Structural insights into the orthosteric inhibition of P2X receptors by non-ATP analog antagonists

P2X receptors are extracellular ATP-gated ion channels that form homo- or heterotrimers and consist of seven subtypes. They are expressed in various tissues, including neuronal and nonneuronal cells, and play critical roles in physiological processes such as neurotransmission, inflammation, pain, and cancer. As a result, P2X receptors have attracted considerable interest as drug targets, and various competitive inhibitors have been developed. However, although several P2X receptor structures from different subtypes have been reported, the limited structural information of P2X receptors in complex with competitive antagonists hampers the understanding of orthosteric inhibition, hindering the further design and optimization of those antagonists for drug discovery. We determined the cryogenic electron microscopy (cryo-EM) structures of the mammalian P2X7 receptor in complex with two classical competitive antagonists of pyridoxal-5'-phosphate derivatives, pyridoxal-5'-phosphate-6-(2'-naphthylazo-6'-nitro-4',8'-disulfonate) (PPNDS) and pyridoxal phosphate-6-azophenyl-2′,5′-disulfonic acid (PPADS), and performed structure-based mutational analysis by patch-clamp recording as well as molecular dynamics (MD) simulations. Our structures revealed the orthosteric site for PPADS/PPNDS, and structural comparison with the previously reported apo- and ATP-bound structures showed how PPADS/PPNDS binding inhibits the conformational changes associated with channel activation. In addition, structure-based mutational analysis identified key residues involved in the PPNDS sensitivity of P2X1 and P2X3, which are known to have higher affinity for PPADS/PPNDS than other P2X subtypes.


Introduction
ATP not only serves as a cellular energy currency but also plays a key role in signal transmission for cellular stimulation between cell surface receptors 1 .P2X receptors are the family of cation channels activated by extracellular ATP and are widely expressed in the mammalian nervous, respiratory, reproductive, and immune systems 2 3 4 5 .There are seven subtypes (P2X1-P2X7) in the family, each of which plays distinct roles in physiological and pathophysiological functions via homo-or heterotrimerization 6,7 .In recent years, there has been growing interest in the development of drugs targeting the P2X family due to its involvement in various physiological and pathological conditions, and various antagonists have been developed 8 .Some have progressed to clinical trials 9 , and Gefapixant, a P2X3 receptor antagonist for chronic cough, was already on the market after the clinical study 10 .
-4 -Several attempts have been made to optimize pyridoxal phosphate derivatives as P2X antagonists 22,23,24,25 .For example, the introduction of bulky aromatic groups at the carbon linker in PPADS was attempted to improve in the subtype specificity profiles, potentially opening up new avenues for targeted drug development and therapeutic intervention 23 .Despite the recent significant increase in structural information on P2X receptors 26,27,28,29,30,31,32,33,34 , the lack of structural information on P2X receptors in complex with pyridoxal phosphate derivative inhibitors has hampered the rational design and optimization of pyridoxal phosphate derivatives for drug discovery targeting P2X receptors.
In this work, we determined the cryogenic electron microscopy (cryo-EM) structures of the panda P2X7 receptor in complex with PPADS and PPNDS.The structures revealed the orthosteric binding site for these pyridoxal phosphate derivatives, and structural comparison with the previously determined apo and ATP-bound P2X7 receptors 31,34 showed PPADS/PPNDS-dependent structural rearrangement at the orthosteric binding site for channel inactivation.Further mutational analysis by electrophysiological recording identified key residues of human P2X1 and P2X3 that show high sensitivity to pyridoxal phosphate derivative inhibitors.

