See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/20533846 Direct measurement of ACh release from exposed frog nerve terminals: Constraints on interpretation of... Article in The Journal of Physiology · January 1990 DOI: 10.1113/jphysiol.1989.sp017871 · Source: PubMed CITATIONS READS 29 32 4 authors, including: Alan D Grinnell Stephen D Meriney 108 PUBLICATIONS 3,444 58 PUBLICATIONS 1,379 CITATIONS University of California, Los An… CITATIONS SEE PROFILE University of Pittsburgh SEE PROFILE All content following this page was uploaded by Stephen D Meriney on 08 December 2016. The user has requested enhancement of the downloaded file. 225 Journal of Physiology (1989), 419, pp. 225-251 With 1 plate and 10 text-ftgure8 Printed in Great Britain DIRECT MEASUREMENT OF ACh RELEASE FROM EXPOSED FROG NERVE TERMINALS: CONSTRAINTS ON INTERPRETATION OF NONQUANTAL RELEASE BY A. D. GRINNELL, C. B. GUNDERSEN*, S. D. MERINEYt AND S. H. YOUNG From the Jerry Lewis Neuromuscular Research Center, Department of Physiology and *Department of Pharmacology, UCLA School of Medicine, Los Angeles, CA 90024, USA (Received 9May 1989) SUMMARY 1. Acetylcholine (ACh) release from enzymatically exposed frog motor nerve terminals has been measured directly with closely apposed outside-out clamped patches of Xenopus myocyte membrane, rich in ACh receptor channels. When placed close to the synaptic surface of the terminal, such a membrane patch detects both nerve-evoked patch currents (EPCs) and spontaneous quantal 'miniature' patch currents (MPCs), from a few micrometres length of the terminal, in response to ACh release from the nearest three to five active zones. 2. Chemical measurements of ACh efflux from whole preparations revealed a spontaneous release rate of 4-1 pmol (2 h)-1, and no significant difference in resting efflux between enzyme-treated and control preparations. The ratio of enzymetreated to contralateral control muscle efflux averaged 1-17, indicating that enzyme treatment did not affect spontaneous ACh release. Vesamicol (1-7 ,UM), which blocks the ACh transporter in synaptic vesicles, decreased the spontaneous release of ACh to 67 % of control. 3. In the absence of nerve stimulation, the frequency of single-channel openings recorded by outside-out patch probes adjacent to nerve terminals was very low (1-2 min-'), and little different at a distance of hundreds of micrometres, suggesting that if ACh was continually leaking from the terminal in a non-quantal fashion, the amount being released near active zone regions on the terminal was below the limit of detection with the patches. 4. Direct measurements of the sensitivity of the patches, coupled with calculated ACh flux rates, lead to the conclusion that the amount of ACh released non-quantally from the synaptic surface of the frog nerve terminal is less than one-tenth the amount expected if all non-quantal release is from this region of the terminal membrane. 5. Following a series of single nerve shocks or a 50 Hz train of nerve stimuli, the frequency of asynchronous single-channel openings increased for several seconds. This transient increase in channel openings was not sensitive to movement of the t To whom correspondence should be sent. MS 7676 8 PHY 419 Downloaded from J Physiol (jp.physoc.org) by guest on July 10, 2011 226 A. D. GRINNELL AND OTHERS patch electrode a significant distance (4 /tm) away from the active sites, or to manipulations previously reported to block non-quantal transmitter leakage, including addition of 10 mM-Ca2+ or 1-7 fM-vesamicol to the bath. These channel openings appear to be due to an accumulation of ACh which originated from many evoked quanta, and not the effect of locally increased non-quantal ACh release due to nerve stimulation. 6. We conclude that transmitter leakage at adult frog terminals is either localized to a source other than the synaptic surface of the nerve terminal, or released in a widespread and diffuse fashion from many sources, which may include the nerve terminal. INTRODUCTION In the early 1960s, Straughan (1960) and Mitchell & Silver (1963) reported that only 1-3 % of spontaneous ACh released from the mammalian hemidiaphragm could be attributed to quantal events, and suggested that pre-terminal nerve branches may be leaking ACh. Using standard intracellular recording techniques, Katz & Miledi (1977) described a 40 #tV average hyperpolarization of the endplate membrane of frogs (termed the 'H-effect') following a local curare application, an effect which they interpreted to reflect non-quantal leakage of ACh from the motor nerve terminals themselves. Based on the frequency and size of miniature endplate potentials (MEPPs) from these terminals, they also conclude that spontaneous quantal release comprises only about 1 % of total spontaneous ACh release. Subsequent studies of non-quantal ACh release from amphibian and mammalian nerve-muscle preparations have further expanded these observations (Vyskocil & Illes, 1977; Miledi, Molenaar & Polak, 1980; Vyskocil, Nikolsky & Edwards, 1983; Edwards, Dolezal, Tucek, Zemkova & Vyskocil, 1985; Sun & Poo, 1985; Molenaar, Oen, Polak & van der Laaken 1987; Edwards, Dolezal, Tucek, Zemkova & Vyskocil, 1988). Based on these measurements, most of the spontaneous ACh release from a nerve-muscle preparation is believed to be non-quantal. The existence of low level but continuous transmitter leakage has led many to propose a long-term trophic role for this form of ACh release in the regulation of various muscle characteristics, including acetylcholinesterase induction or regulation (Katz & Miledi, 1977), ACh receptor distribution (Mathers & Thesleff, 1978; Pestronk, Drachman, Stanley, Price & Griffin, 1980; Drachman, Stanley, Pestronk, Griffin & Price, 1982; Stanley & Drachman, 1986), and resting membrane potential (Drachman et al. 1982; Bray, Forrest & Hubbard, 1982). An important assumption in the interpretation of the H-effect is that non-quantal release occurs at the synaptic surface of the nerve terminal which closely apposes the muscle cell. Recently, it has been proposed (Edwards et al. 1985, 1988) that transmitter leakage is mediated via transporter molecules that are normally located in the synaptic vesicle membrane where they function to load ACh (Andersen, King & Parsons, 1983). The transporter molecules are presumed to be incorporated into the presynaptic membrane after vesicular fusion and exocytosis (Heuser & Reese, 1973, 1981) where they would pump Downloaded from J Physiol (jp.physoc.org) by guest on July 10, 2011 ACh RELEASE FROM EXPOSED NERVE TERMINALS 227 cytoplasmic ACh outward across the plasma membrane until they are again retrieved by endocytosis. This hypothesis predicts that evoked vesicular release will increase transmitter leakage. However, this idea is not easily tested with the electrophysiological, pharmacological or biochemical techniques previously used to detect ACh leakage, as these techniques do not provide the necessary combination of high spatial and chemical sensitivity. If one could position a sensitive detector of ACh release in close proximity to the nerve terminal release sites, one could directly test these hypotheses. Enzymatic 'dysjunction' of the neuromuscular contact renders the synaptic surface of the terminal accessible to small probes, such as a patch pipette. This provides an opportunity to study the release of acetylcholine from nerve terminals with higher spatial resolution, and much greater sensitivity (the opening of single channels), than is possible with conventional extracellular or intracellular microelectrodes. Betz & Sakmann (1973) have shown that enzymatic treatment also removes nerve terminal-associated acetylcholinesterase activity, and causes a decrease in the amplitude of spontaneous MEPPs and evoked endplate potentials as the nerve terminal detaches from the postsynaptic muscle cell. However, this enzymatic detachment does not noticeably influence parameters of presynaptic function such as the MEPP frequency or EPP quantal content. Xenopus muscle cells, at 1 or 2 days in culture, have a relatively high uniform density (about 100 1tm-2) of extrajunctional ACh receptors (Kidokoro & Gruener, 1982). In an attempt to measure directly non-quantal ACh release, sensitive ACh probes have been made using excised patches (about 1 ,tm in diameter) of this ACh receptor-rich Xenopus myocyte membrane (see Hume, Role & Fischbach, 1983; Young & Poo, 1983). These probes can be manipulated into close apposition to the exposed terminals for the purpose of measuring ACh release. Based on an extrapolated calibration of the effect of ACh on membrane potential at the motor endplate, Katz & Miledi (1977) predicted that the curare-induced 40 ,V hyperpolarization observed at the average frog nerve-muscle junction could be accounted for by a steady-state concentration of ACh in the cleft of 19 nm. This estimate was made for the confined environment of the synaptic cleft, and probably overestimates the concentration of ACh one would expect to be present several micrometres away from an enzymatically cleaned nerve terminal. Therefore, using their estimate of the non-quantal flux rate, we calculated that a sensing device positioned 2 ,tm away from a linear source of ACh in a 'freely' diffusing medium would encounter an ACh concentration of 8 nM (Carslaw & Jaeger, 1947). Based on the calibration curve for the sensitivity of the patch probes (see Fig. 1), an ACh concentration of 8 nm should open an average of eighteen ACh receptor channels per minute. Since the background frequency of channel openings is less than one per minute (see Results), the probes should detect this release with an adequate signal-to-noise ratio as long as they are within 2 ,um of the source of the leak. We present data to indicate that probes can be positioned within 2 ,tm of the sites of quantal ACh release on the terminal. Under these circumstances, quantal transmitter release is readily detected. However, we have found no evidence for non-quantal leakage of ACh from the synaptic surface of these terminals. 