Proc. Natl. Acad. Sci. USA Vol. 93, pp. 1859-1863, March 1996 Neurobiology

An ATP-activated, ligand-gated on a presynaptic nerve terminal (purinergic /chicken ciliary ganglion/transmitter release/autoreceptor) XIAO PING SUN AND ELIS FRANK STANLEY* Synaptic Mechanisms Section, National Institute of Neurological Disorders and Stroke, Building 36, Room 5A27, National Institutes of Health, Bethesda, MD 20892 Communicated by Thomas S. Reese, National Institute of Neurological Disorders and Stroke, Bethesda, MD, November 21, 1995

ABSTRACT ATP has recently been identified as a fast chamber. Experiments were carried out at room temperature in both the central and peripheral nervous (22-25°C). systems. Several studies have suggested that ATP can also Solutions. The standard internal solution for patch clamp affect the release of classical , including recording contained 120 mM cesium gluconate, 30 mM CsCl, with which it is co-released. We have searched 10 mM EGTA, 2 mM MgCl2, 10 mM Hepes (pH 7.3, adjusted for ATP receptors on a cholinergic presynaptic nerve terminal with CsOH). In some cases, cesium gluconate was substituted using the calyx-type of the chicken ciliary ganglion. by potassium gluconate or CsCl. The standard external solu- ATP was pulsed onto the terminals under voltage clamp and tion contained 150 mM NaCl, 5 mM CaCl2, 2 mM MgCl2, 3 induced a short latency cation current that exhibited inward mM KCI, 10 mM glucose, 10 mM Hepes (pH 7.3, adjusted with rectification and marked desensitization. This current was not NaOH). The osmolarity of all solutions was adjusted to seen with adenosine but was mimicked by several sterically 300-310 mOsm with sucrose. restricted ATP analogs and was blocked by suramin. ATP- Application. ATP and other were prepared activated single ion channels exhibited prominent flickering from a 0.1 M stock solution before each experiment. The stock and had a conductance of -17 pS. Our results demonstrate a solutions were kept at -20°C and were used no longer than 2 ligand-gated P2x-like purinergic receptor on a cholinergic weeks. ATP and other agonists were pressure expelled from a presynaptic nerve terminal. 3- to 5-,.m tip-diameter pipette positioned -50 ,tm from the target cell. The duration of treatment was controlled by the ATP is a prime candidate as a transmitter recording software. in cholinergic nerve terminals and can result in both up- and Current Recording. Conventional whole-cell and single- down-regulation of synaptic transmission. Although some of channel patch-clamp (outside-out) techniques were used to these actions are due to the breakdown product adenosine detect ion currents under voltage clamp (14). The thin-wall (1-3), ATP itself can also affect transmitter release (4, 5). ATP borosilicate glass patch pipettes were fire polished and coated is known to activate nonselective cation channels on with bee wax or Sylgard and had an electrical resistance in the and muscle fibers (6-8) primarily via the P2X purinergic range of 5-10 Mfl. Calyces and neurons were superfused in the receptor subtype, but there is currently no direct evidence recording chamber for 15-30 min before patching. Only fully localizing these channels on presynaptic nerve terminals. We isolated nerve terminals were used to ensure access of the test have previously demonstrated that it is possible to record substance to the transmitter release face. The currents were whole-cell and single-channel currents from the extensive recorded with an Axopatch-1C amplifier () with 2 kHz calyx-type cholinergic presynaptic nerve terminals of the (-3 db) filter. As successful recordings were infrequent and chicken ciliary ganglion (9, 10). We have used this preparation usually short lived, particularly in the whole-cell configuration, to test for the presence of presynaptic ATP-gated ion channels experimental protocols were designed to minimize the re- by treating the calyces with purinergic agonists and blockers. quired recording duration. Some of these results have been presented in abstract form (11, Data Acquisition and Analysis. Current acquisition and 12). analysis were carried out using the pClamp (Axon) suite of programs. Whole-cell currents were recorded with a sample EXPERIMENTAL PROCEDURES interval of 0.2-1 ms. Peak amplitudes of currents are presented as means + SE. Single-channel currents were recorded with a Preparation of Calyx and . Calyx nerve terminals 0.2-ms sample interval. were isolated as described (13) with minor modifications (10). . The following drugs were used: ATP (disodium salt) Briefly, ciliary ganglia were dissected from 15-day-old chicken (Sigma), adenosine 5'-[ca,3-methylene]triphosphate (a,t4- embryos and were incubated in modified Eagle's medium MeATP), adenosine 5'-[13,y-methylene]triphosphate (1,By- (MEM) containing 0.74 mg of collagenase IV per ml (Worth- MeATP), adenosine 5'-[,B-methylthio]triphosphate (Me- ington, Freehold, NJ), 12.3 mg of dispase per ml (Boehringer SATP), suramin (Research Biochemicals, Natick, MA), and Mannheim), 800 units of hyaluronidase per ml (Worthington), adenosine 5'-[+y-thio]triphosphate (ATP[-yS]) (Boehringer and 0.07 mg of trypsin inhibitor II per ml (Sigma) for 2-3 hr Mannheim). at 37°C in 5% CO,/95% air. The digestion was terminated by removing ganglia from the solution and washing with MEM. The ganglia were triturated in the presence of the vital RESULTS dye 4-(4-diethylaminostyl) N-methyl-pyridinium (4-Di-2-ASP; Characteristics ofWhole-Cell Currents Activated by ATP on Molecular Probes, Eugene, OR) at 80 nM and the cell sus- the Presynaptic Calyx. Pressure ejection of ATP (1 ,tM to 1 pension was allowed to adhere on a glass coverslip recording Abbreviations: a,3-MeATP, adenosine 5'-[a,,3-methylene]triphos- The publication costs of this article were defrayed in part by page charge phate; 3,y-MeATP, adenosine 5'-[3,-y-methylene]triphosphate; Me- payment. This article must therefore be hereby marked "advertisement" in SATP, adenosine 5'-[13-methylthio]triphosphate. accordance with 18 U.S.C. §1734 solely to indicate this fact. *To whom reprint requests should be addressed. 1859 Downloaded by guest on September 25, 2021 1860 Neurobiology: Sun and Stanley Proc. Natl. Acad. Sci. USA 93 (1996) mM) onto calyces held at - 70 mV induced a prominent inward the maximum amplitude of the response as a fraction of that current that decayed gradually in 64 of 123 calyces tested (Fig. evoked by 1 mM ATP to generate a dose-response relation- 1A). Similar success rates have been noted previously for ATP ship (Fig. 1B). The data were fit to the Hill equation over the responses in neurons (15). The minimum delay of the current 1-100 ,M range, giving estimates of the Kd and the Hill in the nerve terminal following a pressure-ejected pulse of coefficient of 15 ,uM and 0.92, respectively. ATP was 60 ms. This latency was comparable to that of a Pharmacological Characterization of the Purinoceptor. A nicotine-induced current in postsynaptic ciliary neurons by the variety of purinergic drugs were applied in order to charac- same technique (data not shown) and was attributed to direct terize the receptor activated by ATP under whole-cell voltage activation of a ligand-gated receptor. clamp. We first tested the possibility that a breakdown product The average current elicited by 100 ,uM ATP in a calyx nerve of ATP, rather than ATP itself, was the active agent. No terminal was 55.3 ± 6.3 pA (from 15.2 to 137 pA; n = 26) in currents were observed with adenosine (n = 6) at a concen- extracellular 140 mM NaCl and equimolar intracellular Cs+ or tration of 100 ,tM. Furthermore, slowly hydrolyzable analogs K+ ion solutions. Rates of activation depended on the precise of ATP, including ATP[,yS] (4/9) among others (see below), flow of onto the target, but the most rapid rate were effective. These results rule out an involvement of P1-type observed had a time constant of 122.6 ms. The ATP current purinoceptors. desensitized markedly, which greatly impeded an examination To confirm that the receptors were within the P2-type group, of its (see below). A second application of ATP we applied the antagonist suramin (18). These experiments in the delay range of 16-35 s evoked only 23.7% ±+ 9.7% (n = were hampered by desensitization of the response after a single 6) of the first trial. Recovery was very slow, beyond the dose of ATP that precluded a typical three-trial protocol: practical maintenance of the whole-cell configuration in the control, drug, and washout. However, it was possible to apply calyx, consistent with previous findings (16, 17). ATP in the presence of suramin and then ATP alone (n = 3). Dose-Response Relationship. An ATP response was seen at We also tested suramin during the time course of a single a threshold concentration of 1 ,uM. We applied a range of ATP application of either ATP or a,13-MeATP (n = 4). A clear concentrations onto a series of nerve terminals and expressed blocking action by suramin was observed with both protocols (Fig. 2). A ATP (10 [LM) P2 receptor types are differentiated primarily on the basis of the potency of ATP agonists. At a constant concentration of K 100 ,tM drug, a,3-MeATP (n = 14), ATP (n = 26), ATP[yS] (n = 4) were approximately equipotent, whereas MeSATP (n = 6) was less effective (Fig. 3) and no response was seen with 3,,y-MeATP (n = 6; P < 0.05). These findings are consistent with a designation of the receptor as P2X type. 0.1 nA Ion Selectivity of the ATP-Activated Current. An exhaustive B 0.5 s examination of channel selectivity was not carried out, but we

