Effect of Alpha-Latrotoxin on the Frog Neuromuscular Junction at Low

Home , Axolemma, Bungarotoxin, Latrotoxin

Journal of Physiology (1988), 402, pp. 195-217 195 With 4 plate8 and 5 text-figures Printed in Great Britain

EFFECT OF ae-LATROTOXIN ON THE FROG NEUROMUSCULAR JUNCTION AT LOW TEMPERATURE BY B. CECCARELLI, W. P. HURLBUT* AND N. IEZZI From the Department of Medical Pharmacology, CNR Center of Cytopharmacology and Center for the Study of Peripheral Neuropathies and Neuromuscular Diseases, Via Vanvitelli 32, 20129 Milano, Italy, and the Rockefeller University*, 1230 York Avenue, New York, NY 10021, U.S.A. (Received 13 July 1987)

SUMMARY 1. a-Latrotoxin (a-LTx) was applied to frog cutaneus pectoris muscles bathed at 1-3 °C in either Ringer solution, Ca2"-free Ringer solution with 1 mM-EGTA and 4 mM-Mg2+ or Ringer solution plus 4 mM-Mg2+, and its effects on miniature end-plate potential (MEPP) frequency, nerve terminal ultrastructure and uptake of horseradish peroxidase (HRP) were studied. 2. Large concentrations (2 ,ug/ml) of a-LTx increased MEPP rates to levels above 100/s at all junctions, but the time course of the increases depended upon the divalent cation content of the bathing solution. However, similar numbers ofMEPPs (0-3-407 x 106) were recorded at all junctions during 2 h of secretion. 3. Nerve terminals exposed to a-LTx for 2 h lost 60-75 % oftheir synaptic vesicles and were swollen; their presynaptic membranes were deeply infolded and they often contained many large vesicular structures. Terminals in Ringer solution retained the largest number of synaptic vesicles; terminals in Ringer solution plus Mg2+ swelled the least and contained the largest number of coated vesicles. The average number of synaptic vesicles lost was approximately equal to the average number of MEPPs recorded. 4. Few vesicles became loaded with HRP when this extracellular tracer was present in the bathing solution and the muscles were fixed near the peak of secretion. 5. When the terminals were warmed to 20 °C, those in the Caa2+-free solution with Mg2+ secreted additional quanta and lost almost all their residual vesicles; those in Ringer solution without Mg2+ secreted few additional quanta and retained most of their residual vesicles. 6. These results suggest that recycling was blocked at these terminals and that for each quantum secreted a vesicle became permanently incorporated into the axolemma.

INTRODUCTION The acetylcholine (ACh) in vertebrate motor nerve terminals is stored in two major compartments: the axoplasm and the synaptic vesicles (Whittaker, Michaelson & Kirkland, 1964; Dunant, Gautron, Israel, Lesbats & Manaranche, 1972). ACh is

7-2 196 B. CECCARELLI, W. P. HURLBUT AND N. IEZZI released from a terminal both in a steady 'leak' (Katz & Miledi, 1977; Vyscocil & Illes, 1979; Edwards, Dolezal, Tucek, Zemkova & Vyskocil, 1985), and in a pulsatile manner as multimolecular packets, or quanta (del Castillo & Katz, 1954). The total release, i.e. the sum of the 'leak' and the quantal components, is measurable by biochemical means (Gorio, Hurlbut & Ceccarelli, 1978; Israel & Lesbats, 1981). The 'leak', or non-quantal component of release, is difficult to measure directly since its major known effect is to produce a small, steady depolarization at an end-plate when acetylcholinesterase has been thoroughly inhibited. Therefore, neither its function or origin within the terminal are known (Edwards et al. 1985). Quantal release, on the other hand, is easily detected by electrophysiological means, and its function is well known since it generates the transient end-plate potentials (EPPs) and miniature end-plate potentials (MEPPs) that mediate neuromuscular transmission (Fatt & Katz, 1952; del Castillo & Katz, 1954; Boyd & Martin, 1956). Although recent quick-freeze studies of frog terminals treated with 4-amino- pyridine have demonstrated that vesicles fuse with the presynaptic membrane a few milliseconds after the arrival of a single nerve impulse (Heuser, Reese, Dennis, Jan, Jan & Evans, 1979; Torri-Tarelli, Grohovaz, Fesce & Ceccarelli, 1985), controversy still exists over whether the secreted quanta represent ACh that is released from the interiors of the fused vesicles (Tauc, 1982). An alternative hypothesis is that the quanta represent axoplasmic ACh which diffuses in pulses through channels in the presynaptic membrane that become transiently permeable to it (Tauc, 1982; Israel & Manaranche, 1985). If quanta were released by exocytosis from the interiors of vesicles which have fused with the presynaptic membrane, then one might expect a vesicle to be lost whenever a quantum is secreted. However, most terminals can secrete large numbers of quanta before they lose significant numbers of vesicles (reviewed by Ceccarelli & Hurlbut, 1980a). On the other hand, vesicles are formed from the axolemma of rapidly secreting terminals, and it has been suggested that these newly formed vesicles are recycled vermions of the original ones and can reaccumulate and secrete additional quanta of ACh (Ceccarelli & Hurlbut, 1980a). Thus, the number of vesicles lost should equal the number of quanta secreted only when endocytosis and vesicle recycling are blocked. Endocytosis is inhibited at terminals treated with ouabain, or with black widow spider venom in Ca2+-free Ringer solution, and such terminals secrete spontaneously until they are exhausted of quanta and depleted of vesicles (Clark, Hurlbut & Mauro, 1972; Ceccarelli & Hurlbut, 1980b; Haimann, Torri-Tarelli, Fesce & Ceccarelli, 1985). The total number of MEPPs that occur at such terminals is similar to the number of synaptic vesicles in resting terminals (- 7 x 105) (Haiman et al. 1985; Fesce, Segal, Ceccarelli & Hurlbut, 1986b), but additional comparisons of MEPP counts and vesicle counts need to be made under other conditions which inhibit endocytosis and stimulate quantal secretion. Since low temperature appears to be an effective inhibitor of most endocytotic processes in mammalian cells (Steinman, Mellman, Muller & Cohn, 1983), we have studied the effects of a-latrotoxin (a-LTx), a purified protein component of black widow spider venom (Frontali, Ceccarelli, Gorio, Mauro, Siekevitz, Tzeng & Hurlbut, 1976), on quantal secretion, nerve terminal ultrastructure and uptake of horseradish peroxidase (HRP) at frog neuromuscular junctions at 1-3 'C. We were especially interested in these ac-LATROTOXIN AT LOW TEMPERATURE 197 experiments because Heuser & Miledi (1971) had previously estimated that several million quanta were secreted from terminals treated with La3+ at 4 'C. Evoked and spontaneous quantal release are highly temperature dependent, and the quantal content of the EPP and the rate of occurrence of MEPPs decline to low values at temperatures below 10 'C (see Barrett, Barrett, Botz, Chang & Mahaffey, 1978, for references). Previous studies of the effect of temperature on the action of black widow spider venom or a-LTx showed that binding occurs at temperatures near 0 'C (Valtorta, Madeddu, Meldolesi & Ceccarelli, 1984), but quantal release is small when moderate doses of these agents were used (Rubin, 1976; Pumplin & Reese, 1977; Fritz, 1980). However, we have found that high concentrations of o- LTx (2 ,ug/ml) raise MEPP rates to several hundred per second at temperatures of 1-3 'C and cause a profound depletion of vesicles after 2 h.