Structural determination and functional characterization
To gain insight into the mechanism of P2X receptor inhibition by pyridoxal phosphate derivatives, we used giant panda (Ailuropoda melanoleuca) P2X7, whose structures in complex with allosteric modulators have been reported 29 .Notably, panda -5 -P2X7 shares 85% identity with the human P2X7 receptor and exhibits high and stable expression profiles suitable for structural studies 29 .In whole-cell patch clamp recordings using HEK293 cells transfected with full-length panda P2X7 (pdP2X7), the application of 10 µM PPNDS and 100 µM PPADS blocked approximately 50% of the ATP-dependent currents from pdP2X7 (Fig. S1, A-B and E-F), which correlates well with the properties of P2X7 receptors moderately inhibited by PPADS/PPNDS 17 , 35 .
We then expressed and purified the previously reported crystallization construct of pdP2X7cryst 29 .The purified pdP2X7cryst was reconstituted into lipid nanodiscs, mixed with PPNDS and PPADS, separately, and subsequently subjected to single-particle cryo-EM (Fig. S2-S5).The structures of pdP2X7 in the presence of PPNDS and PPADS were determined at 3.3 Å and 3.6 Å, respectively (Table S1).
The overall structures are similar and show the trimeric architecture of P2X receptors, consisting of the extracellular domain and two transmembrane (TM) helices, with each protomer resembling the dolphin shape, consistent with the previously reported P2X structures 36 (Fig. 1 and Fig. S6, A-B).More importantly, we identified the residual EM densities at the agonist binding site that fit into the shape of PPNDS and PPADS (Fig. 1).
It should be noted that while the pyridoxal phosphate groups in PPNDS and PPADS are shared, the naphthylazo group of PPNDS is significantly bulkier than the azophenyl group of PPADS, which aided our assignment of compound binding poses for each EM density (Fig. 1).Consistent with antagonist binding, the TM domain adapts to the closed conformation of the channel, as similarly observed in the previously reported closed state structures of P2X7 receptors 29 , 34 .

Orthosteric binding site
In the PPNDS-bound and PPADS-bound structures, PPNDS and PPADS molecules bind to essentially the same orthosteric site, consistent with both PPNDS and PPADS being pyridoxal phosphate derivatives (Fig. 2 and 3 and Fig. S7).Furthermore, the residues involved in PPNDS and PPADS largely overlap with the residues at the ATP binding site in the previously reported ATP-bound P2X7 structure (Fig. 4) 34 , and many of the residues are highly conserved among P2X receptors and have been shown to be crucial for P2X activation 27 , 37 , 38 , 39 , which is consistent with both of them being competitive inhibitors.
In both structures, the pyridoxal phosphate group has extensive interactions with the receptor, whereas the other parts of the compounds, a naphthylazo group with two sulfonic acid groups and a nitro group (PPNDS) and an azophenyl group with two sulfonic acid groups (PPADS), have fewer interactions (Fig. 2 and 3).
The phosphate group of PPNDS and PPADS interacts directly with the side chain of Arg294 and possibly also with Lys145, possibly via a water molecule (Fig. 2B and 3B and Fig. S7), although it is difficult to conclude the existence of the water molecule at this site due to the limited resolution of our structures.Furthermore, in the PPADS-bound structure, Lys64 mediates an additional interaction with the phosphate group of Lys64.
These extensive interactions between the phosphate group and the receptor resemble those with the phosphate groups of ATP (Fig. 4A).Furthermore, the Asn292 and Lys311 residues are similarly involved in the interaction with the hydroxyl group of the pyridoxal part of PPNDS and PPADS (Fig. 2B and 3B).
It is interesting to note that while the naphthylazo group of PPNDS and the azophenyl group of PPADS have certain structural differences, the two sulfonic acid groups of both -7 -PPNDS and PPADS form hydrogen bonds with the side chains of the same residues of the receptor, Lys66 and Gln143 (Fig. 2B and 3B).
Finally, to verify the binding mode of the pyridoxal phosphate derivatives, we performed molecular dynamics (MD) simulations of the higher-resolution PPNDS-bound structure embedded in lipids (Fig. S8).The overall structures were stable during the simulations, and PPNDS remained stably bound to the receptor (Fig. S8).