1---I- .I tested whether the ATP receptor current was consistent with . a nonselective cation current, as described for other P2x-type U0 receptors (Fig. 4). With symmetrical cation distributions of I' Kj,/Na0u, (n = 3) or Csin/Naout (n = 2), the current reversed in the range of -10 to +20 mV, confirming that the channel was relatively nonselective between cations. The current ex- hibited inward rectification at hyperpolarized potentials, con- sistent with other reports on this receptor type. Rectification

A

I.0

7 -5 -4 -2 eAT e cA r0 ,(I 10". :x,2 - - 2x, 1f-8'V1 C, ' | t, O --- s J c Z! Z~~~~~~S e : Concentration of ATP (M)

FIG. 1. ATP-induced whole-cell current and dose-response curve. B -,"' (A) Current activated by pressure injection of ATP (10 ,uM; horizontal t -5 bar) onto a presynaptic calyx under whole-cell voltage clamp. Bath, standard solution but with 25 mM tetraethylammonium chloride \% 50 substituted for an equimolar quantity of NaCl; patch pipette, Cs ,A solution. (B) Dose-response curve of the ATP current. It was not Su ra m'- possible to carry out full dose-response curves on each nerve terminal 2 se because of desensitization of the ATP response. We therefore exposed each nerve terminal to a single ATP application and expressed the peak current as a fraction of that obtained with the highest dose (I'max) FIG. 2. Block of ATP channel by suramin. (A) Twin-pulse protocol. at 1 mM. Imax. was 66 + 16 pA. Thus, each point represents the mean Two pulses of drugs were applied from separate pressure-ejection + SE of the fractional peak current amplitude during a single trial in pipettes. The first contained a,4-MeATP together with suramin and each of (N) calyces at that concentration. A smooth line was fit the second contained a,f3-MeATP alone (horizontal bars). Suramin according to the Michaelis-Menten equation I = Imax/(l + ([ATP]/ blocked the initial response to a,3-MeATP. (B) Staggered protocol. Kd) where I is the observed ATP-induced current, Imax is the a,4-MeATP was applied (long horizontal bar) from one pipette and maximum value of the current, Kd is the dissociation constant, and n then suramin was pressure ejected together with a,3-MeATP from a is the Hill coefficient. Kd = 15 ,uM. (Inset) Hill plot of ATP-induced second pipette (short horizontal bar). Note the transient block of the currents at concentrations of 1-100 ,uM. The slope gave a Hill agonist-induced current (n = 4). All drugs were applied at 100 JIM and coefficient (n) of 0.92. The membrane potential was held at -70 mV. the nerve terminal was held at -70 mV. Downloaded by guest on September 25, 2021 Neurobiology: Sun and Stanley Proc. Natl. Acad. Sci. USA 93 (1996) 1861