METHODS General. Cutaneus pectoris muscles were dissected from frogs (Rana pipien8) together with a short length of nerve, a small flap of skin that covered the throat and a short length of rectus abdominis muscle. The preparation was stretched over a lucite lens on the floor of a thin-bottomed lucite recording chamber and pinned into pads of dental wax at each end. The chamber held about 3 ml of bathing solution and could be perfused. The standard bathing solution (Ringer solution) contained (in mM): NaCl, 116; KCl, 2-1; CaCl2, 1-8; phosphate buffer, 3-0; at pH 7-0. The Ca21- free Ringer solution contained no added CaCl2, 1 mm-ethyleneglycol-bis-(,8-aminoethylether)N,N'- tetraacetic acid (EGTA) and 4 mM-MgCl2; the Ringer solution with Mg2+ contained 4 mM-MgCl2 in addition to the other salts. The concentration of NaCl in the modified Ringer solutions was adjusted to keep the tonicity constant. Tetrodotoxin (TTX, 10-7-10-8 g/ml) was added to all solutions to prevent the muscle fibres from twitching. The recording chamber was mounted on a Peltier device and cooled to a temperature of 1-3 'C as measured by a thermistor pressed firmly against the wax pad near the edge of the muscle. Electrophysiology. End-plate regions were impaled by glass micropipettes filled with 3 M-KCI (resistances 10-30 MQ at room temperature), and MEPPs were recorded with a conventional high- input-impedance amplifier and stored on FM magnetic tape (tape speed: 4-75 cm/s, bandwidth 0-600 Hz for experiments at 1-3 'C; speed 9-5 cm/s, bandwidth 0-1250 Hz at room temperature) for later analysis by computer (Digital Equipment Corp., PDP 11/73). One channel of the tape- recorder stored a high-gain AC-coupled record (bandwidth 0-03-600 Hz at 1-3 'C, or 0-15-1250 Hz at room temperature) of the deviations of the membrane potential from its mean, and a second channel stored a low-gain DC record of the absolute value of the membrane potential. A baseline record 3-5 min in duration was collected, then a-LTx was applied and the potentials were recorded continuously for 45-60 min; they were recorded intermittently at later times. In each experiment we used a different muscle and tried to record from a single neuromuscular junction throughout its discharge of quanta. Data were collected at a particular junction as long as the MEPP rate exceeded - 20/s and the individual MEPPs could be distinguished from noise. We were successful in sixteen experiments at low temperature. In two additional experiments reported here the initial impalements were lost after about an hour (i.e. well after the time the rate of secretion had passed through its peak); in these two cases a second junction was impaled and its quantal secretion added to that of the first junction. At the ends of many experiments, several junctions in a given muscle were impaled briefly and their MEPPs recorded (AC coupling, 0-1 s time constant) for later counting on a storage oscilloscope. The Peltier device was switched off, and the MEPPs were recorded intermittently at a single junction while the muscle gradually warmed to 20 'C; then the MEPPs at several additional junctions were recorded. The membrane potentials of the impaled fibres usually declined and reached levels of about -40 mV at the ends of the experiments. However, the fibres that were sampled at the ends of the experiments had normal membrane potentials (about -70 mV). Therefore, the decline in the membrane potentials of the impaled muscle fibres was probably due to local effects of the 198 B. CECCARELLI, W. P. HURLBUT AND N. IEZZI micropipette, and was not due to a general toxin-induced deterioration of the entire muscle. Much of the decline in the membrane potential of the impaled fibres occurred when the bath was drained and the toxin applied. MEPP waveform. The time course of individual MEPPs is closely approximated by the equation (Segal, Ceccarelli, Fesce & Hurlbut, 1985): hw(t) = h{[exp (t/0j)]-[exp-(t/02)]}, (1) where w(t) = MEPP waveform, h = amplitude factor, 01 = time constant of decay and 02= time constant of rise. During the baseline period, h, 0, and 02 were determined by fitting exponential functions to the average waveform obtained by averaging ten to thirty individual MEPPs. When the MEPP rate was high, the time constants were determined from the power spectrum of the fluctuations in membrane potential. Power spectra. The high-gain AC record was put through a low-pass Butterworth filter set to a cut-off frequency of 250 Hz (1250 Hz at room temperature), and its output was sampled by the computer at a rate of 500 Hz (2500 Hz at room temperature). These cut-off and sampling frequencies are slightly closer than is usually recommended (Bendat & Piersol, 1971; Eisenberg, 1983), but the close separation is permissible for end-plate signals which contain almost no power at frequencies above 50 Hz for muscles near 0°C (Fig. 3), or 500 Hz for muscles at room temperature (Fesce et al. 1986b). Average power densities were computed by a standard discrete fast Fourier routine from eight groups of 2048 data points collected from a 32 s length of tape (Segal et al. 1985; Fesce et al. 1986b). The average baseline densities were subtracted, and the resultant increases in density were plotted against frequency on log-log co-ordinates and fitted with a double Lorentzian curve: p(f) = A/[1 + (2nff0j)2] [1 + (2rf0o2)2], (2) where p(f) = power density at the frequency, f, A = low-frequency asymptote of the curve and 01 and 02 are the MEPP time constants. The double Lorentzian is the Fourier transform of eqn (1) and is the expected shape of the power spectrum of the fluctuations when the MEPP rate is stationary (Verveen & De Felice, 1974). When the MEPP rate is not stationary, the power spectrum contains extra frequency components and deviates from the double Lorentzian (Sigworth, 1981; Fesce, Segal & Hurlbut, 1986a). Since these extra components usually lie at the low-frequency end of the spectrum (Fig. 3), the MEPP time constants can be determined by fitting a double Lorentzian to the points beyond - 3 Hz (Fesce et al. 1986b). Computation ofMEPP rate and amplitude and number of quanta. The high-gain AC record was put through the low-pass Butterworth filter (1250 Hz), then through a high-pass RC filter, and the output was sampled by the computer at a rate of 2500 Hz. The RC filter suppresses the low- frequency spectral components which arise from non-stationarities in the rate and reduces the random errors of the measurements (Segal et al. 1985; Fesce et al. 1986 a). A10 s sample of data was collected (25000 digitized points), and the variance, V (2nd central moment or cumulant), the skew, S (3rd central moment or cumulant), and the 4th cumulant (or semi-invariant), C4, were calculated (5 s computation time). (C4 = M4 -3V2, where M4 is the 4th central moment.) This process was repeated every 15 s until the entire experiment was analysed. The DC record was also sampled once during each data collection period and used to correct the moments for non-linear summation. The averages of the baseline values of the cumulants were subtracted from their values at other times, and the resultant net values were used for further computation only if they exceeded their baseline values (i.e. the cumulants had to double before their net values were used to estimate MEPP rate and amplitude). The net values of the cumulants were corrected for non-linear summation (the correction was significant only for rates > 100/s) and the mean MEPP rate, , and amplitude factor,, were calculated from the equations (Rice, 1944; Segal et al. 1985; Fesce etal. 1986a): = (S/I1)/(V/I1), (3) and =(V/II)3/(S/II)2, where I'= f [w'(t)]n dt, and w'(t) = waveform of a filtered MEPP. oc-LATROTOXIN AT LOW TEMPERATURE 199 The integrals I2 and I' were determined by simulating a MEPP with time constants determined from a power spectrum. The simulated MEPP was put through the Butterworth and RC filters, the output waveform was sampled by the computer, and the integrals of its square and cube were computed. The time constants and integrals were computed from power spectra collected at intervals of 5-15 min throughout each experiment, and their values at other times were calculated by interpolation. Estimates of and could be calculated from each 15 s sample of data, but usually three successive estimates of the cumulants were averaged in order to reduce the random errors of the measurements. The number of quanta secreted, Q, was determined by integrating over time. This method selectively measures quantal secretion and is insensitive to non-quantal secretion. Its results are valid in spite of slow, spurious changes in the average membrane potential, non-stationary MEPP rates and non-linear summation of MEPPs (Segal et al. 1985; Fesce et al. 1986a). However, these values of overestimate, and the values of underestimate, the true ones, because the MEPPs recorded at a single junction are not equal in amplitude. These residual errors can be corrected if the shape of the MEPP amplitude distribution is known, or if the ratio: R = (S/I)2/(V/IJ) (C4/I) is computed from the fluctuations, when is stationary (Fesce et al. 1986a). Most experiments were analysed twice using RC filters with different time constants, r, first with T < 02 and then with 02 < r < 01I Results that are independent ofT indicate that the effects of nonstationary MEPP rates have been effectively filtered out and that the MEPP time constants have been determined reliably (Fesce et al. 1986a). MEPP amplitude histograms. The amplitudes of 15-100 MEPPs were measured at the beginnings and ends of most of the experiments, when was low enough that individual MEPPs could be resolved. For each experiment an amplitude histogram, normalized to the mean amplitude, was constructed, and the normalized histograms from all similar experiments were combined. These averaged histograms were fitted by y distributions and the corrected values of and were computed using the y parameter (Fesce et al. 1986 a, b): t/ = (y+3)2/(y + 2) (y + 1), and t/ = (y + 1)/(y + 3), where t and t are the corrected values of and . Preparation of a-latrotoxin. Concentrated stock solutions of a-LTx were purified from the venom glands of Yugoslavian black widow spiders, L. mactans tredecimguttatus, by previously published methods (Frontali et al. 1976; Meldolesi, 1982); they were divided into small aliquots (4-5 ,ug) and stored at -80 'C. Their protein contents were not measured directly, but were inferred from their effects on rat brain synaptosomes loaded with tritiated dopamine. The dilution of a stock that released 50 % of the synaptosome's radioactivity within 6 min at 37 'C was assumed to contain a- LTx at a concentration of 003 ,ug/ml (Meldolesi, 1982). The concentration of protein in the stock solutions from four different purifications ranged from 40 to 50 ,ug/ml. We checked the potency of each stock by determining how rapidly neuromuscular transmission was blocked when a-LTx was applied at a concentration of 2 ,ug/ml to muscles at room temperature. The toxin from three of the purifications blocked transmission at half the junctions within 8-20 min, in good agreement with previous results (Ceccarelli & Hurlbut, 1980 b), and the fourth blocked half the junctions after 40 min. The dose of a-LTx from the different purifications was not adjusted to compensate for their different potencies. When applied to muscles at 2 °C, a-LTx blocked transmission by 50% within 25-35 min (two experiments). The relatively high concentrations of a-LTx that were used in these experiments are commensurate with its relatively weak binding in frog muscle. Thus, for rat brain synaptosomes at 37 'C the toxin's dissociation constant (KD) is 0-01 ,ug/ml (0 7 x 10"10 M), the 50% release dose is 3 x KD, and the dose which provokes 'maximal' dopamine release is 30 x KD (Meldolesi, 1982). At the frog neuromuscular junction the toxin's KD (at 0 C) is 007,g/ml (5 x 10-10M) (Valtorta et al. 1984), and the 'threshold' dose for quantal release (at room temperature) is 3 x KD (Frontali et al. 1976). The concentration of 2,tg/ml is 30 x KD, and corresponds to a 'maximal' dose. Bovine serum albumin was not used as a carrier in our experiments so the effective doses applied to the muscles may have been lower than the calculated ones. 200 B. CECCARELLI, W. P. HURLBUT AND N. IEZZI An aliquot of the stock solution was diluted with 2-2-3 ml of bathing solution and chilled in a syringe on ice. After the baseline data had been collected from a junction at 1-3 °C, the bathing solution was gently sucked from the recording chamber, and the chilled solution of a-LTx was added. The temperature in the bath never rose above 5 °C during this operation, and it returned to 1-3 °C before rose appreciably. Electron microscopy. All muscles were fixed in the recording chambers. Those not soaked in HRP were fixed for 1 h with a cold solution of 2 % OSO4 in 0-1 M-phosphate buffer, pH 7-2. Small pieces of tissue suspected to be rich in end-plates were cut out, block stained with uranyl acetate and then dehydrated and embedded in Epon 812. HRP (1 %) (Sigma type VI) and sperm whale myoglobin (0 5 %) (Sigma type II) were dissolved in 2 ml of bathing solution and applied for 30 min at room temperature to a muscle in the recording chamber. The chamber was then cooled to 1-3 "C and 30 min later an aliquot of a-LTx was added. MEPP rates were monitored during the experiments but were not recorded for analysis because secretion was not allowed to proceed to exhaustion. The muscles were fixed when the MEPP rates were near their peak as judged from the appearance of the oscilloscope traces. At these times the terminals should still have contained many synaptic vesicles, and if recycling occurred it should have been vigorous. The fixative was a cold solution of 1 % glutaraldehyde and 0 5 % formaldehyde (freshly prepared from paraformaldehyde) in 0-1 M-phosphate buffer, pH 7-2. The muscle was fixed for 1 h, and then small pieces were dissected out, treated by Karnovsky's (1967) procedure to reveal sites of peroxidase activity, and then post-fixed with 2% OSO4 in 0-1 M-phosphate buffer, pH 7-2. HRP was used in three experiments with Ringer solution and in three experiments with Ca2+-free Ringer solution. About an hour was required for secretion to reach high rates in these experiments, suggesting that the enzyme reduced the effectiveness of the toxin, presumably by a non-specific interaction between the concentrated proteins. Silver-grey sections (- 40 nm thick) were cut with a diamond knife (Diatome Ltd, Bienne, Switzerland) on a Reichert-Jung ultramicrotome (Reichert, Wien, Austria), double stained with uranyl acetate and lead citrate and examined with an Hitachi H-600 electron microscope (Hitachi Ltd, Tokyo, Japan) whose magnification was routinely checked against calibrated grids (Balzers Union Ltd, Lichtenstein). Morphometry. Prints were made at a final magnification of 40000 x , and a Zeiss MOP I digitizing image analyser (Zeiss, Oberkochen, F.R.G.) was used to measure the following structures in longitudinal or transverse sections of nerve terminals: area of axoplasm, length of plasmalemma, length of infoldings of the plasmalemma, total perimeter of large vesicular structures (LVSs) (i.e. membrane-bound structures with diameters > 50 nm), number of synaptic vesicles (i.e. vesicular structures with diameters < 50 nm) and number of coated vesicles (CVs). Most of the experiments and all the electron microscopy were done at the Center for the Study of Peripheral Neuropathies and Neuromuscular Diseases, University of Milano, Milano, Italy; most of the computer analysis was done at Rockefeller University, New York, NY, U.S.A.