8-2 Downloaded from J Physiol (jp.physoc.org) by guest on July 10, 2011 228 A. D. GRINNELL AND OTHERS METHODS Detached terminal preparation The cutaneous pectoris nerve-muscle preparation was dissected from adult Rana pipiens following anaesthesia with 01 % tricaine methanesulphonate (Sigma, St Louis, MO, USA), and double-pithing. The preparation was subsequently stained with 2 mg ml-' fluorescein isothiocyanate (FITC)-labelled peanut agglutinin (Sigma) as a probe for the junctional matrix surrounding presynaptic nerve terminals (Ko, 1987), and 15 ,g ml-' tetramethylrhodamine isothiocyanate (TRITC)-labelled a-bungarotoxin (Molecular Probes, Eugene, OR, USA) to visualize the postsynaptic ACh receptor-filled gutter and to eliminate muscle contractions. This double-labelled preparation was placed in an enzyme incubation chamber in which both ends of the muscle (attachments to skin and sternum) were protected from enzyme digestion by Vaseline isolation. The nerve terminal-rich region of the preparation was incubated in 01 % type I collagenase (Sigma) for 1 h, followed by a 1 h incubation in 0-01 % type XXVII protease (Sigma) as previously described (Betz & Sakmann, 1973; Kuffler & Yoshikami, 1975a). Following this treatment, the preparation was washed for 45 min in normal frog Ringer solution (in mM): 116 NaCl, 2 KCl, 1P8 CaCl2, 1 MgCl2, 1 NaHCO3; pH 7-2, and pinned to the bottom of a Sylgard-coated recording chamber. Terminal regions were located using Hoffman modulation contrast optics, and the degree of terminal detachment was determined using alternating fluorescein (terminal probe) and rhodamine (postsynaptic probe) excitation filters at 200 x with an Olympus 20 x long working-distance water-immersion objective. Xenopus myoball culture Xenopus myoball cultures were prepared as previously described (Spitzer & Lamborghini, 1976; Anderson, Cohen & Zorychta, 1977). Briefly, the neural tube and surrounding mesodermal tissue were dissected from stage 17-19 (Nieuwkoop & Faber, 1967) Xenopus embryos and dissociated in Ca2+, Mg2+-free saline (58-2 mM-NaCl, 0 7 mM-KCl, 03 mM-EDTA, 250 i.u. each penicillin and streptomycin; pH 7 8). The dissociated cells were plated on clean glass cover-slips and incubated at room temperature (21-23°C) in medium consisting of 85 % saline (in mM: 58-2 NaCl, 0 7 KCl, 04 Ca(NO3)2, 1-3 MgSO4 and 5 HEPES) 10% Leibovitz medium (L-15; GIBCO), 5% fetal calf serum and 04 % penicillin/streptomycin; pH 7.8. Outside-out patch probes Xenopus muscle cell cultures were used as a source of ACh receptor-rich membrane. A small fractured piece of a cover-slip, on which myoballs had been plated1 or 2 days earlier, was placed in the recording chamber near the enzymatically treated nerve terminal preparation, and both were perfused with normal frog Ringer solution throughout all experiments. Outside-out patches of myoball membrane were made as previously described (Young & Poo, 1983). Briefly, a glass micropipette was prepared by two-stage pulling, Sylgard coating, and microforge polishing to a tip diameter of about1,um. The pipette was filled with an internal solution (in mM: 92 KCl, 40 KOH, 1 CaCl2, 11 EGTA,1 MgCl2, 20 HEPES; pH 7 8), and outside-out patches were made from spherical myocytes and immediately used to detect ACh release from exposed terminals. Patch holding potential was set to the myocyte resting potential (-80 to -100 mV). The patch electrode was manipulated to approach the side and lower surface of terminals at cleanly exposed positions. Amplified recordings of patch membrane current (List EPC-7) were stored on magnetic tape (RACAL) for subsequent analysis. In some experiments (-) vesamicol hydrochloride (Research Biochemicals, Natick, MA, USA) or eserine (Sigma) was included in the perfusate. - Calibration of patch probe sensitivity To calibrate the sensitivity of a patch of this membrane, the response of cell-attached and outside-out patches to ACh application was determined. In experiments with the cell-attached patch, the pipette was filled with varying dilutions of ACh (Sigma) in an external solution (in mM: NaCl, 2 CaCl2, 2 KCI, 1 MgCl2, 5 NaOH, 10 HEPES; pH 7 8). Four to six cell-attached recordings were made from different muscle cells at each of the three concentrations of ACh (3, 10 and 30 nM), and single-channel activity was recorded for 2-5 min on magnetic tape (RACAL Store 4 FM) at a corner frequency of 2-5 kHz for subsequent analysis. Outside-out patches were tested with bath application of ACh in external saline. Cell-attached and outside-out patches respond with similar opening frequencies at a given ACh concentration (see Fig. 1). Downloaded from J Physiol (jp.physoc.org) by guest on July 10, 2011 ACh RELEASE FROM EXPOSED NERVE TERMINALS 229 Histology At the conclusion of some recording sessions, photographs or camera lucida drawings of the terminals studied were made at 400 x following Nitroblue Tetrazolium (NBT) staining (Letinsky & DeCino, 1980). The preparation was fixed in 2% glutaraldehyde in normal frog Ringer solution for 30 s, and immersed in an NBT solution (5 mg Nitroblue Tetrazolium (Sigma), 0-5 mg phenozine methosulphate (Sigma) and 6 ml 2% glutaraldehyde in Ringer solution) until details of the terminal arbor could be visualized (about 3 min). Intracellular electrophysiology To determine the effects of enzyme treatment on junctional physiology, the cutaneous pectoris nerve-muscle preparation was labelled with FITC-peanut agglutinin for nerve terminal visualization, but not with a-bungarotoxin in these experiments. All recordings were made in normal frog Ringer solution with 50 /SM-eserine (Sigma) using conventional intracellular microelectrodes (40-60 MCI) filled with 0-6 M-K2SO4. Miniature endplate potentials were stored on magnetic tape before and after enzymatic treatment, played back on an oscilloscope, photographed, and rise times-to-peak measured manually. Chemical measurements of ACh efflux In some animals, the left cutaneous pectoris nerve-muscle preparation was dissected and treated with either collagenase and protease as described above, or 1P7 ,pM-vesamicol. In others, the left cutaneous pectoris muscle nerve was cut, and denervation maintained for 2 weeks with resectioning after 1 week. In all animals, the right cutaneous pectoris was used as a control. Before biochemical measurements were made, all preparations were washed in normal frog Ringer solution containing 20 /tM-eserine (Sigma) for 30-60 min. Subsequently, consecutive 2 h incubations were made in 1 ml eserine-frog Ringer solution aliquots for biochemical assay. At the conclusion of the experiment, all aliquots and homogenized nerve-muscle preparations were assayed for ACh content by gas chromatography/mass spectrometry as previously described (Jenden, Roch & Booth, 1973). Scanning electron microscopy Enzyme-treated and control cutaneous pectoris nerve-muscle preparations were pinned to Sylgard-coated dishes and fixed for 1 h in 2 % glutaraldehyde (in Ringer solution), followed by a wash in several changes of Ringer solution, and a 1 h post-fix in 1 % OS04 (in Ringer solution). Some preparations were subsequently incubated in 8 N-HCl for 15-20 min at 60 °C to remove completely obstructing connective tissue (see Desaki & Uehara, 1981). All tissues were dehydrated through a graded series of ethanol concentrations and critical-point dried. After sputter coating with gold and palladium, specimens were examined in a Etek Autoscan scanning electron microscope at 10 kV. RESULTS Viability of enzymatically detached terminals Following collagenase and protease treatment, various degrees of synaptic detachment were readily apparent (Fig. 2). Plate 1 shows representative scanning electron micrographs of (A) a terminal covered with the normal amount of connective tissue, (B) a terminal exposed by the enzyme treatment used in these experiments, and (C) an acid-cleaned preparation in which all connective tissue had been removed. A scanning electron micrograph of a patch electrode is shown in the middle panel in much the configuration it might have had during an experiment. In most cases, with the enzyme treatment used, terminal displacement from the postsynaptic gutter was small, although a few terminals were displaced by as much as 500 ,um. A striking characteristic of detached terminals was the maintenance of geometric shape of the terminal arbor despite occasional large displacement from the Downloaded from J Physiol (jp.physoc.org) by guest on July 10, 2011 A. D. GRINNELL AND OTHERS 230 postsynaptic gutter (see Fig. 2). Although this may be partially due to intraterminal cytoskeletal rigidity, we have observed the persistence of a connective tissue matrix between most terminal branches despite enzymatic digestion of connections to the postsynaptic cell (Plate 1). Therefore, it is possible that the maintenance of terminal shape observed following enzymatic detachment is predominately due to the presence of connective tissue stabilization of interbranch geometric relationships. 0) C 0. 