Kinetics of antagonist actions at rat P2X_2/3 heteromeric receptors

Abstract

1. Currents through heteromeric P2X_2/3 receptors were evoked by applying α,β-methylene-ATP to human embryonic kidney cells transfected with cDNAs encoding the P2X₂ and P2X₃ subunits. The concentration of α,β-methylene-ATP were ⩽30 μM because higher concentrations can activate homomeric P2X₂ receptors.

2. The kinetics of action of three structurally unrelated antagonists were studied; these were 2′, 3′-O-(2,4,6,trinitrophenyl)-ATP (TNP-ATP), pyridoxal-5-phosphate-6-azophenyl-2′,4′-disulphonate (PPADS) and suramin.

3. The association and dissociation rate constants were determined by pre-applying the antagonist for various periods prior to the co-application of agonist and antagonist, or by changing the solution from one containing only the agonist to one containing both agonist and antagonist.

4. The high affinity of TNP-ATP for the P2X_2/3 receptor (K_D≈2 nM) results from fast binding (k₊₁≈100 μM⁻¹ s⁻¹) rather than slow unbinding (k₋₁≈0.3 s⁻¹). For suramin (K_D≈1 μM) the association rate constant (≈1 μM⁻¹ s⁻¹) was 100 times slower than that of TNP-ATP but the dissociation rate constant was similar (k₋₁ ≈1 s⁻¹). PPADS (K_D≈0.1 μM) associated and dissociated some 100 – 10,000 times more slowly than the other antagonists.

Introduction

Primary afferent neurons involved in the sensation of pain can be identified by the relatively small diameter of their cell bodies, and by their expression of binding sites for the isolectin B4, the vanilloid (capsaicin) receptor VR1, and the P2X₃ subunit (reviewed by Ding et al., 2000; Hamilton & McMahon, 2000; Bland-Ward & Humphrey, 2000; Burnstock, 2000). Gene knock-out experiments indicate that P2X₃ receptors are also involved in sensation of warmth and bladder distension (Souslova et al., 2000; Cockayne et al., 2000). Primary afferent neurons express other P2X receptor subunits, and the combination of P2X₂ and P2X₃ subunits is of particular interest (Vulchanova et al., 1998; Bradbury et al., 1998; Cook et al., 1997). In heterologous expression systems, this combination can form heteromeric P2X receptors (termed P2X_2/3) and these are marked by a unique phenotype with respect to the currents elicited by ATP and analogues (Lewis et al., 1995). Thus, low micromolar concentrations of ATP, though not α,β-methylene-ATP (αβmeATP), elicit currents in cells expressing only the P2X₂ subunits, and these show minimal desensitization over a period of 1 – 2 s. Both αβmeATP and ATP evoke currents in cells expressing only P2X₃ subunits, but when the concentration exceeds the EC₅₀ these currents desensitize fully with applications continued for 1 s. A third phenotype is observed in cells expressing both P2X₂ and P2X₃ receptors; they respond to ATP and αβmeATP, but the current desensitizes very little (Lewis et al., 1995). This phenotype is also observed in nodose ganglion cells (Lewis et al., 1995; Thomas et al., 1998), in subsets of dorsal root ganglion cells (Grubb & Evans, 1999), in some trigeminal ganglion cells (Cook et al., 1997) and in some autonomic ganglion cells (Khakh et al., 1995; Zhong et al., 2000; 2001). It has recently been proposed that these properties are displayed by a set of capsaicin-insensitive primary afferents involved in sensing mechanical allodynia (Tsuda et al., 2000).

The substantiation of a role for ATP in primary afferent transmission depends very much on the availability of selective antagonists (see North & Surprenant, 2000). Those in most widespread use are 2′,3′-O-(2,4,6-trinitrophenyl)-ATP (TNP-ATP), suramin, and pyridoxal-5-phosphate-6-azophenyl-2′,4′-disulphonic acid (PPADS). When studied in heterologous expression systems, TNP-ATP is particularly effective at blocking homomeric P2X₁, homomeric P2X₃ and heteromeric P2X_2/3 receptors (IC₅₀ about 1 nM; Virginio et al., 1998). It is of intermediate effectiveness at heteromeric P2X_1/5 receptors (Surprenant et al., 2000) and relatively ineffective at homomeric P2X₂, P2X₄, and P2X₇ receptors (Virginio et al., 1998). Suramin shows less discrimination among the different receptors that have been studied, although the P2X₄ receptors are less sensitive than the others. PPADS is a submicrolar blocker of P2X₁ receptors (Jacobson et al., 1998), a micromolar blocker of P2X₂, P2X_2/3 and P2X₅ receptors, essentially ineffective at P2X₄ receptors and of intermediate effect at P2X₇ receptors (Buell et al., 1996; North & Surprenant, 2000).