A ATP figuration in order to ensure that current fluctuations could be attributed to the application of the test drug. Furthermore, .N only patches that were electrically silent prior to agonist treatment and that demonstrated a prompt response to ATP treatment were included in the data. ATP-dependent activity `44 was observed in about half, 8/18, of these patches. Single- a,f#-MeATP channel currents were -2 pA at membrane potentials of -120 mV and exhibited bursting behavior with rapid current fluc- tuation during the bursts (Fig. 5). Clear open levels were difficult to discern because of the rapid fluctuations. However, selected traces were used to obtain amplitude measurements 2MeSATP and to estimate the slope conductance, which was -17 pS.

50 sA DISCUSSION The direct recording of presynaptic receptors in situ with 0.5 sec electrophysiological techniques had been precluded by the B 2.0- general small size and inaccessibility of these structures. In this study, we demonstrate the presence of an ATP-activated ion current in a cholinergic presynaptic nerve terminal. This _ 1.5 current reflects the direct activation of ATP-gated ion chan-

-0N nels on the nerve terminals, since the latency after drug *CJ 1.0 application is too short to be attributed to a second messenger E _ pathway (19). Hence, it is possible to record ligand-gated ion 0 ~~~I channels from a nerve z directly presynaptic terminal. 0.5 It has been reported that ATP activates an inward current 26 -1 4 6 ' in many cell types (6, 7, 20). The ATP current recorded in the 0.0 present study was not radically different from these previous ATP a,:- MeATP 2MeSAT0 reports. The ATP receptor was poorly selective between FIG. 3. Effect of purinoceptor agonists. (A) Current responses monovalent cations and rectified markedly at hyperpolarized induced by ATP agonists applied at 100 ,LM (horizontal bar) in nerve potentials. The ATP concentration dependence of the current terminals held at -70 mV. Note the slower activation rate of a,3- was also comparable to other studies. The half-effective ATP MeATP, as described (30). (B) Histogram of peak current amplitudes concentration (EC50) of 15 ,tM is within the published range, normalized to the mean ATP response (55 + 6 pA). Numbers indicate somewhat higher than that in sensory neurons (15, 21) but number of calyces tested, one trial per calyx. Vertical bar is SE. *, P lower than in PC12 cells (22). The Hill coefficient was -1.0, < 0.05; t test compared to ATP. Membrane potential was held at -70 mV. suggesting that the channel is gated by one ATP binding site, a value that is consistent with findings in other reports (7). The was a property of the receptors and not ion distribution, since single-channel conductance of 17 pS was also close to pub- it was still observed in a symmetrical 140 mM Na+ (n = 2) lished values, which range from 13 to 22 pS (23, 24). gradient (data not shown). A number of different receptors are activated by ATP, and ATP-Activated Single-Channel Currents. ATP single chan- these have been classified as P2-type purinoceptors with two nel events were recorded only in the outside-out patch con- major divisions-P2x and P2y. Suramin is a nonselective P2

A 4GmV B -70mV / (pA) - 1 00mV 0_

20.

