v!.2ntri to c.v.2muscarine was proportional to the change in mean

Transcription

1 Journal of Physiology (1996), 491.2, pp Presynaptic muscarinic inhibition in bullfrog sympathethic ganglia Wei-Xing Shen and John P. Horn Department of Neurobiology, University of Pittsburgh School of Medicine, E144 Biomedical Science Tower, Pittsburgh, PA 15261, USA 1. Muscarinic modulation of nicotinic transmission was studied in bullfrog sympathetic ganglia by recording synaptic currents from B and C neurones. 2. Bath-applied muscarine reduced the amplitude of EPSCs recorded at < -2 Hz from B neurones by up to 57 %. The action was reversible, showed no apparent desensitization, and had an EC5 of 12 nm. Muscarine had no effect on EPSCs in C neurones. 3. Currents evoked by ionophoretic application of ACh to B neurones were unchanged by muscarine. Muscarine increased the coefficient of variation (c.v.) of EPSC amplitude. The effect upon the ratio of c v!.2ntri to c.v.2muscarine was proportional to the change in mean EPSC amplitude. 4. Activation of muscarinic receptors by ACh from nerve terminals was observed by comparing trains of EPSCs in normal Ringer solution and atropine. Inhibition of EPSC amplitude by 15-4% was seen as frequency was increased from 1 to 5 Hz. The minimal latency for onset of inhibition was -2 s. Stimulation at 2 Hz did not produce inhibition. 5. The results indicate that presynaptic muscarinic receptors are selectively expressed by a functional subelass of preganglionic sympathetic nerve terminals. Physiological activation of the receptors occurs during repetitive activity. The extent of autoreceptor-mediated inhibition varies as a biphasic function of stimulus frequency. Inhibitory muscarinic autoreceptors appear to modulate ACh release in some, but not all, autonomic ganglia. The most consistent description of this mechanism comes from studies of enteric ganglia using two approaches. In collection experiments upon guinea-pig ileum, the overflow of ACh evoked by nerve stimulation was decreased by oxotremorine and enhanced by atropine (Kilbinger, 1977; Sawynok & Jhamandas, 1977). Interpretation of these results in terms of a presynaptic mechanism is supported by electrophysiological experiments. Muscarinic agonists inhibit nicotinic EPSPs in myenteric and submucosal neurones without altering postsynaptic nicotinic sensitivity to ACh (Morita, North & Tokimasa, 1982; North, Slack & Surprenant, 1985). Presynaptic muscarinic receptors are also activated by ACh that is released during repetitive stimulation of nerves. Similar effects have been reported in enteric neurones of the large intestine (Tamura & Wood, 1989) and colon (Frieling, Cooke & Wood, 1991), but not the antrum (Tack & Wood, 1992). This raises the possibility that muscarinic autoreceptors are selectively expressed by functional subsets of autonomic neurones. The role of muscarinic autoreceptors in ganglia outside the enteric system is controversial. ACh collection experiments in the cat superior cervical ganglion (SCG) (Kato, Collier, Ilson & Wright, 1975) and adrenal gland (Collier, Johnson, Kirpekar & Prat, 1984) suggest muscarinic receptors are not present on preganglionic sympathetic terminals. A similar negative conclusion was reached concerning the parasympathetic pelvic ganglion in the rat (Somogyi & DeGroat, 1993). On the other hand, evidence in favour of presynaptic muscarinic receptors in sympathetic ganglia has been obtained in collection experiments on the guineapig SCG (Capuzzo, Borasio & Fabbri, 1988, 1989) and cat stellate ganglion (Dujic, Roerig, Schedewie, Kampine & Bosnjak, 199). These contradictory conclusions cannot be explained simply by differences in species or preparations because recent work has provided positive evidence for presynaptic muscarinic inhibition in the cat SCG (Bachoo & Polosa, 1992). However, Bachoo & Polosa (1992) concluded that the receptors were not physiological, based on recordings of postganglionic compound action potentials which indicated the receptors were only activated when inhibitors of acetylcholinesterase were present. This is important because ACh collection experiments are often done in the presence of esterase inhibitors. We became interested in presynaptic receptors after finding that repetitive stimulation of bullfrog sympathetic ganglia at 5 Hz causes a strong use-dependent depression of nicotinic EPSCs in B neurones, but not in C neurones (Shen & Horn, 1995a). The effect is due to a reduction in ACh

2 414 W-X. Shen and J P Horn J Physiol release and is not mediated by presynaptic nicotinic receptors. Involvement of muscarinic autoreceptors is suggested by the observation that muscarinic agonists inhibit nicotinic EPSPs and increase their coefficient of variation (c.v.) when recorded from B neurones under conditions of reduced quantal content (Koketsu & Yamada, 1982). However, another study done under normal ionic conditions concluded that atropine did not modulate nicotinic transmission in B neurones during low frequency stimulation (Connor, Levy & Parsons, 1983). The goals of the present experiments were to determine whether presynaptic muscarinic receptors in a sympathetic ganglion can: (i) inhibit ACh release under conditions of normal quantal content; (ii) be activated by the physiological release of ACh; and (iii) account for the difference in ACh release by preganglionic B and C terminals. An abstract