br Electron crystallographic structures of connexins The
Electron crystallographic structures of connexins The structures of gap junctions are good targets for electron microscopy because they exhibit the characteristic features of self-assembly. Early studies examined the three-dimensional (3D) structures of native gap junction pain and inflammation from rodents using electron crystallography, and the proposed movement for channel function was a twist and tilt of each subunit [18,19] (Figure 1a). While this was supported by an atomic force microscopy (AFM) study , a functional study with electrophysiology provided the alternative interpretation that each subunit can move individually, resulting in voltage-dependent closure . Split gap junctions comprising undocked hemichannels isolated from rodents were reconstructed in 3D, showing six protrusions in the extracellular domain . Recombinant Cx43 channels were also studied by electron crystallography at 7 Å resolution, revealing the separate transmembrane helices . The functional state of this structure was unclear, however, as the pore exhibited a widely open pathway when oleamide, a drug known to be a gap junction blocker , was added to the crystallization buffer. The evidence of a density in the pore was observed in the 3D reconstruction of the Cx26M34A mutant (Figure 1b), which exhibits functional, but decreased, permeability and conductance compared with the wild-type channel [,26, 27, 28]. It has been suggested that the N-terminus of Cx26 physically blocks the pore [,29], but this model remains unproven because this type of closure has yet to be shown in any wild-type connexinb.
X-ray crystal structures of Cx26 The atomic structure was first determined for wild-type Cx26 using X-ray crystallography  (Figure 1c). This structure provided answers to many questions. The transmembrane helix assignment was a controversial point proposed by multiple models [30,31], but was explicitly solved. The two extracellular loops have three intramolecular disulfide bonds, forming an anti-parallel beta sheet in E2 and a short alpha helix in E1. The extracellular docking surface between apposed hemichannels was flatter than expected for engagement by the anti-parallel β-strands of the E1 and E2 loops . The N-terminal arrangement in the pore cavity assumes the conformation of a funnel formed by the N-terminal helix. The helical arrangement of the connexin N-terminus is consistent with the proposed model based on NMR structures of the N-terminal peptides [33, 34, 35]. The conformation of the X-ray structure was interpreted to be an open form as there is no obstacle in the pore pathway. This model does not contain the Met1 residue, however, and therefore an alternative interpretation has been presented based on a molecular dynamics (MD) simulation study that the average equilibrated structure generated by MD simulations is more likely to be an open structure of Cx26, and the N-terminus of Cx26 is less stable than observed in the X-ray structure . X-ray crystal structures of Cx26 in the presence and absence of Ca2+ in the crystallization buffer were recently reported . Comparison of the two conditions suggests that Ca2+ ions are coordinated by Glu42, Gly45, and Glu47, which are closely positioned residues in the pore pathway close to the extracellular side. The Ca2+-bound and Ca2+-free structures of Cx26 are mostly identical, and a conformational change, for example, a Ca2+-induced subunit rotation  (Figure 1a), was not observed. The two structures exhibited different electrostatic surface potentials in the pore, suggesting that the electrostatic effects function as a switch to regulate the charge selectivity of permeants depending on Ca2+ binding (Figure 1d). The N-terminus and most of the cytoplasmic domains were not resolved in these structures. It will be of interest in the future to see the N-terminal distribution in the presence and absence of Ca2+ in complete models of connexins. The closure induced by Ca2+ may be similar to the loop-gating evoked in a voltage-dependent manner . While the N-terminal portion, specifically the charge distribution, is considered to function as a sensor for voltage-dependent channel closure , it remains to be elucidated if channel closure by electrostatic distribution also accounts for the voltage-dependent channel closure.