Unfortunately, it is often difficult to compare the results of different experimenters with respect to the antagonists, because they have not been systematically compared in the same preparation. Various times of preincubation with antagonists have been used, and it has often not been determined whether these are sufficient to reach steady-state. There have been no comparative studies of the three antagonists on cells expressing P2X_2/3 receptors. Given the growing evidence that the P2X_2/3 heteromer may be a useful molecular target for drugs altering sensory processing, the aim of the present work was to determine the kinetics and mechanism of action of these three compounds at this particular P2X receptor subtype.

Methods

Cells and chemicals

HEK293 cells were used that had been stably transfected with rat P2X₂ cDNA (homomeric P2X₂) or stably transfected with rat P2X₂ and P2X₃ cDNAs in a bicistronic vector (Evans et al., 1995; Kawashima et al., 1998). The cells expressing heteromeric P2X_2/3 receptors also express homomeric P2X₂ receptors, but show no evidence of homomeric P2X₃ receptors (Lewis et al., 1995; Kawashima et al., 1998; Virginio et al., 1998). Culture media, sera and all cell culture reagents were obtained from Life Technologies (Paisley, U.K.), ATP and αβmeATP were from Sigma, TNP-ATP (2′-(or-3′)-O-(trinitrophenyl)adenosine 5′-triphosphate, trisodium salt) was from Molecular Probes (Eugene, OR, U.S.A.), suramin was from Bayer and PPADS (pyridoxal 5-phosphate-6-azophenyl-2′,4′-disulphonate, tetrasodium salt) was from Tocris (Cookson, U.K.).

Electrophysiology and drug applications

Standard whole-cell patch-clamp recordings were carried out at room temperature 24 – 72 h after passage of stable cell lines. Currents were recorded with an EPC9 amplifier (HEKA Elecktronik, Lambrecht, Germany). Unless otherwise stated, experiments were performed at a holding potential of −60 mV; series resistance, which ranged from 2 – 15 MΩ, was compensated by 60%. Patch pipettes (3 – 7 MΩ) were filled with (mM): NaF 145, EGTA 10, HEPES 10. The extracellular solution contained (mM): NaCl 147, KCl 2, CaCl₂ 2, MgCl₂ 1, HEPES 10 and glucose 13. Solutions were maintained at pH 7.3 and osmolarity 300 – 315 mOsmol l⁻¹. Current-voltage curves were obtained by applying 0.5 or 1 s ramp voltages from −130 to 50 mV. Agonists and antagonists were applied using the RSC 200 fast flow delivery system (Biologic Science Instruments, Grenoble, France). Agonists were normally applied at 2 min intervals. We measured the solution exchange time by observing the membrane current when switching from NMDG-Cl to NaCl in the continuous presence of ATP; the 10 – 90% time was 268±43 ms (n=4). For the experiments with antagonists, αβmeATP was applied at its EC₅₀ concentration; this was determined each day from a sample of cells, and the values ranged from 4 – 9 μM. The antagonists were then applied at concentrations that gave an onset of inhibition which was clearly slower than the time of solution exchange. Concentrations of suramin lower than 0.3 μM were not used because they usually potentiated rather than inhibited the current evoked by αβmeATP.

