Identification of Negative Residues in the P2X3 ATP Receptor Ectodomain as Structural Determinants for Desensitization and the Ca2+-sensing Modulatory Sites*

On nociceptive neurons, one important mechanism to generate pain signals is the activation of P2X3 receptors, which are membrane proteins gated by extracellular ATP. In the presence of the agonist, P2X3 receptors rapidly desensitize and then recover slowly. One unique property of P2X3 receptors is the recovery acceleration by extracellular Ca2+ that can play the role of the gain-setter of receptor function only when P2X3 receptors are desensitized. To study negatively charged sites potentially responsible for this action of Ca2+, we mutated 15 non-conserved aspartate or glutamate residues in the P2X3 receptor ectodomain with alanine and expressed such mutated receptors in human embryonic kidney cells studied with patch clamping. Unlike most mutants, D266A (P2X3 receptor numbering) desensitized very slowly, indicating that this residue is important for generating desensitization. Recovery appeared structurally distinct from desensitization because E111A and D266A had a much faster recovery and D220A and D289A had a much slower one despite their standard desensitization. Furthermore, E161A, E187A, or E270A mutants showed lessened sensitivity to the action of extracellular Ca2+, suggesting that these determinants were important for the effect of this cation on desensitization recovery. This study is the first report identifying several negative residues in the P2X3 receptor ectodomain differentially contributing to the general process of receptor desensitization. At least one residue was important to enable the development of rapid desensitization, whereas others controlled recovery from it or the facilitating action of Ca2+. Thus, these findings outline diverse potential molecular targets to modulate P2X3 receptor function in relation to its functional state.

P2X 3 receptors of nociceptive sensory neurons transduce the action of extracellular ATP into painful signals especially during chronic pain states (1). Similar to other ligand-gated ionotropic receptors, P2X 3 receptors undergo rapid, full desensitization in the continuous presence of their agonist (2). However, a distinctive property of P2X 3 receptors is the prompt re-attainment of function in high extracellular Ca 2ϩ solution that operates by facilitating recovery from desensitization (3,4).
Thus, Ca 2ϩ can exert a profound, rapid action on the ability of P2X 3 receptors to transmit sensory inputs to the central nervous system. However, the precise sites mediating the effect of Ca 2ϩ remain unknown and carry considerable interest for any attempts to manipulate transduction of pain signals.
The large family of ionotropic ATP receptors (P2X 1-7 ) shares a similar topology that comprises two transmembrane domains joined by one large extracellular loop with 10 disulfide bonds and intracellular N-and C-terminal regions (2,5,6). The facilitating action of extracellular Ca 2ϩ is exclusively produced on desensitized P2X 3 receptors, perhaps via extracellular sites (3,4). To identify the receptor region involved in this action and the amino acid sites important for it, we focused on negatively charged residues of the ectodomain of P2X 3 receptor that are not conserved in other P2X receptors. In fact, other subtypes of the P2X receptor family have either very slow desensitization (typical of the P2X 2 receptor class) (7,8) or fast desensitization (e.g. the P2X 1 receptor class) not modulated by high extracellular Ca 2ϩ (9).
Because on P2X 3 receptors the effects of Ca 2ϩ are closely related to desensitization, our approach also provided an opportunity to explore the role of extracellular negative residues in controlling desensitization development and recovery from it as these processes are not completely understood as far as P2X 3 receptors are concerned.
For other P2X receptors, the development of desensitization is believed to be determined by receptor transmembrane and intracellular segments (10 -15). However, recent experiments have indicated that the ectodomain not only controls agonist binding (16) but also desensitization via its coupling to the C-terminal domain (17). Furthermore, the chimeras of P2X 2 receptors containing the N-half of the P2X 3 receptor ectodomain develop desensitization, indicating that this region influences the desensitized conformation state and the process of recovery of receptor function (18).
Because the latter study has narrowed the number of potential ectodomain residues important for desensitization (18), the present investigation based on single mutations of certain negatively charged amino acids allowed us to examine whether sites mediating recovery from desensitization may be distinct from those involved in the onset of desensitization. 3 Receptor-The pCDNA3-rP2X 3 plasmid was kindly provided by Prof. R. A. North (Sheffield University). The alignment of the amino acid sequences of P2X 1 , P2X 2 , and P2X 3 was deduced from NCBI accession numbers P47824 (rP2X 1 ), 2020424A (rP2X 2 ), and CAA62594 (rP2X 3 ) (2).

