Application of 3-isobutyl-1-methylxan- thine (IBMX), a drug that antagnizes the effects ofendogenous adenosine, produces an increase in synaptic strength. This.
Proc. Nati. Acad. Sci. USA Vol. 89, pp. 8586-8590, September 1992 Neurobiology
Adenosine decreases neurotransmitter release at central synapses DAVID A. PRINCE* AND CHARLES F. STEVENS The Salk Institute, 10010 North Torrey Pines Road, La Jolla, CA 92037
Contributed by Charles F. Stevens, May 14, 1992
sion, as has already been done for the neuromuscular junction (12-15). Our approach has been to measure the amplitude of evoked excitatory postsynaptic currents (epscs) and miniature epscs (mepscs) in various concentrations of adenosine from 5 to 100 juM. Further, we have used 3-isobutyl1-methylxanthine (IBMX) as an "anti-adenosine"; these experiments are based on an earlier analysis of IBMX effects in hippocampal slices (16). The postsynaptic contribution of the adenosine effect should be revealed by a change in the size of mepscs; the extent to which' epscs are affected more than mepscs should define the presynaptic component of the adenosine-induced reduction in synaptic transmission. We find that, under the conditions of our experiments, adenosine's action is predominantly presynaptic. In some experiments we simultaneously observed excitatory and inhibitory synaptic currents (ipscs), and we find, in confirmation of the recent reports (17, 18), that the ipscs are unaffected by concentrations of adenosine that cause a marked reduction of the epscs.
Adenosine, at concentrations ranging from 5 ABSTRACT to 100 AM, decreases the efficacy of transm at the perforant path synapses on dentate granule cells. We have used whole cell recording from these cells in slices to determine the mechanism of the reduced synaptic strength. We find that size of miniature excitatory postsynaptic currents (mepscs) is unaffected by adenosine at concentrations up to 100 pM, an observation that indicates adenosine's mode of action is not through a decreased postsynaptic sensitivity to neurotransmitter. A quantal analysis indicates, however, that the quantity of neurotransmitter released is sufficiently diminished by adenosine to account entirely for the adenosine-produced decrease in synaptic strength. Application of 3-isobutyl-1-methylxanthine (IBMX), a drug that antagnizes the effects of endogenous adenosine, produces an increase in synaptic strength. This observation suggests that the resting level of adenosine in our slices is appreciable, and an analysis of the adenosine doseresponse relation is consistent with endogenous adenosine levels of about 10 AM. IBMX application produces only slight chanes in the amplitude of mepscs, whereas a quantal analysis demonstrates that the drug sgnificantly increases the amount of neurotransmitter released. Thus IBMX acts as an "antiadenosine" in our experiments. In some experiments we have been able to record excitatory and inhibitory synaptic currents produced by the same perforant path stimulus. In these instances we find that inhibitory t ssion is unaffected by concentrations of adenosine that produce a marked decrease in the strength of excitatory synapses.
MATERIALS AND METHODS Eighteen Long-Evans or Sprague-Dawley rats, ages 5-22 days postnatal, were anesthetized with pentobarbital (50 mg/kg i.p.) and decapitated; brains were removed and placed in ice-cold modified Ringer solution in which 252 mM sucrose was substituted for the 126 mM NaCl in the standard slice solution (19). Blocks containing bilateral hippocampi were trimmed and glued to the stage of a vibratome (Technical Products International, St. Louis). Coronal or horizontal slices (400 jam) were cut in the sucrose Ringer solution, trimmed, and transferred to an incubation chamber or to an interface-type recording chamber where they were superfused with Ringer solution consisting of (mM) NaHCO3, 26; NaCl, 126; KC1, 3; NaH2PO4, 1.25; CaCk2, 2; MgSO4, 2; dextrose, 10. The superfusion solution had a pH of 7.4 when maintained at 34-35°C and gased with 95% 02/5% CO2. A bipolar tungsten stimulating electrode was placed in stratum molecular adjacent to the recording site in the superior leafof the dentate gyrus. Patch electrodes were pulled from 1.5-mm borosilicate glass (KG33; Garner Glass, Claremont, CA) and had resistances of 3-5 Mfl when filled with a solution containing (mM) cesium gluconate, 130; MgCl2, 2; CaCl2, 1; CsCl, 5; Hepes, 10; EGTA, 11; pH 7.2. In most experiments the intracellular solution also contained QX314 (10 mM), kindly supplied by Astra Pharmaceutical, Worcester, MA. Intracellular solutions had osmolarities ranging from 175 to 190 mosmol. In various experiments adenosine (5-100 1AM), tetrodotoxin (TTX) (1 ItM), IBMX (5-100 ulM), and theophylline (275 ,uM) were applied by bath profusion in standard slice solution. All drugs were obtained from Sigma. The whole cell configuration of the patch clamp technique (20) was used to obtain recordings from dentate gyrus granule