Data analysis

Two methods were used to study the actions of the antagonists. First, current was evoked by an agonist application sustained for 20 – 25 s (always αβmeATP, at the EC₅₀). During this period, the solution was changed to one that contained both agonist and antagonist, for 5 – 10 s. Subtraction of the two currents gave the effect of the antagonist. Second, onset kinetics were measured by applying the antagonist for a variable period before (and also during) a test application of αβmeATP; the peak amplitude of the current evoked by αβmeATP was compared with and without the antagonist. Conversely, the recovery of the peak amplitude of the αβmeATP-evoked current was measured after the washout of the antagonist. The time courses of antagonist action were fitted by exponentials of time constants τ_on and τ_off. The dissociation rate constant (k₋₁) was computed directly as 1/τ_off; the association rate constant (k₊₁) was computed as {(1/τ_on) −k₋₁}/[B] where [B] is the antagonist concentration. Results are presented as mean±s.e.mean, and the numbers refer to the number of individual cells tested. Differences among measured time constants were tested by Kruskal-Wallis's and Dunn's nonparametric tests.

Results

Isolation of responses mediated through heteromeric P2X_2/3 receptors

Cells transfected with both P2X₂ and P2X₃ cDNAs might be expected to express homomeric P2X₂, homomeric P2X₃ and one or more species of heteromeric P2X_2/3 receptors, and this has been shown to be the case previously (Thomas et al., 1998; Liu et al., 2001). We excluded involvement of homomeric P2X₃ receptors because (a) we did not observe any fast-desensitizing component to the currents and (b) responses at the homomeric P2X₃ receptor disappear when the agonist is applied repeatedly at intervals less than 4 min (Lewis et al., 1995). It is also possible that persistent release of ATP from HEK293 cells under these conditions contributes to desensitization of the receptors (Surprenant et al., 2000). To exclude a contribution from homomeric P2X₂ receptors, we determined the concentrations at which αβmeATP could be used, by carrying out control experiments on cells stably transfected only with the P2X₂ receptor. Concentrations of αβmeATP ⩾100 μM evoked inward currents in cells expressing only homomeric P2X₂ receptors (Figure 1). These results mean that it is not possible to use αβmeATP (at ⩾100 μM) to activate selectively heteromeric P2X_2/3 receptors in cells co-expressing P2X₂ and P2X₃ subunits, because of the concomitant activation of homomeric P2X₂ receptors. They also mean that the method of Schild analysis could not be used reliably to test for competitive interaction, because useful dose-ratio's could only be obtained by increasing the αβmeATP concentration into this non-selective range.

Figure 1.

αβmeATP and ATP currents at heteromeric P2X_2/3 receptors and at homomeric P2X₂ receptors. (A) Representative current traces; period of agonist application indicated by bars above traces. (B) Graph compares currents activated by αβmeATP (filled circles) in HEK cell stably transfected with P2X₂ and P2X₃ with those in cells transfected only with the P2X₂ receptor cDNA. Currents are plotted as per cent of that evoked by 30 μM αβmeATP at the heteromeric P2X_2/3 receptor, or 30 μM ATP at the homomeric P2X₂ receptor (this is a maximal concentration).

The current amplitude in response to 30 μM αβmeATP, in cells co-expressing both subunits, ranged from 25 to 90% (54±6%, n=8) of the current evoked by 30 μM ATP. These agonist concentrations are near-maximal for activation of heteromeric P2X_2/3 and homomeric P2X₂ receptors respectively, so they provide a rough estimate of the proportion of heteromeric P2X_2/3 to homomeric P2X₂ receptors in a single cell. We estimate in this way that there is an approximately 45 : 55 ratio of heteromeric P2X_2/3 to homomeric P2X₂ receptors; this is similar to our previous study in which we estimated a 40 : 60 ratio using the same stably-transfected cell line (Thomas et al., 1998). Concentration-response curves for αβmeATP up to the concentration of 30 μM in these cells yielded an EC₅₀ value of 6.5±2.1 μM (n=18; Figure 1). This EC₅₀ value varied somewhat among cells, and in experiments on antagonist kinetics, the actual concentration of αβmeATP used was that determined to be the EC₅₀ on cells tested on the same day. At this concentration, the αβmeATP-evoked current reached its peak amplitude typically in 0.98±0.03 s (n=54), and declined to 86±0.8% (n=54) during a 5 s application. With a 20 s application, the current declined to 39±2.3% (n=38) of control (see Figure 2).

Figure 2.