RESULTS Electrophysiology Experiments at room temperature. Since previous work (Ceccarelli & Hurlbut, 1980b; Fesce et al. 1986b) was done mainly with crude black widow spider venom, we did a series of experiments in which purified a-LTx was applied at concentrations ranging from 0-2 to 2 ,ug/ml to twelve muscles bathed in Ringer solution at room temperature (19-24 "C). The results were similar to those obtained with crude black widow spider venom. The mean MEPP rate, , increased abruptly 1-9 min after the addition of the toxin and rose to peak values that ranged from 450 to 3000/s at individual junctions. It remained above 100/s for at least an hour at eight of ten junctions treated with doses of 1 ,ug/ml or less (doses that are roughly equivalent to the doses of crude black widow spider venom used previously), and these junctions released 0-9-2 x 106 quanta; at the other two fell below 30/s after 30-45 min, and a-LATROTOXIN AT LOW TEMPERATURE 201 only about 045 x 106 quanta were released. No morphology was done on these muscles. Two muscles were treated with 2 ,ug a-LTx/ml, and quantal secretion was intense but brief. One of these muscles was fixed after an hour and examined in the electron microscope; its nerve terminals were swollen and completely depleted of vesicles. This high dose of toxin appears to inhibit recycling even when Ca21 is present. A -aa'mIII 1MJunmunI6ini g I