Structural comparison and inhibition mechanism
To gain insights into the mechanisms of the orthosteric inhibition of P2X receptors by PPNDS and PPADS, we superimposed our structures and the previously determined ATPbound P2X7 structure onto the apo-state P2X7 structure (Fig. 5).As the PPNDS-bound structure and PPADS-bound structure are very similar, with 0.52 Å RMSD values for the Cα atoms of 960 residues, only the comparison with the higher-resolution PPNDS-bound structure is described in the following discussion.
First, the activation of P2X receptors from the apo (closed) state to the ATP-bound (open) state is known to require motions of both the head and left flipper domains 40,41,42 (Fig. 5A), since these motions are coupled to the movement of the lower body domain, which is directly connected to the TM domain for channel opening (Fig. 5A).
In the PPNDS-bound structure, while we observed motion of the head domain similar to that in the ATP-bound structure, there was only a small structural change in the left flipper domain (Fig. 5A).Consequently, there was no structural change in the lower body domain or associated gating motion of the TM domain (Fig. 5A).In the ATP-bound structure, the three phosphate groups of ATP in the U-shaped conformation pushed down -8 -the left flipper (Fig. 4A and 5A), whereas both PPNDS and PPADS possess only one phosphate group, so there was no corresponding downward movement in the left flipper domain (Fig. 2 and 5A).
To summarize, the structural comparison indicates that PPNDS and PPADS inhibit the ATP-dependent activation of P2X receptors by occupying the orthosteric site but preventing the downward movement of the left flipper domain, which is required for channel opening (Fig. 5B).

Structure-based mutational analysis
To analyze the mechanism of P2X receptor binding to the pyridoxal-5'-phosphate derivatives, we performed structure-based mutational analysis by whole-cell patch-clamp recording of pdP2X7 (Fig. 6 and Fig. S1).Since most of the residues involved in PPNDS and PPADS binding overlap with the conserved residues involved in ATP binding (Fig. 4B), we did not generate mutants targeting these residues (Lys64, Lys66, Asn292, Arg294, and Lys311), as such mutations are known to severely affect or abolish ATP-dependent gating of P2X receptors 27,37,38,39 .Instead, we aimed to mutate the residues surrounding the ATP-binding site, which differ among P2X receptor subtypes, as such residues may be important for the subtype-specific differences in the affinity of the pyridoxal-5'phosphate derivative to P2X receptors.To design such mutants, we compared the residues surrounding the ATP-binding site in our structure (Gln143, Val173, Ile214, Gln248 and Tyr288) with those in human P2X1 (hP2X1) (Fig. 4B), which possesses high affinity for both PPNDS and PPADS 21,35 , and performed structural comparison of our structures with the AlphaFold-based structural model of human P2X1 43 (Fig. 6A).Based on these comparisons, we mutated these residues to the corresponding amino acid residues in hP2X1 (Q143K, V173D, I214K, Q248K, and Y288S) or to alanine (Q143A, Q248A, and Y288A) and performed whole-cell patch clamp recording to analyze the effect of PPNDS on these mutants (Fig. 6B).
In the mutational analysis, among three mutations to the lysine residue, two (Q143K, I214K) showed significantly increased sensitivity to PPNDS (Fig. 6B).In addition, the Q248K mutant showed slightly higher sensitivity to PPNDS, but not as significant an increase as the other two mutants (Fig. 6B).Interestingly, the mutation of Tyr288 to alanine instead of serine, as observed in hP2X1, significantly increased the affinity for PPNDS (Fig. 6B).In addition, the mutation of Val173 to aspartate, as observed in pdP2X7, significantly decreased the sensitivity to PPNDS (Fig. 6B).The two mutants Q143A and Q248A showed little or no decrease in PPNDS sensitivity (Fig. 6B).
Consistently, in the mutational analysis of the corresponding residues in pdP2X7 (Val173 and Tyr288), the mutation of Val173 to aspartate decreased PPNDS sensitivity, and the mutation of Tyr288 to alanine increased PPNDS sensitivity (Fig. 6B).
In summary, our mutational analysis based on structural comparison and sequence alignment identified several key residues involved in PPNDS sensitivity, particularly the residues involved in the subtype-specific difference in affinity for PPNDS.