of the results has appeared (Shen & Horn, 1995 b). METHODS Intracellular recording The large (12-18 cm) male and female bullfrogs (Rana catesbeiana; Charles D. Sullivan, Co., Nashville, TN, USA) used in these experiments were killed by rapid decapitation followed by double pithing. Paravertebral sympathetic ganglia 7-1 were dissected and prepared for recording, under two-electrode voltage clamp, of nicotinic EPSCs and currents evoked by ionophoresis of ACh (Shen & Horn, 1995a). Sympathetic B and C neurones were identified by innervation patterns and axonal conduction velocities, and were stimulated through separate preganglionic electrodes (Dodd & Horn, 1983a). All experiments were done at room temperature (21-24 C). Grouped data are expressed as the mean + S.D. Dose-response curves Muscarine was bath-applied at 3-4 ml min-' to construct dose-response curves, and was washed out between each dose. For each response, 2-4 EPSCs were collected at Hz and averaged. Depression of EPSC amplitude (Y) was defined as (control - muscarine) (control)-'. The EC5 for muscarine was estimated by fitting the mean response at each dose to a logistic function (eqn (1)) (DeLean, Munson & Rodbard, 1978), where YO is the response in the absence of muscarine, Y. is the maximal response to muscarine, B is the slope factor and C is the concentration of muscarine. The data were fitted to eqn (1) using Igor (Wavemetrics, Lake Oswego, OR, USA), a computer program which employs the Levenberg-Marquardt algorithm in an iterative procedure that searches for parameter values that minimize x2. During the fitting procedure, the value of YO was held at zero. Y= (YO Yj) - (1 + [(C) (EC5)-l]B} + YO. (1) Analysis of fluctuations in EPSC amplitude The origin of muscarinic inhibition was analysed by measuring the c.v. of evoked EPSCs. Trains of EPSCs were recorded at rates of Hz in normal Ringer solution and in muscarine (3 nm-1 /SM). Under steady-state conditions where EPSC amplitude was stationary, the mean amplitude (x), standard deviation (S.D.) and c.v. (eqn (2)) were calculated from samples of 3-2 EPSCs. C.V. = (S.D.) (x (2. (2) At the neuromuscular junction, c.v. is inversely proportional to the quantal content (m) of evoked ACh release (del Castillo & Katz, 1954). Under conditions of low quantal content (m < 3), release obeys Poisson statistics and one can estimate m directly from c.v. using the method of variance; m = (c.v.)-2. For higher quantal contents, release is binomially distributed and the method of variance leads to overestimates of m. Previous studies in frog ganglia indicate that release is quantal and that m > 1 in normal Ringer solution (Blackman, Ginsborg & Ray, 1963; Connor et al. 1983). Release is therefore more likely to behave as a binomial rather than a Poisson process. Nonetheless, the method of variance is sufficiently sensitive to detect the decrease in ACh release produced by shifting B neurones from normal Ringer solution to low Ca2+, high Mg2+ Ringer solution (Connor et al. 1983). The EPSC data were used to construct a plot of r v8. f as introduced by Bekkers & Stevens (199) for studying long-term potentiation. Adopting their analysis, f is defined as a synaptic gain factor which represents potentiation when > 1 and depression when < 1 (eqn (3)), and r is defined as the fractional change in c.v. (eqn 4). f = (rmuarine) (Xcontrol)1X r = (c.v.!control) (C.V!muscarine)Y. (3) (4) If one assumes that ACh release is binomially distributed, then x = am = anp, where a is the average quantal size, N is the number of release sites, and p is the probability of release, and v = a2np(1- p) + a(npcm2, where v is the variance of the evoked EPSC, and cm is the coefficient of variation of the spontaneous miniature EPSC. Substituting these equations into eqn (2) shows that c.vx = (1 + Cm2 - p)(npf1 which permits one to derive the following analysis (Bekkers & Stevens, 199). Muscarinic inhibition could arise from decreases in a, N or p. In principle, the value of a could be decreased by lowering the amount of ACh that is released in a quantum or by reducing the postsynaptic sensitivity of nicotinic receptors. In either event, the value of r is independent of changes in the parameter a and is equal to 1. Alternatively, inhibition arising from a decrease in N or p makes different predictions because of the way they enter into the equation for variance. If inhibition results from an f-fold decrease in N, then the model predicts r = f and a plot of the data should lie on the diagonal. If inhibition results from an f-fold decrease in p, then the model predicts r < f for values of f < 1 and a plot of the data should fall to the right of the diagonal. In the limiting case of a Poisson process where p-*, if inhibition is produced by reduction in either N or p, then the model predicts the data will lie on the diagonal. Analysis of trains of EPSCs Muscarinic modulation of synaptic currents by endogenous ACh was assessed by recording trains of EPSCs in control Ringer solution and atropine. Amplitudes were normalized to the first EPSC in the train with the difference between amplitudes in the presence and absence of atropine taken as the measure of muscarinic inhibition. Solutions and drugs The Ringer solution contained (mm): 115 NaCl, 2 KCl, 1P8 CaCl2, 4 Na-Hepes (ph ). Acetylcholine chloride, atropine sulphate and (±) muscarine chloride were obtained from Sigma.

