PPNDS is an agonist, not an antagonist, for the ATP receptor of Paramecium

Paramecium are motile, eukaryotic unicells that are used as simple model systems for studying the kinds of chemosensory transduction and sensory adaptation mechanisms that are common to neuronal cells. In this respect, Paramecium are often thought of as `free-swimming sensory cells'(Machemer and dePeyer, 1977). As their swimming behaviors have been well correlated with electrophysiological changes (Eckert,1972; Saimi and Kung,1987; Hennessey,1989; Machemer, 1988a,b; Hinrichsen and Schultz, 1988),behavioral bioassays are used as quick and convenient estimates of membrane ion channel functions. Confirmation of actual membrane conductance changes is easily obtained by intracellular membrane potential measurements and voltage clamp procedures (Saimi and Kung,1987; Machemer,1988b). Biochemical and molecular genetic approaches are also routinely used because these cells can be grown as pure clonal cultures in axenic media. The combination of behavioral mutant screening(Kung, 1971; Saimi and Kung, 1987) and genetic transformations by microinjection(Kung et al., 2000) enables both `forward' and `reverse' genetic approaches to understanding receptor-operator ion channel activation and regulation. Because of the ease and relevance of these behavioral bioassays, these cells can be useful in pharmacological screening for drugs that may affect membrane receptors, ion channel functions, chemosensory transduction and adaptation(Hennessey, 1989).

In Paramecium, there are many types of depolarizing stimuli that elicit graded, somatic depolarizations (receptor potentials) and consequent Ca²⁺-based action potentials(Eckert, 1972). These stimuli include ionic stimuli (Machemer, 1988a,b),anterior mechanosensory stimuli (Ogura and Machemer, 1980), heat(Hennessey et al., 1983) and chemorepellents (Francis and Hennessey,1995) such as oxidants(Hennessey et al., 1994), GTP(Clark et al., 1992),polycations (Hennessey et al.,1995) and ATP (Kim et al.,1999). These depolarizing receptor potentials are initiated on the body (somatic) membrane and are conducted passively to the voltage-dependent ciliary Ca²⁺ channels for action potential generation(Eckert, 1972; Machemer, 1988a,b). When an action potential is produced, the Ca²⁺ influx causes elevation of intraciliary Ca²⁺ and Ca²⁺-dependent ciliary reversal. Therefore, a sufficiently strong depolarizing stimulus will cause ciliary reversal and backward swimming. Some depolarizing stimuli (such as Ba²⁺, Na⁺, lysozyme, oxidants and ATP) can cause trains of action potentials and repetitive bouts of backward and forward swimming called `avoiding reactions' (AR; Jennings, 1976). These chemicals act as chemorepellents because the avoiding reactions re-orient the swimming direction and result in movement away from the depolarizing stimulus in a manner very similar to the `tumble' response of motile bacteria to chemorepellents (Adler, 1987; Tso and Adler, 1974; Koshland, 1988). The avoiding reactions are easily observed under a simple dissecting microscope and serve as the basis for many behavioral bioassays.

We have previously described ATP-induced AR, depolarizing receptor potentials, external [³²P]ATP binding, chemorepulsion and chemosensory adaptation in the ciliate Tetrahymena(Kim et al., 1999). We define chemosensory adaptation as a decrease in responsiveness to a ligand as a function of time of exposure to an appropriate concentration of that ligand. This is different from the `adaptation', `accommodation' or `acclimatization'responses previously described in Paramecium because they involve long-term exposure to a high K⁺ stimulus and testing in a different stimulus such as Ba²⁺, Mg²⁺ or heat(Schusterman et al.,1978).