Inhibition of αβmeATP-evoked currents by TNP-ATP, suramin and PPADS at heteromeric P2X_2/3 receptors. In each panel, the two superimposed traces show the currents elicited by αβmeATP (at EC₅₀: filled bar); in one of the two traces, the antagonist was also applied during the αβmeATP application (period indicated by the open bar). Subtraction of traces such as these was used to estimate the onset and offset kinetics.

Onset and offset of antagonist action

In the first series of experiments (antagonist co-application), αβmeATP was applied for periods of 20 or 25 s. This elicited an inward current that declined during the application. The decline, which we term slow desensitization (to distinguish it from the millisecond desensitization observed with homomeric P2X₃ receptors), was variable from cell to cell, but quite reproducible within the same cell (Figure 2). We then applied the antagonists TNP-ATP, suramin or PPADS during the maintained application of αβmeATP. The onset of the inhibition of the current was clearly dependent on the concentration of antagonist, and this experiment provided a range of values for τ_on (Figure 2). The offset of action was well fit by single exponential (τ_off for TNP-ATP was 4.7±1.9 s (3), 3.8±0.7 s (4), 1.9±0.13 s (3) and 2.4±0.13 s (4) for 1, 3, 10 and 30 nM: τ_off for suramin was 0.88±0.13 s (3), 0.75±0.06 s (5), 0.89±0.17 s (4) and 0.84±0.04 s (4) for 0.3, 1, 3 and 10 μM). There was no significant dependence on antagonists concentration (P>0.05). We therefore pooled all the concentrations in our estimate of k₋₁; we used this pooled estimate for the computation of k₊₁ and K_D values (Figure 3; Table 1).

Figure 3.

Summary data for the onset and offset of antagonist inhibition from experiments such as those shown in Figure 2. n=3 – 5 for each point. Forward rate constants (closed symbols) (k₊₁) are in μM⁻¹ s⁻¹. Backward rate constants (open symbols) (k₋₁) are in s⁻¹. Backward rate constants for PPADS (open triangles) are from the experiments shown in Figure 5.

Table 1.

Association rate constants (k₊₁), dissociation rate constants (k₋₁) and dissociation equilibrium constants (K_D = k₋₁/k₊₁) for the antagonism of αbmeATP by TNP-ATP, suramin and PPADS at P2X_2/3 heteromeric receptors. For TNP-ATP and suramin, k₋₁ was measured as the mean value of 1/τ_off for all concentrations tested

Figure 4 illustrates the results of the second series of experiments (antagonist pre-application). Each of the traces shows the current elicited by αβmeATP applied for 5 s at the EC₅₀ concentration; the application of αβmeATP was begun after a variable period of antagonist application (indicated above each trace in s). At the lowest concentration of TNP-ATP (1 nM), simultaneous application of TNF-ATP and αβmeATP (i.e. 0 s pre-application in Figure 4) caused minimal effect on the peak current, but it did inhibit the current measured at 5 s; at 10 nM, TNP-ATP profoundly reduced the current during the 5 s co-application. The altered time course of the agonist-induced current could be well accounted for the time course of onset of the antagonist, as determined from the co-application experiments described above. Suramin (10 μM) appears to bind to the receptor almost as quickly as αβmeATP under these conditions, because even their simultaneous application there was a marked reduction in the initial peak response to agonist (Figure 4); however, at this concentration the rate of onset of action of suramin is unreliable because it is limited by the solution exchange time (see Methods).

Figure 4.

Onset kinetics for TNP-ATP (A) and suramin (B), determined by antagonist pre-application. The currents shown were evoked by co-applying αβmeATP and the antagonist, after a variable period of antagonist application (indicated in s above each trace). The first trace in each set shows the control response in the absence of antagonist.

The onset of the inhibition by PPADS was measured in a similar manner, and the results are plotted in Figure 5A. The offset kinetics of PPADS was estimated by observing the recovery of the response to αβmeATP from its fully depressed state. The time course of recovery was the same at all concentrations tested (1, 3, 10 and 30 μM), and was well fit by an exponential of time constant 550 s (Figure 5B; Table 1). Note, however, that the recovery from inhibition by PPADS was never complete, reaching a steady level at which the current evoked by αβmeATP was about 50% of its initial value (Figure 5B). Figure 5 also shows that there was no change in the overall time course of the response to αβmeATP during the washout of PPADS, suggesting that antagonism by PPADS did not obviously change the slow desensitization of the receptor.