Discussion
In this work, we determined the cryo-EM structure of pdP2X7 in complex with two classical non-ATP analog inhibitors, PPNDS and PPADS, of pyridoxal phosphate derivatives (Fig. 1) and revealed their orthosteric binding site (Fig. 2 and 3), which has high overlap with the ATP binding site (Fig. 4).In the cryo-EM structures, the phosphate group of PPNDS and PPADS appears to occupy the position of the -phosphate group of ATP in the ATP-bound structure (Figs.2-4).Structural comparison with the previously reported apo (closed) and ATP-bound (open) structures showed that, unlike ATP binding, the binding of PPNDS and PPADS does not induce the downward movement of the left flipper, an important movement for channel activation, providing mechanistic insights into channel inhibition by pyridoxal phosphate derivatives (Fig. 5).Finally, structurebased mutational analyses revealed several key residues important for PPNDS sensitivity, particularly for the subtype-specific difference in sensitivity (Fig. 6).
Besides PPADS and PPNDS in this study, TNP-ATP is well known as a classical -11 -competitive inhibitor for P2X receptors 21 .In the previously reported TNP-ATP bound P2X7 structure, TNP-ATP binding induces the conformational changes of the left flipper region but not of the head domain 31 .In contrast, PPADS and PPNDS binding induces the conformational changes of the head domain but not of the left flipper region (Fig. 5).

These contrasts would highlight the uniqueness of the competitive inhibition mechanism by pyridoxal phosphate derivatives as well as the diversity of competitive inhibition mechanisms of P2X receptors.
There are several previous reports characterizing the P2X receptor binding site for pyridoxal phosphate derivatives, especially PPADS 15,16,18,19 .Among them, the mutation of Glu249 (Lys249 in hP2X1) to lysine in the rat P2X4 (rP2X4) receptor was first determined to confer PPADS sensitivity on the PPADS-insensitive P2X4 receptor 18 .In pdP2X7, the Glu249 residue in rP2X4 corresponds to Gln248 (Fig. 4B), which is proximal to the hydroxyl moieties of the pyridoxal phosphate group of PPADS and PPNDS (Fig. 6A).More recently, the combination of docking simulation and electrophysiology showed that Lys70, Asp170, Lys190 and Lys249 participate in PPADS binding in hP2X1 15 .Among these four residues, Lys70 and Lys190 are directly involved in ATP binding 27 , and Lys249 corresponds to Glu249 in rP2X4, as shown in a previous study 18 .In addition, the mutation of Asp170 to cysteine increased sensitivity to PPADS 15 .
Consistently, in our structure, the Val173 residue in pdP2X7, corresponding to Asp170 in hP2X1, is located proximal to the pyridoxal phosphate group of PPADS, and hP2X1 D170A also showed an increased density relative to PPNDS, whose pyridoxal phosphate group is in common with that of PPADS.It should be noted that while a recent study by docking simulation and electrophysiology identified several residues involved in PPADS binding, as mentioned above, the orientation of PPADS in their docking model is very different from that in our cryo-EM structure.In the docking model, compared to our cryo-EM structure, PPADS shows an almost 180-degree rotation relative to the axis of rotation parallel to the membrane with the azide group of PPADS as the fulcrum 15 .This difference in the orientation of the compound may be why, in contrast to the results in our study, hP2X1 Lys140 and Lys215 are not predicted to be involved in PPADS binding in the previous docking model.
In the past, several types of pyridoxal phosphate derivatives have been identified as P2X inhibitors 13,23 , but most of them share a common weakness in subtype specificity, where they tend to show high affinity for both P2X1 and P2X3.For example, MRS 2257 was identified as the most active PPADS analog from the screening but has IC50 values of 5 nM and 22 nM for P2X1 and P2X3, respectively 22 .In addition, another compound termed 36j, derived from PPADS, shows improved subtype specificity for P2X3 (IC50: 60 nM for P2X3) but still shows a moderate inhibitory effect on P2X1 at 10 µM 23 .The probable reason it is difficult to obtain competitive inhibitors of P2X receptors, including pyridoxal phosphate derivatives, with subtype specificity is that the residues directly involved in ATP binding are strictly conserved among the subtypes (Fig. 4B).To overcome this situation, our work might facilitate the rational design of pyridoxal phosphate derivatives with strict subtype specificity for P2X receptors because we have not only defined the binding mode of pyridoxal phosphate derivatives to P2X receptors but also newly identified a subtype-specific residue for pyridoxal phosphate derivatives.
For example, Lys215 and Lys249 in hP2X1 are important for PPNDS sensitivity (Fig. 6C) but are not conserved in other P2X subtypes, including P2X3 (Fig. 4B), providing a clue for the design of more subtype-specific pyridoxal phosphate derivatives.
In conclusion, our structural and functional analyses provided mechanistic insights into -13 -the orthosteric inhibition mechanism of P2X receptors by the classical pyridoxal phosphate derivative P2X antagonist and identified key residues involved in compound sensitivity, especially the differential sensitivity between P2X subtypes, thereby potentially facilitating the development of subtype-specific compounds targeting P2X receptors, which have attracted widespread interest as therapeutic targets.