Synaptic strength, the efficacy with which a synapse transmits information, plays a crucial role in determining what computations a neural circuit will perform. Not surprisingly, the brain has a variety of ways at its disposal for dynamically adjusting synaptic strength, one of the most potent of which involves adenosine (1, 2). Adenosine acts on at least two, but probably more, types of receptors (3) and produces a variety of effects, including inhibition (through Al receptors) and activation (A2 receptors) of adenylate cyclase (4) and the activation of potassium channels (5) and inhibition of calcium channels (6), probably through the direct action of G proteins. The potent inhibitory effects of adenosine on normal synaptic transmission and pathophysiological states such as epileptogeneses where excitatory synaptic events are enhanced could well involve one or several of these actions (7). For example, presynaptic activation of a potassium current and/or inhibition of a calcium current could decrease transmitter release, and inhibition of the adenylate cyclase, through the Al receptor, could decrease the sensitivity of the postsynaptic membrane (8-11). These possibilities are, of course, neither exclusive nor exhaustive. The goal of the work reported here was to determine to what extent pre- and postsynaptic effects account for the adenosine blockade of central excitatory synaptic transmis-
Abbreviations: epsc, excitatory postsynaptic current; ipsc, inhibitory postsynaptic current; mepsc, miniature excitatory postsynaptic current; TTX, tetrodotoxin; IBMX, 3-isobutyl-1-methylxanthine. *Present address: Department of Neurology, Room M016, Stanford University Medical Center, Stanford, CA 94305-5300.
The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. §1734 solely to indicate this fact.
8586
Neurobiology: Prince and Stevens cells in stratum granulosum using an Axoclamp 2A amplifier (Axon Instruments, Burlingame, CA) in the continuous single electrode-voltage clamp mode. Neurons had an initial membrane potential (Vm) of -60 to -70 mV and input resistances of 175-300 Mfl. Electrode series resistance was estimated from responses to small hyperpolarizing voltage commands and data from neurons in which significant increases in access resistance occurred during the course of the recording were discarded. A superfusion system with low dead space and a flow rate of 1.5 ml/min allowed multiple solution changes during the study of a cell. Recorded currents were filtered at 1 or 3 kHz and all data were recorded on videotape (Neurodata DR484) for subsequent analysis. Once a stable Gf1 seal and whole cell recording were obtained, the intensity of extracellular stimulation was adjusted so that occasional stimuli failed to evoke epscs when delivered at rates of 1 or 0.5 Hz.
RESULTS Basic Observations on the Adenosine Effect. Our conclusions are based on the analysis of data from 30 dentate granule cells in 20 slices. The electrical properties of the neurons remained stable throughout the recording session, which ranged in duration from 30 to 90 min. Typical inhibitory effects of adenosine are illustrated in Fig. 1, where the average response to the application of 0, 50, and 100 gM adenosine is shown. The peak amplitude of the epscs was reversibly reduced to 51% of the control in 50 ItM and to 20% in 100 ,uM adenosine, without significant changes in the steady-state current at the holding potential of -60 mV. Is enough endogenous adenosine present in our slices to be tonically active (21)? To examine this question we used IBMX, a drug that is supposed, among other actions, to block adenosine receptors without activating them (22, 23). If resting adenosine concentration is sufficiently high to decrease synaptic transmission, this blocker should antagonize the tonic effect of endogenous adenosine and produce an increase in the synaptic strength. We find that IBMX at concentrations of 50-100 ,uM causes the peak synaptic current to increase an average of 1.8 times control values (range, 1.7-1.9 for four cells). A typical effect of IBMX is illustrated in Fig. 2, where 50 gM drug produced a 1.8 times increase in the peak current. Apparently 50 ,uM is a saturating dose because concentrations of 75 and 100 ,uM exhibited no larger effects than did 50 gM. In a single experiment we found that 375 ,uM theopholine increased peak synaptic currents 3.4 times. In so far as the entire effect of the methylxanthines is 50
Proc. Natl. Acad. Sci. USA 89 (1992)
8587
100
50
[
a
Time
(ms)
-4 c m L L
]3
C
-50
-100
FIG. 2. Superimposed traces of average synaptic current as a function of time for control bathing solution (A) and 50 ,uM IBMX (B). Holding potential = -60 mV.
