Conductance of P2X₄ purinergic receptor is determined by conformational equilibrium in the transmembrane region

Significance

Partial agonists of ligand-gated ion channels reportedly offer clinical advantages over antagonists and full agonists in antidepressant and smoking-cessation treatment. In the cases of P2X purinergic receptors, the currents evoked by α,β-methylene ATP are lower than the currents evoked by ATP. Here, our NMR analyses revealed that the transmembrane region and the membrane side of the lower body exist in conformational equilibrium between the closed and open conformations, with slower exchange rates than the chemical shift difference (<100 s^-1), and that the small population of the open conformation of zebrafish P2X₄ purinergic receptor causes the partial activation in the α,β-methylene ATP-bound state. These findings provide insights into the mechanism underlying the partial activation of P2X₄ receptors and other ligand-gated ion channels.

Abstract

Ligand-gated ion channels are partially activated by their ligands, resulting in currents lower than the currents evoked by the physiological full agonists. In the case of P2X purinergic receptors, a cation-selective pore in the transmembrane region expands upon ATP binding to the extracellular ATP-binding site, and the currents evoked by α,β-methylene ATP are lower than the currents evoked by ATP. However, the mechanism underlying the partial activation of the P2X receptors is unknown although the crystal structures of zebrafish P2X₄ receptor in the apo and ATP-bound states are available. Here, we observed the NMR signals from M339 and M351, which were introduced in the transmembrane region, and the endogenous alanine and methionine residues of the zebrafish P2X₄ purinergic receptor in the apo, ATP-bound, and α,β-methylene ATP-bound states. Our NMR analyses revealed that, in the α,β-methylene ATP-bound state, M339, M351, and the residues that connect the ATP-binding site and the transmembrane region, M325 and A330, exist in conformational equilibrium between closed and open conformations, with slower exchange rates than the chemical shift difference (<100 s⁻¹), suggesting that the small population of the open conformation causes the partial activation in this state. Our NMR analyses also revealed that the transmembrane region adopts the open conformation in the state bound to the inhibitor trinitrophenyl-ATP, and thus the antagonism is due to the closure of ion pathways, except for the pore in the transmembrane region: i.e., the lateral cation access in the extracellular region.

In chemical neurotransmission, various neurotransmitters bind to ligand-gated ion channels expressed in the plasma membrane of postsynaptic cells, such as the NMDA, AMPA, and P2X receptors, leading to changes in membrane potential and the concentration of intracellular ions. Each ligand for a ligand-gated ion channel has a distinct ability to evoke currents (1), and the ligands are classified according to the evoked current level: such as, full agonists, partial agonists, and antagonists. Partial agonists of ligand-gated ion channels reportedly offer clinical advantages over antagonists and full agonists in antidepressant and smoking-cessation treatment (2, 3).

Two mechanisms have been proposed for the partial activation of the ligand-gated ion channels: the equilibrium between the open and closed conformations and the distinct conformation of the partial agonist-bound states from the closed and open conformations (4, 5). In the crystal structures of the extracellular region of the AMPA receptor, in which the distances between the two extracellular domains are changed upon agonist binding, the interdomain distances in the partial agonist-bound states correlated with the conductance level, suggesting that the AMPA receptor adopts specific intermediately permeable conformations (4, 6).

The P2X receptors are a family of cation channels gated by extracellular ATP (1, 7–9) and are involved in many physiological and pathophysiological processes (10–12). Seven subtypes of the P2X receptors have been identified in mammals (13), and they share ∼40% sequence identity. The P2X₄ receptor is involved in the pathogenesis of chronic neuropathic, inflammatory pain and the endothelial cell-mediated control of vascular tone (11, 14, 15). Compared with ATP, α,β-methylene ATP (α,β-meATP), in which the oxygen atom linking the α- and β-phosphorous atoms of ATP is replaced by a methylene group (Fig. S1A), reportedly induces a lower maximum current in cells expressing the mouse, rat, and human P2X₄ receptors and other P2X receptors (16, 17).

Fig. S1.

Characterization of the P2X₄ receptor. (A) Chemical structures of ATP and α,β-meATP. (B and C) TEVC recordings of ATP- and α,β-meATP-evoked currents from rat P2X₄ receptor expressed in Xenopus oocytes, respectively. In B, the currents were evoked twice by ATP (30 μM, 1 min, black bar). In C, the currents were firstly evoked by ATP (30 μM, 1 min, black bar) and subsequently by α,β-meATP (300 μM, 1 min, black bar). (D) TEVC recording of the ATP-evoked current (30 μM, 30 s, black bar) from the N-terminally EGFP-tagged ΔzfP2X₄–A′ construct expressed in Xenopus oocytes. (E) Size exclusion chromatogram of purified EGFP-tagged ΔzfP2X₄–A′ in rHDLs. Elution volumes corresponding to 17.0, 12.2, 10.4, and 7.1 nm Stokes diameters were determined by thyroglobulin, ferritin, catalase, and BSA, respectively. V₀ and 1CV are void volume and single column volume, respectively. (F) SDS/PAGE analyses of purified ΔzfP2X₄–A′ embedded in rHDLs. The samples were analyzed by 12% SDS/PAGE with Coomassie Brilliant Blue staining. (G) Measurement of [³H]ATP saturation binding to the purified ΔzfP2X₄–A′ in rHDLs. (H and I) Estimation of the effects of deuteration based on the crystal structures of zfP2X₄ (PDB ID code 4DW1) and the deuteration incorporation rates. The plots on the Left (without deuteration) and the Right (with deuteration) are the sums of the inverse sixth power of the distances between pseudoatoms centered on the methyl hydrogens of M108, M249, M268, or M325 and each hydrogen atom in the crystal structure of zfP2X₄ (sums of the r⁻⁶) and the sums of the r⁻⁶ multiplied by [1 − (deuterium incorporation rates)] of each hydrogen atom, respectively. The graphs in H and I were calculated from the crystal structure in the apo state (PDB ID code 4DW0) and that in the ATP-bound state (PDB ID code 4DW1), respectively. Sums of the r⁻⁶ of each methionine methyl group and Hαβγ of the intraresidue methionine (green), Hαβγ of the interresidue methionine (light green), Hαβ of tyrosine (light violet), Hδεζη of tryptophan (orange), Hαβδεζ of phenylalanine (pink), Hαβγ of valine (blue), Hαβγδ of leucine (light blue), Hαβγδ of isoleucine (cyan), Hαβγ of threonine (light cyan), Hαβ of alanine (red), Hαβγδ of arginine (dark blue), Hα of glycine (dark green), and Hαβ of serine (magenta) residues, and the other hydrogens connected to carbon atoms (other unexchangeable hydrogens, light gray) are shown with colors. Hydrogen atoms connected to nitrogen, oxygen, or sulfur atoms were not considered in these calculations because these hydrogens should be exchanged with deuterium in D₂O. The deuterium incorporation rates of the hydrogen atoms within each methionine residue (intraresidue) and the deuterium incorporation rates of other methionine residues (interresidue) were set to 98% and 85%, respectively, because the methionine residues would be derived from 85% of [α-, β-, γ-98% ²H-, methyl-¹³C]-methionine and 15% of nonlabeled methionine in the medium.

The crystal structures of zebrafish P2X₄ receptor (zfP2X₄) (18, 19), together with mutational analyses (20–26), provided the structural basis for the channel opening of P2X receptors upon ATP binding. In the crystal structures, zfP2X₄ forms a homotrimer (27, 28), in which the transmembrane region of each subunit is composed of two helices (19). In the crystal structure of zfP2X₄ in the ATP-bound state, three ATP molecules are bound to the intersubunit nucleotide binding pockets. In addition, the region that connects the ATP-binding site and the transmembrane region, which is referred to as the “lower body” (Fig. 1 A and B), is expanded by ∼10 Å in the ATP-bound state, and a pore is formed in the transmembrane region, which is proposed to expand by the iris-like movement of the transmembrane helices (18). However, the mechanism underlying the partial activation of P2X receptors is unknown because the structures of the P2X receptors have not been examined in the partial agonist-bound states.

Fig. 1.