Chemosensory adaptation has been shown in responses to 15 min exposure to micromolar quantities of GTP, ATP and polycationic chemorepellents [such as lysozyme and PACAP (pituitary adenylate cyclase activating peptide)] in Tetrahymena (Kuruvulla et al., 1997; Kim et al., 1999; Mace et al., 2000), and similar adaptation has been shown to lysozyme and GTP in Paramecium(Kim et al., 1997). Our working model is that there are three separate chemorepellent receptors and associated pathways (the GTP, `polycation' and ATP reception systems), because cross-adaptation is not seen among the three systems. For example, adaptation to 10 μmol l<sup>-1</sup> GTP for 15 min causes a loss of GTP-induced AR,GTP receptor potentials and surface [<sup>32</sup>P]GTP binding in Paramecium without affecting [<sup>3</sup>H]lysozyme binding,lysozyme receptor potentials and lysozyme-induced AR(Kim et al., 1997). In Tetrahymena, no cross-adaptation was seen between the ligand binding,behavior or receptor potentials associated with the responses to the three representative chemorepellents GTP, ATP and lysozyme(Kuruvilla et al., 1997; Kim et al., 1999). Cross-adaptation is seen when two ligands utilize the same receptor, causing loss of responsiveness to both ligands following adaptation to one of them. Such cross-adaptation has been shown in the responses to lysozyme and PACAP in Tetrahymena, suggesting that they both activate the same `polycation receptor' (Mace et al., 2000)or its adaptation pathway. Therefore, behavioral cross-adaptation can be used as a first-screen bioassay to see whether two ligands may activate either the same receptor or a common receptor adaptation pathway.

In vertebrates, ATP receptors are involved in neurotransmission, regulation of blood flow and sensory transduction of signals for pain, tissue stretch and temperature (Burnstock, 1996, 2000; Cook et al., 1997; Dubyak and El-Moatassim, 1993; Harden et al., 1995; Ralevic and Burnstock, 1998). Although ATP can be released from many types of cells, an extracellular ATP signal often represents cell lysis and acts as a signal for cell and organ damage and abnormal stretching or distention. It is this type of cell-lysis-produced ATP that has recently been studied as a pain (nociceptive)signal (Burnstock, 1996; Cook et al., 1997; McCleskey and Gold, 1999; Souslova et al., 2000). Response adaptation and receptor desensitization has been shown in several types of ATP receptor systems (Burnstock,2000), and agonist-induced receptor internalization of native P2X1 has recently been demonstrated in a vertebrate preparation(Ennion and Evans, 2001). We have proposed that ATP receptors in the ciliates Paramecium and Tetrahymena are also acting as cell lysis detectors, providing a general `blood in the water' danger signal so that these cells can avoid whatever the situation was that caused nearby cell lysis. Therefore, the ciliate ATP receptors are chemorepellent receptors.

There are many types of external purinergic receptors in vertebrates. In general, responses to external purines are categorized as either P1 or P2,with P1 referring primarily to responses to ligands such as adenosine while P2 ligands are generally nucleoside triphosphates such as ATP(Abbracchio et al., 1993; Khakh et al., 2001). The P2X receptors are also ion channels (ionotropic), while P2Y receptors are G-protein coupled (metabotropic) (Khakh et al., 2001). There are currently seven different P2X receptor genes cloned, and each of these subtypes of receptors (P2X1-P2X7) has indicative agonist and antagonist specificities(Khakh et al., 2001). The non-hydrolyzable ATP analog β,γ-methylene ATP is often used as the preferred agonist in ATP receptor studies because ecto-ATPases and other extracellular hydrolases are often present to hydrolyze external ATP. It has been suggested that ecto-ATPases may be involved in the deactivation of purinergic agonists in much the same way as acetylcholinesterase inactivates acetylcholine signals (Kennedy et al.,1997; Hennessey et al.,1997). While ATP is a potent agonist for P2X receptors, the non-hydrolyzable analogs, such as β,γ-methylene ATP, are often much more potent (Burnstock, 2000; Khakh et al., 2001).