_. -200o- 3^N 500 pA 1 S

N A...... -80 -40 U 40 cx,.-MeATP Vm ( rV)

FIG. 4. Current-voltage relationship of ATP response. Voltage ramps were applied to nerve terminals in whole-terminal clamp with a 140 mM Ki./Nao.t ion distribution. (A) Top trace, membrane potential protocol. Terminal was held at -70 mV and voltage ramps were applied (from -100 mV to +40 mV, +467 mV/s, and back -78 mV/s). Two different rates were used to rule out capacitative artifacts. Recorded currents before (middle trace) and during (lower trace) application of a,,-MeATP (100 j±M). (B) Control current trace was subtracted from that during drug treatment and was plotted against voltage for both the fast depolarizing ramp (trace a) and the slower hyperpolarizing ramp (trace b). Current reversed at about +20 mV in both cases and exhibited inward rectification. Downloaded by guest on September 25, 2021 1862 Neurobiology: Sun and Stanley Proc. Natl. Acad. Sci. USA 93 (1996) A ATP 100 FM .I

0.5 s

Co a)

.0a) E z

-'-- 0 -2 Amplitude (pA)

B -- "I/ rn 'sf (pA) 0 - C'11o . - C0 r i,'' tt ~~-1 -21 v *. 0 ) 1~ 1 0 c

0

-3

2.0 pAL. -80 -40 0 200 in<; 1 20 Vrm (mV)

FIG. 5. Characteristics of single-channel current fluctuations activated by ATP. (A) (Upper) Single-channel recording from nerve terminal in an outside-out patch configuration at a holding potential of -120 mV. There was no channel activity prior to application of ATP. (Lower) Current sample amplitude distribution after application ATP for 2 s fit by two Gaussian distributions. Difference in peaks gave an open channel amplitude of 2 pA. K+ solution in the pipette and Na+ solution with 5 mM Ca2+ in the bath. (B) Single-channel current-voltage relationship. (Left) Single-channel current traces at different holding potentials. (Right) Channel amplitudes plotted against holding potential; data were pooled from 5 outside-out recordings. Straight line was fit by regresgion analysis (correlation coefficient, 0.726; P < 0.02) and gave a slope conductance of 17 pS.