Mutagenesis of the P2X
The non-conserved, negatively charged residues of the extracellular domain of the P2X 3 receptor were mutated to the neutral amino acid alanine (Fig. 1). Single point mutations were introduced using the QuikChange mutagenesis kit (Stratagene, La Jolla, CA). Each mutated plasmid P2X 3 DNA was obtained with a single PCR reaction using specifically designed mirror-image oligonucleotides containing the mutation of interest. "Sense" and "antisense" oligonucleotides (Roche Applied Science) used for the mutagenesis are listed in Table I. For all of the mutants, the introduction of the correct mutation and the absence of spontaneous mutations were confirmed by automated DNA sequencing.
Cell Culture and Transfection-HEK 1 293T cells, supplied by the in-house SISSA cell bank, were maintained in culture in Dulbecco's modified Eagle's medium-Glutamax medium supplemented with 10% fetal calf serum and penicillin/streptomycin. For each transient transfection experiment, 5 ϫ 10 Ϫ5 cells were plated and transfected 24 h later with the calcium/phosphate method using 1 g of high quality purified P2X 3 plasmid DNA (Sigma), either WT-or point-mutated. Transfected cells were used for further experiments 48 or 72 h later. Correct cell expression was confirmed with immunofluorescence and Western immunoblot assays as described previously (19).
Drugs and Their Application-␣,␤-Methyleneadenosine 5Ј-triphosphate (␣,␤-meATP; lithium salt) was used as a selective P2X 3 agonist to avoid activation of P2Y receptors natively expressed by HEK 293T cells (20). Unless otherwise stated, we used 100 M ␣,␤-meATP as a routine test concentration to evoke responses of maximal amplitude. All of the drugs were applied via a rapid superfusion system (Rapid Solution Changer RSC-200, BioLogic Science Instruments, Grenoble, France) placed 100 m near the cell. Time for the solution exchange at the cell membrane level was ϳ30 ms. The ␣,␤-meATP applications were 2-s long. For construction of dose-response plots, it was necessary to minimize receptor desensitization, which developed rapidly especially at high doses of agonist and prevented response reproducibility also in view of slow recovery. Thus, the agonist concentrations of Ն100 M were applied sequentially every 6 min, whereas for lower concentrations, the applications were spaced at 2-5-min intervals.
All of the chemicals including enzymes for cell culture were from Sigma. Culture mediums were obtained from Invitrogen, and the antibody against P2X 3 was from Neuromics, whereas the secondary antirabbit antibody was from Sigma. G418 was from Invitrogen.
Data Analysis-All of the data are presented as the mean Ϯ S.E. (n ϭ number of cells) with statistical significance assessed with Student's t test (for parametric data) or Mann-Whitney rank sum test (for nonparametric data). The best fits of the data obtained with a sigmoid function (Origin software, version 6.0) were compared with respective control fits using SigmaStat (Jandel Scientific, version 2.0). A value of p Ͻ 0.05 was accepted as indicative of statistically significant difference. The fitting function for recovery from desensitization as a function of time was as reported earlier (19).

Desensitization of Wild Type P2X 3 Receptors and Its
Sensitivity to High Extracellular Ca 2ϩ -The basic properties of activation and desensitization of native P2X 3 receptors expressed by HEK 293T cells are shown in Fig. 2A. The inward current induced by 100 M ␣,␤-meATP peaked and then fully decayed 1 The abbreviations used are: HEK, human embryonic kidney; WT, wild type; ␣,␤-meATP, ␣,␤-methyleneadenosine 5Ј-triphosphate.

TABLE I Oligonucleotides used for the mutagenesis reaction
Codons containing the mutation are underlined. In this table and in  Tables II and III, amino acid numbers refer to the P2X receptor numbering.

D53A
5Ј 5Ј with biexponential time course to base line (average data of current amplitude and decay are in Table II), indicating full desensitization. Response recovery was complete after a 6-min washout and partial (60 Ϯ 4%; n ϭ 21) when agonist applications were spaced at 30-s intervals. In this case, the protocol of applying agonist pulses at varying intervals after the desensitizing response enabled quantification of recovery from desensitization, which showed the typical sigmoidal time course (Fig.  2B) previously observed with native P2X 3 receptors (19). Membrane currents evoked when P2X 3 receptors were desensitized could be readily enhanced by a high extracellular Ca 2ϩ solution (50 Ϯ 6%; n ϭ 11, Fig. 2A) in a prompt and persistent fashion (Fig. 2C). Hence, the P2X 3 receptors expressed by HEK 293T cells appeared to display all of the main properties (including sensitivity to high extracellular Ca 2ϩ ) of native receptors of dorsal root ganglion nociceptors (3,4).