To determine whether the properties of the ATP receptors of Paramecium and Tetrahymena are similar to other known ATP receptors, we followed a classical pharmacological approach to see if these receptors fit into any of the known ATP receptor classifications. As the drug pyridoxal-phosphate naphthylazo-nitro-disulfate (PPNDS) has been described as a specific P2X1 receptor antagonist(Lambrecht et al., 2000), we tested it to see if it would block the responses of Paramecium to external ATP and its analogs. Surprisingly, PPNDS produced the same depolarizing responses as β,γ-methylene ATP, suggesting that PPNDS is not an ATP receptor antagonist in Paramecium but an agonist. We propose that the ability of PPNDS to cause inhibition of ATP responses is due,at least in Paramecium, to time-dependent, agonist-induced ATP receptor adaptation.

The trichocyst non-discharge mutant, nd6, of Paramecium tetraurelia (Sonneborn,1975) was used primarily in this study to avoid trichocyst discharge during electrophysiological recording. Cells were grown to late log phase (4 days) in the axenic medium of Soldo and Van Wagtendonk(1969).

The `Paramecium wash buffer' solution contained 1.0 mmol l^-1 CaCl₂, 1.0 mmol l^-1 MOPS, 0.5 mmol l^-1 MgCl₂, pH 7.2 with Tris-base. As PPNDS is a tetrasodium salt, cells were preincubated for 15 min in the wash buffer solution with the same Na⁺ concentration before exposure to PPNDS to adapt them to the Na⁺. Similar Na⁺ balancing was done with β,γ-methylene ATP because it is a disodium salt. The solution used for the in vivo [³²P]ATP binding assays contained 10 mmol l^-1 Tris-base, 0.5 mmol l^-1 MOPS, 1.0 mmol l^-1 tartrate, 10 μmol l^-1 EGTA, pH 7.2. PPNDS was obtained from Tocris Cookson Inc., Ellisville, MO, USA, but all other chemicals were purchased from Sigma-Aldrich, St Louis, MO, USA.

The chemorepellent behavioral assay used was the same as previously described (Kim et al., 1997). In this assay, individual cells were manually transferred to a test solution using a micropipet and observed under the dissecting microscope for swimming behavior. Cells showing any significant backward swimming events (`avoiding reactions') during the first 5 s after transfer were scored as positive responders. The avoiding reaction is defined by at least one backward movement of at least one body length. Ten cells were observed for each trial, and the mean ± S.D. was calculated for three trials (30 cells total) and expressed as the percentage of cells showing avoiding reactions (% AR).

In the adaptation experiments (as in all experiments), cells were first pre-incubated in the appropriate wash solution for 15-30 min before testing. Cells were then placed in either 25 μmol l^-1β,γ-methylene ATP (a non-hydrolyzable analog of ATP) or 100 μmol l^-1 PPNDS for varying amounts of time. Cells were transferred to the same solution without the repellent for 20 s (as a brief wash) and then transferred into the original test solution or a new test solution to assay for repellent responses.

One-electrode intracellular membrane potential measurement procedures were similar to those described previously (Kim et al., 1997). The electrophysiological recordings were taken in the behavioral solutions detailed above with 500 mmol l^-1 KCl electrodes (resistances of 100-200 mΩ). The membrane potentials were recorded under continuous perfusion conditions at a rate of 15-20 ml min^-1, and bath volume was approximately 1 ml. In the adaptation experiments, a cell was impaled in the repellent solution and exposed to repellent perfusion for varying times. After a 15-20 s buffer perfusion wash,the cell was re-exposed to the repellent perfusion.