Ci161 W L l

Fig. 1. Strip chart records of MEPPs at junctions treated with a-LTx at 1-3 °C in the different bathing solutions: Ca2+-free Ringer solution with 4 mM-Mg2+ (A), Ringer solution (B), or Ringer solution plus 4 mM-Mg2+ (C). Time bar = 1 min; note that the time scale in C is compressed 50 %. The gain is the same in all records (in A it was reduced 50 % where the envelope abruptly contracts), and they are arranged so as to align the times (arrows) when began to increase. These times were: 10-6 min (A), 6-1 min (B) and 9-5 min (C), after a-LTx had been added. The short gap in A marks a 3 min interval when the tape-recorder was turned off during the original experiment; it was turned on again just as was beginning to rise. Note that MEPPs occur in bursts when Ca2+ is present. These records were made by playing back the tape-recorder at 4 x (A and B) or 8 x (C) the original recording speed, and the MEPPs are attenuated. Their amplitudes (mV), determined by averaging ten to thirty events collected during the baseline period, were: 0 33 (A), 0-32 (B) and 0 30 (C). Temperature (°C): 1-6 (A), 1-5 (B) and 2-5 (C).

Ringer solution with 4 mM-Mg2+ was applied to seven of the muscles after had subsided from its peak and was in a quasi-steady state. At four junctions increased by 100-500/s after the Mg2+ had been added, at two junctions it rose marginally by 50-100/s, and at one it decreased. Since these results suggested that Mg2+ could increase even when Ca2+ was present, we studied junctions bathed at 1-3 °C in Ringer solution with Ca2+ (1-8 mM) alone, Mg2+ (4 mM) alone, or Ca2+ and Mg2+ Experiments at low temperatures (1-3 °C). Low concentrations (0 4-1 jug/ml) of a- LTx often did not provoke a massive discharge of quanta when applied to muscles 202 B. CECCARELLI, W. P. HURLBUT AND N. IEZZI

1000

x ~~~A 100 H

x 0*5 10 _

x - 1 n

10 V1) x 'a co

0UCL wL0-

0 CS 1*0

ioo F

)x XX / 0*5

101

X/ 0 1 ,, .... 0 30 60 90 120 Time in a-LTx (min) Fig. 2. Time course of the changes in and Q at single junctions in three different muscles treated with a-LTx at 1-3 °C in different bathing solutions: Ca2l-free Ringer solution with 4 mM-Mg2+ (A), Ringer solution (B) or Ringer solution with 4 mM-Mg2+ (C). Ordinates: MEPPs/s (X), log scale; or quanta secreted (continuous line), linear scale. Abscissae: time (in minutes) after adding a-LTx. The circled crosses are values of determined by counting individual MEPPs recorded during the final stages of the experiments. These estimates are expected to be higher than those provided by fluctuation analysis because the amplitudes of the MEPPs are distributed over a range of values (see text). Same junctions as in Fig. 1. Some symbols are omitted from the regions of the graphs where the data are densely clustered. in cold Ringer solution. The MEPP rates at individual junctions increased after delays that ranged from 10 to 50 min and reached peak values that were relatively low (- 100/s); they often rose to considerably higher levels when the muscles were warmed later to room temperature. These results are consistent with the observations of others who studied the effects of moderate doses of black widow spider venom or a-LTx on junctions in the cold (Rubin, 1976; Pumplin & Reese, 1977; Fritz, 1980). Since we wanted to raise to levels that would quickly exhaust the terminals, we used high concentrations (P18-2 ,ug/ml) of a-LTx in most of our experiments. Figures 1-4 illustrate results from three individual junctions in three different muscles, each soaked in one of the three different bathing solutions, and Table 1 summarizes the averaged results from six similar experiments in each of the three solutions. a-LATROTOXIN AT LOW TEMPERATURE 203

Ps XXXXX AAA A 1*0 L xx A - x xx 4 xx 0*5 x x :0 N x x u 0 0EL. 0 1-0 '

0-5

Q o -aa .0I O w 8 ,0 C-

4 4 - 0 @go.* I

0%, 0 0-25 2-5 25 250 Frequency (Hz) Fig. 3. Power spectra of the fluctuations in membrane potential at junctions treated with a-LTx at 1-3 °C in three different bathing solutions: Ca2+-free Ringer solution with 4 mm- Mg2+ (A), Ringer solution (B) or Ringer solution with 4 mM-Mg2+ (C). Ordinates: power/ unit bandwidth (arbitrary units), log scale; or power (X) (normalized), linear scale. Abscissae: Fourier frequency (Hz), log scale. Baseline spectrum (a); difference spectrum (experimental minus baseline) (O); double Lorentzian (continuous curve). These experimental data were collected shortly after the MEPP rates had begun to rise. 1 min

(A), - 1 min (B) or - 10 min (C). The spectrum in A fits a double Lorentzian along its entire length, but those in B and C deviate from it at frequencies below - 2 Hz. Note that over 90 % of the total power (integral of the power spectrum = variance) in A is contained in the frequency band between 2-5 and 50 Hz whereas nearly half of the variance in B is contributed by frequencies below 2-5 Hz. These extra low-frequency components arise from the fluctuating MEPP rate, and they are removed by the high-pass RC filters when the data are analysed. The integral of the power spectrum is not shown in C because it would be similar to that in B. The baseline is shown to illustrate the large magnitude of the signal. The time constants (01 and 02) of the three double Lorentzian curves are: 22-6 and 3-6 ms (A), 29-1 and 4-2 ms (B) and 19-1 and 3-4 ms (C). The time constants determined from the average of ten to thirty MEPPs collected during the baseline period were: 23-1 and 3-7 ms (A), 33 and 4-4 ms (B) and 32-1 and 3-8 ms (C). Same junctions as in Fig. 1. Some symbols are omitted from the regions of the graphs where the data are densely clustered. 204 B. CECCARELLI, W. P. HURLBUT AND N. IEZZI

