Competitive Antagonism of Recombinant P2X_2/3Receptors by 2′,3′-O-(2,4,6-Trinitrophenyl) Adenosine 5′-Triphosphate (TNP-ATP)

Materials and Methods

Cell Culture.

Stably transfected 1321N1 human astrocytoma cells expressing either rat P2X_2/3(rP2X_2/3) or human P2X_2a/3(hP2X_2a/3) receptors have previously been described (Bianchi et al., 1999; Burgard et al., 1999; Lynch et al., 1999). Briefly, these heteromultimeric cell lines were constructed by transfecting rP2X₂ or hP2X_2a cDNA into stably transfected rat or human P2X₃-expressing cells using standard lipid-mediated transfection methods. Cell lines were maintained in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum and antibiotics as follows: hP2X₃, 300 μg ml⁻¹ G418; hP2X_2a, 100 μg ml⁻¹ hygromycin; and hP2X_2a/3, 150 μg ml⁻¹G418 and 75 μg ml⁻¹ hygromycin.

Electrophysiology.

Patch-clamp recordings were performed as described previously (Burgard et al., 1999). Briefly, 1321N1 cells expressing rP2X_2/3 receptors were plated on polyethylenimine-coated coverslips and grown to approximately 50% confluence. Whole-cell patch-clamp recordings were obtained using a modified extracellular saline consisting of 155 mM NaCl, 5 mM KCl, 2 mM CaCl₂, 1 mM MgCl₂, 10 mM HEPES, and 12 mM glucose. The intracellular patch pipette solution consisted of 140 mM potassium aspartate, 20 mM NaCl, 10 mM EGTA, and 5 mM HEPES. All cells were voltage-clamped at −60 mV, and series resistance compensated 85 to 90% using an Axopatch 200B amplifier (Axon Instruments, Foster City, CA).

Drugs were applied to the cells using a piezoelectric-driven glass theta tube positioned near the cell. The solution interface between barrels was swept across the cell. Solution exchange times were often recorded by application of a 90% extracellular saline (10% water added) across an open pipette tip. A single exponential function was fitted to the resulting junction current transient, giving an open-tip response time constant (τ). Alternatively, whole-cell exchange τs were recorded in whole-cell patch mode by application of α,β-meATP (10 μM) to activate rP2X_2/3 receptors, and switching into and out of a high (55 mM) potassium-containing solution. The resulting potassium-generated current relaxation through open P2X receptors gave an indication of the whole-cell exchange time. During experiments, agonists were usually applied every 30 to 60 s. Agonist applications were kept short (most <3 s) to minimize pore dilation of P2X_2/3 receptors (Khakh et al., 1999). No evidence of pore dilation was observed.

Responses were acquired and digitized at 3 kHz, and analyzed using pClamp software (Axon Instruments). Current amplitudes were always measured at the end of the agonist application pulse. Current on- and off-responses were fitted with a single exponential function, and the resulting τs were calculated. Agonist concentration-response curves were fitted by least-squares regression to the logistic equation:where Y, min, and max represent the measured, minimum, and maximum responses, respectively; EC₅₀ is the ligand concentration giving half-maximal response; X is the concentration of ligand used; and n_H is the Hill coefficient (Prism; GraphPad Software, San Diego, CA). Antagonist concentration-response curves were fitted in the same manner to determine IC₅₀ values. The inhibition constant (K _i) for the competitive antagonist TNP-ATP was estimated from its IC₅₀ value using the Cheng-Prusoff relationship (Cheng and Prusoff, 1973; Craig, 1993):K _i = IC₅₀ / (1 + A/EC₅₀), where A represents the concentration of agonist, and EC₅₀ represents the agonist EC₅₀.