The in vivo [³²P]ATP binding assays were performed in a solution without added Ca²⁺ to prevent ATP hydrolysis by the Ca²⁺-dependent ecto-ATPase(Smith et al., 1997; Hennessey et al., 1997). In the [³²P]ATP binding assay, 0.2 ml of packed cells were washed three times by centrifugation (700 g for 1 min) in the solution and then 100 μl aliquots were withdrawn for binding assays. Each 100 μl aliquot was mixed with [³²P]ATP and binding solution to a final volume of 200 μl. This was vortexed gently and two 20 μl samples removed and added to 2.5 ml of Ecoscint scintillation fluid for scintillation counting. The average of these readings represented the sum of the bound and free [³²P]ATP. The remaining sample was pelleted by centrifugation(500 g for 1 min) and 20 μl of the supernatant was removed for scintillation counting. This represented the amount of free[³²P]ATP. The amount bound was determined by subtraction because[bound] = [bound + free] - [free]. In the binding assay involving PPNDS, 0.2 ml of packed cells were washed three times by centrifugation in this solution and diluted to a final volume of 2.0 ml. After a 15 min incubation in this binding solution, PPNDS was added to a concentration of 100 μmol l^-1. Cells were incubated in 100 μmol l^-1 PPNDS for various times, and 100 μl aliquots were withdrawn and quickly washed by centrifugation in 1 ml of binding solution. The resulting cell pellet was vortexed with [³²P]ATP and binding solution. The resulting mixture was pelleted by centrifugation (500 g for 1 min). This pellet was extracted and washed via centrifugation in binding solution then placed into 2.5 ml Ecoscint for scintillation counting. This reading represented the amount of [³²P]ATP bound to cells. In the competitive inhibition studies, [³²P]ATP binding was performed in the presence of a 50-fold concentration excess (compared with the concentration of [³²P]ATP present) of either cold (unlabelled)β,γ-methylene ATP or PPNDS. This binding mixture was pelleted by centrifugation (500 g for 1 min) and the pellet was extracted and placed into 2.5 ml Ecoscint for scintillation counting.

β,γ-methylene ATP is an effective, non-toxic, chemorepellent in Paramecium. This non-hydrolyzable ATP analog caused repetitive avoiding reactions (AR) at micromolar concentrations, indicative of a depolarizing chemorepellent (Fig. 1B). It is important to note that these responses were immediate,the cells showing ATP-induced AR as soon as they came out of the transfer pipet. The EC₅₀ value (the concentration at which 50% of the cells tested showed AR) was approximately 12 μmol l^-1 forβ,γ-methylene ATP, and maximal responses were seen at approximately 25 μmol l^-1 (Fig. 1C). The electrophysiological responses to 25 μmol l^-1 β,γ-methylene ATP were transient receptor potentials(Fig. 2) that showed a mean amplitude of 13.7±0.6 mV and a mean duration of 26.5±7.2 s(N=10 each). The faster spikes, which are often seen riding on top of the receptor potential (see Fig. 2, Cell 1), are consistent with the graded, Ca²⁺-based action potentials associated with the AR(Eckert, 1972). It was often possible to observe visible jerks of the cell on the electrode correlating with the appearance of these spikes, verifying that they represent Ca²⁺-based action potentials.

Paramecium showed saturable external [³²P]ATP binding in the in vivo binding assays(Fig. 3A). A Scatchard plot analysis of amount bound and free (unbound) [³²P]ATP suggested a single set of external binding sites within the concentration range of 5-30 nmol l^-1 (Fig. 3B). The apparent K_d was 23.1 nmol l^-1, and the B_max was 112 pmol l^-1. Taking into consideration the number of cells in the assay, the number of functional surface receptors was estimated to be approximately 7.023×10⁵receptors cell^-1.

Continued exposure of cells to 25 μmol l^-1β,γ-methylene ATP for 15 min resulted in a time-dependent behavioral adaptation (Fig. 4A). This type of long-term adaptation was seen by exposing cells to 25 μmol l^-1 β,γ-methylene ATP for varying amounts of time, transferring the cells to the wash buffer for 20 s and then retesting the cells in 25 μmol l^-1 β,γ-methylene ATP. This is different from the short-term adaptation that was seen when cells were observed continuously after transfer to 25 μmol l^-1β,γ-methylene ATP for several minutes (C. R. Wood and T. M. Hennessey, manuscript in preparation). Short-term adaptation was seen as a decrease in the frequency of AR within the first 10-30 s after transfer toβ,γ-methylene ATP and it roughly correlated with the duration of the receptor potential. For example, a cell transferred to 25 μmol l^-1 β,γ-methylene ATP would show immediate, repetitive AR but the frequency of AR decreased over time until, after approximately 30 s, the cell was swimming forward most of the time with very few AR. This is short-term adaptation. However, if this cell (which had been exposed to 25μmol l^-1 β,γ-methylene ATP for 30 s) was transferred to wash buffer for 20 s and re-exposed to 25 μmol l^-1β,γ-methylene ATP, it would show AR because it had not been inβ,γ-methylene ATP long enough to initiate long-term adaptation. Electrophysiological analysis confirmed that receptor potentials could still be generated upon re-exposure to 25 μmol l^-1β,γ-methylene ATP even after 8 min of adaptation but the amplitudes were significantly decreased (Fig. 4B). Exposing cells to 25 μmol l^-1β,γ-methylene ATP for 15-20 min caused almost a complete loss of both AR (Fig. 4A) and the receptor potential (Fig. 4B),defining conditions for long-term adaptation.