to block adenosine receptors, we conclude that the resting concentration of adenosine in our slices is sufficient to decrease synaptic currents to a little more than half of their maximum value. We have used six cells to determine the adenosine doseresponse relation over a range of concentrations from 5 to 100 ,uM. The decrement produced by the adenosine is illustrated in Fig. 3, where the peak amplitude of synaptic current, expressed as a fraction of the control peak current (no adenosine added to the bath), is plotted as a function of adenosine concentration. Since IBMX can increase synaptic currents by a factor of about 1.8, we assume that the resting levels of adenosine in the slice are adequate to reduce the peak current to 56% of its maximum value, and the amplitude of the control currents (with no exogenously added adenosine) is taken to be this value. The smooth curve is a least squares fit to the simple binding equationflc) = 1/[1 + (co + c)/K], whereftc) is the peak synaptic current (relative to its maximum value), co is the resting adenosine concentration in the slice, c is the adenosine concentration in the bathing medium, and K is the apparent dissociation constant for the adenosine receptors mediating this effect. The resting endogenous adenosine concentration is co = 11.1 uM and the dissociation constant is 14.5 ,uM for the smooth curve in the figure. We conclude, then, that our slices have a resting adenosine concentration of =10 ,AM and that the adenosine receptors that control this depressive effect of adenosine can be characterized by a dissociation constant of -15 ,uM. Adenosine Acts Presynapticafly. We turn now to the question of how adenosine and IBMX act. After the effects of 1 a)
Q
N
Time
(ms)
a)
L L-
co 0)
0 0L
-10
CD Qr a)
Ci -30
0
o
100
0
Adenosine concentration (PM) FIG. 1. Superimposed traces of average synaptic current (pA) as (ms) for three concentrations of adenosine: 100 AM (A), 50 ,M (B), and 0 AM (C). The holding potential for the cell was -60 mV.
a function of time
FIG. 3. Dose-response curve relating peak synaptic current (response size) as a function of adenosine concentration (,uM) in the bathing solution. The smooth curve is fitted according to the equation given in the text.
Proc. Natl. Acad. Sci. USA 89 (1992)
Neurobiology: Prince and Stevens
8588
adenosine on epscs were examined in the cell illustrated in Fig. 1, TTX (1 ,uM) was added to the bath and the average peak amplitude of mepscs was determined. Although G protein-coupled postsynaptic increases in K conductance probably would be blocked under our experimental conditions (Cs in the recording pipette), one might still propose that the adenosine acts on postsynaptic receptors to decrease the response to glutamate. For example, Al receptors might inhibit adenylate cyclase and thus decrease the level of phosphorylation of nonreceptors for N-methyl-D-aspartate, thereby decreasing the response of the postsynaptic membrane to glutamate. Were this the case, then the size of mepscs should decrease; if the only action of adenosine were to diminish postsynaptic glutamate sensitivity, then the decrease of mepsc amplitude should equal the change in the peak evoked synaptic currents we have described above. As illustrated in Fig. 4, mepsc amplitude was unchanged even at a 100 ,uM adenosine concentration. In five other experiments, we have examined the effects of adenosine, at concentrations ranging from 20 to 100 ,uM, on mepsc amplitude. We find that, if anything, the amplitude was slightly increased, but certainly was not decreased to the same extent as synaptic strength. The average ratio of mepsc aptitude in adenosine compared to control was 1.04, with a range of 0.79-1.09. No dependence of the effect on epsc size with dose was apparent; specifically, the extremes in the range of effects (0.79 and 1.09) occurred with the lowest adenosine concentration (20 uM). The adenosine-produced decrease in synaptic strength thus appears, in these experiments, to be purely presynaptic. To confirm this conclusion, and to determine what presynaptic factor is responsible for the adenosine effect, we have carried out a quantal analysis on the distribution of evoked epsc sizes. As illustrated in Fig. 5, the standard formalism (25), modified to take account of the observed rather large event-to-event variation in mepsc amplitude (24, 26), accounts quantitatively for the observed fluctuations in epsc amplitude for adenosine concentrations up to 100 AM. According to this analysis, the mean mepsc amplitude was unaffected by the presence of adenosine (as we also observed directly), but the probability of release decreased dramatically. Thus, a reduction in the release probability can account for the entire adenosine effect. The usual notion is that low doses of methylxanthines act to block adenosine receptors, and we have followed this supposition in our analysis of the adenosine dose-response curve illustrated in Fig. 3. If this hypothesis is correct, then the mepsc amplitude should be unaffected by IBMX at concentrations that have a pronounced effect on the mean 10
4
Time
(ms)
c L L
-5
[
-10
FIG. 4. Superimposed average mepscs for 0 and 100 ,uM adenosine as a function of time. Holding potential = -60 mV. These data, obtained with 1 ,uM TTX in the bathing solution, are from the same cell that provided the traces for Fig. 1.
