A conformational intermediate in glutamate receptor activation

Abstract

Ionotropic glutamate receptors (iGluRs) transduce the chemical signal of neurotransmitter release into membrane depolarization at excitatory synapses in the brain. The opening of the transmembrane ion channel of these ligand-gated receptors is driven by conformational transitions that are induced by the association of glutamate molecules to the ligand-binding domains (LBDs). Here, we describe the crystal structure of a GluA2 LBD tetramer in a configuration that involves an ∼30° rotation of the LBD dimers relative to the crystal structure of the full-length receptor. The configuration is stabilized by an engineered disulfide crosslink. Biochemical and electrophysiological studies on full-length receptors incorporating either this crosslink or an engineered metal bridge show that this LBD configuration corresponds to an intermediate state of receptor activation. GluA2 activation therefore involves a combination of both intra-LBD (cleft closure) and inter-LBD dimer conformational transitions. Overall, these results provide a comprehensive structural characterization of an iGluR intermediate state.

Figure 1. A Disulfide-Bridged GluA2-L483Y-A665C LBD Tetramer

(A) A 2F_o–F_c electron density envelope for the LBD tetramer contoured at 2 σ is shown. Subunits A, B, C, and D are shown in green, red, blue, and yellow, respectively. DNQX is shown in purple. (B) A 2F_o–F_c electron density for A655C contoured at 1.5 σ is shown. The view is from “above” the LBD layer. Subunits A and C are covalently linked by the disulfide bond at position 655, which is situated between helices F and G. (C and D) The LBD tetramer from the crystal structure of the full-length GluA2 receptor (Sobolevsky et al., 2009) is presented. The antagonist ZK200775 is shown in purple. The A655 Cα-Cα distance between subunits A and C, indicated by the line, is 8.0 Å. (E) An illustration of the transition predicted by the lowest-frequency normal mode going between the OA (right) and CA (left) conformations is shown. The OA conformation corresponds to the conformation shown in (C), and the CA conformation corresponds to the conformation shown in (A). The angle that describes the relative orientation between dimer pairs A–D and B–C has its vertex at the center of mass of the Cα atoms of residue 665 in subunits A and C. The rays of the angle (cyan arrows) pass through subunits A and C at the Cα atom of Leu748. See also Figure S1.

Figure 2. Trapping of the Full-Length GluA2-A665C Receptor

(A) Activation and desensitization of WT GluA2 were not affected by oxidizing conditions. Single-exponential fits to the desensitization of responses to 10 mM glutamate in 1 mM DTT (control; k_des = 156 s⁻¹, triangles and Wash [washout]; k_des = 164 s⁻¹, gray circles) and 10 μM CuPhen (k_des = 167 s⁻¹, open circles) are illustrated. (B) Activation and desensitization of the R661C mutant were similar in oxidizing conditions (10 μM CuPhen, open circles, k_des = 239 s⁻¹) and in reducing conditions (1 mM DTT; control, triangles, k_des = 239 s⁻¹; wash, gray circles, k_des = 207 s⁻¹). (C) The A665C mutant shows a strong reduction of the peak current (~90%) under oxidizing conditions with a full recovery of the current after the re-application of 1 mM DTT. The desensitization rate was similar before and after the application of CuPhen (control k_des = 149 s⁻¹, triangles; k_des = 175 s⁻¹ after washout in 1 mM DTT, gray circles). The inset shows the fit (dashed line) to the current in oxidizing conditions. (D) Bar graph representing the effect of CuPhen on the peak current in WT and mutant receptors. Individual patches are shown as open circles. The strongest inhibition in oxidizing conditions was for A665C (peak current 11% ± 4% of that in reducing conditions; p<0.0001 versus WT (***); one-way ANOVA with Dunnett’s post hoc test). For R661C, the probability of no difference versus control was 0.95; for V666C, p<0.001 (**); for the other mutants, p<0.01 (*). The data for the I664C mutant are taken from Plested and Mayer (2009). See also Figure S2.