Competitive inhibition was determined using Schild analysis (Arunlakshana and Schild, 1959). The log concentration ratio log (A′ / A − 1) was plotted against TNP-ATP concentration. A and A′ represent agonist EC₅₀ values obtained in the absence and presence of TNP-ATP, respectively. pA₂ values were determined from least-squares linear regression fitted to the Schild plots. Throughout the text, data are expressed as mean ± S.E.M.

Calcium Imaging.

Inhibition of hP2X_2a/3 or rP2X_2/3receptors by TNP-ATP was also studied by measuring changes in cytosolic calcium levels. The fluorescent dye Fluo-4 was used as an indicator of the relative levels of intracellular calcium in a 96-well format using a fluorescence imaging plate reader (Molecular Devices, Sunnyvale, CA), as described previously (Bianchi et al., 1999). TNP-ATP (50 μl of 4× concentration) was added 3 min before the addition of α,β-meATP (50 μl of 4× concentration, final volume = 200 μl). Fluorescence intensities were measured at 1- to 5-s intervals throughout each experimental run. Data shown are based on the peak increase in relative fluorescence units compared with basal fluorescence. Concentration-response curves are shown as a percentage of the maximum α,β-meATP-mediated fluorescence signal measured in the absence of TNP-ATP.

All reagents were obtained from Sigma Chemical Co. (St. Louis, MO.) unless otherwise noted. TNP-ATP and Fluo-4 were purchased from Molecular Probes (Eugene, OR), and suramin from Research Biochemicals International (Natick, MA).

Results

Drug Application.

The chemical structures of ATP and TNP-ATP are shown in Fig. 1A. For TNP-ATP, the nucleotide has been modified to include a trinitrophenyl group on the 2′,3′ dihydroxy position. The structural similarity of TNP-ATP to ATP suggested that both molecules may interact with the same binding site on the P2X receptor.

To determine the mechanism of action of TNP-ATP, a rapid application system was used to apply agonists and antagonists to 1321N1 cells expressing rP2X_2/3 receptors (underMaterials and Methods). Solution exchange times were determined under a variety of conditions (Fig. 1B). Exchange of the solution interface across an open pipette tip was achieved in approximately 1 to 2 ms (1.6 ± 0.1 ms, n = 10). This provided an indication of how fast the solution interface could travel across a pipette. In contrast, whole-cell exchange time constants were 20.6 ± 0.3 ms (n = 5), indicating that diffusion around an entire cell was approximately 10 to 20 times slower than diffusion across an open pipette tip. Similar activation (32 ± 5.5 ms, Fig. 1B) and deactivation (86 ± 13 ms) time constants (τ) for 10 μM α,β-meATP-induced currents were also recorded (n = 20), indicating that activation and deactivation rates were close to being diffusion-limited. Because α,β-meATP exchange rates approached the limits of diffusion, no quantitative conclusions were drawn regarding the actual activation and deactivation rates of P2X receptors. However, comparisons were made to the slower rates of receptor antagonism by TNP-ATP (see below).

Competitive Inhibition by TNP-ATP.

TNP-ATP inhibited P2X_2/3 receptors in a concentration-dependent manner as assayed either by patch-clamp (Fig.2, A and B) or calcium imaging (Fig. 2B) techniques. As can be seen in Fig. 2A, activation of P2X_2/3 receptors by α,β-meATP resulted in a non- or very slowly desensitizing inward current. Because these receptors were relatively nondesensitizing, agonist could be applied and removed as frequently as every 15 s without a decrease in current amplitude. Concentration-dependent inhibition by TNP-ATP was evident from patch-clamp recordings both as an increase in the amount of inhibition (gradual decrease in current amplitude), as well as an increase in the rate at which inhibition developed. Both agonist and antagonist were coapplied in Fig. 2A, and the extent of inhibition was measured at the end of the application. Calculated IC₅₀ values for TNP-ATP were obtained from inhibition curves for both rP2X_2/3 and hP2X_2a/3 receptors, using two different functional assays (Fig. 2B). For rP2X_2/3receptors, patch-clamp recordings revealed an IC₅₀ value of 5.5 ± 1.1 nM (n = 3–11 cells/concentration), and intracellular calcium responses revealed an IC₅₀ value of 3.0 ± 1.1 nM (n = 5 experiments). For hP2X_2a/3 receptors, intracellular calcium responses revealed an IC₅₀ value of 4.6 ± 1.2 nM (n = 3 experiments). It should be noted that in these series of experiments, the IC₅₀ values for TNP-ATP were almost identical, even when measured at different receptors and under different assay conditions.

