Agonist trapped in ATP-binding sites of the P2X2 receptor

Design of the Tools and of the Reaction Protocol.

The NCS-ATP was synthesized in pure form according to an adapted procedure (26) (Fig. S1A) and evoked robust, stable, and reversible inward currents upon washing in HEK-293 cells expressing the WT rat P2X2 receptor as monitored by whole-cell patch-clamp electrophysiology (Fig. 1C, Inset). Dose-response relationships showed that NCS-ATP displayed lower potency but higher efficacy than ATP (Fig. 1C). The probe was stable in buffers (< 4% degradation in 4 h) and reacted selectively and rapidly (in less than 1 min) with N-acetyl-L-cysteine methyl ester as previously described (26) (Fig. S1B). Thus, the agonist property of NCS-ATP along with its stability in water and exquisite reactivity toward cysteine residue prompted us to further use this ligand as a thiol-reactive agonist to trap engineered ATP sites in the P2X2 receptors.

On the basis of our P2X2 homology model (5), we individually mutated into cysteine 26 positions that protrude in the putative binding cavity, including those previously identified conserved residues (12–19) (Fig. 1D). All cysteine mutants, except for K69C, K71C, and K308C, were functional (Table S1).

The reactivity was monitored by systematically exposing each receptor mutant to fixed NCS-ATP reaction conditions (30 μM for 2 min). After a 5-min wash, ATP responses were recorded and compared to those measured before treatment (Fig. 2A). Remarkably, only the N140C mutant displayed robust and irreversible reduction of ATP responses (Fig. 2B). As expected, exposure to the reducing agent DTT restored ATP responses and no irreversible inhibition was observed when ATP, instead of NCS-ATP, was used (Fig. 2 and Fig. S2 A and B). On the other hand, the reversible inhibitions observed in WT, E84C, and Q138C mutants that showed little sensitivity to DTT are likely due to desensitization following continued application of agonist.

Surprisingly, a potentiation of ATP responses was observed for L186C mutant following the application of NCS-ATP, even after extensive washout (Fig. 2). Exposure to 10 s DTT did not restore ATP responses; however, increasing the time exposure to DDT up to 30 s almost completely restored ATP responses (Fig. S2D). Control experiments performed with ATP further confirmed that the irreversible functional effect is owed to the presence of the thiol-reactive group (Fig. S2 C and D). Overall, these results strongly suggest that N140C and L186C mutants are labeled with NCS-ATP.

To confirm that these mutants are water-accessible we used thiol-reactive methanethiosulfonate (MTS) derivatives and showed that these compounds profoundly and rapidly modified ATP responses in N140C and L186C mutants, but not in the WT receptor, showing that the side chains of N140 and L186 are quite accessible to solvent, a fact that is consistent with a water-exposed binding pocket (Fig. S3). Interestingly, the positively charged 2-aminoethyl MTS hydrobromide (MTSEA) potentiated ATP responses in L186C mutant, whereas the negatively charged sodium (2-sulfonatoethyl) MTS (MTSES) abolished responses (Fig. S3C), suggesting a charge-dependent modification.

NCS-ATP Targets the ATP-Binding Sites.

Although NCS-ATP shares high structural similarity with ATP, which strongly argues for a common binding site for the two agonists, we were still concerned about the possibility that the probe might target sites that are not relevant to ATP. In the following, we present four lines of evidence that NCS-ATP specifically targets the ATP-binding site.

First, if modification arises from affinity labeling as suggested in Fig. 1A, then the apparent tethering reaction rates k_app should be limited to NCS-ATP occupancy and thereof to its concentration. As expected, we found that k_app values saturated with increasing NCS-ATP concentration for both N140C (Fig. 3A) and L186C mutants (Fig. 4A).

Second, it is known that adenosine (Ado) is a poor P2X ligand (27), and consequently its site occupancy should be lower than that of ATP. Therefore, the apparent rates of the tethering reaction by NCS-Ado should be slower than those determined by NCS-ATP. We thus synthesized NCS-Ado (Fig. S1A), verified that it was not an agonist and had no effect on ATP response in the WT P2X2 receptor, and showed that, as expected, the tethering reactions occurred for both mutants much more slowly than those produced by NCS-ATP at the same concentration (Figs. 3A and 4A). In L186C mutant, tethered NCS-Ado, observed at high concentration, also resulted in a potentiation of ATP response, suggesting that the adenosine moiety is sufficient to mediate the potentiation effect (Fig. S4).

Third, the tethering reactions observed for both mutants were largely reduced or prevented by the presence of a saturating concentration of ATP or of the competitive P2X2 antagonist 2′,3′-O-(2,4,6-trinitrophenyl)-ATP (28) (TNP-ATP), indicating pharmacological protection (Figs. 3B and 4B). Interestingly, a concentration of TNP-ATP that abolished ATP response in L186C slowed rates of NCS-ATP-induced potentiation, but did not abolish it (Fig. 4B), suggesting that the ability of the antagonist TNP-ATP to prevent the tethering reaction in L186C is not as strong as that of ATP.

Fourth, in support of the results obtained from the MTS derivatives, we found that negatively charged mutations largely decreased ATP potency (N140D and L186C, for which the side chain may be partially negatively charged at pH 7.3) or even abolished ATP function (L186E mutant, for which we verified that cell-surface expression appeared normal compared to WT, Fig. S5), suggesting that repulsive electrostatic interactions occur between the negatively charged mutant residues and the negatively charged ATP (Tables S1 and S2). By contrast, aromatic mutations (N140W and L186F) increased by about 10 times ATP potency, a fact that suggests a possible π-π interaction between the substituted aromatic residues and the adenine ring of ATP (Tables S1 and S2).

