Structural analysis of the MD

Structural analysis of the MD-open state unveils that the transmembrane domain features a symmetric organization of the pore-lining helices M2 similar to the semi-open structure but with a local asymmetry at the constriction point, which is critical for chloride permeation. The atomistic simulations show that such an asymmetry originates from the occurrence of local structural defects, e.g., the loss of α-helical H-bonding in a single M2 helix that stabilizes a flipped proline at position −2′, which facilitates ion permeation by hindering optimal packing of residues at the constriction point that reduces the hydrophobicity of the pore. This observation, along with the computational electrophysiology results, supports the conclusion that a fully symmetric organization of the −2′ prolines as the one seen in the semi-open cryo-EM structure is inconsistent with the physiologically active state of the GlyR α1. Whether the symmetric organization of the semi-open pore is a structural biology artifact due to averaging out of electron densities during molecular reconstruction, or is physiologically relevant and consistent with (fast) desensitization, is presently unclear and requires further investigation. Also, since cationic c-di-AMP mg feature charged or polar residues at the constriction point (Cymes and Grosman, 2016, Sine et al., 2010), which are likely to promote a more symmetric organization of the gate due to electrostatic interactions, we predict that a local asymmetry of the pore is likely to be functionally relevant in anionic channels where the pore-lining residues at the gate (i.e., at position −2′) are hydrophobic. The characterization of a structurally stable and ion-permeable channel at physiological conditions (i.e., MD-open) allows for an original interpretation of structures in anionic pLGICs. Despite the distinct permeability of MD-open versus semi-open measured by computational electrophysiology, the overall configuration of the ion channel in these structures is strikingly similar (see Figures 3B–3D and Table S1). This observation thus suggests that previously characterized open-channel forms in GLIC at pH 4 (Bocquet et al., 2009), GluCl with glutamate and ivermectin bound (Hibbs and Gouaux, 2011), and GlyR with glycine and ivermectin bound provide reasonable models for the active state, which might be accurate enough for exploring gating or drug-discovery purposes but are likely to fail in capturing the details controlling ion conductance and selectivity due to the absence of local asymmetries in the structural organization of the pore. If so, based on the conclusions that MD-open provides a reasonable representation of the physiologically active state and that the wide-open structure is likely to be a cryo-EM artifact, the comparative analysis of the experimental structures of the anionic channels GlyR, GABAAR, and GluCl by HOLE (Figure 5) reveals the existence of three major clusters: (1) a closed-pore configuration constricted at position 9′ consistent with the resting state (GlyR α3 [Huang et al., 2015] and GlyR α1 [Du et al., 2015] with strychnine bound, and GluCl apo [Althoff et al., 2014]); (2) a closed-pore configuration constricted at position −2′ most consistent with desensitization (GlyR α3 with glycine and AM3607 bound [Huang et al., 2017a], GlyRα3 with glycine, AM3607, and ivermectin bound [Huang et al., 2017b], and GABAAR in complex with benzamidine [Miller and Aricescu, 2014]); and (3) an open-pore configuration consistent with the active state (GlyR α1 with glycine and ivermectin bound [Du et al., 2015], and GluCl with L-Glu and ivermectin bound [Hibbs and Gouaux, 2011] or ivermectin alone [Hibbs and Gouaux, 2011]). Interestingly, this annotation supports the conclusion that pore constriction or the gate is located differently in resting versus desensitized channels, which is consistent with recent mutagenesis and kinetic experiments (Gielen et al., 2015). Also, it suggests that pore closing by desensitization versus channel deactivation or ungating (Martin et al., 2017) would involve the polar reorientation of the pore-lining helices M2 in opposite directions, i.e., an inward untilting to straighten for deactivation, and further tilting in the outward direction for desensitization.