Nanodisc reconstitution
Soybean polar lipid extract (Avanti Polar Lipids, USA) was dissolved in chloroform, dried under a nitrogen stream, and then resuspended in reconstitution buffer (20 mg/ml soybean polar lipid, 20 mM HEPES pH 7.0, and 150 mM NaCl).Following a 1-hour incubation at room temperature, the lipid suspension was subjected to sonication for 5 minutes until the lipids reached a near-transparent state.Subsequently, DDM (Anatrace, USA) was added at a final concentration of 0.4% and incubated at room temperature overnight.The mEGFP-fusion pdP2X7, MSP2N2 protein, and soybean polar lipid were combined in a molar ratio of 1:3:180.This mixture was then incubated at 4 °C for 1 hour and further subjected to a 4-hour incubation with bio-beads (Bio-Rad, USA).After incubation, the bio-beads were removed via filtration, and the nanodisc fractions containing mEGFP-fusion pdP2X7 were bound to Ni-NTA (Qiagen, USA) resin preequilibrated with wash buffer (20 mM HEPES pH 7.5, 150 mM NaCl, 30 mM imidazole) and subsequently eluted using elution buffer (20 mM HEPES pH 7.5, 150 mM NaCl, 300 mM imidazole).To cleave the N-terminal EGFP, the elution was mixed with HRV3C protease and incubated at room temperature for 1 hour, followed by overnight incubation at 4 °C.The nanodisc-reconstituted pdP2X7 protein was separated through size-exclusion chromatography using a Superdex 200 Increase 10/300 column (Cytiva, USA) preequilibrated with SEC buffer (20 mM HEPES pH 7.5 and 150 mM NaCl) and subsequently concentrated to 0.9 mg/ml.P2X7 antagonists (PPNDS or PPADS) were added to the nanodisc-reconstituted pdP2X7 at a final concentration of 50 μM and incubated on ice for 1 hour before cryo-EM grid preparation.