In summary, 25 μmoll^-1 β,γ-methylene ATP can elicit a reliable, transient receptor potential that is the basis for the ATP-induced AR. Long-term behavioral and electrophysiological adaptation is complete after 15-20 min of exposure to 25 μmoll^-1β,γ-methylene ATP.

As there are many types of ATP receptors and they are often characterized by their sensitivities to agonists and antagonists, we decided to see whether a specific P2X1 inhibitor, PPNDS(Lambrecht et al., 2000),could eliminate the ATP responses of Paramecium. Instead of inhibiting the type of responses we were assaying, PPNDS caused the same responses as β,γ-methylene ATP. PPNDS produced the same kind of AR as β,γ-methylene ATP in the concentration range commonly used in the literature for inhibition of vertebrate P2X1 responses(Lambrecht et al., 2000). The EC₅₀ for PPNDS was approximately 70 μmoll^-1, and maximal responses were seen at approximately 120 μmoll^-1(Fig. 5A). Addition of 100μmoll^-1 PPNDS to β,γ-methylene-ATP-containing solutions did not produce any immediate inhibition of AR because PPNDS itself produced the same AR as β,γ-methylene ATP.

PPNDS also produced similar transient depolarizations(Fig. 5B) with similar amplitudes and durations to those seen with β,γ-methylene ATP. The`receptor potentials' induced by 100 μmoll^-1 PPNDS had a mean amplitude of 13.2±2.2 mV and a mean duration of 29.2±7.3 s. Using a standard t-test, there was no significant difference between the amplitudes (t=0.46, P>0.1, d.f.=4) or the durations(t=0.38, P>0.1, d.f.=4) of the β,γ-methylene ATP-induced depolarizations and the PPNDS-induced depolarizations. Exposure of cells to PPNDS for 15-20 min produced longterm adaptation as represented by a time-dependent loss of both the PPNDS receptor potential(Fig. 5C) and the PPNDS AR(Fig. 6B). Therefore, the behavioral and electophysiological responses to PPNDS are similar to the responses to β,γ-methylene ATP and they both show the same type of long-term adaptation.

Behavioral cross-adaptation was observed for β,γ-methylene ATP and PPNDS. Cells that were behaviorally adapted to 25 μmoll^-1β,γ-methylene ATP for 15 min lost AR after retesting in either 100μmoll^-1 PPNDS or 25 μmoll^-1β,γ-methylene ATP (Fig. 6A). Similarly, cells that were behaviorally adapted to 100μmoll^-1 PPNDS lost AR after retesting in either 100μmoll^-1 PPNDS or 25 μmoll^-1β,γ-methylene ATP (Fig. 6B). By contrast, cells that were behaviorally adapted toβ,γ-methylene ATP or PPNDS still retained full responsiveness to another non-toxic, depolarizing chemorepellent, 10 μmoll^-1 GTP(Clark et al., 1992). These adapted cells also showed normal AR in the standard ionic depolarizing test solutions (Saimi and Kung,1987) such as 40 mmoll^-1 K⁺, 10 mmoll^-1 Na⁺ and 8 mmoll^-1 Ba²⁺. This suggests that the adaptation is specific for the ATP-sensing pathway in these cells and is not due to an overall desensitization.