Figure 3. The A665C Mutant Traps Dimers

(A) The GluA2-C89S-C190A-C436S-C425S-C528S-C589S-C815S (7 × Cys (−)) construct shows similar activation and desensitization to WT (k_des = 210 s⁻¹), but functional expression was strongly decreased. (B) In western blots under oxidizing conditions, subunit dimers (D) were strongly enriched, compared with the monomer band (M), for the 7 × Cys (−) A665C mutant, but not for the 7x(−) background. All reactions were performed in the presence of 500 μM glutamate and 100 μM CTZ. MW, molecular weight; Con, control; Red, reducing; Ox, oxidizing. (C) Western blot analysis of 7 × Cys (−) A665C mutant following in vivo crosslinking. Cells were exposed to oxidizing conditions for 30 m in DNQX (10 μM) or glutamate (500 μM and 10 mM). All conditions included CTZ. (D) Fraction of dimer formation in reducing (open bars) and oxidizing (filled bars) conditions. The dimer fraction in oxidizing conditions with 500 μM glutamate was 30% ± 4% for the 7 ×Cys (−) A665C mutant (n = 11 blots). The mutant R661C showed no significant increase in dimer formation over the background (14% ± 2%; n = 5; p = 0.99 versus GluA2 7 × Cys (−); n = 6). A665C showed a significant decrease in dimer formation in 10 mM glutamate (14% ± 3%; n = 6; p = 0.026 versus 500 μM). There was no significant difference in dimer formation in the presence of DNQX (p = 0.41 versus 500 μM glutamate; n = 5; Dunnett’s post hoc test). See also Figure S3.

Figure 4. Engineered Metal Bridges Confirm the CA Conformation in the Full-Length Receptor during Activation

(A and B) The engineered HHH metal bridge is formed by the triple-histidine mutation G437H-K439H-D456H (blue spheres at α carbons). Metal bridges cannot form when the LBDs are in the OA conformation (A) but can form when they are in the CA conformation (B). The direction of view is the same as in Figure 1E. Only subunits A and B are shown for clarity. (C) Zinc (1 μM) selectively inhibits a triple-histidine mutant that is predicted to compose an interdimer metal bridge in the CA conformation, but not WT or a mutant with half the site. (D) Summary of inhibition by 1 μM zinc for WT, half-sites, controls, and the four triple mutants that show inhibition. (E) The double-cysteine mutant (K439C-D456C) was not expressed as a functional glutamate receptor, but western blot analysis of total cell lysate revealed redox-sensitive trapping of dimers. (F) Fraction of dimer formation in reducing (green bar) and oxidizing conditions (purple bars). Dimer formation was observed for the 7 × Cys (–) K439C-D456C (KD) mutant (43% ± 5% in oxidizing conditions; p<0.05 versus K439C single mutant (*) and p<0.01 versus reducing conditions (**)). The dashed line shows dimer formation for the 7 × Cys (–) WT construct in oxidizing conditions. The number of blots is indicated in each bar.

Figure 5. The A665C Crosslink Traps an Activation Intermediate that Is Not Saturated with Agonist