EM data acquisition
For both the PPNDS-bound and PPADS-bound pdP2X7 samples, a total of 2.5 μl of the nanodisc-reconstituted pdP2X7 was applied to a glow-discharged holey carbon-film grid (Quantifoil, Au 1.2/1.3,300 mesh, USA).The grid was then blotted using a Vitrobot (Thermo Fisher Scientific, USA) system with a 3.0-second blotting time at 100% humidity and 4 °C, followed by plunge-freezing in liquid ethane.Cryo-EM data collection was carried out using a 300 kV Titan Krios microscope (Thermo Fisher Scientific, USA) equipped with a K3 direct electron detector (Gatan Inc., USA).The specimen stage temperature was maintained at 80 K. Movies were recorded using beam-image shift data collection methods 44 in superresolution mode, with a pixel size of 0.41 Å (physical pixel size of 0.83 Å), a magnification of 29,000, and defocus values ranging from -1.3 µm to -2.0 µm.The dose rate was set to 20 e-s -1 , and each movie consisted of 40 frames with an exposure of 50 e-Å -2 , resulting in each movie being 1.724 s long.

Image processing
A total of 9,664 and 4,692 movies for the PPNDS-bound and PPADS-bound pdP2X7 samples, respectively, were motion-corrected and binned with MotionCor2 45 with a patch The figures were generated by UCSF Chimera 48 .

Model building
The initial models of pdP2X7 were manually built starting from the previously reported pdP2X7 structure (PDB ID: 5U1L).Manual model building was performed using Coot 49 .
Real-space refinement was performed using PHENIX 50 .All structure figures were generated using PyMOL (https://pymol.org/).For the predicted structure of human P2X1, the previously generated model using AlphaFold and ColabFold was used 43,51,52 .The sequence alignment figure was generated using Clustal Omega 53 and ESPript 3.0 54 .

Electrophysiology
-17 -Human embryonic kidney 293 (HEK293) cells were purchased from Shanghai Institutes for Biological Sciences and cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum (FBS), 1% penicillin-streptomycin, and 1% GlutaMAX™ at 37 °C in a humidified atmosphere of 5% CO2 and 95% air 55,56 .Plasmids harboring hP2X1, hP2X3 or pdP2X7 were transfected into cells by calcium phosphate transfection 57 .Currents of hP2X1 and hP2X3 were recorded using nystatin (Sangon Biotech, China) perforated recordings to prevent rundown in current during multiple dose applications of ATP.Nystatin (0.15 mg/mL) was diluted with a high-potassium internal intracellular solution containing 75 mM K2SO4, 55 mM KCl2, 5 mM MgSO4, and 10 mM HEPES (pH 7.4).Currents of PdP2X7 receptors were recorded using a conventional whole-cell patch configuration.After 24-48 h of transfection, HEK293 cells were recorded at room temperature (25 ± 2 °C) using an Axopatch 200B amplifier (Molecular Devices, USA) with a holding potential of -60 mV.Current data were sampled at 10 kHz, filtered at 2 kHz, and analyzed using PCLAMP 10 (Molecular Devices, USA) for analysis.HEK293 cells were bathed in standard extracellular solution (SS) containing 2 mM CaCl2, 1 mM MgCl2, 10 mM HEPES, 150 mM NaCl, 5 mM KCl, and 10 mM glucose with the pH adjusted to 7.4.For conventional whole-cell recordings, the pipette solutions consisted of 120 mM KCl, 30 mM NaCl, 0.5 mM CaCl2, 1 mM MgCl2, 10 mM HEPES, and 5 mM EGTA with pH adjusted to 7.4.ATP and other compounds were dissolved in SS for P2X1 and P2X3 and applied to Y-tubes.For pdP2X7, ATP and other compounds were dissolved in 0Ca, 0Mg solution containing 150 mM NaCl, 10 mM glucose, 10 mM HEPES, 5 mM KCl, and 10 mM EGTA with the pH adjusted to 7.4 58 .PPNDS was purchased from APE ×BIO, and PPADS was purchased from MCE.The standard solution and 0Ca, 0Mg solution were formulated with compounds from Aladdin, and internal solutions were -18 -formulated with compounds from Sigma-Aldrich 59 .All electrophysiological recordings were analyzed using Clampfit 10.6 (Molecular Devices, USA).Pooled data are expressed as the mean and standard error (s.e.m.).Statistical comparisons were made using Bonferroni's post hoc test (ANOVA).* * p < 0.01 and * * * * p < 0.0001 were considered significant.