i
0
1.4
0
-o>
1~~~~~~~~~~~~~~~~~
0
1.4
100,uM adenosine
1111111111111111111111111111111111
I
1.4
0
Response Size (nA) FIG. 5. Cumulative probability of response size as a function of response size (nA) for three adenosine concentrations (0, 50, and 100 ,uM as indicated). The smooth curves are fitted according to a binomial distribution with a mepsc distribution as described by Bekkers et al. (24). For these fits the number of release sites was set at 6; the release probabilities were 0.85 (0 gM adenosine), 0.4 (50 jzM adenosine), and 0.14 (100 gM adenosine). The average mepsc amplitude was 0.133 nA throughout. These data are from the cell that provided traces for Figs. 1 and 4.
amplitude of evoked release, but a quantal analysis should reveal an increase in the release probability. The effects of IBMX on mepsc amplitude are difficult to study because this drug appears to decrease the frequency of spontaneous releases. For two cells from which an adequate sample of mepscs could be obtained after IBMX application, 50 ,M drug resulted in mepsc sizes of 1.18 and 0.75 (average, 0.97) as compared to control; this concentration of IBMX produced an 80% increase in the average size of the evoked synaptic currents. A quantal analysis has been carried out on three neurons treated with IBMX concentrations of 50 and 75 AM. In all three cases the distribution of evoked synaptic current amplitudes after drug application could be predicted from the quantal parameters obtained under control conditions by simply increasing the release probability appropriately. Thus, this methylxanthine appears to act as an antiadenosine. In most experiments, we blocked ipscs with 10 ,M bicuculine, but on some occasions the effects of inhibitory currents were eliminated simply by holding the cell's membrane potential at the ipsc equilibrium potential. In these
Neurobiology: Prince and Stevens
Proc. Natl. Acad. Sci. USA 89 (1992)
same experiments, we also examined the synaptic currents at holding potentials more positive than the inhibitory equilibrium potential so that epscs and ipscs could be seen in the same trace (epsc was inward and ipsc was outward). By making use of the fact that failures in synaptic transmission occurred and that ipscs are of significantly longer duration than epscs, we were able to look at the effects of adenosine on excitatory and inhibitory synaptic transmission separately. Fig. 6 shows that the inhibition of the epsc by adenosine is not accompanied by the simultaneous decrease in inhibitory transmission.
DISCUSSION Although adenosine potentially can alter synaptic strength through a rather large variety of mechanisms, we find that, in our experiments, the adenosine effect is dominated by changes in the amount of transmitter released. We can account quantitatively for this effect in terms of decreased release probabilities, but we cannot exclude the possibility that some fraction of the adenosine-produced change, perhaps all, is due to a decrease in the number of release sites (14). The same mechanism appears to operate when the effect of resting adenosine levels is antagonized with IBMX. If our in vitro observations can be extrapolated to the in vivo case, we would suggest that the normal modulation of synaptic strength by adenosine occurs through modifications in the amount of neurotransmitter release. Because the regulation of adenosine levels in the slice might well differ from that in 25
cL
L
(ms)
Time
C) -15
-25 25
15
4-)
-
5
c
C)
20
L
L ]
-5 -
E
40
80
60
Time
100
(ms)
FIG. 6. Superimposed traces of average evoked sy'naptic current function of time. The inward current (E) is fr om excitatory synapses and the outward current (I) is from inhibittory synapses. (Upper) Traces taken before adding adenosine and alfter washing it away. (Lower) Trace from cell in 20 gM adenosine. T]his record was taken between the two superimposed records in Upp)er. as a
8589
the intact brain, or because interplay of other neuromodulatory mechanisms might differ between slices and brain, our tentative conclusion requires direct experimental verification in vivo.
Although we have detected no postsynaptic effects of adenosine, such effects could well be present in other circumstances. We studied each neuron for at least 30 min, and usually longer; during this time considerable exchange between the pipette solution in the cytoplasm can occur, and our pipettes contained neither GTP nor ATP. Thus, GTP concentrations could have been decreased in the neurons we studied so that postsynaptic adenosine responses that depend on GTP might have been diminished or eliminated. Because the decrease in synaptic strength that we report is comparable to that seen earlier in situations where the postsynaptic adenosine response should have been intact (1, 2), we believe that the presynaptic effect we have studied is the primary one.