10 C : 110 -§§ 0 ACh concentration (nM) Fig. 1. Sensitivity of Xenopu8 muscle membrane patches to ACh. The mean (±S.E.M.) frequency of single channels that opened in the patch is plotted on a log-log scale with the concentration of ACh. The (O) represent the frequency of openings detected with the cellattached configuration using varying concentrations of ACh in the pipette (n = 5, 4 and 6 for 3, 10 and 30 nM-ACh respectively). The (M) represent the frequency of openings detected using an outside-out patch with varying concentrations of ACh in the bath (n = 5, 4 and 2 for 3, 10 and 30 nM-ACh respectively). Given the variability in the number of openings observed at any one concentration (indicated by standard error bars), it does not appear that the sensitivity of the patch of membrane differs in these two configurations. The dashed line represents the frequency of openings that is two standard deviations above the background frequency recorded with an outside-out patch probe hundreds of micrometres above the muscle in a freely perfusing bath. Only 4-5 % of patches would be expected to display this number of openings in normal frog Ringer solution. Extending the observations of Betz & Sakmann (1973) on the condition of partially detached terminals, postsynaptic intracellular recording techniques were used to observe the properties of MEPPs before and after the enzyme treatment at the same terminals. Since the enzymatic treatment has been shown to remove endplate-associated acetycholinesterase activity, eserine was included in the Ringer solution of both detached and intact preparations as a control for the effects of acetylcholinesterase loss itself. MEPP rise times were measured as an indicator of the degree of synaptic detachment. 'Detached' terminals produced MEPPs that were recorded in the postsynaptic membrane at generally lower amplitudes, and with a Downloaded from J Physiol (jp.physoc.org) by guest on July 10, 2011 ACh RELEASE FROM EXPOSED NERVE TERMINALS 231 wide variety of rise times, suggesting that different terminal branches were detached to varying degrees, and that detached portions were continuing to release quanta. Figure 3 represents a typical example where the mean rise time of MEPPs from an enzymatically treated terminal was significantly longer (3-13 + 012 ms, n = 48; mean +S.E.M.) than from a control terminal (2-18 +±004 ms, n = 54; mean+ S.E. M., P <0-005; Student's t test). These observations, coupled with those of Betz & Sakmann (1973), demonstrate that the 'detached' nerve terminal is functionally viable following this enzymatic treatment protocol. A 20 gm Fig. 2. Drawing-tube tracings of the variable degrees of nerve terminal detachment following collagenase and protease treatment. The fluorescein-labelled nerve terminal extracellular matrix is shown outlined, while the rhodamine-labelled postsynaptic ACh receptors are shown stippled. A, the nerve terminal is only slightly detached from its postsynaptic site of attachment. B, the nerve terminal is displaced by a large amount from ACh receptor clusters on the muscle cell. Calibration = 20 ,um. We recorded physiological data with patch probes only from those terminals which had not significantly detached from the postsynaptic muscle cell. The patch electrode was manipulated into close proximity to a visually identified nerve terminal until it appeared to touch, but not displace, the terminal. Although the pipette glass was in very close apposition to the nerve terminal, the portion of the membrane patch detecting ACh release was probably 1-2 ,um away from the terminal surface, due to the ' Q' shape of the membrane patch (see Sakmann & Neher, 1983). Even though the patch probe was consistently brushed and pushed against muscle cells and connective tissue, the patch seal was rarely broken during any of these procedures. Characterization of the sensitivity of Xenopus patch probes To determine the sensitivity of patch probes made from the extrajunctional muscle of cultured Xenopus muscle cells, cell-attached patches were filled with 30, 10 or 3 nmACh, and the frequency of the single-channel openings in the patch was recorded. Downloaded from J Physiol (jp.physoc.org) by guest on July 10, 2011 A. D. GRINNELL AND OTHERS 232 In addition, outside-out patches were exposed to low concentrations of ACh in the bath to confirm the maintenance of ACh sensitivity in this configuration. Figure 1 shows the calibration curve one obtains when the data are plotted on a log-log scale. A level (dashed line, Fig. 1) at which the single-channel opening frequency was two 40 35300 e 25 - 20 o20 0) .0 E z 15 105 0 1.0 2.0 3.0 4.0 5.0 6.0 Rise time (ms) Fig. 3. Distribution of MEPP rise times in control and enzyme-treated preparations. In both, 50 /uM-eserine was added to the perfusate. MEPPs from control terminals (open bars) have a narrow distribution of rise times whose mean (arrow) is significantly faster than the mean (arrow) of the widely distributed MEPP rise times recorded from enzymetreated nerve terminals (hatched bars). standard deviations above the background frequency recorded by the outside-out patch probe in a freely perfusing Ringer solution bath (07+0±6 openings min-', mean + S.D. for 24 min in four experiments) was defined as being significantly higher than background. Thus, using this detection system, any single-channel opening frequency greater than 1-8 min-' is significantly different from background. Statistically, only 4-5 % of membrane patches bathed in normal frog Ringer solution will exhibit single-channel openings at this frequency. Outside-out patch recordings of quantal ACh release High-resolution monitoring of ACh release was achieved with outside-out patches of ACh receptor-rich membrane. At a succession of positions along the length of any particular nerve terminal, nerve impulse-evoked patch currents could be observed at most locations (Fig. 4). Similar recordings were attempted from terminals that had not been enzyme-treated, but no ACh receptor channel activity was observed. This probably reflects significant diffusion barriers and limited access to the terminal in the intact preparation. The amplitude of ACh-induced current varied at each Downloaded from J Physiol (jp.physoc.org) by guest on July 10, 2011 ACh RELEASE FROM EXPOSED NERVE TERMINALS 233 enzyme-treated recording site. This variation is due to differences in the diffusion barriers between the probe and the nerve terminal, the proximity of the patch electrode to the sites of release (presumably, active zones), and the patch sensitivity to ACh (see variability in Fig. 1). On average, the peak of the evoked current at good /N ~ ~ ~ ~ ~ 6/ 110 pA _ 10 ms Fig. 4. Detection of nerve-evoked quantal ACh release from many positions along the length of an identified nerve terminal. In most positions (2-6), summed current responses were recorded with the patch probe. In some cases (1, 7), only single-channel responses were observed, presumably due to either distance or barriers to diffusion between the probe and the nerve terminal. The biphasic event at the beginning of each trace is the stimulus artifact. probe positions (which exhibited rise times of 1-3 ms) was 28-5 pA (n = 54). This corresponds to about eleven simultaneously open ACh channels at the peak of the response. In some recording positions only single-channel openings were observed, while in a few cases, the peak of the nerve-evoked current reached as much as 150 pA, corresponding to approximately sixty simultaneously open channels at the peak. In large responses such as this, total channel openings reached 112. Only those electrode positions which produced evoked currents with fast rise times (1-3 ms) were used for data analysis. Despite the lack of acetyleholinesterase activity following enzyme treatment, the spatial resolution of recorded release was very sharp. These 'exposed' terminals released ACh into an environment which lacked many of the diffusion barriers of the intact synaptic cleft. Diffusion of a quantum of ACh released from a point source into a three-dimensional environment would reduce the concentration of ACh to very low levels within a few micrometres (Ogston, 1955). In view of this, separate experiments were performed to determine the spatial resolution of the probe and to assess the length of nerve terminal from which action potentials could evoke release of ACh such that multichannel summed responses could be recorded at a single probe Downloaded from J Physiol (jp.physoc.org) by guest on July 10, 2011 234 A. D. GRINNELL AND OTHERS position. For these experiments, it was necessary to reduce the probability that adjacent active zones would simultaneously release quanta of ACh. With a low Ca2+, Mg2+-supplemented Ringer solution (1 mM-CaCl2, 2-5 mM-MgCl2) the probability of quantal ACh release from individual release sites was diminished such that the likelihood of quantal release from any given active zone near the patch probe was much less than one. Thus any measured EPC represented the effects of a single quantum released from one of the several release sites at different distances from the probe. The resulting EPCs differed in latency, rise time and amplitude. Plots of the number of channel openings in individual EPC responses at single probe positions in two experiments, recorded at various latencies after the stimulus artifact, are shown in Fig. 5. The number of channel openings was calculated by dividing the total EPC charge flux by the independently determined charge flux of single-channel openings (average current amplitude x average channel open time) seen in that particular patch of membrane. From the distinct grouping of events recorded using this technique (Fig. 5), we estimate that our patch electrode was, in most cases, sampling quantal ACh release from primarily three discrete active zones. Presumably, the largest responses are from the nearest active zone, intermediate responses from the ones to either side, and relatively small and slow responses from the active zones still further away on either side. There are some events on the graph which appear to fall between groups. The presence of such events is not surprising since the patch probe tended to be located at one side of the terminal, and evoked release originating from various positions along the length of a single active zone would be expected to arrive at the patch probe with some degree of variability in latency and amplitude. The results of Fig. 5 indicate that the patches detected release from only the immediately adjacent portion of the 'exposed' terminal. Consistent with this is the fact that spontaneous quantal events (MPCs) were rarely observed. To increase spontaneous