3 J Physiol Presynaptic musc. zrinic inhibition 415 RESULTS Muscarinic inhibition of nicotinic EPSCs is specific for B neurones Bullfrog sympathetic neurones are each innervated by a primary preganglionic fibre that produces a strong suprathreshold nicotinic EPSP, and in addition, some receive secondary subthreshold inputs (Dodd & Horn, 1983a; Ivanoff & Smith, 1995). Only primary synaptic inputs to B and C neurones were included in this study. Synaptic currents were recorded at frequencies < -2 Hz in order to maintain constant the average EPSC amplitude in normal Ringer solution. At frequencies > 5 Hz, EPSCs in B and C neurones undergo transient changes in amplitude that vary with frequency and cell type (Shen & Horn, 1995a). Figure 1A and B illustrates the effect of bath-applied 3 nm muscarine on a B neurone clamped at a membrane potential (Vm) of -5 mv. It produced a slow inward shift in holding current (2-2 na) and a 3% reduction in EPSC amplitude. Both effects reached steady state in < 3 min and reversed upon washing. The slow inward current produced by muscarine arises primarily from the postsynaptic suppression of M-current (Adams, Brown & Constanti, 1982). Shifting the holding potential to -8 mv deactivates M-current and eliminates the slow current. Hyperpolarization of three B cells had no effect upon inhibition of EPSCs. Thus the effects of muscarine upon holding current and EPSC amplitude are independent. Inhibition of the nicotinic EPSC by muscarine (3 nm to 1O /LM) was observed in twenty-six out of twenty-six B neurones. In two B cells exposed for 2 min to muscarine (3-1,uM), synaptic currents remained depressed and the effect did not desensitize. Muscarine did not alter the time constant (T) for exponential decay of the EPSC. In ten B cells at Vm = -5 mv, = 5X ms in normal Ringer solution and r = P41 ms in 3 nm muscarine (P = -32, paired t test). In contrast to its consistent action on B neurones, muscarine ( 3-3,uM) had no effect upon nicotinic EPSCs in four C neurones. EPSC amplitude was 11P6 + 1P9 na in control Ringer solution and na in 1 usm muscarine (Vm = -5 mv; P = 54, paired t test). Figure 1C and D illustrates EPSCs and a plot of their amplitudes in a C cell that was stimulated at -1 Hz and exposed to 1 /UM muscarine. A 3 nm muscarine B :3 a) 15- CP III I I. I -. Iİ -,EitI,, '1- l. - j % ~~~I _I % % I ^ % 2 mqin. % %, I --' r- 2 Imsla 1 na - co C 1 UM muscarine D : :3 - I % I I % I In - * * 8 ms 1 na E a CL, O Figure 1. Effect of muscarine on EPSCs in a B neurone (A and B) and a C neurone (Cand D) In A, EPSCs were recorded at -17 Hz from a B neurone at -5 mv (upper trace). Bath application of muscarine (horizontal bar) produced an inward shift in holding current and a reduction in EPSC amplitude. Means of 1 EPSCs during the control, muscarine and recovery periods are shown in the lower traces. Stimulus artifacts are marked (*). The plot in B shows the time course of changes in EPSC amplitude. Panels C and D are from a similar experiment on a C neurone stimulated at -1 Hz (Vm = -5 mv).

