Tight junctions (TJ) consist of networks of linear contacts or sealing strands between adjoining cells, forming a permeability barrier to control the flow of molecules in the intercellular space. TJ strands are formed by a chain of transmembrane proteins on either side of the contacting membranes, maintained by both cis (intra-membrane) and trans (inter-membrane) interactions. The molecular architecture and mechanism of TJ strand assembly is just beginning to be elucidated. Claudins, a large family of transmembrane proteins, are the main components of the TJs. Multiple claudins (including 9, 11 and 14) are expressed in the inner ear and are required for normal hearing. We have shown that in OHCs claudin-14 and claudin-9/6 segregate into morphologically distinguishable TJ subdomains. This contrasts with the common view that most claudins co-assemble into heteromeric strands. A recent report on the crystal structure of claudin-15 has provided new insights into the architecture of TJs, revealing that the two extracellular loops arrange into a characteristic β-sheet fold allowing for specific cis-interactions between adjacent claudins. No trans-interactions, supporting the transcellular strand architecture, were identified. Our long term goal is to attain a molecular-level description of TJ assembly and a complete ultrastructural model of the TJ strand backbone in the inner ear. These findings will collectively provide the basis to understand how various claudins, and TJ accessory proteins expressed in hair cells and supporting cells, assemble to form the homotypic and heterotypic TJs of the inner ear. To study TJ molecular architecture, we have been using in silico structures and homology modeling/molecular simulation to build plausible 3D models as well as a combination of Molecular Dynamics simulations and mutagenesis approach to assess the formation of TJ strands from heterologously expressed, fluorescently tagged claudin in COS7 cells by fluorescence and freeze-fracture imaging. Based on the claudin-15 crystal structure and homology modeling/molecular simulation, we performed atomic-level conformation screen of over 2,500 claudin strand candidate structures and identified by energy minimization an optimal octameric unitary model (below) consisting of a double row of tightly associated claudins forming a single strand in each membrane. The model has several key supporting features including: tightly packed hydrophobic residues form a hydrophobic wall leaving most charged residues distributed on the loop region facing the aqueous environment;the conserved amino acid W137 stabilizes the interactions between claudins across from each other by π-πstacking (attractive noncovalent interactions between aromatic rings);the inter-molecular π-πstacking of conserved F146F147 result in concerted cis- and trans- interactions. To validate this model, several amino acid substitutions were designed and tested in molecular dynamics simulations before introducing them in a GFP-tagged claudin-15 cDNA vector for expression in COS7 cells to verify their implications for TJ strand formation. As predicted by our model, the replacement F146F147/LL abolished TJ strand formation, while F146F147/WW retained the ability to form TJ strands. Mutations in NMII have been linked to hearing loss in humans (Seri et al., 2000). We recently showed that bipolar NMII filaments integrate with peri-junctional actin to form a belt of muscle-like sarcomeres encircling the apex of hair- and supportingcells (Ebrahim et al., 2013). The sarcomeres appear to be linked across adjacent cells, forming a transcellular, contractile network. This novel sarcomeric network, which occurs universally across different epithelial tissues, provides a well-defined target to investigate the role of NMII in junctional homeostasis. We now plan to investigate the ultrastructural relationship of the sarcomeric belt with other components of the AJC, its mechanisms of assembly and turnover, isoform-specific properties, and cellular functions. We have preliminary data to suggest that in the organ of Corti both the expression level and the characteristic sarcomeric pattern of the predominantly expressed NMIIC appear to diminish from early (P0-P4) to later postnatal (>P7) stages, correlating with early postnatal development of the organ. Additionally, we have data showing a significant redistribution and/or increase in expression levels of NMIIA at the AJC of NMIIC-/- mice, suggesting that the absence of an auditory phenotype in these mice may be due to a degree of compensation by NMIIA. However, epithelia from NMIIC-/- mice show significant differences in cell packing and geometry indicative of reduced junctional tension (in the organ of Corti these present as planar cell polarity defects). Collectively these data suggest that compensation by NMIIA is incomplete, likely due to differences in isoform-specific kinetic and biochemical properties. Other functional data from model cell lines and intestinal epithelia show the specific recruitment of NMIIA, the most kinetically active isoform, to sites of epithelial rearrangement during apoptosis. We are also currently investigating differential inter-cellular isoform distribution patterns that we hypothesize confer differential dynamic properties across heterotypic junctions. Finally, we have the first evidence that we know of, for the existence of NMII heterofilaments (bipolar filaments comprising multiple isoforms) in vivo. We postulate that such heterofilaments provide a regulatory mechanism to further modulate the resultant kinetic properties of the sarcomeric belt in different functional contexts.
Showing the most recent 10 out of 24 publications