Effect of Single Amino Acid Mutations on the Amplitude of P2X 3 Receptor Currents and Their Desensitization-We examined the extracellular loop for negatively charged sites potentially involved in mediating the action of Ca 2ϩ and Mg 2ϩ (3,4) and identified 15 non-conserved amino acids (Fig. 1), which were subjected to mutation. Table II compares the functional properties of the WT and mutated P2X 3 receptors following the application of 100 M ␣,␤-meATP. Although the majority of  High extracellular Ca 2ϩ (10 mM) strongly potentiated current peaks almost to control level. B, to measure recovery from desensitization, the ␣,␤-meATP current amplitude (as percentage of non-desensitized current peak) was plotted versus time of testing subsequent agonist application. Full recovery was attained after ϳ100 s (time for 50% recovery was 24 s). Data are from 6 -11 cells. C, time course of the enhancing action of high extracellular Ca 2ϩ on ␣,␤-meATP currents tested at 30-s intervals. The action of Ca 2ϩ was rapidly lost on a washout with control solution (n ϭ 11). mutants generated current amplitudes very similar to the WT (Table II), the responses recorded from mutants D53A, E57A, D220A, and D266A had a significantly smaller amplitude. Examples of control currents (pulse 1 ) recorded from mutants E111A, D220A, and D266A are given in Fig. 3A, b-d. Although the E111A response was closely similar to that of WT, D220A generated responses of very limited size (note larger current calibration) but with a shape analogous to WT. Conversely, the D266A response was considerably slower because it produced currents with monoexponential decay, suggesting that, in this case, the onset of receptor desensitization was largely impaired. All of the other mutants had 1,2 values close to those of WT (see Table II), indicating that, in those cases, receptor desensitization developed unabatedly.
Mutated Receptors Display Different Recovery from Desensitization-Mutants could be classified into three groups as far as recovery from desensitization was concerned. Fig. 3A shows examples of current recovery (compare pulse 2 records taken 30 s after the corresponding pulse 1 traces) after ␣,␤-meATPinduced desensitization. The first group (Fig. 3B) comprised 10 mutants (of 15) with recovery time not significantly different from the WT recovery. The second group (Fig. 3C) showed significantly faster recovery and included E111A (with onset of desensitization similar to the WT P2X 3 ; see also Fig. 3Ab) and D266A (with very slow onset of desensitization; Fig. 3Ad). The third group comprised mutants with much slower recovery (Fig. 3C), namely D220A (Fig. 3Ac) and E289A, both of them with onset characteristics close to WT.
In summary, by single point mutations of certain non-conserved, negatively charged amino acids in the extracellular loop of P2X 3 receptor, we observed that Glu-111, Asp-220, Asp-266, and Glu-289 differentially contributed to the process of recovery from desensitization.
Effect of Single Point Mutations on the Ability by Extracellular Ca 2ϩ to Facilitate Recovery from Desensitization-Using the same protocol shown in Fig. 2A, we investigated how single point mutations of the extracellular loop of P2X 3 receptors might affect the effectiveness of high extracellular Ca 2ϩ to facilitate responses to ␣,␤-meATP. On WT receptors activated by repeated applications of ␣,␤-meATP to produce stable, low amplitude currents, the application of 10 mM Ca 2ϩ potentiated (ϳ150%) the agonist-induced current with respect to the one in B, plots of recovery from desensitization (for details see Fig. 2B) for a range of mutated P2X 3 receptors. This data group comprised mutants with recovery not significantly different from that of WT (n ϭ 4 -8). C, plots of recovery from desensitization to include mutants with a significantly faster (D266A, E111A) or slower (D220A and E289A) recovery time course than that of WT (n ϭ 4 -12; *, p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.001). standard saline solution as exemplified in Fig. 4A.