4 416 W-X. Shen a'nd J P Horn J Physiol T J6 C.) c) a- w. CL o (I) Figure 2. Dose-response relation for muscarinic inhibition of the EPSC in B neurones The effect upon EPSC amplitude is expressed as (control - muscarine) (control)-'. Each point is the mean + S.E.M. of data from 4-8 cells. The line was fitted to the means with a logistic equation. - - a. soul a I v w a Vogl I a I v,1 4...I. N [Muscarine] (jim) 1 The maximal effect of muscarine upon EPSC amplitude in B cells was determined by constructing a dose-response relation (Fig. 2). Eight B neurones were stimulated at low rates and systematically exposed to increasing concentrations of muscarine. The cells were allowed to recover for 6-15 min between each dose. The threshold for inhibition of the EPSC was 1-3 nm. Based on a fit of mean data to a logistic equation, the EC5 was nm and the slope factor (B) was 1P3 + 1 (Fig. 2). Maximal inhibition occurred at 3,UM where EPSC amplitude was depressed by %. Tests for presynaptic inhibition of the EPSC Inhibition of the nicotinic EPSC in B neurones could arise from a reduction in postsynaptic sensitivity to ACh or from a reduction in ACh release by preganglionic terminals. The possibility of reduced transmitter sensitivity was examined by measuring the effect of muscarine on membrane currents induced by focal ionophoresis of ACh. One of four experiments that yielded similar results is illustrated in Fig. 3. Exposure to 1 /UM muscarine for 5 min produced a small inward shift in holding current but did not alter the amplitude of ACh-evoked membrane currents elicited at -2 Hz. In this cell, nicotinic EPSCs were depressed by muscarine in < 2 min and thus the onset of the effect was similar to that in other B neurones that were studied. Therefore, the insensitivity of nicotinic ACh responses to muscarine cannot be explained by inadequate access to the receptors mediating inhibition. If muscarinic inhibition of the fast EPSC is caused by a reduction in the quantal content (m) of ACh release, then the effect should be accompanied by an increase in the coefficient of variation (c.v.) of EPSC amplitude (del Castillo & Katz, 1954; Connor et al. 1983). In fifteen out of fifteen B neurones, exposure to muscarine in concentrations ranging from 3 nm to 1,uM increased the c.v. of the EPSC. A Control 1 jim muscarine Wash B C - 'D = 3- L E co a 2- a s 5 1M muscarine SnAr 4- _ M& Figure 3. Muscarine does not alter the sensitivity of postsynaptic nicotinic receptors to ACh A, exposure of a B neurone for 5 min to muscarine did not change the amplitude of currents evoked by ionophoretic pulses of 1 M ACh (15 na, 13 ms) at 2 Hz. Each trace is the mean of 5 ACh responses (truncated current artifacts are present at the start of each response). Muscarine produced a small inward shift in holding current (dashed line). B, a plot of individual ACh responses during the experiment shown in A

5 J Physiol Presynaptic muscarinic inhibition 417 The effect of 3 UM muscarine upon the distribution of EPSCs in one cell is illustrated by the change in probability density histograms (Fig. 4A). In this experiment, muscarine decreased the mean EPSC amplitude from 14-4 to 5-6 na. The data were well described by Gaussian functions (X2 goodness-of-fit test, P < X25 for control, P < 1 for muscarine). When the Gaussian fits were scaled to the same mean and amplitude (inset in Fig. 4A), it was clear from a comparison of their widths that muscarine increased the c.v. of EPSCs. A similar effect of muscarine upon the distribution of EPSCs was observed in two other neurones in which we were able to collect samples of > 1 EPSCs. When calculated directly from the amplitude data shown in Fig. 4A, c.v. increased from X12 to X21 upon addition of muscarine. Using the method of variance, this corresponds to a decrease in m from 68 to 24. In a group of five neurones, c.v. increased from to after addition of 3,CM muscarine. When estimates of quantal content in normal Ringer solution were made from the variance of EPSCs, they ranged from 68 to 21 (mean = , n = 15). This agrees with a previous report (Connor et al. 1983) and suggests that release under normal conditions is unlikely to obey Poisson statistics. An alternative analysis which does not depend on such an assumption is to plot r, the fractional change in c.v!, vs. f the synaptic gain factor (Bekkers & Stevens, 199). Figure 4B shows that the data were clustered near the diagonal (r = f), thereby indicating that the effect of muscarine was presynaptic. Closer examination suggests the data deviate to the right of the diagonal. Using linear regression, the slope was , which is significantly different from unity (P < 5, paired t test), and the y-intercept was - 1. This provides additional evidence that release under these conditions is not a Poisson process and suggests that the effect of muscarine is unlikely to arise purely from a decrease in N Presynaptic modulation during repetitive stimulation In order to determine whether presynaptic muscarinic receptors function as autoreceptors, trains of forty EPSCs were compared in normal Ringer solution and atropine. The dose of atropine was 3 nm, which is sufficient to block muscarinic receptors without blocking open nicotinic channels in the postsynaptic membrane (Connor et al. 1983). As observed in previous work (Shen & Horn, 1995a) the profile of EPSC amplitudes in normal Ringer solution varied with stimulus frequency (Fig. 5A). At 1 Hz, EPSCs became depressed during the initial part of the train and remained so. At 5 Hz, the pattern was biphasic, with transient facilitation followed by depression. At 2 Hz, EPSC amplitude became facilitated and remained elevated. In atropine (Fig. 5B), the depression at 1 and 5 Hz was lessened. By contrast, atropine had no effect on the 2 Hz train. The time course and magnitude of the effects are more easily seen in plots of grouped data showing the difference between EPSCs in normal Ringer solution and atropine (Fig. 5C). At 1 Hz, EPSC amplitude progressively decremented after a latency of 4 s and reached a plateau where the magnitude of inhibition approached 15%. At Muscarinef\ A mu e -2 3 um muscarine 1\ Control B 1- -._ -15 ) ( -1-5 Control N / / EPSC amplitude 2 [I I I I I ---I f (synaptic gain factor) Figure 4.Muscarine increases the coefficient of variation (c.v.) of the EPSC A, probability density distributions of EPSC amplitude from a B neurone in control Ringer solution (n = 191) and 3/uM muscarine (n = 125). Smooth lines are fits of Gaussian distributions (mean + S.D. = na in control and na in muscarine). In the inset, the two fitted distributions were scaled to the same height and mean to show the increase in c.v. produced by muscarine. B, plot of r vs. f (15 cells, 1 dose of muscarine per experiment). If the mechanism of inhibition was a reduction in quantal size, then the model predicts r= 1 (horizontal dashed line). If inhibition was presynaptic, then the model predicts the data would fall on or to the right of the diagonal (continuous line). The dotted line was drawn by linear regression analysis of the data.