To evaluate the competitive nature of TNP-ATP antagonism, competition curves were constructed for TNP-ATP inhibition of rP2X_2/3 (electrophysiology, Fig.3A) and hP2X_2a/3 (calcium imaging, Fig. 3B) receptors. As can be seen from both sets of graphs, increasing concentrations of α,β-meATP were able to fully overcome inhibition produced by TNP-ATP over a range of concentrations. A similar rightward shift of the α,β-meATP concentration-response curve was observed when calcium imaging experiments were performed using rP2X_2/3 receptors (data not shown). Schild analysis of both rat and human receptors (Fig. 3, A and B, insets) revealed plots best fitted with a regression line of slope 1.0 (hP2X_2a/3) or 1.1 (rP2X_2/3). pA₂ values obtained from Schild analysis were −8.2 (6 nM) for hP2X_2a/3 and −8.7 (2 nM) for rP2X_2/3. Parallel right-shifted competition curves with a Schild plot of slope = 1 were consistent with competitive antagonism of α,β-meATP with TNP-ATP at both human and rat P2X_2/3 receptors

Assuming competitive binding at the P2X_2/3receptor, K _i values were calculated from TNP-ATP IC₅₀ values using the Cheng-Prusoff relationship (under Materials and Methods). Agonist concentrations of 3 μM (hP2X_2a/3) and 10 μM (rP2X_2/3) α,β-meATP were used in these experiments. The EC₅₀ values for α,β-meATP were 3.4 μM (rP2X_2/3, Bianchi et al., 1999) and 1.3 μM (hP2X_2a/3; H. McDonald and E. C. Burgard, unpublished observations), as determined from agonist concentration-response curves. CalculatedK _i values for TNP-ATP were 1.4 nM for both rP2X_2/3 and hP2X_2a/3. This was very similar to IC₅₀ as well as calculated pA₂ values, in that all were between 1 and 10 nM. The high affinity of TNP-ATP for P2X_2/3 receptors was independent of species or assay conditions.

In contrast, calcium imaging experiments performed using the hP2X₃ receptor (Fig. 3C) revealed TNP-ATP competition curves characterized by a decrease in the maximal agonist-induced current, with a rightward shift in the agonist EC₅₀ value. The apparent inability of α,β-meATP to overcome TNP-ATP block is in agreement with a published report (Virginio et al., 1998) that describes TNP-ATP as a noncompetitive antagonist at desensitizing P2X₃receptors. However, this apparent discrepancy could be explained by differences in desensitization rates between the two receptor subtypes (see below).

Kinetics of TNP-ATP Inhibition.

If TNP-ATP and α,β-meATP compete for the same binding site at P2X_2/3receptors, then analysis of a competitive interaction should be performed when both ligands have achieved steady-state binding. This assumes that both ligands have had a sufficient amount of time to bind to (or unbind from) the receptor. The agonist-induced activation and deactivation kinetics of P2X_2/3 receptors approached the experimental limits of solution diffusion (Fig. 1B), suggesting that 10 μM α,β-meATP probably reached steady-state binding in tens of milliseconds. However, it appeared that TNP-ATP on- and off-kinetics were much slower. To evaluate the on- and off-kinetics of TNP-ATP, three different drug application protocols were used in patch-clamp experiments to apply TNP-ATP to cells expressing rP2X_2/3 receptors (Fig.4). TNP-ATP was either 1) preapplied, to measure the steady-state effects of TNP-ATP on subsequent activation by agonist; 2) postapplied, to measure both the on- and off-kinetics of TNP-ATP in the presence of agonist; or 3) coapplied, to measure TNP-ATP kinetics during simultaneous application with agonist.