We constructed models of the cysteine mutants, which likely represent the closed state of the receptor (8, 9), to dock NCS-ATP. We found two energetically favorable poses that fit with the experimental data, in which the NCS moiety is located closely (between 4.3 to 4.6 Å) to the sulfhydryl groups of N140C (Fig. 3E) and L186C (Fig. 4E) from two adjacent subunits. Of note, the residues previously found to be important for ATP function (12–19) are in close proximity to docked NCS-ATP. Our results thus demonstrate that the intersubunit cavities found in the X-ray structure (9) are the ATP-binding sites.

The ATP-Binding Sites Are Trapped in Covalently Agonist-Bound States That Differ in Their Ability to Gate the Channel.

NCS-ATP labeled two cysteine mutants, N140C and L186C, that are separated by about 18 Å in the closed state of the receptor (Figs. 3E and 4E). Yet a crucial difference should be noted; whereas the irreversible attachment at N140C resulted in a decrease of both ATP efficacy and nominal open probability (NP_o) of channels activated by ATP in outside-out patches (Fig. 3 C and D), locking at L186C trapped the receptor in a state that is more sensitive to ATP with a concomitant increase of NP_o (Fig. 4 C and D), both without changing the single-channel conductance. Importantly, both mutants labeled covalently by NCS-ATP did not produce spontaneous channel gating following washout of patches (Fig. S6), a fact that confirms the absence of irreversible sustained inward currents following washout of cells (Figs. 3A and 4A). Finally, the unitary conductance of channels activated by NCS-ATP was similar to that activated by ATP, both in mutant and WT receptors (Fig. S7). Overall, these findings suggest that (i) the covalent attachment of the agonist does not affect ion permeation, (ii) attached ATP by itself is not sufficient to gate the channel, and (iii) the ATP-binding sites were trapped in covalently agonist-bound states that differ in their ability to gate the channel.

Chemical Synthesis.

Synthesis of NCS-ATP was adapted from ref. 26 and is detailed in SI Materials and Methods. NCS-Ado was obtained using a similar procedure as for NCS-ATP. All chemical products were characterized by ¹H-, ³¹P-NMR, and UV spectrum.

Complementary DNA Construction and Site-Directed Mutagenesis.

The pcDNA-based expression plasmids, mutagenesis, and sequencing procedure have been described previously (5).

Cell Culture and Transfection.

HEK-293 cells were cultured and transiently transfected with the rat P2X2 constructs (0.01–0.3 μg) and a green fluorescent protein cDNA construct (0.3 μg), as described previously (5).

Cell-Surface Labeling.

Cell-surface expression of L186E mutant was determined as described previously (5) using the thiol-cleavable reagent sulfosuccinimidyl-2-(biotinamido)ethyl-1,3-dithiopropionate (Sulfo-NHS-SS-Biotin, Pierce).

Whole-Cell Recordings.

Patch pipettes (3–5 MΩ) contained (in mM) 140 KCl, 5 MgCl₂, 5 EGTA, 10 Hepes, pH 7.3. External solution contained (in mM) 140 NaCl, 2.8 KCl, 2 CaCl₂, 2 MgCl₂, 10 glucose, 10 Hepes, pH 7.3. The holding potential was -60 mV. Dose-response relationship experiments and drug applications were carried out as described previously (5). The apparent reaction rate constant k_app was determined by fitting data with a single exponential equation. K_D and k_modification were determined by fitting data with the following equation:

[1]

in which L is the concentration of NCS-ATP, k_modification is the covalent modification rate constant, and K_D is the reversible dissociation constant of NCS-ATP.

Single-Channel Recordings.

Single-channel recordings using outside-out configuration were made from HEK-293 cells at room temperature 24–72 h after transfection. Recording pipettes pulled from borosilicate glass (Harvard Apparatus) were coated with Sylgard 184 (Dow Corning Co.) and fire polished to yield resistances of 6–20 MΩ. The holding potential was -120 mV. The extracellular solution contained (in mM) 147 NaCl, 2 KCl, 1 CaCl₂, 1 MgCl₂, 10 Hepes, and 13 glucose, pH 7.3. The intracellular solution contained (in mM) 147 NaF, 10 Hepes, and 10 EGTA, pH 7.3. The osmolarity was approximately 300 mosmol kg^-1 for all the solutions. Following NCS-ATP exposure, patches were always washed by the external solution for at least 15 s before recordings.

Data were acquired with a patch-clamp amplifier (HEKA EPC 10) using PATCHMASTER software (HEKA Co.), sampled at 4–10 kHz, and low-pass filtered at 2.9 kHz. For off-line analysis, data were refiltered to give a cascaded filter cutoff frequency of 1–2 kHz. Channel events were detected by using a 50% threshold criterion (30) and idealized with TAC (Bruxton Co.). Nominal open probability (NP_o) was determined with TACFit (Bruxton Co.) and was defined as the fraction of time for which the channels are open, where N is the number of active channels in the patch and P_o is the single-channel open probability. As previously observed for WT P2X2 receptor under ATP activation (10), flickery bursts were also visible with ill-defined conductance levels. Such flickering openings were treated as previously specified (10): Each burst was defined as a single conductance state possessing a lot of noise.

Molecular Modeling.

Homology modeling of cysteine mutants was performed with MODELLER 9.7 (31). Docking was performed with the software Autodock vina (32).