quantal discharge from the nerve terminal and enhance the probability of observing MPCs in a particular recording position, the preparation was perfused with hyperosmotic (50 mM-sucrose) Ringer solution. Under these conditions, several MPCs could be recorded at one position during a 3-5 min observation period. The amplitude of MPCs in normal and hypertonic Ringer solutions was very similar to the amplitude of most EPCs recorded at a single patch position (Fig. 6). Occasionally, an EPC of about twice the amplitude of the average MPC was observed. MPC rise time was generally faster than the rise time of large, multi-quantal EPCs. This is presumably due to synchronous release of two quanta from sites at different distances from the patch probe (see below and Fig. 7). In addition, the total number of channel openings in the observed MPCs (integral of MPC current divided by the mean single-channel current integral) was slightly smaller than that of the EPCs. The explanation for this is apparent when the time course of evoked and spontaneous quantal events are compared (Fig. 7). Spontaneous events (MPCs) from the most proximal active zone had a fast rise time and smooth exponential decay (Fig. 7A and a). Evoked events (EPCs) with similar rise times decayed with a variable time course, and often displayed multiple peaks (Fig. 7E and e), presumably due to contributions from nearby active zones whose ACh molecules arrive more slowly than the ACh from the most proximal active zone. We cannot, however, rule out the possibility Downloaded from J Physiol (jp.physoc.org) by guest on July 10, 2011 ACh RELEASE FROM EXPOSED NERVE TERMINALS 235 A 35 n o 30 0. c 25- (I) C 202E._ O 15. / / 0 C c j 10/ ,-, 5- * * 4) v0 0 *0 0 v * I 0 /4 * 20 15 Latency (ms) B o 35 0 0 10 5 0 / . c 0 XL 30 .' 25 n 0) *'C 20 0 o 15 C C , 1 s 6-'0 I- 5. 0, .1 0 * 0 5 - 15 Latency (ms) 10 . 20 Fig. 5. Characterization of the spatial resolution of the patch probe at two terminal positions. Recordings in low Ca2+ decreased the probability of release such that nerveevoked ACh release detected by the probe originates from no more than one release site per stimulus. Under these conditions, when the latency of the onset of the current response was plotted against the total number of single-channel openings, the evoked responses fell into three distinct groups. A and B are two examples of this, which suggest that the patch probes were detecting evoked release from three to five active zones along the length of the terminal. that evoked vesicular release resulted in two vesicles being released from opposite ends of a single active zone (about 1 ,tm in length). The patch probe would be unable to distinguish this type of spatial separation from release at two active zones (also about 1 ,um apart). Despite this possibility, sharply rising EPCs recorded at any given site were rarely seen to occur in multiples of a smaller sharply rising MPC size. This Downloaded from J Physiol (jp.physoc.org) by guest on July 10, 2011 A. D. GRINNELL AND OTHERS 236 implies that only one quantum is released from any given active zone, even under conditions that promote full quantal discharge (i.e. using normal frog Ringer solution). Outside-out patch recordings of non-quantal A Ch release Detection of ACh release with single-channel sensitivity provides the opportunity to study directly the non-quantal leakage of transmitter that is thought to occur 20 0 c 0 ._ 4_- > .0 o E z 0 10 20 30 40 50 Peak amplitude (pA) 60 Fig. 6. An example of the distribution of spontaneous (hatched bars) and nerve-evoked (open bars) quantal current amplitudes detected by the patch probe under conditions which promote increased MEPP frequency (hypertonic Ringer solution). On average, one quantum of ACh is released synchronously in close proximity to a single probe position, but occasionally, two may be released. from motor nerve terminals. As noted in the Introduction, the findings of Katz & Miledi (1977) lead to the prediction of a maintained ACh concentration of 19 nM in the intact synaptic cleft. It is possible to estimate the concentration one might expect in an enzymatically treated preparation using the Katz & Miledi (1977) estimates of the non-quantal flux rate (4-74 x 10-17 mol cm-' s-), and the flux equation for a linear source: ACh concentration = (1 15(q)/17T D) log (1/r), where q is the flux rate, D is the diffusion coefficient and r is the distance from the source (Carslaw & Jaeger, 1947). Using this approach, it can be calculated that a Downloaded from J Physiol (jp.physoc.org) by guest on July 10, 2011 ACh RELEASE FROM EXPOSED NERVE TERMINALS 237 sensing device positioned in close apposition to, but with the sensing membrane about 2 ,sm away from, a linear source of ACh in a freely diffusing medium would encounter an ACh concentration of 8 nm. Based on the calibration curve for the ACh sensitivity of patch probes made from Xenopus myoballs (Fig. 1), this concentration of ACh is detectable above background, and should open an average of 18 channels min-'. A a B b cAf-~~~~cV D d E e 110 pA 10 Ms Fig. 7. Quantal transmitter release from a single position at an enzyme-treated nerve terminal. A,A, two examples of spontaneous quantal release from a nearby active zone. B,b-E,e, representative pairs of responses to nerve-evoked quantal release, which exemplify variation in amplitude and time course, presumably dependent on probe proximity to different active zones and diffusion barriers between the patch probe and the nerve terminal. Calibration = 10 pA, 10 ins. To measure accurately the frequency of single-channel openings at sites along the nerve terminal, and to avoid the contamination of this signal by quantal ACh release, we evoked quantal ACh release (as a test of electrode position), no more than once every 3 min. We recorded single-channel activity between stimuli for 3-9 min at each position. Since pipette position is especially critical for these recordings, we only used positions with relatively large EPCs (mean peak amplitude of 35-4 pA or fourteen simultaneously open channels) presumably reflecting either particularly close apposition to release sites, or unusually sensitive patch probes. A total of 73 min of recordings was made from these positions (n = 12). At such terminal positions, the background single-channel frequency ranged between 0 and 2-7 channel openings min-', and averaged only 1-6 ± 05 (mean ±S.E.M.). This value is not significantly different (P < 0-1) from the background frequency recorded in the bath at a point far from the muscle (0-7 ± 0-3 channel openings min-'). More importantly, this frequency Downloaded from J Physiol (jp.physoc.org) by guest on July 10, 2011 A. D. GRINNELL AND OTHERS 238 is more than 10 times lower than the frequency one would predict (18 min-') based on the observations of Katz & Miledi (1977). Chemical measurements of non-quantal A Ch release Direct chemical measurements were used to compare the rate of spontaneous ACh release from the frog cutaneous pectoris nerve-muscle preparation with that measured for other preparations (Miledi et al. 1980; Molenaar et al. 1987). These measurements also were designed to determine whether the enzymatic treatment of these preparations might have disrupted the process of non-quantal release independently of its effects on quantal ACh release. Measurements of spontaneous ACh release were made from thirty-nine cutaneous pectoris nerve-muscle preparations that had not been treated with enzymes, and the mean (±S.E.M.) efflux was 4-1 + 0-6 pmol (2 h)-1. Since the spontaneous quantal release rate from each of the approximately 800 nerve terminals in this muscle is about 1 s-1, one can calculate that over a 2 h incubation, spontaneous quantal release will contribute about 0-06 pmol of ACh to the bath. This implies that most of the ACh released spontaneously from the frog cutaneous pectoris is coming from a source other than the quantal events at synaptic sites. This finding is consistent with results obtained using frog sartorius (Katz & Miledi, 1977; Miledi et al. 1980) and mammalian nerve-muscle preparations (Straughan, 1960; Mitchell & Silver, 1963; Molenaar et al. 1987). To determine if collagenase and protease treatment affected the spontaneous ACh release from this preparation, paired measurements (each pair from one animal) were made from control and enzyme-treated cutaneous pectoris muscles (see Table 1). The average (mean + S.E.M.) release of ACh was similar for both enzyme-treated (4 7 + 0 7 pmol (2 h)-1; n = 8) and control preparations (3-8 + 0 9 pmol (2 h)-1; n = 8). Since some variability exists among preparations, paired values from enzyme-treated and contralateral control preparations were compared (enzyme-treated value divided by control value) and an average ratio of 1-17 + 0-12 was obtained. Therefore, these data indicate that enzymatic digestion does not significantly alter the non-quantal release process. Since it has been reported that vesamicol blocks non-quantal ACh release from the mammalian nerve-muscle preparation (Edwards et al. 1985), we investigated the effect of vesamicol on the total spontaneously released ACh. Consistent with previous reports in mammals, 1-7 /SM-vesamicol decreased the mean rate of ACh efflux from the nerve-muscle preparation such that when paired with contralateral control muscles, the efflux ratio of vesamicol-treated to control preparations was 0-67 + 0-16 (n = 5). Therefore, there appears to be a portion of the spontaneous ACh release that is sensitive to vesamicol blockade. The patch clamp method can also be used to measure the bath accumulation of ACh. In one experiment, the cutaneous pectoris preparation was bathed for 2 h in a static bath containing 1-5 ml of external solution (see Methods) with 20 /SM-eserine. Then, patch pipettes were filled with samples of the bathing solution taken after 1 min or 2 h of incubation and cell-attached patches were made onto Xenopus myoballs. The ratio of single ACh channel openings recorded when the pipette was filled with the 2 h vs. the 1 min incubation solution was 22/1. The frequency of single- Downloaded from J Physiol (jp.physoc.org) by guest on July 10, 2011 ACh RELEASE FROM EXPOSED NERVE TERMINALS 1-4 0 CO CN r4) 14 +1 0 9 0 +1 +1 m M- Ca1 t 0t -. -4 0 C)4) Cq1- -4) 05 4-) 0 4) CO 4.) .m) - o V- 4.. ea +1 o; 2 - r- 0 ¢ CO C> 1 Q~ -¢04 4) = '_ CS e- _ +1 +1 '4 _O a 0 .