6 418 5 Hz, the latency of inhibition decreased to 2 s and its magnitude increased to -4%. One explanation for the lack of inhibition during 2 Hz trains of 4 EPSCs is that they were too short, given that 2 s may be the minimum latency for activation of presynaptic receptors (Fig. 5C). In two other B neurones, atropine had no effect upon EPSC amplitudes during trains of 12 shocks at 2 Hz. DISCUSSION The results address three issues raised in the introduction. First, they show that activation of presynaptic muscarinic receptors under normal ionic conditions can selectively inhibit the release of ACh by preganglionic sympathetic B neurones. Second, they demonstrate that presynaptic muscarinic receptors can be physiologically activated by ACh that is released during repetitive stimulation. Lastly, they show that presynaptic muscarinic inhibition provides an explanation for some of the frequency-dependent differences in ACh release by preganglionic B and C neurones. A B Control Atropine I Hz I W-X. Shen and J P Horn Previous intracellular studies of presynaptic muscarinic inhibition in autonomic ganglia have relied on recordings of membrane potential. The first study of this kind was on B neurones in bullfrog sympathetic ganglia. Koketsu & Yamada (1982) found that bethanecol reduced the amplitude of subthreshold nicotinic EPSPs recorded in low Ca!+, high Mg2+ Ringer solution and in (+)-tubocurarine. They concluded the action was presynaptic because bethanecol did not affect spontaneous miniature EPSPs or nicotinic responses to ACh applied by ionophoresis and because it increased the coefficient of variation of EPSPs evoked in low Ca!+, high Mg2+ Ringer solution. Finally, they reported that in low Ca2+, high Mg2+, atropine augmented the amplitude of EPSPs evoked at --2 Hz. This pioneering work was more comprehensive in its scope than subsequent experiments on mammalian enteric neurones, which showed that the amplitudes of subthreshold nicotinic EPSPs were decreased by muscarinic agonists (Morita et al. 1982; North et al. 1985; Tamura & Wood, 1989; Frieling et al. 1991) and that atropinesensitive inhibition of EPSPs occurred during repetitive 5 Hz -~~~"?TT IE 2 Hz EIf 1 na l na 2 na 1 s 2 s -5 s C n -2-.N~ 'a -o '' & --2 c E) ca] Hz D Hz _., e,.,_ amm mn m Trmm n nmtr J Physiol jlw &J,r- x- -2 -* Hz --8 l * i i zv-i W Figure 5. Muscarinic autoinhibition during trains of EPSCs Trains of 4 EPSCs were recorded at 3 frequencies in normal Ringer solution (A) and in 3 nm atropine (B). The 1 and 5 Hz data are from one B neurone and the 2 Hz data are from a second cell. In control Ringer solution, EPSCs undergo different changes in amplitude that are characteristic of each frequency. Effects of atropine were measured by normalizing the amplitudes in each train to the first EPSC and subtracting the atropine from control data. C, plots of averaged data (+ S.E.M.) show the time course and magnitude of muscarinic autoinhibition (1 and 5 Hz, n = 8 cells; 2 Hz, n = 4 cells).

