Tight junctions (TJ) consist of networks of linear sealing strands between adjoining cells, forming cell-cell contacts in epithelial cell sheets that regulate the ion permeation through the intercellular space. Claudins, a large family of transmembrane proteins, are the main components of TJs. A crystal structure of claudin-15 monomers has been solved by X-ray crystallography and a linear double stranded model has been proposed based on the rigid crystalline contacts. However, it has remained largely unknown what the crucial domains involved in the polymerization that forms the elaborate TJ networks are, and how claudins interact with other TJ proteins to achieve barrier and pore functions. Multiple claudins are expressed in the inner ear and are required for normal hearing. We have previously shown that in outer hair cells (OHCs) claudin-14 and claudin-9/6 segregate into morphologically distinguishable TJ subdomains. While the TJ strands in the claudin 14 domain are mostly parallel and unbranched, in the claudin 9/6 rich domain they form a highly interconnected network. This contrasts with the common view that most claudins co-assemble into heteromeric strands. Our long term goal is to attain a molecular-level description of assembly and dynamics of the homotypic and heterotypic TJs of the inner ear. We have developed several methods of producing high-resolution freeze fracture replicas to image TJ strands in their native membrane environment. The first method utilizes liquid crystal metal alloys to circumvent the intrinsic tendency of metals to crystalize and form a granular texture. We further developed this technique to not require any heavy metal. By using phase-plate based phase-contrast field emission TEM with direct electron detection we were able to produce high contrast and high resolution freeze fracture images of pure carbon replicas. As before, we were able to confirm the claudin oligomer has a double lined structure. Analysis of the TJ network using carbon replicas also showed new details on claudin strand branching and intersections. Branching sites often consisted of a triangular structure where the double lines are continuous between all strands; in other cases the strands appear to be in close proximity but with no direct interactions suggesting simple strand encounters. Live cell imaging of GFP-tagged claudin showed the TJ network form these intersections by one claudin strand colliding with another, often maintaining contact for extended times suggesting some forms of lateral interactions. We also observe end-to-end annealing of TJ strands. We have not been able to observe clear TJ strand branching. These novel observations provide insight into how claudin strands form the complex and dynamic TJ branched network required for proper paracellular barrier function. To identify and characterize interacting claudin interfaces we use a combination of computational methods and microscopy. Previous computationally determined docking of claudin 15 monomers had shown a clear interface with an approximate 15 degrees rotation from the model based on the X-ray structure. This interface uses the critical residues Ser67 and Glu157 we identified previously. We further validated this interface by swapping the Ser67 and Glu157. In single mutation controls the Glu157Ser failed to form TJ strands, however the double mutant Ser67Glu and Glu157Ser rescued strand formation providing evidence for the interaction between the two residues. We performed curvature analysis of TJ strands to better understand the flexibility/rigidity of claudins. These results showed claudin strands exist in a large distribution of curved configurations with no preferred handedness. The curvature distribution measured suggest that at the monomer level the angles formed between monomers are up to 25o in the axial direction. Live cell imaging confirmed that the strands are very dynamic, changing curvature on the order of seconds. The curvature data supports that the computationally determined interface and the linear crystalline interface are the same oligomerization sites at different curvatures. Fully understanding TJ dynamics and structure can provide a basis to understand how different claudins interact, how other TJ proteins associate with claudins and ultimately reveal potential targets for TJ modulation. We had previously reported that the apical junction in the cochlear epithelium is organized in a fashion similar to muscle sarcomeres, in that non-muscle myosin II (NMII), constituting the hair cell's equivalent of the M-line, alternates with actin and alpha-actinin, which are localized at the hair cell's equivalent of the Z-disk. We further demonstrated that, like muscle sarcomeres, this sarcomeric NMII network is also contractile, generating tensions both along the junctions and also across cell-cell boundaries, so that ultimately forces are propagated across the epithelia. This mechanism likely drives morphogenetic changes during development. We have since determined that this sarcomeric NMII network is ubiquitous across epithelial apical junctions, and are using the dynamic gut epithelia as a model system to further investigate the mechanisms of NMII isoform-specific function. We have further uncovered the that NMII also organizes as linear sarcomeric arrays across the mid-apical actin network, sometimes lining up across multiple cell boundaries. These observations imply a potential integration/coordination between the chains of sarcomeres that are part of the mid-apical network and those that are part of the trancellular junctional network to form a putative active transepithelial contractile system. We also find that NMII isoforms compete to localize at apical junctions, and this competition is possibly an underlying mechanism for the differential distribution of NMII isoforms. The association of NMII isoforms with various pathologies, including deafness, highlight the importance of these new findings on NMII organization, dynamics and function.
Showing the most recent 10 out of 24 publications