NMR resonances from the endogenous methionine residues of zfP2X₄ in rHDL. (A and B) Distribution of the methionine residues in the ΔzfP2X₄–A′. One subunit from the crystal structure of zfP2X₄ in the apo form (A) (PDB ID code 4DW0) and one from the ATP-bound form (B) (PDB ID code 4DW1) are shown in ribbons. The lower body and the right flipper are yellow. The A330 residues, the methionine residues, and the residues in which methionine mutations were introduced, L339 and L351, are depicted by green sticks. ATP is depicted by red sticks. Dummy atoms generated by Orientations of Proteins in Membranes (OPM), which represent membrane boundary planes, are gray. (C) Overlaid ¹H-¹³C HMQC spectra of [²H-11AA, α, β-²H, methyl-¹³C-Met]ΔzfP2X₄-A′, embedded in rHDLs, in the apo state (black) and the ATP-bound state (red). The regions with resonances from methionine residues are shown, and the assigned resonances are indicated. The centers of the resonances are indicated with dots. Cross-sections at lines through the centers of each resonance in the ATP-bound state and the cross-sections of the spectra using [α, β-²H, methyl-¹³C-Met]ΔzfP2X₄-A′ are shown on the top of the overlaid spectra. The intensities of the cross-sections were normalized by the concentration of ΔzfP2X₄-A′ and the conditions of the NMR measurements.

The P2X₄ receptor used in the previous crystallographic studies was solubilized by detergents, which are widely used for structural investigations of membrane proteins, but the P2X₄ receptor is embedded in lipid bilayers under physiological conditions. It was recently reported that reconstituted high-density lipoproteins (rHDLs), which are also known as nanodiscs (29), can accommodate membrane proteins within a 10-nm-diameter disk-shaped lipid bilayer (30). The rHDLs reportedly provide a lipid environment with more native-like properties, compared with liposomes, in terms of the lateral pressure and curvature profiles because detergent micelles have strong curvature and different lateral pressure profiles from lipid membranes (31). Our NMR analyses of a G protein-coupled receptor (GPCR) and an ion channel in rHDL lipid bilayers revealed that the population and the exchange rates of the conformational equilibrium determine their signal transduction and ion transport activities (32–34) and that the population of the active conformation of the GPCR in rHDLs correlated better with the signaling levels than that in detergent micelles (32). Therefore, NMR investigations of membrane proteins in the lipid bilayer environments of rHDLs are necessary for accurate measurements of the exchange rates and the populations in conformational equilibrium.

Here, we used NMR to observe the conformational equilibrium of the alanine and methionine residues of zfP2X₄ bound to α,β-meATP in rHDLs. Based on the conformational equilibrium, we discuss the mechanism underlying the partial activation of P2X receptors.

Results

Steady-State Current Evoked by α,β-meATP.

The amplitude of the plateau current determines the responsiveness of the P2X receptors in the sustained presence of ATP (11). In addition, NMR spectra are recorded after prolonged (>30 min) exposure to the ligands. Therefore, the currents in the sustained presence of ATP or α,β-meATP were determined by whole cell two-electrode voltage-clamp (TEVC) analyses using Xenopus oocytes expressing the rat P2X₄ receptor. As a result, the currents in the sustained presence of α,β-meATP were ∼20% of the currents in the sustained presence of ATP (Fig. S1 B and C).

Preparation and Characterization of Zebrafish P2X₄ Embedded in Reconstituted High-Density Lipoproteins.

A truncated zebrafish P2X₄ receptor (zfP2X₄) construct (35, 36) with the M364L mutation, with residues S28 through K365, was expressed in a baculovirus–insect cell expression system. Hereafter, this construct is referred to as ΔzfP2X₄-A′. TEVC analyses using Xenopus oocytes expressing ΔzfP2X₄-A′ revealed that ΔzfP2X₄-A′ retains the full ATP-dependent channel activity (Fig. S1D). To examine the conformational dynamics of ΔzfP2X₄-A′ embedded in lipid bilayers, purified ΔzfP2X₄-A′ in n-dodecyl-β-d-maltopyranoside (DDM) micelles was reconstituted into rHDLs, and further purified with three chromatography steps. The size exclusion chromatography revealed that purified ΔzfP2X₄-A′ in rHDLs is monodisperse, with a Stokes diameter of 12 nm (Fig. S1E), in good agreement with the previously reported rHDL size (37). As judged from the SDS/PAGE analysis, the purity of ΔzfP2X₄-A′ in rHDLs was >90%, and the ratio of zfP2X₄ and MSP1 was consistent with the 3:2 stoichiometry that would be expected if each rHDL particle is composed of two MSP1 molecules and a trimer of zfP2X₄ (Fig. S1F). The dose-dependent curve observed in the ³H-ATP binding assays revealed that the EC₅₀ value and the Hill coefficient were 93 nM and 1.5, respectively (Fig. S1G), which are comparable with the previously reported affinity and cooperativity of the interaction between zfP2X₄ and ATP (18, 19).

NMR Resonances from Endogenous Methionine Residues in the Extracellular Region.

ΔzfP2X₄-A′ possesses five methionine residues in the extracellular region, and they adopt distinctly different conformations between the apo and ATP-bound crystal structures (Fig. 1 A and B). M108, M249, and M256 exist in the region previously referred to as the “right flipper” or the extracellular side of the lower body, and M325 exists on the membrane side of the lower body (Fig. 1 A and B). Therefore, we used the ¹³C selective labeling of methionine methyl groups to investigate the structure of ΔzfP2X₄-A′. To observe the NMR resonances from the huge trimeric ΔzfP2X₄-A′ (∼250 kDa in rHDLs), ΔzfP2X₄-A′ was deuterated for sensitivity enhancement. We selected the deuterated amino acids, based on the previously reported labeling efficiencies (32) and the ¹H–¹H distances between the observed methionine methyl groups and each amino acid residue in the crystal structures of zfP2X₄. Our calculation revealed that, with the deuteration of the alanine, phenylalanine, glycine, isoleucine, leucine, proline, arginine, serine, threonine, valine, and tryptophan residues, the ¹H–¹H dipole–dipole interactions of the methyl groups in ΔzfP2X₄-A′ would be ∼30% of the dipole–dipole interactions of nondeuterated ΔzfP2X₄-A′ (Fig. S1 H and I), and the sensitivities of the resonances from these residues would be increased by about fivefold upon deuteration. Therefore, we prepared ΔzfP2X₄-A′ in which the methionine methyl groups were labeled with ¹³C and the 11 types of amino acid residues were deuterated. Hereafter, the ΔzfP2X₄-A′ obtained by this method is referred to as [²H-11AA, αβ-²H-, methyl-¹³C-Met] ΔzfP2X₄-A′.

In the ¹H-¹³C HMQC spectra of [²H-11AA, αβ-²H-, methyl-¹³C-Met] ΔzfP2X₄-A′ in rHDLs in the apo state, signals that apparently correspond to the five methionine residues were observed (Fig. 1C). In the spectra of the ATP-bound states, all resonances markedly shifted, compared with the resonances in the apo state (Fig. 1C). For the assignment of the methionine resonances, we mutated the methionine residues. For example, M108 was assigned by introducing the M108L mutation. Assignments were established using the spectra in DDM micelles, considering that chemical shifts in rHDLs were similar to the chemical shifts in DDM micelles. In the apo and ATP-bound states, one resonance was absent in the spectra of the M108L mutant, revealing that these resonances are from M108 (Fig. S2 A and B). The resonances from the other methionine residues in the apo and ATP-bound states were assigned in a similar manner (Fig. 1C and Fig. S2 C–J).

Fig. S2.

Assignment of the resonances from M108, M249, M256, M268, and M325. Overlaid ¹H-¹³C HMQC spectra of ΔzfP2X₄–A′ (black or red) and its mutants (light green or green) in the apo state (A, C, E, G, and I) and the ATP-bound state (B, D, F, H, and J). The spectra of [²H-11AA, α,β,γ-²H, methy-¹³C-Met] zfP2X₄ in DDM micelles and [²H-11AA, α,β-²H, methy-¹³C-Met] zfP2X₄ in rHDLs were used for the assignments in (A–G) and (H–J), respectively. The regions with resonances from methionine residues are shown, and the assigned resonances are indicated. Cross-sections at the dotted lines are shown on the top of the overlaid spectra. The observation of multiple resonances for M268 in the apo and ATP-bound states may be due to the local conformational equilibrium triggered by the cis–trans isomerization of P269 and/or the isomerization of the disulfide bond (50) between C264 and C273, along with the ring current shift induced by W272.