The preapplication protocol is the conventional application method used to study antagonist potency and efficacy. Using this method, preapplication of TNP-ATP (≥30 s) allowed the antagonist to reach a steady-state interaction with the receptor before agonist was applied. As shown in Fig. 4, preapplication of TNP-ATP (100 nM) produced a decrease in both amplitude and apparent activation rate of the α,β-meATP (10 μM) response. This response resembled that of a low α,β-meATP concentration, and is consistent with the profile of a competitive interaction at the receptor. Because the α,β-meATP on-kinetics are fast, the slow activation seen in the presence of TNP-ATP is an indirect reflection of the slow off-kinetics of TNP-ATP (Benveniste et al., 1990; see below) because the competitive antagonist must unbind for agonist to activate the receptor. Here 100 nM TNP-ATP also inhibited the α,β-meATP-induced response to <10% of control, consistent with the concentration-response curve in Fig. 2B. In addition, the first agonist response was always inhibited after TNP-ATP application, so there was no evidence for a use-dependent mechanism of action of TNP-ATP. Although the preapplication method reveals the extent of steady-state inhibition by an antagonist, it provides only indirect information about the antagonist off-kinetics, and no information regarding the on-kinetics of antagonist interactions with the receptor.

To investigate the on- and off-kinetics of antagonists, the postapplication protocol was used because it allowed the antagonist kinetics to be measured entirely during steady-state receptor activation by agonist. As shown in Fig. 4, on- and off-response kinetics were recorded when rP2X_2/3 receptors were first activated by α,β-meATP (10 μM), and then TNP-ATP (100 nM) was applied and removed during steady-state receptor activation (postapplication). All on- and off-responses were adequately described by a single exponential function. Using this protocol, the 100 nM TNP-ATP on-response time constant τ_on was 320 ± 74 ms (n = 9), and the off-response time constant (τ_off) was 2390 ± 315 ms (n = 7). Even at this high IC₉₀concentration of TNP-ATP, it is evident that the on- and off-kinetics are ≥10-fold slower than 10 μM α,β-meATP kinetics. Antagonist steady state was reached after a sufficiently long TNP-ATP application, and the extent of inhibition was measured under these conditions. One minor drawback to this approach was the presence in some cells of very slow α,β-meATP-induced desensitization during “steady-state” agonist application. Agonist was often applied for >10 s before applying TNP-ATP, and the steady-state current amplitude decreased slightly in a proportion of cells. Although this did not affect either the TNP-ATP kinetics or extent of inhibition, it did shift the baseline amplitude of the agonist response during TNP-ATP application. This effect can be seen in Fig. 4 (postapplication trace, asterisk).

Another approach used to study TNP-ATP kinetics was to apply both agonist and antagonist simultaneously, and then measure the effect of the antagonist under these conditions. This coapplication protocol eliminated any potential effects of slow agonist-induced desensitization on TNP-ATP kinetic measurements. When TNP-ATP was coapplied with α,β-meATP (Figs. 2 and 4, coapplication), currents activated rapidly, but a time-dependent inhibition of current developed (relatively slowly). Both the extent of block and the apparent rate at which inhibition occurred increased with increasing concentrations of TNP-ATP. Because agonist steady state was reached within the first 50 ms, it was assumed that antagonist on-kinetics could be reliably estimated if they were slower than 50 to 100 ms. Indeed, the α,β-meATP activation τ_on was the same whether applied alone (26 ± 10 ms, n = 5) or coapplied with 100 nM TNP-ATP (27 ± 12 ms, n = 5), indicating that TNP-ATP inhibition developed more slowly than α,β-meATP activation. As shown in Fig. 4, the 100 nM TNP-ATP τ_on for both coapplication (387 ± 91 ms,n = 5) and postapplication (320 ± 74 ms,n = 9) protocols was similar. Although the overall τ_on was slower at lower TNP-ATP concentrations (10 and 30 nM), these TNP-ATP τ_on were also not significantly different between co- and postapplication protocols. Because the TNP-ATP on-kinetics were slow, and a relatively long (≥2 s) application was used, a reasonable estimate of both τ_on and extent of inhibition was obtained using the coapplication protocol.