0-6 -20 exXxAIx x x A_

0-3 7& 4x, x . -50

o -80 0-6 -20 5 E E -7Cu 0 c -clCu 0*3 -50 a wc;. "I -

a._/ .0 LL E 0) -80 2

0-3 -50 xx ,

0 -80 0 30 60 90 120 Time in a-LTx (min) Fig. 4. Time courses of the changes in membrane potential (continuous line) or (X) at the three junctions illustrated in Fig. 2. A, muscle in Ca2+-free solution; B, muscle in Ringer solution; C, muscle in Ringer solution plus 4 mM-Mg2+. Abscissae: membrane potential or in mV. Ordinates: time after adding a-LTx (2 ,ug/ml). The circled crosses are values of calculated from the mean amplitude of MEPPs recorded at the beginnings or ends of the experiments. These estimates are expected to be lower than those provided by fluctuation analysis because the amplitudes are distributed over a range of values (see text). The junction in B suffered a sharp drop in membrane potential at about 70 min. Some symbols are omitted from the regions of the graphs where the data are densely clustered.

The most straightforward results were obtained when a-LTx was applied in a Ca2+-free Ringer solution with 1 mM-EGTA and 4 mM-Mg2+. In these six experiments rose gradually after delays that ranged from 4 to 17 min (Fig. 1A), reached maximum rates of 200-500/s some 6-11 min later and then declined exponentially to levels below 30/s after 1-5 h (Fig. 2A, Table 1). The total number of MEPPs recorded, which equals the total number of quanta secreted, ranged from 0-23 to 0-56 x 101 (Table 1). The average MEPP rates determined by sampling junctions in the cold at the ends of the experiments was about 20/s (Table 2), in good agreement with the average results from fluctuation analysis (Table 1). When the muscles were warmed to room temperature, the MEPP rates rose to levels below 100/s and then fell. The number of quanta secreted after the muscles were warmed was not measured. a-LATROTOXIN AT LOW TEMPERATURE 205

0 0 - +1- +1: 0 022 X C@2 +1 z C 2 0 s7E +1 lf2 c10 C0 +o -

+1 +1 '4- = 0 3 0 0 1- *- C2 CO2 if 0 C: - - C@ 02 I- X r. 02 +1 * +1 3@ -4 c-1 I-d 0 02 c es I-'l/& +1 +1 +1 0 +1 * .-- if 0 z +l 02 m Cw es S 0 +^ F, *tQ +1 +1 +1 -6 cs eq cD 0 0 pp IF 1- ; 0 '-4 to xo ._ 0 O .I X +I +1 +1 z cI C;O r+ it 1:: +I +l Ev 6 5 . o M- 02 02 +1C-~ +1 ce C;t 0 uD +l d1 .-q r +1 C-~ 4- 00. *10 +I +1 +l ~Ca o - 0 0O rA 4:)t

.-I 102 0 2) H -I 4o =e. Ev0 *. +1 +1+1- 0ce es _02 4 4+D +1 00 Cl 0 - o 9 C': *6c,Qm -_R--( .E~4 o-l c=c0- _m

*_.. ~ d "- +1 +l G _o 4 I-L HS r- 02 +I .. w *ifO +1 or 02

0 0 0+ ._- cq ew 0 0 02S Go + . .0 + + ce I o4 ._ ._ + O + 02 i + (L 4

.." + 206 B. CECCARELLI, W. P. HURLBUT AND N. IEZZI The power spectrum of the fluctuations in membrane potential at these junctions was shaped almost exactly like a double Lorentzian (Fig. 3A). The smooth increase in generated almost no extra frequency components to distort the spectrum from the shape expected for MEPPs that rise and decay exponentially with time. When the bathing solution was Ringer solution (1P8 mM-Ca2+ and no Mg2+), secretion always began with a burst of MEPPs (Fig. 1 B) after delays that ranged from 5 to 28 min in six experiments. The rate usually (five of the six experiments) did not pass through a clear maximum, but fluctuated about average levels near 100/s (Fig. 2B). The large, rapid fluctuations in added extra components to the power spectrum so that its shape deviated at low frequencies from a double Lorentzian (Fig. 3B). When began to subside after 60-90 min of secretion, it fell quickly to very low levels (Fig. 2B). The total number of quanta secreted by these terminals ranged from 0-24 to 0-53 x 106 (Table 1). The final MEPP rates at these junctions were lower than those in the Ca2+-free solution (Table 2). When these muscles were warmed to room temperature, rose briefly to peak values below 30/s and then fell within a few minutes to levels below 10/s (Table 2). The final rate at the junction illustrated in Fig. 2B (14/s by fluctuation analysis, 25/s by direct count) is higher than that at the junction illustrated in Fig. 2A (7/s by both methods). We impaled seven other junctions in the muscle of Fig. 2B at the end of the experiment, and their MEPP rates ranged from 23 to 1/s. The junction in Fig. 2B happened to be the most active one examined in this muscle, or in any of the four other muscles in Ringer solution that were tested in this way. We did not impale other junctions in the muscle of Fig. 2A, while it was in the cold, but this junction was among the least active of all those tested at the ends of the experiments in Ca2+-free Ringer solution. When the bathing solution contained both 1-8 mM-Ca2+ and 4mM-Mg2+, the pattern of secretion exhibited a mixture of the characteristics described above. Secretion began with a burst of MEPPs after delays of 4-10 min in six experiments (Fig. 1 C). The bursting pattern of secretion was maintained throughout much of the discharge and distorted the power spectra (Fig. 3C), but it was superimposed upon a steady rise in rate so that passed through a clear maximum that ranged from 150 to 700/s at six individual junctions in six experiments (Table 1). The MEPP rates then declined gradually over the next hour at five of these six junctions (Fig. 2C); at one junction fell quickly from its peak and secretion stopped almost completely after only 20 min. The total number of quanta released by these terminals ranged from 0-17 x 106 (at the one which had stopped secreting) to 0 74 x 106. The final values of measured in these experiments were similar to those measured in the absence of Ca2+ and were greater than those measured in the absence of Mg2+ (Table 2). When these muscles were warmed to room temperature, rose to peak values that ranged from 8 to 250/s. When junctions were sampled 10-20 min later the rates varied greatly, ranging from <1 to > 200/s. Figure 4 illustrates the time courses of the changes in membrane potential and at the three junctions illustrated in Fig. 2. The changes in are small and seem to follow the decreases in membrane potential; they are affected little by the enormous changes. in, as would be expected for preformed quanta that were not renewed during secretion. Table1 summarizes the overall changes in at the six junctions followed in each of the three experimental solutions. MEPP amplitude histograms. The results in Table1 should be corrected for the ac-LATROTOXIN AT LOW TEMPERATURE 207

20 - L. 20~~~~~~~I L.-0 0i00 0

I em 0 w 20-8 0-~~~~~~ 0) C" 0) &- ~~~~~~~~~0 0)0 0- Q.-e - - C~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ c