Behavioral adaptation to 100 μmoll^-1 PPNDS was associated with a time-dependent decrease in external [³²P]ATP binding sites(Fig. 7A). Adaptation to 25μmoll^-1 β,γ-methylene ATP for 15 min caused a similar loss of external [³²P]ATP binding(Fig. 7B). This showed that adaptation to either β,γ-methylene or PPNDS caused the loss of external [³²P]ATP binding. These changes in [³²P]ATP binding are not permanent but reversible. Cells were adapted to 100μmoll^-1 PPNDS for 15 min and were then transferred to the same solution without PPNDS for an additional 15 min. These `de-adapted' cells regained nearly 100% [<sup>32</sup>P]ATP binding, as did cells that were previously adapted to β,γ-methylene ATP for 15 min and then washed free of β,γ-methylene ATP and `de-adapted'(Fig. 7B). Similar reversibility (`de-adaptation') was seen in the behavioral and electrophysiological changes described when adapted cells were incubated in the ligand-free wash buffer for 15 min. We therefore do not consider this time-dependent, PPNDS-induced loss of [³²P]ATP binding to be true antagonist-type inhibition because the well-known ATP receptor agonistβ,γ-methylene ATP has the same effect. Bothβ,γ-methylene ATP and PPNDS are agonists in this system and prolonged (15 min) exposure to either ligand causes the same behavioral adaptation for the same reasons.

If both β,γ-methylene ATP and PPNDS activate the same receptor,they may bind to the same active site. To test this hypothesis, we added a 50-fold excess of PPNDS to the in vivo [³²P]ATP binding mixture and immediately measured the amount of [³²P]ATP bound. Excess PPNDS caused an immediate decrease in [³²P]ATP binding to whole cells (Fig. 8),suggesting that PPNDS may compete with [³²P]ATP for binding to the ATP receptor. Similarly, cold (non-radioactive) β,γ-methylene ATP caused immediate inhibition of [³²P]ATP binding when added in a 50-fold excess (Fig. 8). These observations are consistent with the idea that β,γ-methylene ATP and PPNDS may compete for [³²P]ATP binding to the active site of the ATP receptor; however, future purification of the ATP receptor is necessary to confirm such competitive inhibition. In this respect, bothβ,γ-methylene ATP and PPNDS are inhibitors of [³²P]ATP binding when they are present in the in vivo [³²P]ATP assay, but it must be remembered that cells were washed free of externalβ,γ-methylene ATP and PPNDS in all previous experiments where long-term, time-dependent adaptation was seen (see Figs 4,5,6,7,8).

As we had previously characterized the ATP avoidance behavior, depolarizing receptor potential, in vivo external [³²P]ATP binding and chemosensory adaptation in Tetrahymena(Kim et al., 1999), our first goal was to verify that similar responses were reliably seen in Paramecium. Although electrophysiological analysis of ATP responses is possible in Tetrahymena (Kim et al., 1999; Hennessey and Kuruvilla, 1999), we chose Paramecium for further pharmacological and electrophysiological studies because their larger size enables easier and more reliable electrophysiological results and greater ease of viewing in behavioral assays. As with Tetrahymena(Kim et al., 1999), in vivo [³²P]ATP binding assays with Paramecium showed[³²P]ATP binding kinetics that suggest a single class of binding sites within the concentration range tested and an apparent K_d in the nanomolar range(Fig. 4), although Paramecium were generally less sensitive to β,γ-methylene ATP in behavioral and electrophysiological assays compared with Tetrahymena (Kim et al.,1999). However, the β,γ-methylene ATP-induced receptor potential was larger and more clearly transient in Paramecium(Fig. 2) than in Tetrahymena (Kim et al.,1999).