All three protocols resulted in similar τ_onand/or τ_off values for TNP-ATP. It is important to note that for concentrations of TNP-ATP ≤ IC₉₀, τ_on and τ_off kinetics were ≥10 times slower than α,β-meATP kinetics. At these concentrations, TNP-ATP inhibition developed slowly, and recovery from inhibition was correspondingly slow.

Of the three protocols, measurements using the postapplication method best represented the true on- and off-kinetics of TNP-ATP inhibition. Using this protocol, the concentration dependence of τ_on and τ_off was further investigated. As shown in Fig.5A, the τ_ondecreased with increasing concentrations of TNP-ATP. This concentration-dependent increase in the apparent rate of inhibition was consistent with a mass action increase in binding probability. However, the τ_off did not change with increasing TNP-ATP concentrations, indicating that the off-response kinetics were concentration-independent. The on- and off-kinetics of the P2X receptor antagonist suramin are shown in Fig. 5B. In comparison, suramin exhibits an IC₅₀ value of approximately 800 nM at rat P2X_2/3 receptors (Bianchi et al., 1999). In this cell at equimolar concentrations (10 μM), suramin displayed much faster off-kinetics (τ_off = 775 ms) than TNP-ATP (τ_off = 5951 ms), consistent with its lower affinity for the P2X_2/3 receptor. A summary of the effects of TNP-ATP concentration on τ_onand τ_off from 15 cells is shown in Fig.6. At all concentrations of TNP-ATP, the mean τ_off values were very slow (>2000 ms). It is reasonable to assume that the high affinity of TNP-ATP for rP2X_2/3 receptors is predominantly determined by its slow τ_off.

In contrast to rP2X_2/3 receptors, rP2X₃ receptors exhibit rapid desensitization in the presence of agonist (Lewis et al., 1995). The relationship between receptor desensitization and TNP-ATP time course can be seen in the traces in Fig. 7. Here α,β-meATP induced a nondesensitizing inward current at rP2X_2/3 receptors (α,β-meATP control trace). Preapplication of TNP-ATP (α,β-meATP + TNP-ATP trace) produced a decrease in the α,β-meATP activation rate due to competition of α,β-meATP with TNP-ATP. The slow τ_off of TNP-ATP is evident here because it takes >2 s to approach a steady-state interaction between agonist and antagonist. Only after this time can a competitive interaction be observed where agonist can replace the slowly dissociating antagonist at the receptor. The relatively nondesensitizing rP2X_2/3 receptor remains open long enough to detect the competition. In contrast, the rapid desensitization of rP2X₃ (light P2X₃ trace) receptors occurs before TNP-ATP can dissociate from the rP2X_2/3 receptor, precluding a steady-state competition to be recorded. Apparent noncompetitive inhibition would be observed at rapidly desensitizing P2X₃ receptors, due to the inability of TNP-ATP to dissociate before the current desensitized.

Discussion

TNP-ATP is a trinitrophenyl analog of ATP that binds to a variety of ATP-binding sites. When used as a fluorescent probe, relatively high concentrations have been used to label ATP-binding sites and to probe enzyme activities (Hiratsuka and Uchida, 1973; Watanabe and Inesi, 1982; Mockett et al., 1994). However, the nanomolar affinity of TNP-ATP for P2X₁, P2X₃, and P2X_2/3 receptors (Virginio et al., 1998) suggests that TNP-ATP may be a selective antagonist for these P2X receptors over other ATP-binding sites. Using these criteria, TNP-ATP is currently the most potent and selective P2X receptor antagonist available.