A -*J@@

0 0 1 2 3 Amplitude/mean amplitude Fig. 5. MEPP amplitude histograms at 1-3 'C. Ordinate: percentage of MEPPs. Abscissae: MEPP amplitude normalized to the mean amplitude. A, histogram of MEPPs recorded in all three solutions during baseline periods or before began to rise; B, histogram of MEPPs recorded in Ringer solution 1-5-2 h after addition of a-LTx, C, histogram of MEPPs recorded in Ringer solution plus 4 mM-Mg2+ 1-S2 h after addition of a-LTx. The dotted curves are y distributions calculated from the means and variances of the experimental histograms. The number of MEPPs and y parameters of the three histograms are: A, 381 MEPPs, y = 10-3; B, 957 MEPPs, y = 7-7 and C, 631 MEPPs, y= 4-6. The mean values (± S.D.) of the coefficients of variation (c.v.) and the y parameters of the individual histograms from which the combined histograms were constructed are: for A, c.v. = 0-31 + 0 05, y = 10-5 + 4 0, N = 10; for B, c.v. = 0-34 + 0-06, y = 8-1+2-9, N = 5; for C, c.v. = 0-41+0-09, y = 5-8+3-3, N = 6, where N is the number of histograms that were combined in each case. distribution of MEPP amplitudes, which often broadens during the course of secretion (Kriebel & Gross, 1974; Kriebel & Florey, 1983; Fesce et al. 1986b). This broadening is illustrated in Fig. 5. Figure 5A presents the combined normalized histograms of the distribution of MEPP amplitudes obtained in all solutions before a-LTx was applied, and Fig. 5B and C presents the combined normalized histograms obtained at the ends of the experiments in Ringer solution (B) or Ringer solution with 4 mM-Mg2+ (C). The histograms are fitted by y distributions. The distributions broaden in both cases, but the changes are smaller than at room temperature (Fesce et al. 1986b). The y parameters decline from the initial value of 10-3 to final values of 7.7 and 4-6 respectively; the correction factors for and change from their initial values of 1-26 and 0-85 to final values of 1-36 and 0-81, or 1-56 and 0-74, respectively. The MEPPs recorded at the ends of all but one of the experiments in 208 B. CECCARELLI, W. P. HURLBUT AND N. IEZZI the Ca2+-free solution were too small for amplitude histograms to be constructed. In that experiment the y parameter declined to 3-8, and the correction factors for and were 1-66 and 0-71, respectively. The average of the initial and final values of the correction factors was used to correct the data in Table 1, and the corrected estimates are listed in Table 5. The ratio, R, was also monitored in each experiment, but the only useful results are those obtained in the Ca2+-free solution, since is quasi-stationary only in this solution (Fesce et al. 1986a). In those experiments the average value of R declined from 0-9, the value characteristic of the initial distribution of amplitudes, to a final value of about 0O8, which is consistent with the observed changes in the amplitude distributions. However, the final estimates of R were highly scattered because the MEPPs were very small, and the cumulants often were barely significant. Therefore R is not used to correct the estimates of and . Morphology Plate 1 shows longitudinal sections, and Plate 3A and B shows transverse sections of portions of neuromuscular junctions in muscles bathed for 2 h at 2 °C in Ringer solution (Plate IA), Ca2+-free Ringer solution (Plates 1 B and 3A) or Ringer solution with Mg2+ (Plates 1 C and 3B). The terminals look normal. Each is covered by a Schwann cell process and contains mitochondria, elements of smooth endoplasmic reticulum, neurofilaments, large numbers of synaptic vesicles and a few large dense core vesicles. The axolemma is smooth except where fingers of Schwann cell cytoplasm interdigitate between the presynaptic membrane and the crests of the post-junctional folds. Plate 2 shows longitudinal sections, and Plates 3C and 4A and B show transverse sections of terminals treated with a-LTx (2 ,ag/ml) for 2 h at 2°C in Ringer solution (Plate 2A), Ca2+-free Ringer solution (Plates 2:B and 3C) or Ringer solution plus Mg2+ (Plates 2C and 4A and B). The terminals and their mitochondria are swollen; they contain many large vesicular structures, their axolemmas are deeply infolded and they have lost most of their synaptic vesicles. Table 3 summarizes the results of a morphometric analysis of longitudinal sections of portions (10-15,um long) of thirty control or thirty-eight experimental terminals from several different muscles. Table 4 summarizes)the results from transverse sections of eighty-one control or 267 experimental terminals from several different muscles. The data are internally consistent; if the values in Table 4 are divided by the mean cross-sectional area of the terminals, then the results agree well with the corresponding values in Table 3. The control terminals can be lumped together since the differences among them are not significant. The terminals treated with a-LTx lost 60-75% of their vesicles, and those in Ringer solution tended to lose the fewest. These reductions are real and not dilution artifacts resulting from the swelling of the terminals. The infoldings tended to be most extensive in terminals bathed in solutions with Ca2+, and the large vesicular structures tended to be most numerous in terminals bathed in the Ca2+-free solution. The swelling was least in terminals bathed in solutions with both Mg2+ and Ca2+, and the number of coated vesicles was greatest in these terminals. Plate 3 D and E shows transverse sections of junctions treated at 2°C withoX-LTx in Ringer solution (D), or Ca2+-free Ringer solution (E), plus HRP. These terminals were fixed about 60 min after the toxin had been applied, when was judged to a-LATROTOXIN AT LOW TEMPERATURE 209