(A) Responses to 10 mM glutamate in 1 mM DTT (green bars) flanking an application of 10 μM CuPhen (red bars) in the DNQX-bound resting state (1 s; top trace) and open state (3 s; middle trace) reveal negligible trapping. A substantial trapping relaxation was observed in the presence of an intermediate glutamate concentration (τ = 0.5s; 500 μM; lower trace, red circles). CTZ (100 μM) was present throughout. (B) The active fraction of receptors after oxidization in the presence of 10 μM DNQX (purple), 10 mM glutamate (blue), and 500 μM glutamate (yellow, exponential fit) is plotted against the trapping interval. The open red circles represent the mean of exponential fits to the trapping relaxation (as in A). Arrows indicate intervals for the traces in (A) (n = 3–6 patches per point). (C) Trapping in oxidizing conditions at different concentrations of glutamate is presented. In the top panel, three records from the same patch show trapping in CuPhen (red bar) with 10 μM (light-blue trace), 500 μM (orange trace), and 10 mM glutamate (dark-blue trace). Robust trapping only occurs in 500 μM glutamate (expanded scale, middle panel). The open red circles are a mono-exponential fit to the relaxation (same fit as in A, lower trace, and B). The arrows and symbols indicate the measurements plotted in (D). WT GluA2 was not modified by oxidization (lower panel, red bar) in 500 μM glutamate. (D) Concentration-response curves in CuPhen and 100 μM CTZ for WT GluA2 (blue open circles; EC₅₀ = 270 ± 140 μM), GluA2 A665C following reduction (green triangles; EC₅₀ = 530 ± 160 μM), and following trapping (red squares; EC₅₀ = 1.9 ± 0.2 mM). The difference between the reduced and oxidized (trapped) A665C mutant is significant (p < 0.001). Plotting the active fraction current after trapping (yellow diamonds) versus glutamate concentration reveals maximal trapping at 248 μM glutamate (n = 5). See also Figures S4, S5, and S6.

Figure 6. State Dependence of Lobe 1 Metal Sites

(A) The half-maximal inhibitory concentration of zinc for the mutant G437H-K439H-D456H (HHH, yellow circles) was 95 ± 30 nM (n = 4). (B) Voltage-independent inhibition by zinc (Zn²⁺; 1 μM, red bar) at intermediate glutamate concentration (500 μM, −60 and +60 mV) is shown. The lower dashed line shows the average size of the control responses to 10 mM glutamate in EDTA (green bars). The lower trace shows a mono-exponential fit to trapping relaxation (red circles, τ = 220 ms) and the slow component of a double exponential fit to current in EDTA (green bar) following trapping (green circles, loss of zinc trapping, τ = 550 ms; amplitude 33%). (C) State dependence of inhibition by 1 μM zinc (n = 3–6 patches per point). Zinc preferentially traps a partially glutamate-bound state (yellow, 500 μM Glu + CTZ). (D) WT receptors are not trapped by 1 μM zinc in the presence of 500 μM Glu plus CTZ. (E) Concentration-response curves in 1 μM zinc and 100 μM CTZ for WT GluA2 (blue open circles; EC₅₀ = 140 ± 10 μM), GluA2 HHH before trapping by zinc (green triangles; EC₅₀ = 230 ± 27 μM), and GluA2 HHH following trapping (red squares; EC₅₀ = 424 ± 30 μM). The difference between the apparent affinities for glutamate in the presence and absence of zinc was significant (p<0.01). Plotting the active fraction current after trapping (yellow diamonds) versus glutamate concentration reveals maximal trapping at 368 μM glutamate (n = 3). (F) Top view of the CA conformation of the LBD tetramer modeled with two interdimer triple-histidine mutants coordinating zinc ions is presented. The right panel shows the boxed region from the left panel on an expanded scale.

Figure 7. Proposed Structural Transitions Involved in AMPA Receptor Activation

Subunits A, B, C, and D are colored green, red, blue, and yellow, respectively. The LBDs are arranged in the OA conformation when the receptor is in a state of rest. When the LBDs of subunits B and D bind glutamate (orange sphere), the receptor enters a state of partial activation. In this state, the LBDs may adopt either the OA or CA conformation. The conformation trapped by either the disulfide bridge formed by the A665C mutation or the metal bridge formed by one of the triple-residue mutations is indicated by an asterisk. When glutamate is bound to all LBDs, the receptor enters a state of full activation. We have no information about the LBD conformation in this state. It is unknown whether the ATD layer undergoes large-scale structural rearrangements during gating transitions, and it is therefore omitted from this figure. See also Figure S7 and Table S1.