Observation and Assignments of the NMR Resonances from A330.

ΔzfP2X₄-A′ possesses 17 alanine residues, and A330, as well as M325, exists on the transmembrane side of the lower body. It is possible that the resonance from A330 will be separated from the other resonances from the alanine residues because the Y198 ring current should induce a strong upfield shift in the A330 methyl group, based on the crystal structures of zfP2X₄ in the apo and ATP-bound states (Fig. 2 A and B). Therefore, we used the alanine methyl groups to investigate the conformation of the extracellular region of zfP2X₄.

Fig. 2.

NMR resonances from A330, M339, and M351. (A and B) Distribution of the alanine residues in ΔzfP2X₄–A′. One of the subunits from the crystal structure of zfP2X₄ in the apo form (A) (PDB ID code 4DW0) and one from the ATP-bound form (B) (PDB ID code 4DW1) are shown in ribbons. Alanine and tyrosine residues are depicted by green and white sticks, respectively. (C) Overlaid ¹H-¹³C HMQC spectra of [²H-5AA, α-²H, methyl-¹³C-Ala] ΔzfP2X₄–A′ without the M364L mutation, embedded in rHDLs, in the apo state (black) and the ATP-bound state (red). (D and E) Position of L339 in the zfP2X₄. (F) Overlaid ¹H-¹³C HMQC spectra of [²H-11AA, α, β-²H, methyl-¹³C-Met] ΔzfP2X₄–A′/L339M, embedded in rHDLs, in the apo state (black) and the ATP-bound state (red). (G and H) Position of L351 in zfP2X₄. (I) Overlaid ¹H-¹³C HMQC spectra of [²H-11AA, α, β-²H, methyl-¹³C-Met] ΔzfP2X₄–A′/L351M, embedded in rHDLs, in the apo (black) and ATP-bound (red) states. In D, E, G, and H, the transmembrane helices of the crystal structures of zfP2X₄ in the apo and ATP-bound states, respectively, viewed from the intracellular side, are shown in ribbons. One of the subunits is colored blue, and L339 (D and E) or L351 (G and H) in this subunit is depicted by a green stick model. In C, F, and I, only the regions with the A330, M339, and M351 methyl resonances are shown, respectively, and the centers of the resonances from these residues are indicated with dots. The full spectra are shown in Figs. S3 and S4.

The preparation of the deuterated and alanine selectively labeled protein, using the insect cell–baculovirus expression system, was accomplished by modification of the aforementioned method for the preparation of the deuterated and methionine selectively labeled protein. We added deuterated amino acids and algal amino acid mixtures, as well as [α-²H-, methyl-¹³C]-alanine, to the amino acid-deficient medium. The ¹³C labeling efficiencies were calculated from NMR analyses of thioredoxin, which was used as a test case. As a result, the alanine methyl groups were selectively labeled with ¹³C with an efficiency of ∼30%.

For the observation of the A330 methyl resonances of zfP2X₄, we selected the deuterated amino acids in the same manner as described above for the methionine methyl groups. Our calculation revealed that, in the case of the deuteration of the cysteine, phenylalanine, glycine, leucine, and tyrosine residues and alanine Hα, the ¹H–¹H dipole–dipole interactions of the A330 methyl groups in ΔzfP2X₄-A′ would be ∼25% of the dipole–dipole interactions of nondeuterated ΔzfP2X₄-A′. Hereafter, the ΔzfP2X₄-A′ obtained by this method is referred to as [²H-5AA, α-²H-, methyl-¹³C-Ala] ΔzfP2X₄-A′.

In the ¹H-¹³C heteronuclear multiple quantum coherence (HMQC) spectra of apo [²H-5AA, α-²H-, methyl-¹³C-Ala] ΔP2X₄-A′ in rHDLs, the signals that apparently correspond to the 17 alanine residues were observed (Fig. S3A). In the spectra of the ATP-bound states, several resonances markedly shifted, compared with the resonances in the apo state (Fig. S3B). For the resonance assignments, the spectra of the A330G mutant were recorded, according to the comparison with rat P2X₇, in which A330 corresponds to G325. As a result, the most upfield-shifted resonance was absent in the spectra of the A330G mutant, revealing that this resonance is from A330 (Fig. S3C). The resonance from A330 markedly shifted upon ATP binding (Fig. 2C).

Fig. S3.

Observation and assignments of the resonances from A330. (A and B) ¹H-¹³C HMQC spectra of [²H-5AA, α-²H, methy-¹³C-Ala] ΔzfP2X₄–A′, embedded in rHDLs, in the apo state (A, black) and the ATP-bound state (B, red). (C). Overlaid ¹H-¹³C HMQC spectra of apo [²H-5AA, α-²H, methy-¹³C-Ala] ΔzfP2X₄–A′ (black) and ΔzfP2X₄–A′/A330G (light green) in DDM micelles. Only the regions with alanine methyl signals are shown, and the assigned resonances were indicated. In A and B, the regions with ¹H chemical shifts of 0.75–0.9 and 1.1–1.3 ppm, which contain t₁ noise derived from the intense lipid signals, are colored gray. In C, the regions with ¹H chemical shifts of 0.8–0.9 and 1.2–1.4 ppm, which contain t₁ noise derived from the intense DDM signals, are colored gray. The chemical shifts of the A330 resonance in DDM micelles were almost identical to the chemical shifts of the resonance in rHDLs. In C, cross-sections at lines through the centers of the resonances from A330 are shown on the top of the overlaid spectra.

NMR Resonances from Methionine Residues Introduced in the Transmembrane Region.

To investigate the structure of the transmembrane region, we introduced methionine residues to the extracellular and intracellular sides of the transmembrane helix 2 (TM2) by the L339M and L351M mutations, respectively (Figs. 1 A and B and 2 D and E and G and H), according to the comparison with the human P2X₄ receptor, in which L339 and L351 correspond to M336 and M348, respectively. L351 is a pore-forming residue (Fig. 2H) and is close to a glycine residue, G350 in zfP2X₄, which was proposed to function as a gating hinge (18, 38). We confirmed that the L339M and L351M mutants retained full ATP-binding and ATP-dependent cation channel activities (Fig. S4 A and D).

Fig. S4.

NMR resonances from M339 and M351. (A) TEVC recording of ATP-evoked current (10 mM, 30 s, black bar) from the N-terminally EGFP-tagged ΔzfP2X₄–A′/L339M construct expressed in Xenopus oocytes. (B) Overlaid ¹H-¹³C HMQC spectra of [²H-11AA, α,β-²H, methy-¹³C-Met]ΔzfP2X₄–A′ (black) and [²H-11AA, α,β-²H, methy-¹³C-Met]ΔzfP2X₄–A′/L339M (light green), embedded in rHDLs, in the apo state. (C) Overlaid ¹H-¹³C HMQC spectra of [²H-11AA, α,β-²H, methy-¹³C-Met]ΔzfP2X₄–A′ (red) and [²H-11AA, α,β-²H, methy-¹³C-Met]ΔzfP2X₄–A′/L339M (green), embedded in rHDLs, in the ATP-bound state. (D) TEVC recording of ATP-evoked current (10 mM, 30 s, black bar) from the N-terminally EGFP-tagged ΔzfP2X₄–A′/L351M construct expressed in Xenopus oocytes. (E) Overlaid ¹H-¹³C HMQC spectra of [²H-11AA, α,β-²H, methy-¹³C-Met]ΔzfP2X₄–A′ (black) and [²H-11AA, α,β-²H, methy-¹³C-Met]ΔzfP2X₄–A′/L351M (light green), embedded in rHDLs, in the apo state. (F) Overlaid ¹H-¹³C HMQC spectra of [²H-11AA, α,β-²H, methy-¹³C-Met]ΔzfP2X₄–A′ (red) and [²H-11AA, α,β-²H, methy-¹³C-Met]ΔzfP2X₄–A′/L351M (green), embedded in rHDLs, in the ATP-bound state. In B, C, E, and F, the regions with resonances from methionine residues are shown, and the assigned resonances are indicated.