7 J; Physiol Presynaptic muscarinic inhibition 419 stimulation at 1-2 Hz (Morita et al. 1982). In these studies the conclusion that inhibition is presynaptic was based solely on the observation that nicotinic responses to exogenous ACh were insensitive to muscarinic agonists (Morita et al. 1982). Although the existing evidence provides a good case for presynaptic muscarinic inhibition, the interpretation of postsynaptic potentials is complicated by several factors. EPSP amplitude is influenced not only by ACh release and postsynaptic nicotinic receptor activation; but also by: (i) activation of voltage-dependent conductances; (ii) effects of membrane depolarization on ionic driving forces; and (iii) interactions with postsynaptic muscarinic mechanisms. All three factors are relevant in bullfrog sympathetic B neurones. In normal Ringer solution, the safety factor for nicotinic transmission is high (Shen & Horn, 1995a) and stimulation of primary nicotinic synapses on B neurones invariably leads to an action potential which obscures the EPSP (Shen & Horn, 1995a). Koketsu & Yamada (1982) avoided this problem by reducing EPSP amplitude with low Ca!+, high Mg2+ Ringer solution. They also claimed to observe presynaptic inhibition in (+)-tubocurarine but did not describe the details of these experiments. However, even when EPSPs are reduced below action potential threshold using these approaches, the analysis of fluctuations in large EPSPs is still hampered by the reduction of ionic driving forces (Martin, 1955). In addition, ACh activates postsynaptic muscarinic receptors that increase input resistance. Koketsu & Yamada (1982) did not account for the ability of this effect to potentiate nicotinic EPSPs and nicotinic responses to exogenous ACh (Schulman & Weight, 1977). These problems are circumvented by recording synaptic currents under voltage clamp; voltage-dependent conductances are not activated, the driving force on synaptic currents is constant, and the effect of postsynaptic muscarinic receptor activation upon holding current can be resolved from changes in fast EPSC amplitude (Fig. 1A). Our results provide direct evidence for presynaptic muscarinic inhibition under normal ionic conditions. Bath application of muscarine depressed EPSCs in B, but not C neurones (Fig. 1). A presynaptic origin of the effect is supported by two lines of evidence. First, muscarine did not alter the postsynaptic sensitivity of nicotinic receptors to ACh (Fig. 3). Second, muscarine increased the c.v. of EPSCs (Fig. 4A). This would be expected for a decrease in quantal content over a broad range of release conditions (del Castillo & Katz, 1954) and happens when EPSCs are recorded from B neurones during a shift from normal to low Ca2P, high Mg2+ Ringer solution (Connor et at. 1983). In a plot (Fig. 4B) of r, the fractional change in c.v!, vs. f the synaptic gain factor (Fig. 4B), the data points fell near the diagonal, thereby indicating muscarine inhibits release (Bekkers & Stevens, 199). We were unable to record the spontaneous miniature EPSCs required for a full quantal analysis. It is therefore impossible to infer whether muscarine reduces N, p or both. However, the deviation of the data to the right of the diagonal in the graph of r vs. f (Fig. 4B) indicates that a pure decrease in N cannot account for the results (Bekkers & Stevens, 199). Presynaptic muscarinic receptors are activated to a significant extent by the ACh that is released during repetitive stimulation. During a 5 Hz train, muscarinic inhibition reduced EPSC amplitude by -4% (Fig. 5). By contrast, saturating concentrations of muscarine reduced the EPSC by 57 % (Fig. 2). Activation of muscarinic autoreceptors varies with stimulus frequency. It increases between 1 and 5 Hz and disappears at 2 Hz (Fig. 5). This corresponds with the frequency-dependent depression of nicotinic transmission observed in the B system when the safety factor is lowered using (+)-tubocurarine (Shen & Horn, 1995a). The lack of a similar effect in the C system can be explained by the absence of presynaptic inhibition. However, presynaptic muscarinic inhibition cannot account for the depression of transmission observed at 2 Hz (Shen & Horn, 1995a). This implies that the action potential threshold increases during high frequency stimulation. It is difficult to reconcile the finding that muscarinic inhibition reduced EPSC amplitude by 15% during 1 Hz stimulation (Fig. 5) with the earlier report that atropine increased quantal content by 3% during -2 Hz stimulation (Koketsu & Yamada, 1982). When Connor et al. (1983) measured the variance of nicotinic EPSCs in B neurones, they concluded that presynaptic muscarinic receptors did not modulate ACh release during -38 Hz stimulation. One explanation for the disparity with Koketsu & Yamada's (1982) data is that the frequency dependence of inhibition may decrease in low Ca!+, high Mg2~ M2+. The minimum latency of muscarinic autoinhibition was -2 s during 5 Hz stimulation (Fig. 5) and provides a clue as to possible ionic mechanisms. Muscarinic receptors on postganglionic neurones can be coupled to several conductances. They inhibit N-type Ca!+ channels (Wanke, Ferroni, Malgaroli, Ambrosini, Pozzan & Meldolesi, 1987), activate an inwardly rectifying K+ conductance (Dodd & Horn, 1983 b; Selyanko, Smith & Zidichouski, 199), and inhibit M-type K+ channels (Adams et al. 1982). Muscarinic suppression of M-current is too slow to account for presynaptic inhibition (Jones, 1991) and is mediated by a pharmacologically distinct receptor (North et al. 1985). Moreover, the EC5 for presynaptic inhibition (12 nm, Fig. 2) is lower than for M-current suppression (7 nm) (Jones, 1985). Membrane-delimited inhibition of Ca2P currents and activation of K+ currents are both rapid (Bernheim, Beech & Hille, 1991; Clapham, 1994) and provide plausible mechanisms for presynaptic inhibition.