blocker and was highly effective on the nerve terminal ATP- where specific P2 receptor agonists were used, release was induced current. Of the P2 receptor family, only the P2X type found to be facilitated. ATP enhanced both spontaneous (4) has been demonstrated to be associated with a ligand-gated and evoked (5) transmitter release. cation channel. Purinergic receptors are catagorized primarily The biological function of presynaptic P2x receptors is not on the basis of agonist potency (25). The potency order of P2X known. It has been suggested that they are involved in the agonists is preparation dependent due to the apparent exis- interactions between the nerve terminal and its target during tence of several receptor subtypes, but a recent review suggests development (4). They could also play a similar role in the a general order of ca,3-MeATP f3,y-MeATP > ATP close relationship between the terminal and its surrounding MeSATP (20). Since it was not practical to carry out a full glial cells. Perhaps the most intriguing possibility is that ATP, dose-response analysis on the calyx nerve terminal, we com- co-released with acetylcholine (26), enhances transmitter re- pared the agonists at a single concentration. ATP and a,4- lease via activation of presynaptic P2X autoreceptors. This MeATP were prominent agonists, while MeSATP was less action could be mediated via the influx of calcium ions, which effective, results that are in general agreement with the above are known to be highly permeable through this receptor type potency sequence. Less expected, however, was the lack of (27). Since the presynaptic P2x receptors are activated rela- effect of 3;iy-MeATP, a finding that raises the possibility that tively slowly (time constant, 120 ms), compared, for example, the presynaptic ATP receptor is a distinct P2X variant. to postsynaptic nicotinic receptors, it seems unlikely that Previous studies have examined the effect of ATP on significant recruitment would occur as a result of the ATP transmitter release, but in general these did not distinguish the released during a single nerve impulse. However, peripheral action of ATP itself from the inhibitory effects of its break- (28), including the chicken ciliary ganglion calyx down product adenosine via P1 receptors. In the few cases synapse (29), have been demonstrated to exhibit a long- Downloaded by guest on September 25, 2021 Neurobiology: Sun and Stanley Proc. Natl. Acad. Sci. USA 93 (1996) 1863 duration potentiation during prolonged stimulus trains. The 15. Bean, B. P. (1990) J. Neurosci. 10, 1-10. mechanism of this effect is not known, but does not appear to 16. Hume, R. I. & Honig, M. G. (1986) J. Neurosci. 6, 681-690. involve cholinergic transmission (28, 29). Whether P2x auto- 17. Thomas, S. A. & Hume, R. I. (1990) J. Physiol. (London) 43Q, 373-388. receptors are involved in this phenomenon will require further 18. Dunn, P. M. & Blakeley, A. G. H. (1988) Br. J. Pharmacol. 93, study. 243-245. 19. Inoue, R. & Brading, A. F. (1990) Br. J. Pharmacol. 100, 619-625. 1. Silinsky, E. M., Hunt, J. M., Solsona, C. S. & Hirsh, J. K. (1990) 20. Abbracchio, M. P. & Burnstock, G. (1994) Pharmacol. Therapeut. Ann. N.Y Acad. Sci. 603, 324-334. 64, 445-475. 2. Silinsky, E. M. & Ginsborg, V. (1983) Nature (London) 305, 21. Krishtal, 0. A., Marchenko, S. M. & Pidoplichko, V. I. (1983) 327-328. Neurosci. Lett. 35, 41-45. 3. Bennett, M. R. & Ho, S. (1991)J. Physiol. (London) 440,513-527. 22. Nakazawa, K., Inoue, K., Fujimori, K. & Takanaka, A. (1991) 4. Fu, W. M. & Poo, M. M. (1991) Neuron 6, 837-843. Pflugers Arch. 418, 214-219. 5. Sperlagh, B. & Vizi, E. S. (1991) J. Neurochem. 56, 1466-1470. 23. Silinsky, E. M. & Gerzanich, V. (1993) J. Physiol. (London) 464, 6. Surprenant, A., Buell, G. & North, R. A. (1995) Trends Neurosci. 197-212. 18, 224-229. 24. Nakazawa, K., Inoue, K., Fujimori, K. & Takanaka, A. (1990) 7. Bean, B. P. (1992) Trends Physiol. Sci. 13, 87-90. Neurosci. Lett. 119, 5-8. 8. Dubyak, G. R. & El-Moatassim, C. (1993) Am. J. Physiol. 265, 25. Burnstock, G. & Kennedy, C. (1985) Gen. Pharmacol. 16, 433- C577-C606. 440. 9. Stanley, E. F. (1989) Brain Res. 505, 341-345. 26. Zimmerman, H. (1994) Trends Neurosci. 17, 420-426. 10. Stanley, E. F. (1991) Neuron 7, 585-591. 27. Edwards, F. A. (1994) Curr. Opin. Neurobiol. 4, 347-352. 11. Sun, X. P. & Stanley, E. F. (1993) Soc. Neurosci. Abstr. 19, 901. 28. Dolphin, A. C. (1985) Trends Neurosci. 8, 376-378. 12. Sun, X. P. & Stanley, E. F. (1994) Soc. Neurosci. Abstr. 20, 1120. 29. Scott, T. R. D. & Bennett, M. R. (1993) Br. J. Pharmacol. 110, 13. Stanley, E. & Goping, G. (1991) J. Neurosci. 11, 985-993. 461-469. 14. Hamill, 0. P., Marty, A., Neher, E., Sakmann, B. & Sigworth, 30. Khakh, B. S., Surprenant, A. & Humphrey, P. P. A. (1995) Br. J. F. J. (1981) Pflugers Arch. 391, 85-100. Pharmacol. 115, 177-185. Downloaded by guest on September 25, 2021