~~~~~1 . -o _ 3 +1 +1 C) 4- . t to xt = c3 ._~~~~4 4)d 0E - 020 C) xo = 4O - CO 02 *<Ll _ S~~4 > ._~~~~4 4) 14 C.) 2 eq CO H 1 CN _ o +1 +1 - eq 44 Eq .J.D 4-1w~ ~ ~ ~ 1 o 2Eq VV o o 0 4 P 9 P1 OD 0 0 02 02 N 4 4 0 0 0 S Z V ;sV tV4 vV 4- v Downloaded from J Physiol (jp.physoc.org) by guest on July 10, 2011 239 240 A. D. GRINNELL AND OTHERS channel openings recorded when the pipette was filled with the 2 h incubation solution averaged 11P3± 16 (mean+S.E.M., n=4), which corresponds to a concentration of approximately 6 nM-ACh (see Fig. 1). This concentration of ACh in a volume of 1-5 ml equals about 9 pmol of ACh. Although this value is about twice that detected by direct chemical measurement, it is within the range of variability observed. A Ch transporter On the basis of the experiments on the mammalian diaphragm, Edwards et al. (1985) suggested that the non-quantal leak of ACh is mediated by vesicle membrane transport proteins that are inserted into the plasma membrane by vesicle fusion during transmitter release. The problem in testing this hypothesis is to distinguish non-quantal ACh release from the accumulation of ACh released quantally. Several methods have been tried: if non-quantal release were due to the incorporation of ACh transporting molecules into the terminal membrane by vesicular exocytosis, one would predict that the frequency of single-channel openings that follow a nerve-evoked quantal event would be proportional to the size of that event (the number of locally released quanta). A plot of the frequency of single-channel openings between 100 and 1500 ms following the stimulus shows that there is no apparent correlation between this value and the size of the initial summed response (Fig. 8A). Secondly, the decay in single-channel frequency following a locally released quantum can be compared with the behaviour following the failure of the releasing sites near the patch probe to release a quantum. At a stimulus frequency of 0 75 Hz, and in a low-Ca2" Ringer solution in which the probability of release near the patch probe was lowered, the number of channel openings associated with nearby vesicular release (fast-rising summed channel openings) was higher than that observed following a 'failure' (no summed response detected at the probe) only during the first 100 ms following the stimulus (Fig. 8B). There was no significant difference in the frequency of singlechannel openings between 100 and 1500 ms following the stimulus (Fig. 8B). This result, coupled with the absence of a correlation between evoked response amplitudes and spontaneous openings, suggests that ACh released locally by vesicle fusion has diffused away to such an extent (within 100 ms) that it is not as significant as the cloud of ACh diffusing into the area from many other quantal release sites along the length of the terminal. One might expect that with repetitive stimulation, ACh transporter molecules would build up in the plasma membrane, resulting in a higher post-stimulus frequency of single-channel openings. In a test of this prediction, the time course of decay in frequency of single-channel openings was found to be essentially the same after a series of five or of twenty stimuli at 0 75 Hz (Fig. 9). This suggests that, at this frequency of release, either the transporters inserted into the membrane do not significantly increase the amount of ACh present near the terminal, or that the transporters are being recovered from the membrane as fast as they are being inserted. In contrast, after stimulating at 20 Hz, there were many more openings and the rate declined only gradually over a period of 5-8 s to the levels seen after the Downloaded from J Physiol (jp.physoc.org) by guest on July 10, 2011 ACh RELEASE FROM EXPOSED NERVE TERMINALS 241 A 120. 100 0. . E0) . . 80 6 60 0 X w40 * S 20. * 0 8 12 16 20 24 28 32 36 40 44 48 Number of single-channel openings post-stimulus 4 B CL S , 3.0 2.0 ._ o * 0 L. 1.5 .0 E : 10I 1.0 0.5 - 0.0. 0 200 400 800 1000 600 Time post-stimulus (ms) 1200 1400 Fig. 8. The transient single-channel frequency increase observed following single nerve stimuli does not correlate with the size of EPCs originating near the patch probe. A, a plot of EPC size v8. the number of single-channel openings observed following the EPC reveals a lack of correlation. B, when a plot is constructed of the mean (± standard error bars, when larger than the symbol size) number of channels that open at 100 ms intervals following either a local quantal event (@) or a failure (0) locally, only the first 100 ms time period shows a significant (*P < 0005) increase with local quantal release. The subsequent decay in single-channel frequency describes a curve which probably represents diffusion of evoked ACh release from a larger terminal environment than the active zone region from which the EPCs originate. same number of stimuli at the lower frequency (Fig. 9). This higher rate could have been due to a transient build up of ACh transporters in the membrane. On the other hand, the massive vesicular release of ACh (undoubtably facilitated at this frequency) would result in a large cloud of ACh diffusing away from the terminal, and an increase in the frequency of spontaneous quantal events (MEPPs) from the terminal (Lev-Tov & Rahamimoff, 1980), thereby obscuring the measurement of any leak. Downloaded from J Physiol (jp.physoc.org) by guest on July 10, 2011 A. D. GRINNELL AND OTHERS 242 In order to assess the potential contribution of non-quantal release to the posttetanic increase of ACh, measurements were done under conditions thought to modify non-quantal release. Data in Fig. 10 illustrate the decay in frequency of single-channel openings observed under different conditions at increasing time after a 50 Hz tetanus in two representative preparations. 0 80 70 Co L 0) a 60 0 30 C m 0 20 en O 10 0 0 2 0 0 2 2 i O° 000 ° Q 4 6 8 Time post-stimulus (s) ;0io0 10 12 Fig. 9. Transient increase in single-channel opening frequency following nerve stimulation at 0-75 and 20 Hz. The decay in single-channel frequency following five (A) or twenty (LiO) nerve stimuli at 075 Hz is similar. Twenty nerve stimuli at 20 Hz (0) increases the number of single-channel openings following the stimulus. None of the experimental conditions noticeably affected the decay in the frequency of single-channel events observed following the tetanus. The conditions were as follows: (1) application of vesamicol. Vesamicol is a blocker of the vesicle membrane transport protein that is responsible for loading ACh into synaptic vesicles (Andersen et al. 1983), and that has been implicated in the mediation of nonquantal ACh leakage (Edwards et al. 1985; see Introduction). Vesamicol, at a concentration (1-7 /LM) which blocks non-quantal release in mammals (Edwards et al. 1985), and which blocks a portion of the spontaneous ACh release measured biochemically in the frog cutaneous pectoris preparation (see above), had no obvious effect on the time course of decline in single-channel opening frequency following a tetanus (see Fig. lOA). Although the absolute number of openings following a tetanus in vesamicol tended to be lower than in control (vesamicol reduces the ACh content of vesicles), the time constants of decay in the frequency of openings might be expected to differ if non-quantal release contributes significantly to the total amount of ACh present. In three experiments testing this prediction, the time constants were not significantly different (mean + S.D. equalled 6-8 + 1f6 s in control vs. 74 + 0-8 s in vesamicol; P» 005 by Student's t test). (2) Increased calcium levels. An increased level of calcium in the Ringer solution bathing the preparation has been reported to selectively block the non-quantal leakage of transmitter (Edwards et al. 1985; Sun & Poo, 1985). In our experiments, Downloaded from J Physiol (jp.physoc.org) by guest on July 10, 2011 ACh RELEASE FROM EXPOSED NERVE TERMINALS 243 10 mM-calcium did not have any significant effect (P > 0 05 by one-way analysis of variance using Tukey's test) on the decay curves of single-channel frequency (Fig. lOB). The mean time constant (± S.D.) from three preparations was 6-0 + 1 1 s in control vs. 7-5 + 29 s in 10 mM-calcium. If calcium was blocking a significant nonA 80 C -0c cn Cu C 0 o0 4~ 0 -* OO 0 % 20 L 0 10 0 20 Time B 30 40 50 post-stimulus (s) 100- m) C & E(D 80 . 