin phosphorylation by these kinases vs. increased junction turnover induced by MLCK and/or rho kinase38, 39 remains to be determined.

Claudin phosphorylation associated with tight junction disassembly is also enhanced by EphA2, which is recruited to bind to claudin-4 by forming a complex with ephrin40. A direct role for EphA2 in interfering with claudin assembly is intriguing, given the role of ephrins in inducing epithelial to mesenchyme transition (EMT)41, a process where tight junction protein expression is downregulated42, 43.


Although the network of tight junctions is constantly remodeled, the different components have different stability within the strands themselves44, 45. Based on fluorescence recovery after photobleaching (FRAP), the majority of claudins are stably integrated into tight junctions, reflecting the strength of intercellular binding and association with the cytoskeleton44, 45. In contrast, occludin and ZO-1 are more dynamic45. The mobile fraction of occludin is nearly three times greater than that of claudin. ZO-1 is highly dynamic as well, although instead of diffusing along the plane of the membrane, it dissociates from tight junctions and diffuses into the cytoplasm. This is consistent with models where ZO-1 interactions with occludin and claudins are required to stabilize these proteins in junctions. In fact, the rate of diffusion for the rapidly mobile fractions of claudins and occludin are comparable45.

Tight junctions are also remodeled at a more macroscopic level through strand breaks and reformation44. Clathrin-mediated endocytosis plays an important role in this process46, 47. Claudins are internalized by a unique mechanism, where the tightly opposed membranes of the tight junction are endocytosed together into one of the adjoining cells, a mechanism also used for gap junction internalization48. During internalization, the claudins separate away from occludin, JAM and ZO-1 and generate claudin-enriched vesicles, which has the potential to regulate the claudin composition of tight junctions.


In response to pathogens and during inflammation, increased claudin internalization can result in a deleterious increase in paracellular permeability. Toxins such as E.coli cytotoxic necrotizing factor-1, H. Pylori associated factors and Clostridium perfringens enterotoxin (CPE) has been found to induce claudin internalization49, 50. The effect of CPE is due to a direct interaction with the second extracellular loop domain of claudin-3 and claudin-4, which can also inhibit tight junction reformation in addition to enhancing endocytosis51.

Host factors and cytokines can also influence tight junction turnover and claudin expression, presumably in part to accommodate neurophil migration across epithelial barriers52. For instance, interferon-γ increases claudin endocytosis and tight junction permeability53. In addition to interferon-γ, other inflammatory cytokines such as tumor-necrosis factor (TNF)-α and interleukin (IL)-13 downregulate claudins and induce a marked increase in paracellular permeability by epithelial cells in culture39, 54. The notion of pathogenic consequences for claudin misregulation are underscored by the findings that inflammatory bowel disease (IBD) causes a strong increase in the expression of claudin-2 while normal colons have low levels of claudin-239, 55. A role for altered claudin expression in predisposition to IBD is also suggested by studies of JAM-A- deficient mice, which show increased claudin-10 and claudin-15 expression in the colonic mucosa and increased sensitivity to dextran sulfate sodium- induced colitis56. Alterations in claudin expression has also correlated with pathological failures in barrier function in other systems. For instance, in chronic alcohol abuse, the lung barrier is compromised which is accompanied by decreased claudin-1 and claudin-7 and increased claudin-5 expression38, 57. In these examples, the correlation of increased claudin content reducing the tight junction barrier is consistent with the notion of “leaky” claudins3, although the mechanism underlying this remains obscure at present.


The function of specific claudins can be elucidated from mutations leading to human diseases, many which have been mimicked by mouse knockout models. For instance, mutations in human claudin-1 cause neonatal icthyosis and sclerosing cholangitis, consistent with high levels of claudin-1 expression in the skin and in cholangiocytes of the bile duct58 and claudin-1 null mice have a neonatal lethal phenotype due to skin permeability defects58, 59. Claudin-14 mutations are associated with nonsyndromic deafness in humans and mice60, which suggests a role for claudin-14 in the cation-restrictive barrier that maintains the normal endolymph ionic concentration that bathes the outer hair cells of the cochlea in the ear61.