8 42 W-X. Shen and J P Horn J Physiol The absence of presynaptic muscarinic inhibition during 2 Hz stimulation has precedent in collection experiments on myenteric plexus (Wessler, Eschenbruch, Halim & Kilbinger, 1987). A possible explanation is that intraterminal Ca2P accumulates during high frequency stimulation and saturates the release process, thereby making it insensitive to partial inhibition of Ca2P entry. This hypothesis is consistent with the observation that oxotremorine inhibits ACh overflow from myenteric plexus during high frequency stimulation in low extracellular Ca2P (Wessler et al. 1987). Another possibility is that intense repetitive depolarization disinhibits presynaptic Ca2P channels (Bean, 1989; Boland & Bean, 1993; Elmslie, Kammermeier & Jones, 1994). Finally, the present results suggest two explanations for the contradictory observations in collection experiments on mammalian sympathetic ganglia (cf. the introduction). We have shown that presynaptic muscarinic receptors are selectively expressed by preganglionic B cells (Fig. 1) which comprise 4% of the neurones innervating ganglia 9 and 1 (Horn & Stofer, 1988). If inhibition only occurs in a subpopulation of mammalian preganglionic terminals, then it would lessen the effect of muscarinic activation upon ACh overflow from the entire ganglion. Our results also show that the magnitude of inhibition is critically dependent upon stimulus parameters. It remains for future work to determine the pathway between presynaptic muscarinic receptors and transmitter release, and the significance of this mechanism for ganglionic integration. ADAMS, P. R., BROWN, D. A. & CONSTANTI, A. (1982). M-currents and other potassium currents in bullfrog sympathetic neurones. Journal of Physiology 33, BACHOO, M. & POLOSA, C. (1992). An AF-DX 116 sensitive inhibitory mechanism modulates nicotinic and muscarinic transmission in cat superior cervical ganglion in the presence of anticholinesterase. Canadian Journal of Physiology and Pharmacology 7, BEAN, B. P. (1989). Neurotransmitter inhibition of neuronal calcium currents by changes in channel voltage dependence. Nature 34, BEKKERS, J. M. & STEVENS, C. F. (199). Presynaptic mechanism for long-term potentiation in the hippocampus. Nature 346, BERNHEIM, L., BEECH, D. J. & HILLE, B. (1991). A diffusible second messenger mediates one of the pathways coupling receptors to calcium channels in rat sympathetic neurons. Neuron 6, BLACKMAN, J. G., GINSBORG, B. L. & RAY, C. (1963). On the quantal release of the transmitter at a sympathetic synapse. Journal of Physiology 167, BOLAND, L. M. & BEAN, B. P. (1993). Modulation of N-type calcium channels in bullfrog sympathetic neurons by luteinizing hormonereleasing hormone: kinetics and voltage dependence. Journal of Neuroscience 13, CAPUZZO, A., BoRASio, P. G. & FABBRI, E. (1988). Effect of oxotremorine and RMI 1233 A on [3H]acetylcholine release and adenylate cyclase activity in guinea pig superior cervical ganglion. Neurochemical Research 13, CAPUZZO, A., BoRAsio, P. G. & FABBRI, E. (1989). Presynaptic muscarinic receptors in guinea pig superior cervical ganglion. Neuroscience Letters 14, CLAPHAM, D. E. (1994). Direct G protein activation of ion channels? Annual Review of Neuroscience 17, COLLIER, B., JOHNSON, G., KIRPEKAR, S. M. & PRAT, J. (1984). The release of acetylcholine and of catecholamine from the cat's adrenal gland. Neuroscience 13, CONNOR, E. A., LEVY, S. M. &PARSONS, R. L. (1983). Kinetic analysis of atropine-induced alterations in bullfrog ganglionic fast synaptic currents. Journal of Physiology 337, DEL CASTILLO, J. & KATZ, B. (1954). Quantal components of the endplate potential. Journal of Physiology 124, DELEAN, A., MUNSON, P. J. & RODBARD, D. (1978). Simultaneous analysis of families of sigmoidal curves: application to bioassay, radioligand assay, and physiological dose-response curves. American Journal of Physiology 235, E DODD, J. &HORN, J. P. (1983a). A reclassification of B and C neurones in the ninth and tenth paravertebral sympathetic ganglia of the bullfrog. Journal of Physiology 334, DODD, J. & HORN, J. P. (1983b). Muscarinic inhibition of sympathetic C neurones in the bullfrog. Journal of Physiology 334, Dujic, Z., ROERIG, D. L., SCHEDEWIE, H. K., KAMPINE, J. P. & BOSNJAK, Z. J. (199). Presynaptic