Using Paramecium, we have shown that long-term behavioral adaptation to β,γ-methylene ATP occurs over a time span of approximately 15 min (Fig. 4). While the initial response of a cell to 25 μmol l^-1β,γ-methylene ATP is a depolarizing receptor potential (and consequent AR), cells adapted for 15 min in 25 μmol l^-1β,γ-methylene ATP swim forward and show no β,γ-methylene ATP-induced receptor potentials, even after a 20s wash and retest. As with Tetrahymena (Kim et al.,1999), behavioral adaptation to β,γ-methylene ATP in Paramecium is correlated with a loss in extracellular[³²P]ATP binding (Fig. 7). The loss of [³²P]ATP binding prevents the generation of the β,γ-methylene ATP-induced receptor potential, and the result is no β,γ-methylene ATP-induced AR. We conclude that Paramecium offer a unique opportunity to use quick, easy and humane behavioral bioassays to screen the effects of purinergic agonists and antagonists and then follow these behavioral screens with further electrophysiological and [³²P]ATP binding analysis.

The current literature suggests that one of the most specific anti-P2X1 antagonists to use for such a purpose is PPNDS(Lambrecht et al., 2000), a derivative of PPADS (pyridoxalphosphate-6-azophenyl-2,4-disulfonic acid) with reportedly higher specificity and potency than PPADS(Khakh et al., 2001). It has been common in such studies to also expose a cell to a variety of agonists and antagonists and to categorize the receptor on the basis of the spectrum of sensitivities to these ligands (Khakh et al., 2001). If PPNDS is an antagonist of the ATP receptor of Paramecium, it should inhibit β,γ-methylene ATP-induced AR.

PPNDS, like β,γ-methylene ATP, is an effective agonist for the ATP responses of Paramecium. PPNDS causes the same depolarizing responses as β,γ-methylene ATP. The behavioral EC₅₀ is higher for PPNDS than for β,γ-methylene ATP and it requires >100μmol l^-1 PPNDS for maximal stimulation. This is the approximate PPNDS concentration used in the previous inhibition studies(Lambrecht et al., 2000). PPNDS at 100 μmol l^-1 produces transient, depolarizing receptor potentials (Fig. 5B) that are indistinguishable from those elicited by 25 μmol l^-1β,γ-methylene ATP (Fig. 2) in both amplitude and duration.

Prolonged (15 min) exposure to 100 μmol l^-1 PPNDS produces chemosensory adaptation, seen as a loss of PPNDS-induced AR and PPNDS receptor potentials. Similarly, adaptation to β,γ-methylene ATP causes adaptation and loss of β,γ-methylene ATP-induced AR andβ,γ-methylene ATP receptor potentials. This type of loss in responsiveness has also sometimes been called desensitization or downregulation in the sensory literature but we chose the term adaptation to encompass the behavioral, electrophysiological and receptor binding events. The ATP receptor in rat vas deferens, which is a P2X1-like receptor, undergoes similar agonist-induced desensitization(Ennion and Evans, 2001).

Cross-adaptation is seen between PPNDS and β,γ-methylene ATP responses, suggesting that they may be activating the same receptor. Cells that have been adapted to 100 μmol l^-1 PPNDS also lose responsiveness to β,γ-methylene ATP (and vice versa). The time courses of losses in β,γ-methylene ATP-induced AR are similar for both PPNDS and β,γ-methylene ATP adaptation(Fig. 6A). The opposite is also true (Fig. 6B). Adaptation to either PPNDS or β,γ-methylene ATP causes loss of external[³²P]ATP binding (Fig. 7A), which is reversible by a process we call `de-adaptation'(Fig. 7B). Therefore, prolonged(15 min) exposure to 100 μmol l^-1 PPNDS causes a loss of responsiveness to β,γ-methylene ATP because of cross-adaptation of these responses. This is similar to the cross-adaptation we have seen between ATP and β,γ-methylene ATP in both Paramecium (C. R. Wood and T. M. Hennessey, unpublished observations) and Tetrahymena(Kim et al., 1999). This adaptation was specific for ATP and PPNDS responses because we did not see any cross-adaptation between PPNDS and GTP responses or β,γ-methylene ATP and GTP responses in either Tetrahymena(Kim et al., 1999) or the present study. Paramecium that were adapted to either PPNDS orβ,γ-methylene ATP were fully responsive to the other chemorepellents GTP and lysozyme as well as to ionic stimuli (data not shown). This is consistent with our previous hypothesis that there are at least three separate chemorepellent pathways in Paramecium and each pathway shows independent adaptation without affecting the sensitivities of the other two pathways. We further proposed (Kim et al.,1997; Kuruvilla et al.,1997) that behavioral cross-adaptation is a convenient way to test whether different ligands may activate the same receptor (or receptor pathway).