0a- uln 660 -. - o C ° E 400) *0 A o cr 0s 20 0 10 40 30 20 Time post-stimulus (s) 50 Fig. 10. The decay in single-channel opening frequency following a 50 Hz nerve stimulus. A, application of 1-7 m-vesamicol (O) does not alter the normal decay (M; values multiplied by 0 9 to correct for decreased release with vesamicol treatment during the tetanus) in single-channel opening frequency. B, application of 10 mm-Ca21 (M), or movement of the probe 4 ,um away from the terminal (A) does not change the decay curve observed in normal Ringer solution at a close terminal position (O ) quantal leak, the predicted time constant would be shorter than control. (3) Lastly, if local non-quantal leakage contributes significantly to the frequency of single channels observed, moving the probe away from the active region of the terminal (about 4 ,um), which eliminates summed current responses to evoked release, should also significantly reduce detection of any local non-quantal leakage. However, Downloaded from J Physiol (jp.physoc.org) by guest on July 10, 2011 244 A. D. GRINNELL AND OTHERS removing the probe 4 ,um from the synaptic surface of the terminal did not significantly change the time constant of decay of single-channel opening frequency following the end of the stimulus (Fig. lOB). The mean time constant from three experiments was 6-0 + 1 1 s in control vs. 6-6 + 2 7 s at 4 ,um distance (P > 0 05 by oneway analysis of variance using Tukey's test). In general, none of the manipulations thought to influence non-quantal ACh release had a significant effect on the time constant of decay in single-channel opening frequency after the tetanus. DISCUSSION Using the patch probe, nerve-evoked quantal release of ACh could be detected at most pipette positions along the terminal branches of identified, clearly visible terminals. However, adjustments of probe position were often necessary to optimize a given response, and in some positions it was not possible to obtain large responses (see Fig. 4). We suspect that failures to record significant release were most probably the result of intervening connective tissue. This is supported by scanning electron micrographs of enzyme-treated preparations (Plate 1). At most initial pipette positions, nerve stimulation only occasionally evoked the release of a single quantum of ACh in close enough proximity to the patch that a fastrising EPC was measured, although there was commonly a slow barrage of channel openings indicative of quantal release at greater distances from the probe. At good recording positions (usually obtained by pipette manipulation), grouping of responses by amplitude and latency suggests that these release events occur from discrete spots, presumably active zones. The range of evoked patch current amplitudes observed at any given latency might be explained by differing quantal sizes, by release from different portions of an active zone (closer or further from the probe), or by release of different numbers of quanta. We saw only occasional instances in which an EPC amplitude was twice the average MPC size, however, and consider it probable that a given active zone normally releases one quantum or none. An average terminal has about 500 active zones, and release is widespread along the length of the terminal (Katz & Miledi, 1965; D'Alonzo & Grinnell, 1985). Since full quantal content is only about 100 quanta, one would not expect more than one vesicle of transmitter to be liberated from any particular active zone, except in rare instances. Given ACh sensing probes with single-channel resolution, we expected to be able to detect and study the properties of non-quantal leakage of ACh from motor nerve terminals. Based on the observations of Katz & Miledi (1977) and the sensitivity of the probes (Fig. 1), an average of 18 channel openings min-' would be predicted at probe positions close to quantal release sites in the terminal. The observed frequency of single-channel opening (1-6+005 min-1) was more than 10 times lower than expected. Indeed, this value is not significantly different from that recorded with the probe far removed from the muscle in the bath (0 7 +0 3 min-1). From the calibration curve of Fig. 1, the frequency of openings observed close to quantal release sites (1-6 min-') would correspond to a local ACh concentration in the probe environment of Downloaded from J Physiol (jp.physoc.org) by guest on July 10, 2011 ACh RELEASE FROM EXPOSED NERVE TERMINALS 245 about 1 nM. However, a mean difference of only a single ACh receptor channel opening per minute between terminal and bath pipette locations is too small to be considered meaningful. Calculating predicted levels of non-quantal release It is possible that we have incorrectly estimated the frequency of single-channel openings expected in the patch probe due to nerve terminal release of non-quantal ACh, but we consider this to be unlikely, since we can calculate the expected frequency by two additional methods and still anticipate detection by the patch probe. Most of our computations are based on previous measurements of non-quantal release in the frog sartorius (Katz & Miledi, 1977; Miledi et al. 1980). Our direct chemical measurements of the resting efflux from the sartorius (4-2 + 0 5 pmol (2 h)-', n = 25; mean + s.E.M.) and cutaneous pectoris preparations (4-1 + 0-6 pmol (2 h)-1, n = 39), indicate that the rate of spontaneous ACh release is nearly identical from both (see Table 1). However, since the sartorius has more than twice as many nerve terminals as the cutaneous pectoris (of essentially the same average size; A. A. Herrera and A. D. Grinnell, unpublished observations), estimates of release per terminal for the cutaneous pectoris that are based on previously reported work in the sartorius are conservative. Katz & Miledi (1977) estimated that the average 40 1tV H-effect would correspond to an average endplate current of 133 pA. Using this average current as a reference, the non-quantal leakage of ACh should be clearly detectable as a background barrage of single-channel openings in the patch probe, based on the following considerations: EPC measurements suggest that the patch probes recorded primarily from three active zones, thus, we assume that the probe was monitoring non-quantal release from the portion of the terminal corresponding to about 0-6 % of the terminal. Assuming that, on average, the junctional receptors at normal resting potential conduct about 2-5 pA of current for approximately 1 ms (Anderson & Stevens, 1973), 0-6 % of the 133 pA average whole terminal current equals 0-8 pA, which corresponds to 19200 single-channel openings every minute. Thus if one had a sensor as close to the release sites, as extensive in size, and as sensitive to ACh as the postsynaptic junctional membrane, one would expect to see approximately 19000 single-channel openings per minute. The patch probe does not fulfill any of these criteria; it is probably located 2-10 times further from the release sites and is a much smaller patch of membrane with a lower density of ACh receptors. However, all of the factors that affect signal size can be taken into account by comparing the size of MEPCs recorded at an adult synapse (which corresponds to about 4000 total channel openings, or 1800 simultaneous openings at the peak, based on the data of Gage & McBurney (1972) and Glavinovic (1986)) with the size of the average MPC recorded by the patch probes (an average of twenty-two total openings, or ten simultaneous openings at the peak). This signal loss can be represented by a capture ratio equal to the patch MPC amplitude divided by the normal synaptic MEPC amplitude; using either total channel openings (22 divided by 4000) or peak channel openings (10 divided by 1800), the same capture ratio (0-006) is obtained. Since embryonic receptors in the patch probe have a longer mean open time and larger conductance, Downloaded from J Physiol (jp.physoc.org) by guest on July 10, 2011 246 A. D. GRINNELL AND OTHERS it is preferable to convert current estimates into frequency of openings. In order to estimate the frequency of single-channel openings observed with a patch probe, the frequency expected at the adult junction can be multiplied by the capture ratio (19000 x 0 006), yielding a prediction of about 115 openings min-'. This estimate is higher than our earlier prediction (18 min-'), probably because it assumes that the synaptic cleft does not concentrate the ACh that leaks out in nonquantal fashion. This factor is difficult to estimate, but one can avoid these assumptions by using the conclusions of biochemical measurements to derive a flux rate for the Carslaw & Jaeger (1947) linear source flux equation. Based on biochemical assays, the non-quantal flux rate is approximately 100 times the spontaneous quantal release (see Results; Katz & Miledi, 1977; Miledi et al. 1980; Molenaar et al. 1987). Approximately one quantum is released per second from an average resting frog nerve terminal. Therefore, if the number of ACh molecules in a quantum (about 7500; Kuffler & Yoshikami, 1975b) is multiplied by 100, and then divided by the number of micrometres in a terminal (about 500) and Avogadro's number, the number of moles of ACh released non-quantally per micrometre per second from the nerve terminal can be derived. This number can now be used as the flux rate in the linear source equation given by Carslaw & Jaeger (1947). Using this method of estimating, one predicts a concentration of 4-3 nm at a distance of 2 ,m from the terminal. This concentration of ACh should open an average of 5 channels min-' in our patch probe (see Fig. 1). This estimate is probably a conservative one since it assumes 'free' diffusion, and we know that the terminal is not completely cleaned of connective tissue (see Plate 1). However, even this conservative estimate is more than 3 times higher than the frequency of single channels which opened when the patch probe was positioned near nerve terminal active zones. Using any of the estimations presented above, therefore, we conclude that if nonquantal ACh were predominantly released from the synaptic surface of the nerve terminal, we should have been able to detect a higher frequency of openings than was observed with our patch probe. Alternative explanations for inability to detect non-quantal ACh leak Non-quantal release may be highly localized. If the leakage of ACh is episodic and randomly distributed at only a few locations on the terminal arbor, it is possible that records were never obtained from an appropriate position. However, this is unlikely since such highly localized episodic leak behaviour would need to be locally very large. The probability of observing such an episode would be larger (about 100-fold, based on biochemical measurements) than the probability of observing MPCs, which occasionally were observed in the absence of any enhancement of spontaneous release frequency. We collected data from a total of ninety-six positions on eighteen cutaneous pectoris