In the ¹H-¹³C HMQC spectra of [²H-11AA, αβ-²H-, methyl-¹³C-Met] ΔzfP2X₄-A′/L339M and ΔzfP2X₄-A′/L351M in the apo and ATP-bound states, one resonance was additionally observed in each spectrum, revealing that these resonances are from M339 and M351 (Fig. S4 B and C and E and F). The remaining signals did not exhibit significant changes upon introducing the L339M and L351M mutations, suggesting that the conformation of the extracellular region of ΔzfP2X₄-A′ was not affected by these mutations. Both of the resonances from M339 and M351 remarkably shifted upon ATP binding (Fig. 2 F and I and Fig. S4 B and C and E and F).

Conformational Equilibrium in the α,β-meATP–Bound State.

We used the methionine and alanine methyl groups to investigate the conformation of the zfP2X₄ bound to a partial agonist, α,β-meATP. Our TEVC analyses revealed that the α,β-meATP–evoked currents from ΔzfP2X₄-A′ expressed in Xenopus oocytes were ∼30% of the currents induced by ATP (Fig. S5 A–C) and that α,β-meATP competed with ³H-ATP binding to ΔzfP2X₄-A′ in a concentration-dependent manner, with a K_i of 2.7 μM (Fig. S5D).

Fig. S5.

Currents of ΔzfP2X₄–A′ evoked by α,β-meATP and the affinity of α,β-meATP and ΔzfP2X₄–A′. (A and B) TEVC recordings of ATP-evoked (A) or α,β-meATP-evoked (B) current (10 mM, 10 s, black bars) from N-terminally EGFP-tagged ΔzfP2X₄–A′/L351M construct expressed in Xenopus oocytes. (C) Plots of the currents evoked in A and B, relative to the currents evoked by ATP. (D). Plots of [³H]ATP bound to ΔzfP2X₄–A′ in DDM micelles, in the presence of increasing concentrations of α,β-meATP. (E and F) ¹H-¹³C HMQC spectra of [²H-11AA, α,β-²H, methy-¹³C-Met]ΔzfP2X₄–A′/L339M (E) and [²H-11AA, α,β-²H, methy-¹³C-Met]ΔzfP2X₄–A′/L351M (F) in rHDLs bound to α,β-meATP. (G) ¹H-¹³C HMQC spectra of [²H-5AA, α-²H, methy-¹³C-Ala] ΔzfP2X₄–A′ in rHDLs bound to α,β-meATP. (H–J) ¹H-¹³C HMQC spectra of [²H-11AA, α,β-²H, methy-¹³C-Met]ΔzfP2X₄–A′/L339M in rHDL in the presence of 0.1 mM (H), 0.5 mM (I), and 1 mM (J) α,β-meATP. In E–G, the regions with resonances from methionine residues are shown, and the assigned resonances are indicated. In G, the regions with ¹H chemical shifts of 0.75–0.9 and 1.1–1.3 ppm, which contain t₁ noise derived from the intense lipid signals, are colored gray. In (H–J), only the regions with M339 resonances are shown.

To investigate the structure in the transmembrane region of zfP2X₄ bound to α,β-meATP, the ¹H-¹³C HMQC spectra of [²H-5AA, α-²H-, methyl-¹³C-Ala] ΔP2X₄-A′, [²H-11AA, αβ-²H-, methyl-¹³C-Met] ΔzfP2X₄-A′/L339M, and [²H-11AA, αβ-²H-, methyl-¹³C-Met] ΔzfP2X₄-A′/L351M in rHDLs were observed, under the condition where >99% of ΔzfP2X₄-A′ bound to α,β-meATP (Fig. 3 and Fig. S5 E–G). As a result, two resonances each were observed for A330, M339, and M351, which exist in the transmembrane region and on the membrane side of the lower body (Fig. 1 A and B), and their chemical shifts were almost identical to the chemical shifts in the apo and ATP-bound states (Fig. 3 A–C). Two resonances were also observed for M325, which exists on the membrane side of the lower body (Fig. 1 A and B), and the chemical shift of one of the resonances was almost identical to that in the ATP-bound state (Fig. 3D). Two resonances with chemical shifts almost identical to the chemical shifts in the apo state and the chemical shifts in ATP-bound state were also observed for M268 (Fig. 3E). In the spectra of [²H-11AA, αβ-²H-, methyl-¹³C-Met] ΔzfP2X₄-A′/L339M with a lower concentration of α,β-meATP, the M339 signals were observed at the same chemical shifts as the signals at higher ligand concentrations (Fig. S5 H–J), suggesting that the NMR signals are not significantly affected by the exchange between the free and bound states or the nonspecific effects of the ligands.

Fig. 3.

Resonances from M268, M325, A330, M339, and M351 in the α,β-meATP–bound state. ¹H-¹³C HMQC spectra of [²H-5AA, α-²H, methyl-¹³C-Ala] ΔzfP2X₄–A′ (A), [²H-11AA, α,β-²H, methy-¹³C-Met] ΔzfP2X₄–A′/L339M (B and D–H), and [²H-11AA, α,β-²H, methy-¹³C-Met] ΔzfP2X₄–A′/L351M (C), embedded in rHDL, in the α,β-meATP–bound state recorded at 310 K (cyan) and the spectra recorded at 318 K (purple) are shown. The centers of the resonances from these residues are indicated with dots. Only the regions with A330, M339, M351, M325, M268, M108, M249, and M256 methyl signals are shown in A, B, C, D, E, F, G, and H, respectively. Dashed lines represent the ¹H or ¹³C chemical shifts of the resonances in the α,β-meATP–bound states, observed at 310 K. The full spectra in the α,β-meATP–bound state recorded at 310 K are shown in Fig. S5. In E, the chemical shifts of the M268 signals shown in the Middle in the apo state are similar to the chemical shifts in the ATP-bound state, and the M268 signal in the apo state was not observed in the Bottom.

To examine the resonances from M325, M339, and M351 in the α,β-meATP–bound state undergoing conformational exchange, we also recorded the spectra at a higher temperature, 318 K (Fig. 3 B–D). As a result, the intensity ratios of the resonances from M325, M339, and M351 with chemical shifts identical to the chemical shifts in the apo state, relative to the resonances with chemical shifts identical to the chemical shifts in ATP-bound state, were decreased with the increase of temperature.

In the cases of the resonances from M108, M249, and M256, which are on the right flipper or the extracellular side of the lower body (Fig. 1 A and B), the chemical shifts in the α,β-meATP–bound state were almost identical to the chemical shifts in the ATP-bound state (Fig. 3 F–H).

NMR Resonances in the TNP-ATP–Bound State.

To examine how different ligands produce distinct structural changes in different parts of P2X receptors, we recorded the NMR spectra of zfP2X₄ bound to trinitrophenyl ATP (TNP-ATP), which is composed of ATP and a trinitrophenyl group attached to the 2′ and 3′ hydroxyl of the ribose moiety (Fig. S6A) and is reportedly a representative non–subtype-selective antagonist of P2X receptors that competes with ATP (10, 39–41). We confirmed that the ATP-evoked currents of ΔzfP2X₄-A′ were inhibited by TNP-ATP and that TNP-ATP did not evoke currents in oocytes expressing ΔzfP2X₄-A′ (Fig. S6 B and C). In the ¹H-¹³C HMQC spectra of [²H-5AA, αβ-²H-, methyl-¹³C-Met, α-²H-, methyl-¹³C-Ala] ΔP2X₄-A′/L339M in the TNP-ATP–bound state, two resonances from A330 with ¹H chemical shifts different from that in the ATP-bound state were observed (Fig. S7A). In the ¹H-¹³C HMQC spectra of [²H-11AA, αβ-²H-, methyl-¹³C-Met] ΔzfP2X₄-A′/L339M and [²H-11AA, αβ-²H-, methyl-¹³C-Met] ΔzfP2X₄-A′/L351M in the TNP-ATP–bound state, the chemical shifts of the resonances from M339 and M351, as well as the resonances from the endogenous methionine residues, were almost identical to the chemical shifts in the ATP-bound state (Fig. S7 B and C).

Fig. S6.