Mutations in claudin-16 (also known as paracellulin-1) lead to recessive renal hypomagnesaemia62 and claudin-19 mutations show a similar deficiency in regulating magnesium reabsorption63. Although it was initially hypothesized that claudin-16 directly regulated paracellular magnesium permeability, in fact claudin-16 interacts with claudin-19 to increase paracellular selectivity to sodium64. This leads to a model where a gradient of claudin-16/claudin-19 expression in the thick ascending limb of the kidney establishes a NaCl concentration gradient which creates a lumen positive transepithelial diffusion potential that drives paracellular magnesium reabsorption, as opposed to a direct change in magnesium permeability, underscoring that role for claudins in regulating ion permeability can be indirect65.


Epithelial-mesenchymal transition (EMT) is a process where epithelial cells differentiate towards a more fibroblastic phenotype. EMT is involved in normal embryonic development and repair of epithelial injury; however, it is also implicated in carcinogenesis66, 67. Consistent with a decrease in epithelial gene expression, EMT is associated with changes in claudin expression and regulation. For instance, overexpression of the transcription factor SNAIL induces EMT and downregulates transcription of tight junction proteins including claudins and occludin42, 43. This is due in part to E-box motifs in their promoter where SNAIL can bind and repress transcription43, 68. SNAIL has also been shown to downregulate claudin-1 without affecting transcription69, however, this may be a cell specific effect, because repression of claudin-1 transcription has been observed in other systems, suggesting that it might also regulate claudin translation68.

Histological screens of human tumors have revealed changes in claudin expression associated with tumor phenotype. This is consistent with EMT as a contributing factor in tumorigenesis, because tight junctions are often altered in human carcinomas70. Some trends are beginning to emerge for some claudins. For instance, tumors seem to be generally associated with decreased claudin-7 expression71, 72. However, there are studies linking either decreased73 or increased74 claudin-1 expression by tumors. Claudin-3 and claudin-4 seem to be generally upregulated, particularly in aggressively metatastic tumors, although this is not always the case72, 75. The physiologic role for upregulated claudin-3 and -4 in some human tumors is not fully understood, but this provides a fortuitous target, because it enables drugs based on Clostridium toxin which binds to the extracellular loop of these claudins to be developed as potential anti-tumor agents76–78. As the structure of the loop binding domain of Clostridium toxin is now known, this could potentially enable other claudin specific interacting proteins to be designed79.


Functional roles for claudins in regulating tissue barrier function have are becoming more solidly established, with claudin heterogeneity having both direct and indirect influences on paracellular ion permeability. There also remain several open questions in understanding how claudins are regulated. Some of the critical factors required for claudin assembly into tight junctions, such as roles for occludin and scaffold proteins are now known, but the mechanism of claudin assembly is not well understood. In particular, the intracellular compartments where claudins are assembled have not been identified and determinants which influence heteromeric and heterotypic claudin-claudin interactions have just begun to be elucidated. This last point is particularly critical to define how claudin multiplicity can influence paracellular permeability.

Finally, correlations between claudin expression and metastasis raise the intriguing notion of roles for claudins in cell growth control. Whether changes in claudin expression are a cause or a consequence of carcinogenesis is not clear. The observed discrepancies in claudin expression in tumorigenesis could be related to tissue-specific differences in claudin function or tissue microenvironmental features. Nonetheless, a role for claudins in regulation of cell growth would be comparable to roles found for other cell adhesion molecules in cancer, such as cadherins. Whether this is the case, remains to be determined, but this suggests that claudins play a multiple roles beyond acting as a simple paracellular barrier.


Supported by Emory Alcohol and Lung Biology Center (NIH P50-AA013757), NIH HL083120 (MK) and NIH T32GM0 08367 (MKF).