Most mutants displayed the same degree of Ca 2ϩ modulation as the WT P2X 3 (see open bars in Fig. 4B). However, six mutants did not (Fig. 4B, filled bars). In particular, E161A, E187A, and E270A generated responses significantly less sensitive to high Ca 2ϩ (despite the fact that their recovery from desensitization was similar to that of WT). Even more strikingly, D266A, D220A, and E111A completely lost sensitivity to high Ca 2ϩ . In the latter case, we wondered whether the lack of sensitivity to high Ca 2ϩ was related to differential recovery from desensitization (see Fig. 3C). To this end, we applied ␣,␤-meATP at different intervals because, for each mutant, the response amplitude became ϳ50% of the control (obtained at a 6-min interval) as in the case of WT receptors.
On E111A, 10 mM Ca 2ϩ produced clear potentiation when responses were evoked by applying ␣,␤-meATP every 10 s instead of every 30 s (Fig. 5A). The same approach was used for D220A receptors tested at 120-s intervals and D266A receptors tested at 5-s intervals. These results summarized in the histograms of Fig. 5B indicate that currents generated by E111A and D220A could then display strong Ca 2ϩ -dependent potentiation, whereas those produced by D266A remained Ca 2ϩ -insensitive.
These data show that Glu-161, Glu-187, and Glu-270 contributed to the facilitating action by Ca 2ϩ . For E111A, D220A, and D266A, their low sensitivity to Ca 2ϩ seemed to be secondary to primary changes in receptor desensitization properties.

Potency and Efficacy of the Agonists on Mutants with
Changed Ca 2ϩ Sensitivity-To explore this issue, we first constructed an ␣,␤-meATP dose-response relation for WT P2X 3 receptors, which provided EC 50 (1.4 Ϯ 0.4 M; n ϭ 6) and n H (1.2) values very similar to those of native receptors of rat dorsal root ganglion neurons (19). Fig. 6A shows that E111A, E161A, E187A, and E270A produced maximal current amplitudes not significantly different from WT. Although the EC 50 values of these four mutants were relatively close to those of WT, they all remained significantly larger but with similar n H values (Table III). On the other hand, Fig. 6B shows that D266A and D220A had lower potency and efficacy than WT (see Tables II and III). The reduced ability of D266A and D220A to respond to ␣,␤-meATP could not be overcome even by applying a 10-fold larger agonist concentration (1 mM), suggesting that such mutations probably impaired channel gating rather than FIG. 5. Changes in agonist application timing can restore the effect of high Ca 2؉ . A, on E111A P2X 3 receptors, Ca 2ϩ was ineffective when ␣,␤-meATP was applied at a 30-s interval (top); however, on the same cell, Ca 2ϩ potentiation could be observed when the agonist was applied every 10 s with consequently smaller amplitude of P2X 3 receptor-mediated current. B, histograms summarizing data for mutants tested with the same concentration of ␣,␤-meATP at different intervals. In the case of E111A, the enhancing action of Ca 2ϩ was similar to the WT for the 10-s interval. For D220A, the facilitation by Ca 2ϩ could be restored when the agonist was applied every 2 min. However, for D266A, neither a 30-nor 5-s interval manifested any enhancing action by Ca 2ϩ (n ϭ 3-5; *, p Ͻ 0.05).
agonist binding (see comparable data with P2X 2 receptors) (21). Finally, we tested E293A because this mutant had standard desensitization properties and sensitivity to Ca 2ϩ (Table II and Fig. 4B). Also in this case, the EC 50 value was slightly but significantly larger than that for the WT (Table III). Fig. 6C shows an example of Western blot analysis of WT, D220A, and D266A P2X 3 receptor expression. It is apparent that, in each case, there was the comparable pattern of P2X 3 receptor protein, indicating that the large disparity in agonist sensitivity could not be simply due to insufficient receptor expression. Similar levels of P2X 3 receptor expression were also detected for all of the other mutants (data not shown).

DISCUSSION
The principal finding of this study is the identification of non-conserved, negatively charged residues in the extracellular loop of P2X 3 receptors involved in sensing Ca 2ϩ . Furthermore, we observed distinct ectodomain residues responsible for desensitization development or for recovery from it. Thus, the present data might suggest new targets for novel analgesic strategies based on discrete regulation of P2X 3 receptor states.