Inhibition of the ATP-evoked current of zfP2X₄ by TNP-ATP. (A) Chemical structure of TNP-ATP. (B and C) TEVC recording of ATP-evoked and TNP-ATP–evoked currents from the N-terminally EGFP-tagged ΔzfP2X₄–A′/L351M construct expressed in Xenopus oocytes. In B, the currents were first evoked by ATP (30 μM, 10 s, black bar) and subsequently by TNP-ATP (50 μM, 10 s, black bar). In C, the currents were first evoked by ATP and subsequently by the mixture of ATP and TNP-ATP (10 and 50 μM, respectively, 10 s, black bar).

Fig. S7.

Resonances from methionine residues of zfP2X₄ bound to TNP-ATP. (A) ¹H-¹³C HMQC spectra of [²H-5AA, αβ-²H-, methyl-¹³C-Met, α-²H, methyl-¹³C-Ala] ΔzfP2X₄–A′/L339M, embedded in rHDLs, in the TNP-ATP–bound state (purple), and the spectra of ΔzfP2X₄–A′ in the ATP-bound state (red) and the apo state (black), recorded at 310 K. (B) ¹H-¹³C HMQC spectra of [²H-11AA, α,β-²H, methy-¹³C-Met] ΔzfP2X₄–A′/L339M, embedded in rHDLs, in the TNP-ATP–bound state (purple), the ATP-bound state (red), and the apo state (black) recorded at 310 K. (C) ¹H-¹³C HMQC spectra of [²H-11AA, α,β-²H, methy-¹³C-Met] ΔzfP2X₄–A′/L351M, embedded in rHDLs, in the TNP-ATP–bound state (purple), the ATP-bound state (red), and the apo state (black) recorded at 310 K. Only the regions with the A330, M339, M325, M351, M108, M249, or M256 methyl signals are shown. Dashed lines represent the ¹H or ¹³C chemical shifts of the resonances from these residues in the ATP-bound state, and the centers of the resonances from these residues are indicated with dots. The observation of multiple resonances for A330 in the TNP-ATP–bound state may be due to the local conformational equilibrium of the lateral cation pathway, along with the ring current shift induced by Y198.

Discussion

In this study, we established methods for alanine residue-selective labeling and deuteration in the insect cell–baculovirus system. These methods, along with recently developed methods for methionine residue selective labeling (32), enabled us to observe the resonances from ΔzfP2X₄-A′ in rHDLs in the apo, ATP-bound, and α,β-meATP–bound states (Figs. 1–3).

The ¹H chemical shifts of methionine methyl signals depend on the ring current effects from the neighboring aromatic rings: Y45 in the case of M339 (Fig. 2 D and E). The ¹H chemical shift of M339 in the apo state (1.82 ppm) exhibits a larger upfield ring current shift than that in the ATP-bound state (1.91 ppm) (Fig. 2F and Fig. S4 B and C), implying the close contact of the M339 methyl group with Y45 in the apo state. In the crystal structure of apo zfP2X₄, Y45 is in close proximity to L339 whereas no aromatic residue contacts the sidechain of L339 in the crystal structure of zfP2X₄ with ATP (Fig. 2 D and E). These characteristics of the structure are in good agreement with the above-described conformations indicated by the ¹H chemical shifts. Therefore, we conclude that the resonances observed in the apo and ATP-bound states correspond to the closed and open conformations, respectively (Fig. 4 A and B). The ATP-induced chemical shift changes observed for the other methionine resonances (Figs. 1 and 2) are in agreement with the conformational differences of the crystal structures in the apo and ATP-bound states.

Fig. 4.

Schematic diagrams of the conformational equilibrium of the lower body and the transmembrane region of zfP2X₄ in the apo (A), ATP-bound (B), and α,β-meATP–bound (C) states. The conformations of the lower body and the transmembrane region change upon ATP binding. In the α,β-meATP–bound state, the transmembrane region and the region connecting the transmembrane region and the membrane side of the lower body exist in equilibrium between the closed and open conformations, with exchange rates slower than the chemical shift differences (<100 s⁻¹).

In the α,β-meATP–bound state, the observation of two resonances each for A330, M325, M339, and M351 in the α,β-meATP–bound state (Fig. 3), together with the temperature-dependent population shift, suggests that the transmembrane region and the membrane side of the lower body exist in conformational equilibrium between the closed and open conformations, with slower exchange rates than the chemical shift difference (<100 s⁻¹) (Fig. 4C). The population of the open conformation in the state bound to α,β-meATP at 293 K, which was calculated from the signal intensities of the resonances from M339, is ∼35% (see SI Text for details), in good agreement with the currents evoked by α,β-meATP relative to the currents evoked by ATP in the TEVC recordings of ΔzfP2X₄-A′ (Fig. S5C). These results suggest that the small population of the open conformation of zfP2X₄ causes the partial activation in the α,β-meATP–bound state.

The small chemical shift difference between the apo and α,β-meATP–bound states observed for M325 (Fig. 3D) suggests that the closed conformation of the membrane side of the lower body in the α,β-meATP–bound state is slightly different from that in the apo state. The chemical shifts of the resonances from the other methionine resonances suggest that, even in the α,β-meATP–bound state, the conformation of the extracellular side of the lower body is almost identical to that in the ATP-bound state (Fig. 4C). Therefore, the conformational changes of the extracellular and transmembrane sides, which are observed in the crystal structures of zfP2X₄ in the apo and ATP-bound states, are uncoupled in the α,β-meATP–bound state.

In the crystal structure of zfP2X₄ in the ATP-bound state, the regions previously referred to as the “dorsal fin” and the “left flipper” (Fig. S8A) directly bind to ATP and adopt markedly different conformations from the conformations in the apo state, and the conformation of the left flipper is closely associated with the conformations of the membrane side of the lower body and the transmembrane region (Fig. S8B). The residues in the left flipper (N296 and R298) bind to the phosphates of ATP, in which one of the oxygen atoms is replaced by a methylene group in α,β-meATP (Fig. S1A), whereas the ribose and adenine base of ATP bind to the residues on the dorsal fin (L217 and L232) and the extracellular side of the lower body (T189 and L191) (Fig. S8A). Therefore, in the α,β-meATP–bound state, the perturbation of the interaction between the left flipper and ATP would induce changes in the conformational equilibrium in the left flipper, the membrane side of the lower body, and the transmembrane region, whereas the ATP-binding mode and the conformations of the dorsal fin and the extracellular side of the lower body would be similar to the conformations in the ATP-bound state (Fig. S8B).

Fig. S8.

Conformation in the α,β-meATP–bound state. (A) Distribution of the residues that exhibited conformational equilibrium in the α,β-meATP–bound state. One of the subunits from the crystal structure of zfP2X₄ in the ATP-bound form (PDB ID code 4DW1) is shown in ribbons. The dorsal fin and the left flipper are cyan and yellow, respectively. The right flipper and the lower body are light yellow. ATP is depicted by red sticks. T189 and L191, which exist on the extracellular side of the lower body, are depicted by yellow sticks. L217 and L232, which exist on the dorsal fin, are depicted by magenta sticks. N296 and R298, which exist on the left flipper, are depicted by cyan sticks. M325, A330, L339, and L351, which exist in conformational equilibrium in the α,β-meATP–bound state (Fig. 3), are depicted by green sticks. M108, M249, M255, and M268, which had chemical shifts in the α,β-meATP–bound state that were similar to the chemical shifts in the ATP-bound state (Fig. 4), are depicted by black sticks. The oxygen atom that connect the α- and β-phosphates and the oxygen atoms in the 2′ and 3′ positions of ATP are white. Dummy atoms generated by Orientations of Proteins in Membranes (OPM), which represent membrane boundary planes, are gray. (B) Schematic diagrams of the conformation of the extracellular region of zfP2X₄ in the apo, ATP-bound, and α,β-meATP–bound states. One of the subunits and the ligands that bind to this subunit and/or its neighboring subunits are shown. In the ATP-bound state, the dorsal fin and the left flipper directly bind to ATP and adopt markedly different conformations from the conformations in the apo state, and the conformation of the left flipper is closely associated with the conformations of the membrane side of the lower body and the transmembrane region. In the α,β-meATP–bound state, the perturbation of the interaction between the left flipper and ATP would induce changes in the conformational equilibrium in the left flipper, the membrane side of the lower body, and the transmembrane region whereas the ATP-binding mode and the conformations of the dorsal fin and the extracellular side of the lower body would be similar to the ATP-binding mode and the conformations in the ATP-bound state. (C) Schematic diagrams of the conformation of the extracellular region of zfP2X₄ in the TNP-ATP–bound state. In the TNP-ATP–bound state, the conformation and the ATP-binding mode of the dorsal fin would be perturbed by the trinitrophenyl group, leading to the closure of the lateral cation access pathway.