In the present study, we have confirmed the high affinity of TNP-ATP for P2X_2/3 receptors. Potent antagonism (IC₅₀ < 10 nM) of both human and rat P2X_2/3 receptors by TNP-ATP was demonstrated using calcium imaging as well as electrophysiological techniques. TNP-ATP was determined to be a competitive antagonist based on a number of experimental results. First, the inhibition could be overcome by increasing concentrations of agonist. Second, parallel right-shifted agonist concentration-response curves were constructed in the presence of increasing concentrations of TNP-ATP, with no change in the maximal agonist response. Third, Schild analysis revealed a competitive inhibition by TNP-ATP at P2X_2/3 receptors. Fourth, calculated pA₂ values were also <10 nM, and correlated well with the IC₅₀ value and calculated K _i values. A similar competitive antagonism of nondesensitizing P2X responses in cochlear hair cells was reported by Mockett et al. (1994), although the TNP-ATP concentrations used were orders of magnitude higher than those used in the present study. This was presumably due to lower affinity interactions of TNP-ATP with nondesensitizing cochlear P2X₂receptors (Housley et al., 1999). From these two studies, it appears that TNP-ATP is a competitive antagonist of nondesensitizing P2X receptors.

In addition to confirming the high affinity of TNP-ATP for the P2X_2/3 receptor, we have demonstrated that the off-kinetics of TNP-ATP are very slow (>2000 ms), and is concentration-independent. It is proposed that the slow off-kinetics contribute to the high affinity of the antagonist, as well as contribute to its apparent noncompetitive action at P2X₃ receptors. However, to draw conclusions based on apparent antagonist kinetics, a number of experimental assumptions must be met. In relation to the kinetics of a competitive antagonist, the solution exchange rate, the binding and unbinding kinetics of the agonist, and the channel gating kinetics must all be fast (Benveniste et al., 1990; Benveniste and Mayer, 1991). In addition, the channel must be relatively nondesensitizing. Under these assumptions, agonist will reach equilibrium quickly, and the slow kinetics of the antagonist will be measurable. In the present studies, the on- and off-kinetics of 10 μM α,β-meATP approached the limits of whole-cell diffusion for our drug delivery system (τ = 20–30 ms). In comparison, the on-kinetics for TNP-ATP at an IC₉₀ concentration was 10-fold slower, and the corresponding off-kinetics were >25-fold slower than the agonist. Although channel-gating kinetics have not been described adequately for either P2X₃ or P2X_2/3receptors, channel opening and closing rates estimated from either recombinant or native P2X receptors indicate that these parameters are also significantly faster than TNP-ATP kinetics (Krishtal et al., 1988;Bean et al., 1990; Cloues, 1995; Wright and Li, 1995; Evans, 1996; Ding and Sachs, 1999). Having met these assumptions, the TNP-ATP kinetics measured in the present studies are a reasonable measure of apparent antagonist binding and unbinding rates.

Preapplication of a competitive antagonist for a sufficient period of time allows equilibrium to be reached between antagonist and receptor. Subsequent application of a rapidly equilibrating agonist will result in a slow increase in agonist-activated current, reflecting the dissociation kinetics of the antagonist (Benveniste et al., 1990). Preapplication of TNP-ATP significantly slowed the subsequent agonist response and decreased its amplitude. This protocol gave a reliable measure of magnitude of inhibition by TNP-ATP, and allowed an indirect estimate of the off-kinetics of TNP-ATP. A better measure of TNP-ATP on- and off-kinetics was obtained by using the postapplication protocol, where agonist equilibrium was achieved first, and TNP-ATP was subsequently applied. The antagonist on-kinetics reflected theapparent on-rate because these kinetics depend on antagonist concentration, the intrinsic antagonist association and dissociation rates, and the concentration of agonist (Benveniste and Mayer, 1991). On the other hand, the measured TNP-ATP off-kinetics revealed a closer estimate to the true dissociation rate because these were concentration-independent and were sufficiently slow to not be influenced by rapid association and dissociation of agonist. Interestingly, we also found that if TNP-ATP was coapplied with agonist, the response reached a rapid peak, but a progressive decrease in agonist-activated current occurred with a kinetic profile that was similar to the on-kinetics measured using the postapplication protocol. It appeared that coapplication was a rapid way to obtain an estimate of the on-kinetics of TNP-ATP.

