Gating of the proton-gated ion channel from Gloeobacter violaceus at pH 4 as revealed by X-ray crystallography

Abstract

Cryoelectron microscopy and X-ray crystallography have recently been used to generate structural models that likely represent the unliganded closed-channel conformation and the fully liganded open-channel conformation of different members of the nicotinic-receptor superfamily. To characterize the structure of the closed-channel conformation in its liganded state, we identified a number of positions in the loop between transmembrane segments 2 (M2) and 3 (M3) of a proton-gated ortholog from the bacterium Gloeobacter violaceus (GLIC) where mutations to alanine reduce the liganded-gating equilibrium constant, and solved the crystal structures of two such mutants (T25'A and Y27'A) at pH ~4.0. At the level of backbone atoms, the liganded closed-channel model presented here differs from the liganded open-channel structure of GLIC in the pre-M1 linker, the M3-M4 loop, and much more prominently, in the extracellular half of the pore lining, where the more pronounced tilt of the closed-channel M2 α-helices toward the pore's long axis narrows the permeation pathway. On the other hand, no differences between the liganded closed-channel and open-channel models could be detected at the level of the extracellular domain, where conformational changes are expected to underlie the low-to-high proton-affinity switch that drives gating of proton-bound channels. Thus, the liganded closed-channel model is nearly indistinguishable from the recently described "locally closed" structure. However, because cross-linking strategies (which could have stabilized unstable conformations) and mutations involving ionizable side chains (which could have affected proton-gated channel activation) were purposely avoided, we favor the notion that this structure represents one of the end states of liganded gating rather than an unstable intermediate.

Keywords: allostery; electrophysiology.

Fig. 1.

Loss-of-function mutations in the M2–M3 loop of GLIC. (A) Simplified, general kinetic scheme that shows conformational changes and ligand association–dissociation reactions occurring as distinct, separate steps. C, O, and D denote the closed- open-, and desensitized-channel conformations, respectively. n denotes the number of activating ligands (L) bound to the channel in the fully liganded state. In the case of GLIC, n remains unknown. Partially liganded states are not shown. (B) Amino acid sequence of the stretch between Ser-217 and Thr-252 (amino acid numbering as in ref. 9). The prime-numbering system is also indicated; for example, Glu-221 corresponds to position −2′, Ile-232 corresponds to position 9′, and Thr-248 corresponds to position 25′. The M2–M3-loop PKTPYMT sequence (in red) was mutated to alanine, one residue at a time. (C) Macroscopic-current responses to 5-s pulses of pH-4.5 extracellular solution (holding pH value: 7.4) recorded from representative alanine mutants in the outside-out configuration. Four of the seven mutations (at positions 23′, 24′, 26′, and 28′) had little effect on peak-current values; longer pH-4.5 applications revealed that, as is the case for the wild-type channel (Fig. S1D), these transients eventually attain a zero level. The other three alanine mutations (at positions 25′, 27′, and 29′), however, seemed to abolish channel function because no currents could be elicited by the applied pH jumps, even in the whole-cell configuration. (D) Normalized peak-current values corresponding to the outside-out recordings described in C. (E) Addition of the gain-of-function I9′A mutation to the three loss-of-function alanine mutants rescued channel function. The peak-current values measured in outside-out patches were small, and hence, we resorted to the whole-cell configuration. (F) Normalized peak-current values corresponding to the whole-cell recordings described in E.

Fig. 2.

X-ray crystal structure of GLIC T25′A at pH 4.0. (A) Distance between aligned Cα atoms (residue range: 12–316). The Cα atoms of each subunit in the structural model of the T25′A mutant at pH 4.0 were aligned to those of the wild-type GLIC channel at pH 4.0 (PDB ID code 3EHZ; mutant chain A with wild-type chain A, mutant chain B with wild-type chain B, and so on, for all five subunits in the asymmetric unit), and the distances between Cα atoms were calculated and averaged. The symbols corresponding to residues 59 and 82–84 were omitted because the associated electron density in the structural model of the T25′A mutant was exceedingly weak. (B–D) Magnified views of the M2 α-helix and the extracellular M2–M3 loop, the pre-M1 linker, and the M3–M4 loop, respectively. Mesh representations show the 2F₀ − F_c electron-density maps contoured at a level of 1.0σ. Red, T25′A mutant; orange, wild-type GLIC.

Fig. 3.

The pore of the liganded closed-channel conformation. (A) Comparison of the liganded closed-channel (GLIC Y27′A at pH 3.9) and liganded open-channel (GLIC wild type at pH 4.0; PDB ID code 3EHZ) conformations of GLIC at the level of the transmembrane segments and intervening linkers. A circle indicates the long axis of the pore. (B) Comparison of the pore-radius profiles [estimated using HOLE software (19)] of the two constructs compared in A. The pore profile of unliganded ELIC (PDB ID code 2VL0; a different structural model of a bacterial nicotinic-receptor–like channel with a nonconductive pore) is also included. The transmembrane pore extends between M2 positions −2′ and 20′. (C) Comparison of the distances from the Cα atoms to the long axis of the pore (mean ± SEM of all five subunits) for residues in and flanking the M2 segments of the three constructs compared in B. The Cα profile of the unliganded muscle-type AChR from Torpedo fish (PDB ID code 2BG9; yet another model of a nicotinic receptor with a nonconductive pore), in gray, is also included after a correction that accounts for the likely misthreading of the primary sequence in the original model. EC, extracellular; IC, intracellular; TM, transmembrane.

Fig. 4.

Mutations to the C loop have only mild effects on GLIC function. (A) Macroscopic-current responses to 5-s pulses of pH-4.5 extracellular solution (holding pH value: 7.4) recorded from C-loop mutants in the outside-out configuration. In the C-loop Gly construct, all 10 residues of the C loop (ANFALEDRLE) were mutated to Gly. In C-loop Δ6, a subset of 6 residues of the loop (FALEDR) was deleted. In C-loop Δ10, all 10 residues of the loop were deleted. (B) Normalized peak-current values. (C–E) Entry-into-desensitization, activation, and deactivation time constants, respectively.