Although we have interpreted the effects of IBMX in terms of the adenosine receptor blocking action of this drug, IBMX of course also blocks phosphodiesterase at the concentrations we have used (22, 23). We cannot exclude the possibility that the inhibition of phosphodiesterase accounts for the observed effects of IBMX, but we do not favor this interpretation. The IBMX effect appears to be saturated already at 50 AM, whereas the inhibition of phosphodiesterase should not be complete at this concentration. Whatever the mechanism of IBMX action in the context of our experiments, it is just opposite to that of adenosine. We have used a simple binding equation to describe the adenosine dose-response relation. This is only a first-order theory, however, and-although the equation characterizes our data satisfactorily-its physical interpretation must be viewed with great caution. We do not know the number of binding sites for adenosine on its receptor (the theory assumes a single binding site), but, more importantly, we have no information on the functional form relating adenosine receptor occupancy to decrease in synaptic strength (the theory assumes it is linear). Nevertheless, the equation does provide a useful description of the dose-response relationship and provides the quantitative basis for characterizing the adenosine effect in vivo. We find the differential regulation of excitatory and inhibitory transmission by adenosine an interesting and significant phenomenon (17, 18). It suggests that the brain uses some global mechanism to maintain the balance between excitation and inhibition. Such a mechanism could have not only computational but also therapeutic implications. Even if adenosine or adenosine analogs might be unsuitable for the treatment of, say, seizure states, other agents that tap into the same postulated global regulatory mechanisms might provide a path for therapeutic intervention. 1. Dunwiddie, T. V. (1985) Int. Rev. Neurobiol. 27, 63-139. 2. Greene, R. W. & Haas, H. L. (1991) Prog. Neurobiol. 36, 329-341. 3. Stiles, G. L. (1992) J. Biol. Chem. 267, 6451-6454. 4. Van Calker, D., Muller, M. & Hamprecht, B. (1979) J. Neurochem. 33, 999-1005. 5. Trussell, L. 0. & Jackson, M. B. (1985) Proc. Natl. Acad. Sci. USA 82, 4857-4861. 6. MacDonald, R. L., Skerritt, J. H. & Werz, M. A. (1986) J. Physiol. (London) 370, 75-90. 7. Dunwiddie, T. V. & Worth, T. (1982) J. Pharmacol. Exp. Ther. 220, 70-76. 8. Knapp, A. G. & Dowling, J. E. (1987) Nature (London) 325, 437-439. 9. Knapp, A. G., Schmidt, K. F. & Dowling, J. E. (1990) Proc. Nad. Acad. Sci. USA 87, 767-771. 10. Wang, L. Y., Salter, M. W. & MacDonald, J. F. (1991) Science 253, 1132-1135.
8590
Neurobiology:
Prince
and Stevens
11. Greengard, P., Jen, J., Nairn, A. C. & Stevens, C. F. (1991) Science 253, 1135-1138. 12. Ginsborg, B. L. & Hirst, G. D. S. (1972) J. Physiol. (London) 224, 629-645. 13. Ribeiro, J. A. & Walker, J. (1975) Br. J. Pharmacol. 54, 213-218. 14. Branisteanu, D. D., Haulica, I. D., Proca, B. & Nhue, B. G. (1979) Arch. Pharmacol. 306, 273-279. 15. Silinsky, E. M. (1984) J. Physiol. (London) 346, 243-256. 16. Chavez-Noriega, L. E. & Stevens, C. F. (1992) Brain Res. 574, 85-92. 17. Yoon, K.-W. & Rothman, S. M. (1991) J. Neurosci. 11, 13751380. 18. Lambert, N. A. & Teyler, T. J. (1991) Neurosci. Lett. 122, 50-52.
Proc. Nadl. Acad. Sci. USA 89 (1992) 19. Aghajanian, G. K. & Rasmussen, K. (1989) Synapse 3, 331338. 20. Hamill, 0. P., Marty, A., Neher, E., Sakmann, B. & Sigworth, F. J. (1981) Pfluegers Arch. 391, 85-100. 21. Wu, P. H. & Phillis, J. W. (1984) Neurochem. Int. 6, 613-632. 22. Smellie, F. W., Davis, C. W., Daly, J. W. & Wells, J. N. (1979) Life Sci. 24, 2475-2482. 23. Phillis, J. W. & Wu, P. H. (1981) Prog. Neurobiol. 16,187-239. 24. Bekkers, J. M., Richerson, G. B. & Stevens, C. F. (1990) Proc. Nati. Acad. Sci. USA 87, 5359-5362. 25. del Castillo, J. & Katz, B. (1954) J. Physiol. (London) 124, 574-585. 26. Bekkers, J. M. & Stevens, C. F. (1989) Nature (London) 341, 230-233.