If two ligands activate the same receptor by binding to the same site, they should also compete for binding. We have shown that a 50-fold excess of eitherβ,γ-methylene ATP or PPNDS can virtually eliminate all external[³²P]ATP binding in our in vivo assay(Fig. 8) but a 50-fold excess of GTP will not. These results support the hypothesis that both PPNDS andβ,γ-methylene ATP are agonists that share a common binding site on the ATP receptor of Paramecium but GTP does not. The question of whether PPNDS is a true competitive inhibitor for ATP binding to the ATP receptor will be best approached when the ATP receptor of Parameciumis purified for in vitro binding assays.

Preliminary observations in low Ca²⁺ solutions suggest the EC₅₀ for ATP to be approximately 200 μmoll^-1 (S. Zdep, personal communication), which is much higher than the measured K_d value for [³²P]ATP. This discrepancy may be due to the fact that Ca²⁺ is included in the behavioral assays(because it is necessary to show AR) but not in the binding assays. Ca²⁺ can activate the Ca²⁺-dependent ecto-ATPase and raise the EC₅₀ because of ATP hydrolysis(Hennessey et al., 1997). For this reason, Ca²⁺ is omitted from the binding assays to inhibit this ecto-ATPase (which has no known inhibitor in Paramecium). However, such a large discrepancy between the EC₅₀ and K_d values for ATP suggests that the binding data may not always fit quantitatively with the behavioral data for other reasons that we have yet to resolve.

We conclude that PPNDS is an effective ATP receptor agonist (activator) in Paramecium and it is not an antagonist in this system. PPNDS causes the same depolarizing receptor potentials and avoiding reactions as the well-known ATP receptor agonist β,γ-methylene ATP, and cross-adaptation is seen between these two agonists. The in vivo[³²P]ATP binding results are consistent with the hypothesis that PPNDS and β,γ-methylene ATP both compete for the same ATP binding site on the ATP receptor.

As the literature describes PPNDS as an antagonist in vertebrate cells, it is possible that PPNDS only has agonist activity in Parameciumbecause of a difference in the structure of the ciliate ATP receptor. Considering the long evolutionary distance between the ciliates and vertebrates, such a difference in agonist specificity cannot be ruled out. If so, this ciliate ATP receptor may represent a new classification of ATP receptors. This class would be easily recognized by the fact that PPNDS acts as an agonist instead of an antagonist. However, it is possible that PPNDS may also be an agonist in other systems where preincubation with 100μmoll^-1 PPNDS for ≥15 min is necessary for inhibition of an ATP-dependent response. In some of these cases, it is possible that the loss of ATP responses seen after prolonged exposure to PPNDS may be due to the same agonist-induced adaptation (or `desensitization') that might be caused by any agonist for this receptor (such as β,γ-methylene ATP) rather than by the action of a true antagonist.

The distinction between ATP receptor antagonism and agonist-induced adaptation is not a trivial matter when considering the mechanism of action of new, experimental drugs. An example is the type of vertebrate ATP receptor that is a specific type of pain receptor(Cook et al., 1997). For these receptors, ATP released from lysed cells causes pain by activation of ATP receptors on specialized nerve endings. This type of preparation may prove valuable for screening drugs that may act as local analgesics by blocking the ATP receptors. However, it would be important to know that the drug does not cause the pain that you want to prevent. Although sensory adaptation could eventually result in analgesia by prolonged exposure to an agonist, the action of a true antagonist might be preferred.

We thank Mark Kim for his generous contribution of the [³²P]ATP binding data and Alan Siegel for his help in generating the swim path digital photographs. This work was supported by NSF grant MCB-9983060 and NIH grant DK 60869-01.