A330 is located near D61 and Q329, which were proposed to form the lateral cation pathway, because the ATP-evoked currents were affected by the chemical modification of these residues (42). Therefore, the difference between the ¹H chemical shift of A330 in the ATP-bound state and chemical shift in the TNP-ATP–bound state (Fig. S7A) suggests that the conformation of the lateral cation pathway in the ATP-bound state is different from that in the TNP-ATP–bound state. In contrast, the chemical shifts of the resonances from M339 and M351, as well as the resonances from the endogenous methionine residues, were almost identical to the chemical shifts in the ATP-bound state (Fig. S7 B and C), suggesting that the antagonism induced by TNP-ATP is not due to the closure of the pore in the transmembrane region. In the crystal structure of zfP2X₄ in the ATP-bound state, the 2′ and 3′ hydroxyl groups of ATP, to which a trinitrophenyl group is attached in the case of TNP-ATP, are recognized by the dorsal fin (Fig. S8A), and the dorsal fin is proposed to form a lateral cation access pathway through the fenestrations above the ion channel pore (18, 19, 42, 43). Therefore, in the TNP-ATP–bound state, the conformation and the ATP-binding mode of the dorsal fin would be perturbed by the trinitrophenyl group, leading to the closure of the lateral cation access pathway (Fig. S8C).

Previous NMR analyses of membrane proteins embedded in the lipid bilayers of rHDLs revealed that the conformational equilibria and the functions of membrane proteins are affected by the lipid bilayer environments (32, 33). Our NMR spectra of ΔzfP2X₄-A′ in DDM micelles demonstrate that the exchange rate and/or the population of the conformational equilibrium in the α,β-meATP–bound state in DDM micelles are markedly different from the exchange rate and/or the population in rHDLs, and not in agreement with the current evoked by α,β-meATP in the TEVC recordings (Fig. S9) (see SI Text for details). Therefore, we conclude that NMR investigations of zfP2X₄ in the lipid bilayer environments of rHDLs are necessary for accurate measurements of the exchange rates and the populations in conformational equilibrium.

Fig. S9.

NMR spectra of ΔzfP2X₄–A′ in DDM micelles. (A) Measurement of [³H]ATP saturation binding to the purified ΔzfP2X₄–A′ in DDM micelles. (B) Overlaid ¹H-¹³C HMQC spectra of [²H-11AA, α,β,γ-²H, methy-¹³C-Met]ΔzfP2X₄–A′, embedded in DDM micelles, in the apo state (black) and the ATP-bound state (red). The regions with resonances from methionine residues are shown, and the assigned resonances are indicated. (C) ¹H-¹³C HMQC spectra of [²H-11AA, α,β-²H, methy-¹³C-Met]ΔzfP2X₄–A′/L339M and [²H-11AA, α,β-²H, methy-¹³C-Met]ΔzfP2X₄–A′/L351M, embedded in DDM micelles, in the α,β-meATP–bound state (cyan), the ATP-bound state (red), and the apo state (black). Only the regions with the M339 and M351 methyl resonances are shown. The centers of the resonances from M339 and M351 are indicated with dots. Dashed lines represent the ¹H chemical shifts of the resonances from M339 or M351 in the α,β-meATP–bound state.

SI Text

Population of the Open and Closed Conformations.

The ¹H transverse relaxation times of the resonances from M325 in the apo state and the ATP-bound state, which were calculated from the linewidths, are ∼6 ms (linewidth, 50 Hz) and ∼10 ms (linewidth, 30 Hz), respectively, and the ¹H transverse relaxation time of the resonances from M351 in the apo state and the ATP-bound state are ∼10 and ∼6 ms, respectively (Fig. 3 C and D). These linewidths suggest that, in the cases of the resonances from M325 in the apo state and M351 in the ATP-bound state, the signal intensity losses during the 8-ms insensitive nuclei enhanced by polarization transfer (INEPT) evolution time are approximately twice the signal intensity losses in the cases of the resonances from M325 in the ATP-bound state and M351 in the apo state, respectively. In contrast, the linewidths of A330 and M339 in the apo state were similar to the linewidths in the ATP-bound state (Fig. 3 A and B). The population of the open conformation estimated from the signal intensities in the α,β-meATP–bound state at 310 K, along with the aforementioned signal losses during the INEPT evolution time, is ∼60% for M325, A330, M339, and M351.

The NMR spectra in Fig. 3 were recorded at either 310 K or 318 K whereas the TEVC experiments were performed at room temperature (∼293 K). According to the van’t Hoff elation [ln(K) = ΔH/RT − ΔS/R] and the NMR signal intensities at 310 and 318 K, the calculated population of the open conformation at 293 K was 35% although we could neither observe the resonances from zfP2X₄ in rHDLs with sufficient sensitivity at temperatures lower than 310 K, due to line broadening, nor perform the TEVC experiments at 310 K.

Conformational Equilibrium of zfP2X₄ in DDM Micelles.

To examine the effect of the lipid bilayer environment on the conformational equilibrium of zfP2X₄, NMR experiments with zfP2X₄ were performed in DDM micelles. The dose-dependent curve of the [³H]-ATP binding assays of the purified ΔzfP2X₄-A′ in DDM micelles revealed that the EC₅₀ value and the Hill coefficient were 51 nM and 1.9, respectively (Fig. S9A), which are similar to the EC₅₀ value and the Hill coefficient in rHDL, and that more than 90% of the purified ΔzfP2X₄-A′ in DDM retained the ATP-binding activity.

In the ¹H-¹³C HMQC spectra of [²H-10AA, α, β, γ-²H-methyl-¹³C-Met] ΔzfP2X₄–A′-L351M in DDM micelles bound to ATP, the chemical shifts of all of the methionine resonances were different from the chemical shifts in the apo state (Fig. S9B). In the spectra of ΔzfP2X₄–A′-L339M in DDM micelles, the ¹H chemical shift of the resonance from M339 in the apo state (1.80) exhibited a larger upfield shift than in the ATP-bound state (1.91 ppm) (Fig. S9C), suggesting that the resonances observed in the apo and ATP-bound states correspond to the closed and open conformations, respectively.

In the ¹H-¹³C HMQC spectra of [²H-10AA, α, β, γ-²H-methyl-¹³C-Met] ΔzfP2X₄–A′-L351M and ΔzfP2X₄–A′-L339M in DDM micelles bound to α,β-meATP, one resonance each was observed for M351 and L339, respectively, with chemical shifts almost identical to the chemical shifts in the ATP-bound state (Fig. S9C). Therefore, the exchange rate and/or the population of the conformational equilibrium in the α,β-meATP–bound state in DDM micelles are markedly different from the exchange rate and/or the population in rHDLs, and not in agreement with the currents evoked by α,β-meATP relative to currents evoked by ATP in the TEVC recordings of ΔzfP2X₄-A′ (Fig. S5C).

Materials and Methods

See SI Materials and Methods for the reagents (44, 45) and the methods for the expression and purification of zfP2X₄, preparation of zfP2X₄ in rHDLs, [³H] ATP binding assay (46), NMR experiments (47), and the two-electrode voltage clamp electrophysiology (48–50).

SI Materials and Methods

Reagents.

