Computational Assessment of Different Structural Models for Claudin-5 Complexes in Blood–Brain Barrier Tight Junctions

Abstract

The blood–brain barrier (BBB) strictly regulates the exchange of ions and molecules between the blood and the central nervous system. Tight junctions (TJs) are multimeric structures that control the transport through the paracellular spaces between the adjacent brain endothelial cells of the BBB. Claudin-5 (Cldn5) proteins are essential for TJ formation and assemble into multiprotein complexes via cis-interactions within the same cell membrane and trans-interactions across two contiguous cells. Despite the relevant biological function of Cldn5 proteins and their role as targets of brain drug delivery strategies, the molecular details of their assembly within TJs are still unclear. Two different structural models have been recently introduced, in which Cldn5 dimers belonging to opposite cells join to generate paracellular pores. However, a comparison of these models in terms of ionic transport features is still lacking. In this work, we used molecular dynamics simulations and free energy (FE) calculations to assess the two Cldn5 pore models and investigate the thermodynamic properties of water and physiological ions permeating through them. Despite different FE profiles, both structures present single/multiple FE barriers to ionic permeation, while being permissive to water flux. These results reveal that both models are compatible with the physiological role of Cldn5 TJ strands. By identifying the protein–protein surface at the core of TJ Cldn5 assemblies, our computational investigation provides a basis for the rational design of synthetic peptides and other molecules capable of opening paracellular pores in the BBB.

Claudin-10b cation channels in tight junction strands: Octameric-interlocked pore barrels constitute paracellular channels with low water permeability

“…Nevertheless, not only the detailed interfacial conformations, but also the precise orientation of chains relative to each other could slightly but significantly differ, for instance to close the pore region for barrier-forming claudins. Alternative non-JDR arrangements suggested for oligomers of classic claudins [18] , [20] , [21] , [22] might be rather involved in branching of strands.…”

“…Different models of the claudin backbone of TJ strands with embedded paracellular channels have been reported [10] , [14] , [15] , [18] , [19] , [20] , [21] , [22] , [23] , [41] . The models differ regarding the claudins investigated, the input of (structural and functional) in vitro/in situ data and the in silico methodology on which they are based.…”

“…The molecular mechanism of claudin oligo/polymerization resulting in strand and channel formation is largely unclear. Different architectural models have been suggested for CLDN15 and other classic claudins [10] , [14] , [15] , [18] , [19] , [20] , [21] , [22] , [23] . The joined double rows (JDR) model of CLDN15 suggested by Suzuki et al [14] , refined [24] , [25] and expanded to other classic claudins based on sequence homology [15] , [26] , is the one that is supported most by experimental and modeling data [10] .…”

The Impact of Pathogenic and Artificial Mutations on Claudin-5 Selectivity from Molecular Dynamics Simulations

“…3, and compared with analogous profiles for the WT Cldn5 pores as calculated by us in Ref. 24 (data reproduced with permission from ACS). Results show that the two models remain permeable to water after the insertion of the mutation, which induces only a small FE barrier in proximity of the mutated residues in Pore I (Fig.…”

“…By identifying the protein-protein surface at the core of TJ Cldn5 assem-blies, our computational investigation provides a step forward 139 in the establishment of a robust basis for rational design of 140 small molecules capable of modulating the BBB's paracellu-141 lar permeability in a safe and reversible manner. Pore II), as described in our previous work (24). In addition subunit with the help of UCSF Chimera (26), and the initial 156 configurations of the R side-chains were selected from the progressive release of positional restraints, were performed to achieve a stable starting configuration for US runs.…”