modulation of ganglionic ACh release by muscarinic and nicotinic receptors. American Journal of Physiology 259, R ELMSLIE, K. S., KAMMERMEIER, P. J. & JONES, S. W. (1994). Reevaluation of Ca2+ channel types and their modulation in bullfrog sympathetic neurons. Neuron 13, FRIELING, T., COOKE, H. J. & WOOD, J. D. (1991). Synaptic transmission in submucosal ganglia of guinea pig distal colon. American Journal of Physiology 26, G HORN, J. P. & STOFER, W. D. (1988). Spinal origins of preganglionic B and C neurons that innervate paravertebral sympathetic ganglia nine and ten of the bullfrog. Journal of Comparative Neurology 268, IVANOFF, A. Y. & SMITH, P. A. (1995). In vivo activity of B- and C- neurones in the paravertebral sympathetic ganglia of the bullfrog. Journal of Physiology 485, JONES, S. W. (1985). Muscarinic and peptidergic excitation of bull-frog sympathetic neurones. Journal of Physiology 366, JONES, S. W. (1991). Time course of receptor-channel coupling in frog sympathetic neurons. Biophysical Journal 6, KATO, A. C., COLLIER, B., ILSON, D. & WRIGHT, J. M. (1975). The effect of atropine upon acetylcholine release from cat superior cervical ganglia and rat cortical slices: measurement by a radioenzymic method. Canadian Journal of Physiology and Pharmacology 53, KILBINGER, H. (1977). Modulation by oxotremorine and atropine of acetylcholine release evoked by electrical stimulation of the myenteric plexus of the guinea-pig ileum. Naunyn-Schmiedeberg8s Archives of Pharmacology 3, KOKETSU, K. & YAMADA, M. (1982). Presynaptic muscarinic receptors inhibiting active acetylcholine release in the bullfrog sympathetic ganglion. British Journal of Pharmacology 77,

9 J Physiol Presynaptic muscarinic inhibition 421 MARTIN, A. R. (1955). A further study of the statistical composition of the end-plate potential. Journal of Physiology 13, MORITA, K., NORTH, R. A. & TOKIMASA, T. (1982). Muscarinic presynaptic inhibition of synaptic transmission in myenteric plexus of guinea-pig ileum. Journal of Physiology 333, NORTH, R. A., SLACK, B. E. & SURPRENANT, A. (1985). Muscarinic M1 and M2 receptors mediate depolarization and presynaptic inhibition in guinea-pig enteric nervous system. Journal of Physiology 368, SAWYNOK, J. & JHAMANDAS, K. (1977). Muscarinic feedback inhibition of acetylcholine release from the myenteric plexus in the guinea pig ileum and its status after chronic exposure to morphine. Canadian Journal of Physiology and Pharmacology 55, SCHULMAN, J. A. & WEIGHT, F. F. (1976). Synaptic transmission: long-lasting potentiation by a postsynaptic mechanism. Science 194, SELYANKO, A. A., SMITH, P. A. & ZIDICHOUSKI, J. A. (199). Effects of muscarine and adrenaline on neurones from Rana pipiens sympathetic ganglia. Journal of Physiology 425, SHEN, W.-X. & HORN, J. P. (1995a). A presynaptic mechanism accounts for the differential block of nicotinic synapses on sympathetic B and C neurons by d-tubocurarine. Journal of Neuroscience 15, SHEN, W.-X. & HORN, J. P. (1995b). Presynaptic muscarinic inhibition of nicotinic transmission in bullfrog sympathetic ganglia. Society for Neuroscience Abstracts 21, 192. SOMOGYI, G. T. & DEGROAT, W. C. (1993). Modulation of release of [3H]acetylcholine in the major pelvic ganglion of the rat. American Journal of Physiology 264, R TACK, J. F. & WooD, J. D. (1992). Synaptic behaviour in the myenteric plexus of the guinea-pig gastric antrum. Journal of Physiology 445, TAMURA, K. & WOOD, J. D. (1989). Electrical and synaptic properties of myenteric plexus neurones in the terminal large intestine of the guinea-pig. Journal of Physiology 415, WANKE, E., FERRONI, A., MALGAROLI, A., AMBROSINI, A., POZZAN, T. & MELDOLESI, J. (1987). Activation of a muscarinic receptor selectively inhibits a rapidly inactivated Ca2+ current in rat sympathetic neurons. Proceedings of the National Academy of Sciences of the USA 84, WESSLER, I., ESCHENBRUCH, V., HALIM, S. & KILBINGER, H. (1987). Presynaptic effects of scopolamine, oxotremorine, noradrenaline and morphine on [3H]acetylcholine release from the myenteric plexus at different stimulation frequencies and calcium concentrations. Naunyn-Schmiedeberg's Archives of Pharmacology 335, Acknowledgements This work was supported by NIH grant RO1 NS2165. Received 1 August 1995; accepted 27 September 1995.