TNP-ATP was originally characterized as a potent, noncompetitive antagonist at rapidly desensitizing rat P2X₃receptors (Virginio et al., 1998). Likewise, we have observed an apparent noncompetitive block by TNP-ATP at rP2X₃receptors (Fig. 3C). This apparent competitive (rP2X_2/3) versus noncompetitive (rP2X₃) discrepancy between two receptor subtypes could be explained by intrinsic differences in the receptor desensitization rates. For example, a competitive antagonist of nondesensitizing nicotinic acetylcholine receptors appeared to be noncompetitive at rapidly desensitizing receptor subtypes (Alkondon et al., 1992; Briggs and McKenna, 1996). This effect was also due to rapid receptor desensitization occurring before slow dissociation of the high-affinity antagonist.

We propose a model (Fig. 7) to explain why TNP-ATP appears to be a noncompetitive antagonist at rapidly desensitizing P2X₃ receptors. When TNP-ATP is preapplied and allowed to reach equilibrium binding before an agonist is applied, agonist activation produces a P2X₃ response that desensitizes faster than TNP-ATP can dissociate from the receptor. Because rP2X₃ receptors desensitize with an initial rapid time course (τ = 39 ms; Burgard et al., 1999), desensitization will occur before a significant amount of TNP-ATP dissociates. In addition, the desensitization rate of P2X₃ receptors increases with increasing agonist concentration (Chen et al., 1995). Attempts to compete with TNP-ATP for receptor binding by increasing the agonist concentration can only make the agonist response faster, allowing even less time for antagonist dissociation. Under these conditions, competition with TNP-ATP will not be measured. However, competition can be measured at a relatively nondesensitizing receptor such as P2X_2/3.

The possibility that TNP-ATP can exhibit different antagonist actions (competitive versus noncompetitive) at different P2X receptors remains unlikely. P2X_2/3 receptors are formed by heteromultimeric combination of P2X₃ and P2X₂ subunits. Although the desensitization kinetics of P2X_2/3 receptors closely matches the P2X₂ phenotype, the pharmacology of the antagonist suramin, as well as that of agonists ATP and α,β-meATP, resembles that of P2X₃ receptors (Bianchi et al., 1999). The antagonist potency of TNP-ATP at P2X_2/3 receptors also matches that of P2X₃ receptors. Assuming that ATP, α,β-meATP, and TNP-ATP all bind at the same extracellular binding site on P2X_2/3 receptors, we would propose that this binding relationship is similar for P2X₃receptors. TNP-ATP would therefore show competitive antagonism at both P2X₃ and P2X_2/3 receptors. Because of the rapid desensitization of P2X₃receptors, this cannot be measured using standard electrophysiological techniques on wild-type receptors. The development of a sensitive and specific P2X₃ radioligand-binding assay could measure steady-state interactions between TNP-ATP and agonist, and could determine the competitive nature of TNP-ATP binding at P2X₃ receptors. The success of this approach would also depend on the ability of agonist to remain bound to the receptor after desensitization. Alternatively, this issue could be investigated electrophysiologically using a mutant P2X₃ receptor that retains P2X₃ binding properties, but has altered desensitization/gating kinetics. These approaches will require development of new tools for studying antagonist actions at P2X receptors.