Molecular dynamics simulations
The energy-minimized models of the PPNDS-bound pdP2X7 were used as the initial structures for MD simulations.A large 1-palmitoyl-2-oleoyl-sn-glycero-3phosphocholine (POPC, 300 K) bilayer, available in System Builder of DESMOND 57,60 , was built to generate a suitable membrane system based on the OPM database 61 .The systems were dissolved in simple point charge (SPC) water molecules.The DESMOND default relaxation protocol was applied to each system prior to the simulation run. 1) 100 ps simulations in the NVT (constant number (N), volume (V), and temperature (T)) ensemble with Brownian kinetics using a temperature of 10 K with solute heavy atoms constrained; 2) 12 ps simulations in the NVT ensemble using a Berendsen thermostat with a temperature of 10 K and small-time steps with solute heavy atoms constrained; 3) 12 ps simulations in the NPT (constant number (N), pressure (P), and temperature (T)) ensemble using a Berendsen thermostat and barostat for 12 ps simulations at 10 K and 1 atm, with solute heavy atoms constrained; 4) 12 ps simulations in the NPT ensemble using a Berendsen thermostat and barostat at 300 K and 1 atm with solute heavy atoms constrained; and 5) 24 ps simulations in the NPT ensemble using a Berendsen thermostat and barostat at 300 K and 1 atm without constraint.After equilibration, the MD simulations were performed for 0.3 µs.The long-range electrostatic interactions were calculated using the smooth particle grid Ewald method.The trajectory recording interval was set to 200 ps, and the other default parameters of DESMOND were used in the MD simulation runs.All simulations used the all-atom OPLS_2005 force field [62][63][64] , which is used for proteins, ions, lipids and SPC waters.The Simulation Interaction Diagram (SID) module in DESMOND was used for exploring the interaction analysis between PPNDS and pdP2X7.All simulations were performed on a DELL T7920 with NVIDIA TESLTA K40C or CAOWEI 4028GR with NVIDIA TESLTA K80.The simulation system was prepared, and the trajectory was analyzed and visualized on a CORE DELL T7500 graphics workstation with 12 CPUs.

Statistics
Electrophysiological recordings were repeated 5-10 times.Error bars represent the standard error of the mean.Cryo-EM data collection and refinement statistics are summarized in Table S1.

of 5 × 5 ,
producing summed and dose-weighted micrographs with a pixel size of 0.83 Å. Contrast transfer function (CTF) parameters were estimated by CTFFIND 4.1 46 .Particle picking and 2D classification were performed using RELION 3.1 In total, 1,537,753 particles for the PPNDS-bound sample and 1,835,907 particles for the PPADS-bound sample were autopicked and extracted using a box size of 256 × 256 pixels.After 2D classification, nonuniform refinement was performed using cryoSPARCv4.2.1 47 on 633,674 particles for the PPNDS-bound sample and 236,753 particles for the PPADSbound sample, applying C3 symmetry for the final 3D reconstruction.The resulting resolutions of the PPNDS-bound and PPADS-bound pdP2X7 structures were 3.3 Å and 3.6 Å, respectively, as determined by the Fourier shell correlation (FSC) = 0.143 criterion on the corrected FSC curves.The local resolution was estimated using cryoSPARCv4.2.1.The workflows for image processing and for 3D reconstruction are shown in Figs.S2-S5.

Figure. 4 .
Figure. 4. ATP binding site and sequence comparison

Figure. 6 .
Figure. 6. Structure-based mutational analysis Fig. S2.Cryo-EM analysis of PPNDS-bound pdP2X7 (A) The gold -standard Fourier shell correlation curves for the PPNDS -bound data.(B) Angular particle distribution.The heat map of particle projections in each viewing angle.(C) The side view, a top-down view from the extracellular surface and a bottom-up view from the intracellular surface colored by local resolution.