preparations, and sampled many more positions during pipette manipulations, and never observed a prolonged episode of high-frequency singlechannel openings. Also, it is possible that transmitter leakage is not concentrated at active zone regions of the frog nerve terminal. Katz & Miledi (1977) have reported that the Heffect in response to curare application did not differ significantly between normal muscle and preparations that had been denervated for 19-36 days. Biochemical Downloaded from J Physiol (jp.physoc.org) by guest on July 10, 2011 ACh RELEASE FROM EXPOSED NERVE TERMINALS 247 measurements indicate a comparable lack of effect of denervation on ACh leakage in the frog cutaneous pectoris muscle (see Table 1). Two weeks after denervation of the cutaneous pectoris muscle, the ratio of ACh efflux from denervated to contralateral control preparations was 1-13 + 0-13 (mean+ S.E.M.; n = 25), whilst the ratio of denervated to control whole tissue ACh content reflected the loss of the nerve terminals (079 + 006; n = 10). These data suggest that denervated frog nervemuscle preparations continue to leak ACh despite the loss of the nerve terminals. Following denervation, the Schwann cells that remain following the loss of nerve terminals are known to contain substantial quantities of cytoplasmic ACh, and to release ACh in a quantal fashion (Dennis & Miledi, 1974). It is possible that nonquantal release of transmitter at the intact neuromuscular junction originates from a localized source other than the under surface of the nerve terminal (Schwann cell, nodes of Ranvier, muscle). It could also be diffusely liberated from many sources in the environment, which might, or might not, include the nerve terminal. If any of these possibilities are true, high levels of non-quantal release would not be recorded by probes located specifically at the nerve terminal active zone region. The strongest evidence to date for local terminal leakage of ACh in frogs comes from Katz & Miledi (1977). As they pointed out, however, it is possible that the H-effect they observed was due to ACh that originated from another source and accumulated in the bath to a level that would locally activate junctional ACh receptors at the site of curare application. Katz & Miledi (1981) also reported that depolarization by repetitive motor nerve impulses does not detectably increase nonquantal release recorded at the endplate, as would be expected if it were driven by an electrochemical gradient through leaky pores in the presynaptic membrane. They cite these results as evidence that ACh leakage may occur from other sources in the environment. Interestingly, Edwards et al. (1985) have studied a similar curareinduced hyperpolarization in the rat diaphragm, and report that the effect was greater when a smaller volume recording chamber was used without perfusion, than when a larger volume chamber was used. If ACh is coming from any cell type located within the connective tissue capsule covering the junction, it seems likely that ACh could accumulate sufficiently in the cleft region, in the absence of acetylcholinesterase activity, to produce the small hyperpolarization evident with curare application in the endplate region. It is also possible that curare application had a direct effect on muscle membrane potential, contributing to the H-effect. Sun & Poo (1985) have described a much larger H-effect (requiring very high concentrations of curare) at newly formed Xenopus nerve-muscle synapses in vitro. We have found no significant non-quantal release in the same culture system using patch clamp techniques under conditions where single ACh channel openings are detectable (S. H. Young and A. D. Grinnell, unpublished observations), and have no explanation for their observations. Recently, Molenaar & Polak (1987) have suggested that the high level of ACh leakage previously reported may be the result of the use of membrane permeant acetylcholinesterase blockers. These blockers may enter the presynaptic terminal and block cytoplasmic acetylcholinesterase, increasing the cytoplasmic ACh content, thereby increasing the ACh leakage. Since our patch probe experiments were generally done without the use of any anticholinesterases (collagenase removed the Downloaded from J Physiol (jp.physoc.org) by guest on July 10, 2011 248 A. D. GRINNELL AND OTHERS requirement for this treatment by removing the esterases), we added eserine to the bath in some experiments. There was, however, no effect of exogenous eserine on the single-channel frequency at rest, or following a 50 Hz nerve stimulation. The ACh transporter modelfor non-quantal ACh leak Our evidence does not support the hypothesis that the ACh leak derives largely from the ACh transporter molecules that are incorporated into the terminal membrane during vesicle exocytosis and which then pump cytoplasmic ACh out of the terminal. If this hypothesis were correct, one would expect each vesicle fusion event to insert enough carriers to transport about 106 ACh molecules (assuming 104 ACh molecules per quantal event and 100 times more non-quantal than quantal ACh release (see Introduction)). The local concentration of ACh that would result from this predicted non-quantal leakage would depend on the length of time the transporters remained incorporated and functional in the plasma membrane. If this time is less than 100 ms, one would expect to observe a large extended depolarization 'tail' in the postsynaptic muscle cell following each quantal event. This has never been seen. The absence of this phenomenon is also clearly demonstrated by the clean single-exponential, decay of good voltage clamp measurements of MEPCs (Gage & McBurney, 1972). Alternatively, if the transporter remains in the plasma membrane for a significant length of time after the vesicular ACh has diffused away, one would expect the leak to be increased in proportion to the amount of quantal release. There is little information available concerning the recycling time of vesicular membrane components in the plasma membrane following vesicle fusion. Heuser & Reese (1973, 1981) observed a dispersion of large vesicular particles in the presynaptic membrane away from active zones 1 s after the termination of nerve stimulation, and the re-aggregation of those particles over presumptive endocytotic-coated pits in subsequent seconds. They also described a secondary form of endocytosis evident during periods of intense evoked release that appears to randomly invaginate larger portions of the presynaptic membrane. During a prolonged stimulation (1 Hz for min), both of these processes apparently function to some degree and result in a maintained enlargement (approximately 20 %) of the presynaptic membrane area, with an 80% recycling of membrane into the terminal (Heuser & Reese, 1973). In addition, Miller & Heuser (1984) have followed the time course of endocytotic retrieval of the vesicular membrane which contains large intramembranous particles. Following a single nerve stimulus under conditions that promote the release of 3000-6000 vesicles from a single nerve terminal, they observed the first selective vesicular membrane endocytosis at 1 s, the peak of endocytosis at 30 s, and the completion of this process by 90 s. Neher & Marty (1982) have directly measured membrane capacitance changes during mammalian chromaffin vesicle exocytosis and endocytosis. Based on these observations, it appears that chromaffin vesicle membrane recycling takes about 5-10 s. If tetanic stimulation were to leave a significant fraction of the vesicular ACh transporter in the presynaptic membrane as long as 5-10 s after the tetanus when the immediate effects of evoked release have dissipated (see Fig. 10), and if the time course of vesicle recycling described by Miller & Heuser (1984) correlates with the presence of vesicular ACh transporting proteins, a massive leakage of ACh would be expected due to the persistence of an increased number of ACh transporting molecules inserted into the plasma membrane. No such Downloaded from J Physiol (jp.physoc.org) by guest on July 10, 2011 ACh RELEASE FROM EXPOSED NERVE TERMINALS 249 component of ACh release could be distinguished from the accumulation due simply to quantal release, nor did agents such as vesamicol or high calcium, which are reported to inhibit transporter function, have any detectable effect on ACh accumulation and dispersion from terminal regions. It should be pointed out, however, that the ACh leak could still be mediated by ACh transporter molecules in the terminal plasma membrane if they are present at sufficient density that the addition of more transporter molecules due to vesicle exocytosis does not significantly change the amount of transport. In view of our inability to detect a significant leak in the resting condition, this seems unlikely; but the fact that vesamicol does significantly reduce the ACh leakage from the frog neuromuscular preparation suggests that the transporters, or some other vesamicol-sensitive protein, are involved in some way. Our results suggest that non-quantal ACh release from frog nerve-muscle preparations is not localized to the synaptic surface of the terminal that closely apposes the postsynaptic muscle cell. There is convincing biochemical evidence that ACh is released in non-quantal fashion from the nerve-muscle preparation, but we cannot detect it near nerve terminals even during periods of enhanced vesicular exocytosis. This suggests that if the transporters are producing some of the leak, they are doing so at a level too low to be detected, or they may be randomly inserted into the presynaptic membrane (as well as into the vesicular membrane) and not be predominantly associated with quantal transmitter release. Since all previous measurements of non-quantal release have been done with an inhibitor of acetylcholinesterase in the bath, the source of the ACh is not known. Under physiological