All reagents were purchased from Wako Chemicals or Nacalai Tesque, unless otherwise noted. Amino acid-deficient ESF921 medium with modified hydrolysate (ESF921 ΔAA) was purchased from Expression Systems (catalog no. 96–275). Aspartic acid (D) sodium salt, glutamic acid (E) sodium salt, asparagine (N), and glutamine (Q) were dissolved at 6, 8, 6, and 12 mg/mL, respectively, in ESF921 ΔAA (4× DENQ stock). Lysine (K) hydrochloride, hydroxyproline (O), proline (P), arginine (R) hydrochloride, and serine (S) were dissolved at 100, 50, 50, 100, and 20 mg/mL, respectively, in H₂O (100× KOPRS stock). Isoleucine (I), threonine (T), and valine (V) were dissolved at 12, 10, and 10 mg/mL, respectively, in H₂O (20× ITV stock). The [²H]-algal amino acid mixture [Cambridge Isotope Laboratories (CIL)] was dissolved at 25 mg/mL in H₂O with gentle heating, and ultrafiltrated with a Centriconplus-70 (10-kDa molecular mass cutoff; Millipore). The [²H]-alanine and [α,β,β,-²H₃, methyl-¹³C]-methionine were synthesized by the enzymatic deuteration of nonlabeled alanine and [methyl-¹³C]-methionine (CIL), with Escherichia coli cystathionine-γ-synthase, as previously described (44), and dissolved at 100 mg/mL and 30 mg/mL in H₂O, respectively. The [α-²H, methyl-¹³C]-alanine was synthesized by the enzymatic deuteration of [methyl-¹³C]-alanine (ISOTEC) with E. coli tryptophan synthase, as previously described (45), and dissolved at 50 mg/mL in H₂O, respectively. The [ring-²H]-tyrosine (CIL) was dissolved at 7.5 mg/mL in 80 mM NaOH. The [β-²H]-DL-cystine (CIL) was dissolved at 13.5 mg/mL in 250 mM HCl. Cystine dihydrochloride was dissolved at 5.2 mg/mL in 100 mM HCl. The other amino acids were individually dissolved at 8–200 mg/mL in H₂O.

Expression and Purification of zfP2X₄.

The cDNA fragment encoding a truncated zebrafish P2X₄ with residues Ser28 through Lys365 and an N-terminal EGFP fused to an octahistidine-tag, which was kindly provided from Eric Gouaux, Oregon Health & Science University, Portland, OR, was amplified by PCR and cloned into the pFastBac1 vector (Invitrogen) via the BamHI-HindIII sites. Mutations were introduced by a QuikChange site-directed mutagenesis kit (Stratagene). To overcome the overlap of the methionine signals, the M364L mutation was introduced in all constructs, unless otherwise stated.

Sf9 cells (Invitrogen) were routinely maintained at 27 °C in Grace’s supplemented medium (GIBCO) containing 10% (vol/vol) FBS (Biowest), 0.1% Pluronic F-58 (GIBCO), 50 international units per mL penicillin, 50 mg/mL streptomycin, and 0.125 mg/mL amphotericin B (GIBCO, as Antibiotic-Antimycotic). Recombinant baculoviruses were generated and amplified with the Bac-to-Bac system (Invitrogen), according to the manufacturer’s instructions.

The expresSF+ cells (SF+ cells; Protein Sciences) used for the expression of zfP2X₄ were routinely maintained, as previously described (34). For the production of [methyl-¹³C-Met] zfP2X₄, SF+ cells in Sf-900 II medium were centrifuged at 200 × g and resuspended in methionine-depleted ESF921 medium (Expression Systems), at about 2 × 10⁶ cells per mL. For the expression of nonlabeled zfP2X₄, Sf-900 II medium was used instead of the methionine-depleted ESF921 medium. The cells were inoculated with the high-titer virus stock (30-40 mL per 1 L of cells). At 20 h postinfection, 200 mg/L [α,β,β,-²H₃, methyl-¹³C]-methionine and 5 mg/L E-64 (Peptide Institute) were added, and the cells were further cultured at 293 K. The cells were harvested 92 h postinfection by centrifugation at 800 × g, and the resulting cell pellets were stored at −80 °C. For the expression of [²H-11AA, αβ-²H-, methyl-¹³C-Met] zfP2X₄, 25 mL of 4× DEQN stock, 2.5 mL of 1 M NaCl, 0.5 mL of 200 mg/mL lysine monohydrochloride, 1 mL of 50 mg/mL [²H]-glycine (CIL), 0.5 mL of 40 mg/mL [²H]-DL-serine (CIL), 1 mL of 10 mg/mL [ring-²H]-tryptophan (CIL), 1 mL of 30 mg/mL [α,β,β,-²H₃, methyl-¹³C]-methionine, 1 mL of 30 mg/mL histidine, 2.5 mL of 5.2 mg/mL cystine dihydrochloride, and 4 mL of 8 mg/mL tyrosine disodium salt hydrate (Sigma) were added to 50 mL of ESF921 ΔAA. SF+ cells in Sf-900 II medium (GIBCO) were centrifuged at 200 × g and resuspended in the medium described above, at about 2 × 10⁶ cells per mL. The cells were inoculated with 3–4 mL of the high-titer virus stock and maintained at 27 °C. At 20 h postinfection, 0.1 mL of 10 mg/mL β-chloro-l-alanine hydrochloride (Sigma), 0.1 mL of 5 mg/mL E-64, 1.3 mL of 100 mg/mL [²H]-alanine, 1 mL of 50 mg/mL [²H]-glycine, 0.5 mL of 10 mg/mL [²H]-valine (CIL), and 6 mL of 25 mg/mL [²H]-algal amino acid mixture were added, and the temperature was shifted to 20 °C. The cells were harvested at 92 h postinfection by centrifugation at 800 × g, and the resulting cell pellets were stored at −80 °C.

In the case of the preparation of [²H-5AA, α-²H-, methyl-¹³C-Ala] zfP2X₄, the constructs without the L364M mutation were used. For the expression of [²H-5AA, α-²H-, methyl-¹³C-Ala] zfP2X₄, 25 mL of 4× DEQN stock, 1 mL of 1 M NaCl, 1 mL of 100× KOPRS stock, 5 mL of 20× ITV stock, 1 mL of 50 mg/mL [²H]-glycine (CIL), 1 mL of 20 mg/mL methionine, 1 mL of 10 mg/mL tryptophan, 1 mL of 30 mg/mL histidine, 2 mL of 13.5 mg/mL [β-²H]-DL-cystine, 1 mL of 7.5 mg/mL [ring-²H]-tyrosine, and 0.3 mL of 1 M NaOH were added to 50 mL of ESF921 ΔAA. SF+ cells in Sf-900 II medium (GIBCO) were centrifuged at 200 × g and resuspended in the medium described above, at about 2 × 10⁶ cells per mL The cells were inoculated with 3–4 mL of the high-titer virus stock and maintained at 27 °C. At 20 h postinfection, 0.2 mL of 10 mg/mL β-chloro-l-alanine hydrochloride (SIGMA), 0.1 mL of 5 mg/mL E-64, 1 mL of 50 mg/mL [α-²H, methyl-¹³C]-alanine, 1 mL of 50 mg/mL [²H]-glycine, 1 mL of 10 mg/mL [²H]-phenylalanine (CIL), and 1 mL of 10 mg/mL [²H]-leucine (CIL) were added, and the temperature was shifted to 20 °C. The cells were harvested at 92 h postinfection by centrifugation at 800 × g, and the resulting cell pellets were stored at −80 °C.

All of the following procedures were performed either on ice or in the cold room (4 °C) unless otherwise noted. The pellet obtained from 1 L of cell culture was resuspended in 75 mL of buffer A [50 mM Tris, pH 8.0, 150 mM NaCl, 0.25 mM 4-(2-aminoethyl) benzenesulfonyl fluoride hydrochloride (AEBSF), 0.75 μM aprotinin (Wako Chemicals), 10 μM leupeptin hemisulfate (Peptide Institute), 15 μM pepstatin A (Peptide Institute), and 14 μM E-64], supplemented with 2 mM MgCl₂ and 0.07 unit/mL Apyrase (Sigma-Aldrich). The cells were disrupted by nitrogen cavitation (Parr Bomb) at 600 psi for 30 min. The cell lysate was centrifuged at 800 × g for 10 min, and the resulting supernatant was centrifuged at 100,000 × g for 60 min. The membrane pellet was resuspended by Dounce homogenization in buffer A, supplemented with 2 mM MgCl₂, 0.25 unit/mL Apyrase, and 20% (wt/vol) glycerol, and was stored at −80 °C.