---

+l ° - ° - V * * sD~~~~~~~~~4

= O O ° O O _ S~~~

e :. + +1 +1 $1 - _ O

Ei~~~~~++1=ot+1 +

rXt8 ° ° +l +l : -+B° _ E +

X XO Pz

i 0 C>, ao t- ° > 48

o~~~~~~~~~~~~~u . 0

bo~ ~ bo~~~~~,a 0~~ E

aq km xo~~~~~~~~~~~~~~~~~~~~~~~~~ +1~~~~~ +1 210 B. CECCARELLI, W. P. HURLBUT AND N. IEZZI be near its peak. The extracellular spaces are heavily labelled with reaction product, but few vesicles contain it. The numbers of vesicles varied greatly among different junctions, and the ones shown in this Plate were chosen because they still contained many vesicles, thereby improving the chances of finding a labelled one. The almost complete absence oflabelled vesicles indicates that recycling was blocked. This result stands in marked contrast to the situation at room temperature where vigorous recycling has been demonstrated at frog neuromuscular junctions stimulated electrically (Ceccarelli, Hurlbut & Mauro, 1972, 1973; Heuser & Reese, 1973; Ceccarelli & Hurlbut, 1975), by 20 mM-K+ (Ceccarelli, Grohovaz & Hurlbut, 1979), by 1 mM-La3+ (Heuser & Miledi, 1971), by 0 1 mM-La3+ (Segal et al. 1985) or by low doses of black widow spider venom in Ringer solution (Ceccarelli & Hurlbut, 1980b). Plate 4 and Table 4C summarize the results obtained from transverse sections of terminals that had been treated with az-LTx for 2 h at 2 °C and then warmed to 20 °C for 30 min. All terminals lost most of their infoldings; those in Ringer solution retained most of their residual vesicles, those in Ca2+-free Ringer solution lost almost all their residual vesicles, and those in Ringer solution plus Mg2+ lost some of their residual vesicles, most of their coated vesicles and swelled further. The average number of synaptic vesicles per terminal was calculated from the data in Table 4, taking an average terminal length of 600 ,sm (Letinsky, Fishbeck & McMahan, 1976; Valtorta et al. 1984), a thickness of 40 nm for silver-grey sections (Weibel & Paumgartner, 1978) and correcting the vesicle counts for the fact that a given section contains portions of vesicles whose centres do not lie within it. Prolonged exposure to a-LTx does not cause gross changes in the total length of the nerve terminal branches at a single junction. This is shown by studies of the distribution of two proteins which mark specific presynaptic components: one is the extracellularly oriented binding site for a-LTx (Valtorta et al. 1984), the other is the protein synaptophysin (p 38), which resides in the vesicle membrane (Valtorta, Jahn, Fesce, Greengard & Ceccarelli, 1988). The distribution (as revealed by immuno- fluorescence) of antibodies to either marker corresponds closely to the distribution of acetylcholinesterase or binding sites for oc-bungarotoxin (markers for the synaptic gutters) on toxin-treated terminals. Thus, the changes in the volume of a nerve terminal are due entirely to the changes in its diameter. The correction for section thickness is given by the formula (Weibel & Paumgartner, 1978) N = NC/[1 + (D-2H)/T], where N = number of vesicles with centres in the section, NC = number of vesicle profiles counted in the section, T = section thickness, D = vesicle diameter, and H = the height of the cap that must be included in a section for a vesicle to be recognized. The average diameter of the synaptic vesicles measured in our micrographs of control terminals (Plate 1) is 50 + 5 nm. We assumed H to be 5 nm; reducing it to zero or increasing it to 10 nm (i.e. to 40% of the radius) changes the results by ±12 %. The final results are presented in Table 5 together with the average numbers of quanta secreted in the various solutions. The decrease in the number of vesicles agrees quite well with the number of quanta secreted. a-LATROTOXIN AT LOW TEMPERATURE 211

't * * * * * o + to 6 0Co -- ~> 6 +l +1 +1 +1 +1 +1+1 +1 +1+1 cs o Co O- *-1 0 -0- 0 10 to 00 o aq 4 cO to +l +1 +1 +1 +1 +1 0 I- e +1 +1 +1 0r- 0 o Co - V- 0

00 000 C0 00

0 +1 +1 + +1 0 +1 a) o eq 00 - 6 0 0 +1 +1 0 1- j 1.4 Co 0 4- 0 Co Co -

06 0000 00~~04-D 0 os I r - to 00t- cZ p.4 01 +1+1 +1 +I0+I +I c E- _C911Cob4~ 0 _ _o _o o 0 r- +1 +l +1 +l +1 +1 + +1 ++1 cl) +lb co Co 4¢. 0 I-. cO - C- -oC- do

IL4 I-L E. +1 +1 +1 +1 M- 9 - 10 01 49 4o 4Q_ z -- ¢ d1 dQ bO be 10--~~ + +l 4-4 OD E-- CO~ WGSd~ ~ A,>~ i Ca e * * - c co o - +l+l 000aq 0 -+l e ++ll+ - _ es O ; o 4-0) 3 S~~~p4- 00 r Xb e b ° o~~~~~~~0 + 2.4 +l+l +l o +o CQ 4a sf mP _ CO s s o 0 o 0+ ^+0 * -4 ~ ~

-a _ r. V ._ 00GD 000 00 -4oX-- + +000 212 B. CECCARELLI, W. P. HURLBUT AND N. IEZZI

TABLE 5. Vesicles lost and quanta secreted per terminal No. of No. of No. of No. of vesicles, vesicles vesicles quanta controls remaining lost secreted Solution (x 10-3) (X 10-3) (x 10-3) (x 10-3) Ringer solution 730 330 400 500 0 Ca21 + 4 mM-Mg2+ 900 230 670 580 Ringer solution 820 210 610 710 +4 mM-Mg2+ Grand mean 800 260 540 600 Mean values only. The coefficients of variation (c.v. = s.D./mean) of the data in the upper three rows are equal to the c.v.s of the appropriate boxes in Tables 1 and 4. The standard errors (as percentage of mean) of the grand mean values are (from left to right): 6, 8, 10 and 9%.