conditions, in the presence of acetylcholinesterase, non-quantal ACh release may not represent a significant fraction of the total ACh released from the synaptic surface of the frog nerve terminal. Our observations put constraints on hypotheses which propose a mechanism for this form of release, and a trophic function of non-quantal ACh release at the frog nerve-muscle synapse. We thank Professor T. P. Feng for enthusiastic participation in the early stages of this work, and F. Knight, C. Haubach and M. Akers for technical assistance. S. Hulsizer and Drs I. Chow, Y. Harada, M. Jackson and E. Serrano provided insightful discussions and critical review of the manuscript. Supported by grants from the MDA, NSF (BNS 8719613) and NIH (HD 24522; NS 06232; MH17691; NS00827). REFERENCES ANDERSEN, D. C., KING, S. C. & PARSONS, S. M. (1983). Pharmacological characterization of the acetylcholine transport system in purified Torpedo electric organ synaptic vesicles. Molecular Pharmacology 24, 48-54. ANDERSON, C. R. & STEVENS, C. F. (1973). Voltage clamp analysis of acetylcholine produced endplate current fluctuations at frog neuromuscular junction. Journal of Physiology 235, 655-691. ANDERSON, M. J., COHEN, M. W. & ZORYCHTA, E. (1977). Effects of innervation on the distribution of acetylcholine receptors on cultured muscle cells. Journal of Physiology 268, 731-756. BETZ, W. & SAKMANN, B. (1973). Effects of proteolytic enzymes on function and structure of frog neuromuscular junctions. Journal of Physiology 230, 673-688. BRAY, J. J., FORREST, J. W. & HUBBARD, J. I. (1982). Evidence for the role of non-quantal acetylcholine in the maintenance of the membrane potential of rat skeletal muscle. Journal of Physiology 326, 285-296. CARSLAW, H. S. & JAEGER, J. C. (1947). Conduction of Heat in Solids. Oxford: Clarendon Press. Downloaded from J Physiol (jp.physoc.org) by guest on July 10, 2011 250 A. D. GRINNELL AND OTHERS D'ALONZO, A. J. & GRINNELL, A. D. (1985). Profiles of evoked release along the length of frog motor nerve terminals. Journal ofPhysiology 359, 235-258. DENNIS, M. J. & MILEDI, R. (1974). Electrically induced release of acetylcholine from denervated Schwann cells. Journal of Physiology 237, 431-452. DESAKI, J. & UEHARA, Y. (1981). The overall morphology of neuromuscular junctions as revealed by scanning electron microscopy. Journal ofNeurocytology 10, 101-1 10. DRACHMAN, D. B., STANLEY, E. F., PESTRONK, A., GRIFFIN, J. W. & PRICE, D. L. (1982). Neurotrophic regulation of two properties of skeletal muscle by impulse-dependent and spontaneous acetylcholine transmission. Journal of Neuroscience 2, 232-243. EDWARDS, C., DOLEZAL, V., TUCEK, S., ZEMKOVA, H. & VYSKOCIL, F. (1985). Is acetylcholine transport system responsible for non-quantal release of acetylcholine at the rodent myoneural junction? Proceedings of the National Academy of Sciences of the USA 82, 3514-3518. EDWARDS, C., DOLEZAL, V., TUCEK, S., ZEMKOVA, H. & VYSKOCIL, F. (1988). A possible role for the acetylcholine transport system in non-quantal release of acetylcholine at the rodent myoneural junction. Puerto Rico Health Science Journal 7, 71-74. GAGE, P. W. & MCBURNEY, R. N. (1972). Miniature end-plate currents and potentials generated by quanta of acetylcholine in glycerol-treated toad sartorius fibres. Journal of Physiology 226, 79-94. GLAVINOVIC, M. I. (1986). Variability of quantal events in control solution and after cholinesterase blockade in frog. Neuroscience 17, 519-526. HEUSER, J. E. & REESE, T. S. (1973). Evidence for recycling of synaptic vesicle membrane during transmitter release at the frog neuromuscular junction. Journal of Cell Biology 57, 315-344. HEUSER, J. E. & REESE, T. S. (1981). Structural changes after transmitter release at the frog neuromuscular junction. Journal of Cell Biology 88, 564-580. HUME, R. I., ROLE, L. W. & FISCHBACH, G. D. (1983). Acetylcholine release from growth cones detected with patches of acetylcholine receptor-rich membranes. Nature 305, 632-634. JENDEN, D. J., RoCH, M. & BOOTH, R. A. (1973). Simultaneous measurement of endogenous and deuterium labeled tracer variants of choline and acetylcholine in subpicomole quantities by gas chromatography/mass spectrometry. Analytical Biochemistry 55, 438-448. KATZ, B. & MILEDI, R. (1965). Propagation of electrical activity in motor nerve terminals Proceedings of the Royal Society B 161, 453-482. KATZ, B. & MILEDI, R. (1977). Transmitter leakage from motor nerve endings. Proceedings of the Royal Society B 196, 59-72. KATZ, B. & MILEDI, R. (1981). Does the motor nerve impulse evoke 'non-quantal' transmitter release? Proceedings of the Royal Society B 212, 131-137. KIDOKORO, Y. & GREUNER, R. (1982). Distribution and density of a-bungarotoxin binding sites on innervated and noninnervated Xenopus muscle cells in culture. Developmental Biology 91, 78-85. Ko, C.-P. (1987). A lectin, peanut agglutinin, as a probe for the extracellular matrix in living neuromuscular junctions. Journal ofNeurocytology 16, 567-576. KUFFLER, S. W. & YOSHIKAMI, D. (1975a). The distribution of acetylcholine sensitivity at the postsynaptic membrane of vertebrate skeletal twitch muscles: iontophoretic mapping in the micron range. Journal of Physiology 244, 703-730. KUFFLER, S. W. & YOSHIKAMI, D. (1975b). The number of transmitter molecules in a quantum: an estimate from iontophoretic application of acetylcholine at the neuromuscular synapse. Journal ofPhysiology 251, 465-482. LETINSKY, M. S. & DECINO, P. (1980). Histological staining of pre- and postsynaptic components of amphibian neuromuscular junctions. Journal ofNeurocytology 9, 303-320. LEv-Tov, A. & RAHAMIMOFF, R. (1980). A study of tetanic and post-tetanic potentiation of miniature end-plate potentials at the frog neuromuscular junction. Journal of Physiology 309, 247-273. MATHERS, D. A. & THESLEFF, S. (1978). Studies on neurotrophic regulation of murine skeletal muscle. Journal of Physiology 282, 105-114. MILEDI, R., MOLENAAR, P. C. & POLAK, R. L. (1980). The effect of lanthanum ions on acetylcholine in frog muscle. Journal ofPhysiology 309, 199-214. MILLER, T. M. & HEUSER, J. E. (1984). Endocytosis of synaptic vesicle membrane at the frog neuromuscular junction. Journal of Cell Biology 98, 685-698. MITCHELL, J. F. & SILVER, A. (1963). The spontaneous release of acetylcholine from the denervated hemidiaphragm of the rat. Journal ofPhysiology 165, 117-129. Downloaded from J Physiol (jp.physoc.org) by guest on July 10, 2011 The Journal of Physiology, Vol. 419 A. D. GRINNELL AND OTHERS Downloaded from J Physiol (jp.physoc.org) by guest on July 10, 2011 Plate I (Facing p. 251) ACh RELEASE FROM EXPOSED NERVE TERMINALS 251 MOLENAAR, P. C., OEN, B. S., POLAK, R. L. & VAN DER LAAKEN, A. L. (1987). Surplus acetylcholine and acetylcholine release in the rat diaphragm. Journal of Physiology 385, 147-167. MOLENAAR, P. C. & POLAK, R. L. (1987). Is the non-quantal resting release of ACh from isolated rat diaphragm an artifact due to cholinesterase inhibition? In Cellular and Molecular Basis of Cholinergic Function, Ellis Horwood Series in Biomedicine, ed. DOWDALL, M. J. & HAWTHORNE, J. N., pp. 221-225. Chichester: Ellis Horwood. NEHER, E. & MARTY, A. (1982). Discrete changes of cell membrane capacitance observed under conditions of enhanced secretion in bovine adrenal chromaffin cells. Proceedings of the National Academy of Sciences of the USA 79, 6712-6716. NIEUWKOOP, P. D. & FABER, J. (1967). Normal Table of Xenopus laevis (Daudin), 2nd edn. Amsterdam: Elsevier/North-Holland Publishing Co. OGSTON, A. G. (1955). Removal of acetylcholine from a limited volume by diffusion. Journal of Physiology 128, 222-223. PESTRONK, A., DRACHMAN, D. B., STANLEY, E. F., PRICE, D. L. & GRIFFIN, J. W. (1980). Cholinergic transmission regulates extrajunctional acetylcholine receptors. Experimental Neurology 70, 690-696. SAKMANN, B. & NEHER, E. (1983). Geometric parameters of pipettes and membrane patches. In Single Channel Recording, ch. 2, pp. 37-51. New York: Plenum Press. SPITZER, N. C. & LAMBORGHINI, J. E. (1976). The development of the action potential mechanism of amphibian neurons isolated in culture. Proceedings of the National Academy of Sciences of the USA 73, 1641-1645. STANLEY, E. F. & DRACHMAN, D. B. (1986). The effects of nerve section on non-quantal release of ACh from the motor nerve terminal. Brain Research 365, 289-292. STRAUGHAN, D. W. (1960). The release of acetylcholine from mammalian motor nerve endings. British Journal of Pharmacology 15, 417-424. SUN, Y.-A. & POO, M.-M. (1985). Non-quantal release of acetylcholine at a developing neuromuscular synapse in culture. Journal of Neuroscience 5, 634-642. VYSKO0IL, F. & ILLES, P. (1977). Non-quantal release of transmitter at mouse neuromuscular junction and its dependence on the activity of Na+-K+ ATPase. Pflugers Archiv 370, 295-297. VYSKOIL, F., NIKOLSKY, E. & EDWARDS, C. (1983). An analysis of the mechanisms underlying the non-quantal release of acetylcholine at the mouse neuromuscular junction. Neuroscience 9, 429-435. YOUNG, S. H. & Poo, M.-M. (1983). Spontaneous release of transmitter from growth cone of embryonic neurone. Nature 305, 634-637. EXPLANATION OF PLATE Scanning electron micrographs of frog nerve terminals. A, a nerve-muscle preparation before treatment with collagenase or protease is heavily ensheathed with fibrous connective tissue which obscures the view of any nerve terminals, and makes the detection of ACh release with a patch probe impossible. B, enzyme treatment removes a significant portion of the connective tissue (seen as fibrous strands) such that patch electrodes (P) can be brought into close proximity to the nerve terminal (arrow-heads). C, further digestion of the connective tissue around a fixed nerve-muscle preparation (with HCl) reveals details of the pre- and postsynaptic structure. Calibration = 20 ,sm. Downloaded from J Physiol (jp.physoc.org) by guest on July 10, 2011