Figure 5.

Onset and offset of PPADS inhibition. (A) The time course of inhibition by PPADS determined from experiments such as those shown in Figure 4. Time constants (τ_on) for onset of inhibition are 79 s (1 μM), 26 (3 μM), 8.3 (10 μM), 2.7 (30 μM), and 1.4 s (100 μM; not shown in figure), corresponding to k₊₁ of 0.013 μM⁻¹ s⁻¹. (B) The recovery of the response to αβmeATP after washout of PPADS. The values in the graph are the means±s.e.mean for six experiments.

Lack of voltage dependence of antagonist action

Agonist-induced currents over the voltage range −130 to 50 mV were recorded during ramp commands in the absence and presence of increasing concentrations of TNP-ATP (1 – 30 nM) and suramin (0.3 – 10 μM). There was no voltage-dependence in the range −130 to −20 mV; the marked rectification of the current precluded accurate measurements at more positive potentials. Similar experiments were performed using PPADS (1 μM), with pre-application of 30 – 90 s; the percentage inhibition was the same at all voltages.

Discussion

The present experiments have determined the rates of association and dissociation of three antagonists at heteromeric P2X_2/3 receptors. The first important step was the isolation of currents evoked at the heteromeric receptor, because cells transfected with P2X₂ and P2X₃ subunits clearly exhibit both homomeric P2X₂ receptors and heteromeric P2X_2/3 receptors (Thomas et al., 1998). Individual neurons also appear to express both such receptor classes (Thomas et al., 1998; Zhong et al., 2000). If the P2X receptor functions as a trimer (Nicke et al., 1998; Stoop et al., 1999), then two types of heteromeric receptor might be formed, which contain one or two copies of the P2X₃ subunit. We do not know whether one or both such receptor species are activated by αβmeATP in the present experiment; the Hill slope of the agonist dose-response curve (about 1.5; Figure 1) indicates that at least some of the heteromeric receptor(s) must contain more than one αβmeATP binding site. However, we are confident that homomeric P2X₂ receptors would not be activated by αβmeATP within the concentration range that we used (Figure 1). The limitation that the concentration of αβmeATP could not be increased by more than about 10 fold without leading to activation of homomeric P2X₂ receptors precluded the application of methods which depend on surmounting the antagonist action, such as Schild analysis. To be confident that antagonism is competitive, it is necessary to demonstrate that the block is surmountable for larger dose-ratio; a parallel shift in the dose-response curve for smaller dose-ratio is readily observed for non-competitive antagonism (Milner et al., 1982) and can be accounted for by the presence of spare receptors (see Foreman & Johansen, 1996).

When the agonist αβmeATP was applied alone at the EC₅₀ concentration, the current peaked within 1 s; at the termination of the application, the currents declined exponentially with a time constant of about 250 ms (similar results were obtained by Burgard et al., 2000). On the other hand, the slow desensitization that was observed had a time constant of about 10 – 20 s. We used two methods to estimate the on- and off-rates for the antagonists. The first was co-application (Figure 2), where the antagonist was introduced when the channel was already opened by the continuous presence of the agonist. This method was useful for the onset of action of all the agonists, so long as their concentration was low enough (Figure 2); it was also useful for the offset of action of TNP-ATP and suramin. The second was pre-application (Figure 4), where the antagonist was applied alone before changing to co-application of agonist and antagonist. This method was useful for PPADS, but only for the lowest concentrations of TNP-ATP and suramin (Figure 4).

In the case of PPADS, there was reasonable agreement between the k₊₁ measured by co-application (10, 30, 100 μM; Figure 2; Table 1) and that measured by pre-application (Figure 5A; the k₊₁ values for these three concentrations are 0.013 μM⁻¹ s⁻¹). At the lower concentration (3 μM) PPADS appeared to bind some three-times more rapidly to the open (co-application) than the closed (pre-application). The kinetics of the response to agonist was unaltered during the wash-out of PPADS (Figure 5) and this indicates that desensitization is unaffected by the presence of the antagonist.