Single Point Mutations of the P2X 3 Receptor Strongly Affect Its Ca 2ϩ Sensitivity-The strong and rapid desensitization of P2X 3 receptors followed by rather slow recovery (2, 3) is an important process to regulate the activity of such receptors. Because extracellular Ca 2ϩ can powerfully facilitate recovery of receptor function whenever the receptor is desensitized (3), understanding the mechanisms underlying this phenomenon is a major issue, especially because extracellular Ca 2ϩ concentrations can rapidly fluctuate during intense neuronal activity (22) and thus modulate P2X 3 receptor function. Although previous data suggested the action of Ca 2ϩ to be on the extracellular loop of P2X 3 receptors (3, 4), supportive evidence was only indirect. By focusing on non-conserved negatively charged residues that should be potential sites for Ca 2ϩ binding (6), the present data indicated Glu-161, Glu-187, and Glu-270 to be important for the action of Ca 2ϩ . Because the effect of Ca 2ϩ on these mutated receptors was significantly reduced but not abolished, it is probable that such residues acted in concert to fully express the action of Ca 2ϩ . Furthermore, because E161A, E187A, and E270A mutants possessed the onset of desensitization and recovery similar to the WT P2X 3 , this observation suggests that the molecular sites involved in the action of extracellular Ca 2ϩ were structurally and operationally distinct from those controlling desensitization.
Altered Desensitization Properties Can Mask the Action of Ca 2ϩ -Another cause for impaired Ca 2ϩ action was found when testing E111A, D220A, and D266A mutants with standard protocols of desensitization. However, this result was simply the result of the mutant altered properties of desensitization onset and recovery. For instance, E111A displayed very rapid recovery from desensitization, thus depriving Ca 2ϩ of its receptor state target when the normal experimental protocol was used. Applying the agonist at a faster rate to make this process proportionally similar to the WT promptly restored the facilitatory action of Ca 2ϩ . Conversely, on D220A, spacing agonist applications at much longer intervals than in the standard protocol reinstated the facilitatory action by extracellular Ca 2ϩ . Hence, such experiments suggested that neither Glu-111 nor Asp-220 was crucial for the effect of Ca 2ϩ , which, to be fully expressed, required an intermediate level of desensitization to be manifested.  A special case was observed with the D266A mutant (with a rather slow onset of desensitization and recovery), because extracellular Ca 2ϩ could not improve the recovery of D266A, even after changing the extent of the desensitization state.
In conclusion, the finding that slow onset or very fast recovery was accompanied by reduced sensitivity to Ca 2ϩ was consistent with data showing that non-desensitized P2X 3 receptors are almost insensitive to Ca 2ϩ (3,4).
Structural Determinants of Desensitization Onset-Studies of chimeras of P2X 2 /P2X 3 and P2X 1 /P2X 2 receptors have indicated that the onset of desensitization is controlled by a concerted interaction between transmembrane and intracellular domains (10,(15)(16)(17)23). Whereas recent investigations using chimeras demonstrated the role of the ectodomain N-region to stabilize the desensitized receptor conformation (18), this study showed that a single negative residue (Asp-266) was important to control the desensitizing properties of the P2X 3 receptor because its mutation conferred unusually slow onset and rapid recovery. Thus, it seems probable that the multiple regions of P2X 3 receptors are involved in shaping development of desensitization.
In accordance with recent data on P2X 4 receptors (24), this report indicated that a single mutation in the extracellular domain could control agonist potency, which was reduced in the tested mutants, a phenomenon not attributable to insufficient receptor protein expression. However, decreased receptor sensitivity was not necessarily intertwined with the faster development of receptor desensitization because it was possible to observe very slow desensitization together with reduced agonist potency (e.g. D266A), normal desensitization with reduced agonist potency (see Table III), and rather weak efficacy coupled with standard desensitization (D220A).
All together, these results suggested a complex interplay between the agonist binding sites (and activated channels) and the extracellular and intracellular domains controlling the onset of desensitization.
Structural Determinants of Recovery from Desensitization-Our previous work with distinct P2X 3 agonists (19) suggested the onset of desensitization and recovery to be governed by independent mechanisms since agonists with identical receptor activation and onset of desensitization generated different rates of recovery. Desensitization recovery in chimeric P2X 2 / P2X 3 receptors is controlled by the extracellular domain (18,23), indicating a special protein structure regulating this process. Consistent with this notion, we observed that mutations in the extracellular loop, which left the onset of desensitization unaltered with respect to WT, displayed either faster (E111A) or slower (E289A and D220A) recovery from desensitization.
Since our previous work has suggested that desensitization is a multi-step process involving several receptor conformational states (19), it seems plausible to hypothesize that the ability to generate (or exit from) discrete conformations is governed by distinct molecular determinants within the P2X 3 receptor protein.