The membranes obtained from 1 L of cell culture were solubilized in 60 mL of buffer A, supplemented with 20% (wt/vol) glycerol and 2% (wt/vol) n-dodecyl-β-d-maltopyranoside (DDM) (Dojindo) for 1.5 h and were then centrifuged at 140,000 × g for 40 min. The supernatant was batch incubated for 2 h with 2 mL of TALON metal affinity resin (Clontech). The resin was washed with 20 mL of buffer B (50 mM Tris, pH 8.0, 150 mM NaCl, 1 mM DDM), 20 mL of buffer B supplemented with 10 mM imidazole, 40 mL of buffer B supplemented with 20 mM imidazole, and 20 mL of buffer B supplemented with 25 mM imidazole. The protein was eluted with 10 mL of buffer B supplemented with 250 mM imidazole and concentrated using a centrifugal filter device (AmiconUltra-15, 100-kDa molecular mass cutoff; Millipore) to 500–700 μL.

In the cases of the NMR experiments of zfP2X₄ in DDM micelles, 100 U of thrombin (GE Healthcare) in buffer B was then added to the concentrated protein and incubated at room temperature for 4 h. The reaction mixture was further purified by size exclusion chromatography, using Superdex200 10/300 GL (GE Healthcare) or Superdex200 Increase 10/300 GL (GE Healthcare) in buffer C (20 mM Hepes-NaOH, pH 7.0, 80 mM NaCl, 20 mM KCl, 1 mM DDM) or buffer D (20 mM Hepes-NaOH, pH 7.2, 150 mM NaCl, 1 mM DDM) and concentrated using a centrifugal filter device (AmiconUltra-4, 30-kDa molecular mass cutoff; Millipore), while simultaneously exchanging the buffer to buffer E [20 mM Hepes-NaOH, pH 7.0, 80 mM NaCl, 98% (vol/vol) D₂O].

Preparation of zfP2X₄ in rHDLs.

MSP1 was expressed and purified as previously described (33). Solutions of 1-palmitoyl-2-oleoyl-phosphatidylcholine (POPC) (Avanti Polar Lipids) and 1-palmitoyl-2-oleoyl-phosphatidylglycerol (POPG) (Avanti Polar Lipids) in chloroform were mixed at a molar ratio of 3: 2. The solvent was evaporated under a nitrogen atmosphere and dried in vacuo to form a lipid film. The film was solubilized by the addition of buffer F (20 mM Hepes-NaOH, pH 7.2, 150 mM NaCl), supplemented with 100 mM sodium cholate, for a final lipid concentration of 50 mM.

All of the following procedures were performed either on ice or in the cold room (4 °C), unless otherwise noted. The purified zfP2X₄ in the immobilized metal ion affinity chromatography eluate was further purified by size exclusion chromatography using Superdex200 10/300 GL (GE Healthcare) or Superdex200 Increase 10/300 GL (GE Healthcare) in buffer D and concentrated using a centrifugal filter device (AmiconUltra-4, 100-kDa molecular mass cutoff; Millipore) to 400–1,300 μL. The zfP2X₄ trimer, MSP1, and lipids were mixed at a final molar ratio of 1:90:4,500 and incubated for 1 h. To remove the detergents, 67% (wt/vol) of Bio-Beads SM-2 (Bio-Rad) was added to the solution and incubated for 2.5 h with gentle mixing. The supernatant was batch incubated for 2 h with TALON resin. The resin was washed with 60 column volume (CV) of buffer F, followed by 10 CV of buffer F, supplemented with 10 mM imidazole. The zfP2X₄ in rHDLs was eluted from the resin with 3–5 CV of buffer F, supplemented with 200 mM imidazole, and concentrated using a centrifugal filter device (AmiconUltra-15, 10-kDa molecular mass cutoff; Millipore) to 500–700 μL. Thrombin (250 U) in buffer F was added to the concentrated protein and incubated at room temperature for 4 h. The reaction mixture was further purified by size exclusion chromatography, using Superdex200 10/300 GL (GE Healthcare) or Superdex200 Increase 10/300 GL (GE Healthcare) in buffer F. The fractions corresponding to the Stokes diameter of ∼12 nm were applied to the TALON resin and washed with 5 CV of buffer F. The flow-through and the wash were concentrated using a centrifugal filter device (AmiconUltra-4, 10-kDa molecular mass cutoff; Millipore), while simultaneously exchanging the buffer to buffer E.

[³H] ATP Binding Assay.

In the saturation binding assay, solutions containing 10–2,000 nM [2,8-³H] ATP tetrasodium salt (PerkinElmer), with or without 1 mM unlabeled ATP, were added to the purified ΔzfP2X₄-A′. In the competition binding assay, 50 nM [³H] ATP and 0.05–500 μM α,β-meATP were added to 5 nM ΔzfP2X₄-A′. The reaction mixtures were incubated on ice for 1–2 h and then applied to a NAP-5 size-exclusion column (GE Healthcare). The fractions containing ΔzfP2X₄-A′ were mixed with Optiphase Supermix (PerkinElmer), and the radioactivity was detected with a liquid scintillation counter (MicroBeta2; PerkinElmer). In the saturation binding assay, the amount of bound [³H] ATP was estimated from the differences in the counts with and without unlabeled ATP, and the dissociation constants and the Hill constant were determined by nonlinear least square fitting, using Origin 7.0 (Origin Lab). In the competition binding assay, the inhibition constants were determined by nonlinear least square fitting, using Origin 7.0 (Origin Lab) and the Cheng and Prusoff equation (46).

NMR Experiments.

All spectra were recorded with a Bruker Avance 600, 800, or 950 spectrometer equipped with a cryogenic probe. ¹H-¹³C HMQC spectra (47) were recorded for 0.4–17 μM [αβ-²H-, methyl-¹³C-Met], [²H-11AA, αβ-²H-, methyl-¹³C-Met], [²H-11AA, αβγ-²H-, methyl-¹³C-Met], or [²H-5AA, α-²H, methyl-¹³C-Ala] ΔzfP2X₄-A′ trimer and its mutants, in rHDLs and in DDM micelles in buffer E. In the NMR experiments with ΔzfP2X₄-A′ in the ATP-, α,β-methylene ATP–, and TNP-ATP–bound states, ATP, α,β-methylene ATP, and TNP-ATP were added to the samples at a final concentration of 1 mM, 1 mM, and 0.5 mM, respectively. In total, 1,024 × 64–128 complex points were recorded, and 56–512 scans per free induction decay with 1-s interscan delays gave rise to an acquisition time of 3–24 h for each spectrum. The spectra were referenced to 3-(trimethylsilyl)-1-propanesulfonic acid sodium salt in both the ¹H and ¹³C dimensions. All of the spectra were processed and analyzed by Topspin version 2.1 or 3.1 (Bruker).

Two-Electrode Voltage Clamp Electrophysiology.

The cRNAs encoding the rat P2X₄ receptor, ΔzfP2X₄-A′, ΔzfP2X₄-A′/L339M, and ΔzfP2X₄-A′/L351M with N-terminal EGFP fusions were transcribed from pGEMHE plasmids (48), using an mMessage mMachine T7 ULTRA kit (Ambion). Xenopus laevis oocytes were prepared, and RNA (5 ng for zebrafish P2X₄, and 20 ng for rat P2X₄) was injected into the oocytes (49). The oocytes were incubated at 18 °C in Barth’s solution, containing 88 mM NaCl, 1 mM KCl, 2.4 mM NaHCO₃, 10 mM Hepes, 0.33 mM Ca(NO₃)₂, 0.41 mM CaCl₂, and 0.82 mM MgSO₄, pH 7.4, supplemented with 50 µg/mL gentamicin, and were used for recording after 2–4 d. Recording solutions contained 90 mM NaCl, 1 mM KCl, 2 mM MgCl₂, 10 mM Hepes, pH 7.4, and each nucleotide (ATP, α,β-meATP, and TNP-ATP). Nucleotide solutions were freshly prepared each day. Oocytes were held at −80 mV with a bath-clamp amplifier (OC-725C; Warner Co.), and macroscopic currents were recorded and analyzed using the Digidata 1550 and pClamp 10 software (Molecular Devices) (50). We measured n = 4 for all data.