The results of the present work are significant for two kinds of reasons. The first has to do with the practical aspect of using the antagonists experimentally to characterize P2X receptors. A recent paper by Li (2000) on bullfrog dorsal root ganglion cells reported that PPADS blocked the currents evoked by ATP with k₊₁ similar to that of the present study (0.06 μM⁻¹ s⁻¹ measured at 1 μM PPADS). He also reported that PPADS bound more rapidly to the channel when agonist was present than when agonist was absent. However, the rate of PPADS dissociation was somewhat faster than we observed, being described as complete after 8 min. The present work indicates that (Table 1) in the case of PPADS at the P2X_2/3 receptor, the dissociation equilibrium constant is about 0.1 μM, and at this concentration an application of 20 – 30 s would be sufficient for steady-state occupancy.

The second interest in the present results relates to the possible mechanism of inhibition at the P2X_2/3 receptor. Because of the need to isolate the currents through P2X_2/3 receptors, we have been constrained to use a single agonist concentration; we are therefore not able to say whether the antagonism is competitive. With a single agonist concentration, the simulations also do not discriminate between models in which agonist and antagonist can be bound to the same subunit at the same time (non-competitive), and those in which they can not (competitive). On the other hand, the inhibition by TNP-ATP that we have observed is consistent with competition for an overlapping binding site for ATP and TNP-ATP. However, the parts of the binding pocket that interact with the phosphate chain and the purine base must discriminate between ATP and TNP-ATP. The TNP-ATP analogues TNP-ADP, TNP-AMP and GTP-TNP are all antagonists in the nM concentration range, even though ADP, AMP and GTP do not activate or block the receptor (Virginio et al., 1998).

We had originally conjectured that the high affinity for TNP-ATP at the P2X_2/3 receptor (K_D about 3 nM) would result from a slow dissociation rate, and thought that the kinetics of the wash-out of antagonist action might provide information regarding the number of binding sites (Palma et al., 1996). In fact, the dissociation rate is only about 10 times slower than that of αβmeATP, and it is the 10 times faster association rate that accounts for its high affinity at the P2X_2/3 receptor. It is possible that TNP-ATP approaches the binding site more rapidly than αβmeATP because it has three additional negative charges; however, the lack of voltage-dependence to the action of TNP-ATP indicates that the site is probably outside the membrane electrical field. The finding that the association rate constant (k₊₁) for TNP-ATP was faster at lower concentrations most likely reflects negative cooperativity in the binding steps, implying that more than one molecule of TNP-ATP must bind for effective blockade. There has recently been progress in identifying some of the residues that contribute to the ATP binding site for the P2X₁ (Ennion et al., 2000) and P2X₂ (Jiang et al., 2000) receptors. Given the negativity of the TNP moiety, it will be interesting to determine by mutagenesis whether removal of positive charges from the P2X₃ subunit can selectively reduce the affinity for TNP-ATP without changing the effectiveness of ATP.

Although the association rate of suramin is 30 – 100 times slower than that of TNP-ATP, it is about 100 times faster than that of PPADS. Suramin is a larger molecule than PPADS, so it is possible that this difference (as surmised for TNP-ATP) might result from electrostatic forces; there are six negative charges on the suramin molecule compared with three on PPADS (one phosphate, two sulphonates). The dissociation rate of PPADS is also about 10,000 times slower than that of suramin, and recovery remained incomplete after 21 min (about 50% recovery). This is very similar to the situation with the homomeric P2X₂ receptor, which also shows only a partial recovery from inhibition after 20 min washing (Evans et al., 1995; Buell et al., 1996). This has been interpreted as resulting from Schiff base formation with a lysine residue at position 246 of the P2X₂ receptor, because the washout of the antagonism occurs within 10 min (and is complete) for the mutant receptor K245E (Buell et al., 1996). The P2X₃ subunit has threonine rather than lysine at this position, and might therefore behave like P2X₂[K246E]. The present finding that recovery from PPADS inhibition occurred was only partial could result from the sum of two binding steps, one slowly reversible and the other irreversible.

Abbreviations

αβmeATP

α,β-methylene-ATP

HEK

human embryonic kidney

PPADS

pyridoxal-5-phosphate-6-azophenyl-2′,4′-disulphonic acid

TNF-ATP

2′,3′-O-(2,4,6-trinitrophenyl)-ATP