DISCUSSION These results confirm previous findings (Haimann et al. 1985; Fesce et al. 1986b) that the number of synaptic vesicles lost from nerve terminals is approximately equal to the total number of quanta secreted from them, when vesicle recycling is blocked. This result is consistent with the idea that the loss of vesicles results from the secretion of quanta. Our results also show that the morphological effects of the toxin are not confined to changing the number of synaptic vesicles, and that it also swells the terminals and their mitochondria. The loss of vesicles might, therefore, be interpreted as the final outcome of cell damage rather than being specifically related to quantal secretion. However, the pure toxin has no known lipolytic or proteolytic activities (Frontali et al. 1976), and terminals treated with low concentrations of the toxin in Ringer solution at room temperature secrete millions of quanta and recycle their vesicles. Furthermore, vesicle membrane seems not to be lost, but is redistributed into the infoldings of the axolemma or the extra membrane needed to surround the swollen terminals. A redistribution of vesicle membrane from the cytoplasm to the axolemma has been unequivocally demonstrated in recent experiments with antibodies to synaptophysin (p38), a specific integral protein of the vesicle membrane (Valtorta et al. 1988). The peak MEPP rates measured in the experiments reported here are about an order of magnitude less than those measured at room temperature, and the duration of the period of secretion is correspondingly longer. The fact that a rough equality between vesicles lost and quanta secreted is obtained at such widely different MEPP rates strengthens the inference that the two quantities are causally related and not merely equal by chance. A further detail of the correlation between these two quantities at junctions in Ringer solution is that secretion stopped before the terminals were completely depleted of vesicles, so that they released few additional quanta and retained most of their residual vesicles when they were warmed to 20 'C. A block ofsecretion and retention ofvesicles has been observed previously at terminals treated at room temperature with large doses of black widow spider venom in solutions with high concentrations of Ca2+ (Smith, Clark & Kuster, 1977); the a-LATROTOXIN AT LOW TEMPERATURE 213 likely cause of the block in both cases is the accumulation of large concentrations of Ca2+ in the axoplasm (Smith et al. 1977). Low doses of black widow spider venom block vesicle recycling at terminals bathed in Ca2+-free solutions at room temperature, but not at terminals bathed in solutions with the normal concentrations of Ca21 (Ceccarelli & Hurlbut, 1980b; Fesce et al. 1986b). However, it is not known whether that block is total and develops immediately, or whether it is partial and develops slowly and so permits some recycling during the early stages of secretion. This uncertainty would be absent from experiments conducted at low temperature if recycling were totally blocked by this condition. Our results show that recycling is blocked at terminals treated with high concentrations of a-LTx at 1-3 °C, independently of the Ca2+ concentration of the bathing solution, but it is not clear whether the block is due to the low temperature per se or to the high concentration of toxin. Two apparently conflicting sets of observations have been published on the ability of frog motor nerve terminals to secrete at 4 'C. Heuser & Miledi (1971) suggested that vesicles recycled several times at nerve terminals in frog sartorius muscles treated at 4 'C with 1 mM-La3+; they estimated that such terminals secreted 3-5 million quanta at peak rates of 1-2 x 103/s and lost about 75% of their synaptic vesicles within 2-2-5 h. Flory & Kriebel (1983), on the other hand, reported that such terminals secreted at peak rates around 10/s, released only about 5 x 104 quanta during 2 h and lost few synaptic vesicles in this time. Since the terminals in sartorius and cutaneus pectoris muscles ofsimilarly sized frogs are approximately equal in length (Kuno, Turkanis & Weakly, 1971; Letinsky et al. 1976; Herrera, Grinnell & Wolowske, 1985; Valtorta et al. 1984), the large differences among these various estimates of the rates of secretion may arise either from the different methods used to obtain them, or from systematic differences among the animals used. Florey & Kriebel (1983) used small frogs (Kriebel & Florey, 1983) and measured the relatively low rates by counting MEPPs on photographic records. Heuser & Miledi (1971) estimated either by counting extracellularly recorded MEPPs and correcting the result for the limited length of terminal recorded from, or by measuring the membrane potentials of junctions impaled at the peak of secretion and applying Campbell's theorem. We think they seriously overestimated (Segal et al. 1985). They also demonstrated with HRP that synaptic vesicles were recycled in terminals treated with 1 mM-La3+ at 23 °C, but seem to have done no experiments at 4 'C with HRP and La3+. The situation is confused, and there is no consistent evidence for recycling at 4 'C at frog nerve terminals. The addition of 4 mM-Mg2+ to Ringer solution has three effects on nerve terminals treated with oc-LTx at low temperature: it reduces the swelling of the terminals, it increases the number of coated vesicles in them, and it increases the peak MEPP rates. Since black widow spider venom and a-LTx probably increase the permeability of nerve terminals to monovalent and divalent cations (Finkelstein, Rubin & Tzeng, 1976; Misler & Hurlbut, 1979; Nicholls, Rugalo, Scott & Meldolesi, 1982; Fritz & Mauro, 1982; Meldolesi, Scheer, Madeddu & Wanke, 1986; Misler & Falke, 1987), all these effects of extracellular Mg2+ may result from its entry into the axoplasm. The first effect might result from interactions between increased levels of Mg2" and the 214 B. CECCARELLI, W. P. HURLBUT AND N. IEZZI cytoskeleton, and the second from the ability of high concentrations of Mg2+ to stabilize clathrin coats (Woodward & Roth, 1979) and to uncouple the clathrin 'uncoating ATPase' (Braell, Schlossman, Schmid & Rothman, 1984). The third effect might occur because Mg2+ displaces Ca2+ from internal storage sites, thereby increasing its concentration in the axoplasm. However, the average concentration of Ca2+ inside synaptosomes or PC12 cells bathed in Ca2+-free Krebs with Mg2+ is not increased when a-LTx is applied, even though secretion is stimulated (Meldolesi, Huttner, Tsien & Pozzan, 1984). Hence, it seems that Mg2+ does not cause a massive, indiscriminate release of Ca21 from storage sites distributed throughout the axoplasm. If Mg2+ acts by displacing Ca2+, it presumably does so only at select sites near the points of quantal release. If is determined solely by the local concentration of Ca2+ near these points, then our measurements imply that the average concentration of Ca2+ in these regions is higher in terminals bathed in Ca2+- free solutions with 4 mM-Mg2+ than it is in terminals bathed in Ringer solution. Although this possibility cannot be excluded, an alternative is that Ca2+ and Mg2+ act synergistically or independently to increase . The authors thank Professor Jacopo Meldolesi for his interest in and encouragement with this work and for supplying the a-LTx. We also thank Ulla Bastrup and Paolo Tinelli for technical assistance and Toni Weil and Serenella Avogadro for preparing the manuscript. This work was partially supported by an MDA grant (B.C.) and by NIH grant NS-18354 (W.P.H.).

REFERENCES BARRETT, E. F., BARRETT, J. N., BOTZ, Z., CHANG, D. B. & MAHAFFEY, D. (1978). Temperature- sensitive aspects of evoked and spontaneous transmitter release at the frog neuromuscular junction. Journal of Physiology 279, 253-273. BENDAT, J. S. & PIERSOL, A. G. (1971). Random Data: Analysis and Measurement. New York: Wiley-Interscience. BOYD, I. A. & MARTIN, R. (1956). The endplate potential in mammalian muscle. Journal of Physiology 132, 74-91. BRAELL, W. A., SCHLOSSMAN, D. M., SCHMID, S. L. & ROTHMAN, J. E. (1984). Dissociation of clathrin coats coupled to the hydrolysis of ATP: role of an uncoating ATPase. Journal of Cell Biology 99, 734-741. CECCARELLI, B., GROHOVAZ, F. & HURLBUT, W. P. (1979). Freeze-fracture studies of frog neuromuscular junctions during intense release of neurotransmitter. II. Effects of electrical stimulation and high potassium. Journal of Cell Biology 81, 178-192. CECCARELLI, B. & HURLBUT, W. P. (1975). The effects of prolonged repetitive stimulation in hemicholinium on the frog neuromuscular junction. Journal of Physiology 247, 163-188. CECCARELLI, B. & HURLBUT, W. (1980a). Vesicle hypothesis of the release of quanta of acetylcholine. Physiological Reviews 60, 396-441. CECCARELLI, B. & HURLBUT, W. P. (1980b). Ca2+-dependent recycling of synaptic vesicles at the frog neuromuscular junction. Journal of Cell Biology 87, 297-303. CECCARELLI, B., HURLBUT, W. P. & MAURO, A. (1972). Depletion of vesicles from frog neuromuscular junctions by prolonged tetanic stimulation. Journal of Cell Biology 54, 30-38. CECCARELLI, B., HURLBUT, W. P. & MAURO, A. (1973). Turnover of transmitter and synaptic vessicles at the frog neuromuscular junction. Journal of Cell Biology 57, 499-524. CLARK, A. W., HURLBUT, W. P. & MAURO, A. (1972). Changes in the fine structure of the neuromuscular junction of the frog caused by black widow spider venom. Journal ofCell Biology 52, 1-14. DEL CASTILLO, J. & KATZ, B. (1954). Quantal components of the end-plate potential. Journal of Physiology 124, 560-573. a-LATROTOXIN AT LOW TEMPERATURE 215 DUNANT, Y., GAUTRON, J., ISRAEL, M., LESBATS, B. & MANARANCHE, R. (1972). Les compartiments d'acetylcholine de l'organe electrique de la Torpille et leurs modifications par la stimulation. Journal of Neurochemistry 19, 1987-2002. EDWARDS, C., DOLEZAL, V., TUCEK, S., ZEMKOVA, H. & VYSKOCIL, F. (1985). Is an acetylcholine transport system responsible for nonquantal release of acetylcholine at the rodent myoneural junction? Proceedings of the National Academy of Sciences of the U.S.A. 82, 3514-3518. EISENBERG, R. S. (1983). Impedance measurement of the electrical structure of skeletal muscle. In Handbook of Physiology, section 10: Skeletal Muscle, ed. PEACHY, L. D., pp. 301-323. Baltimore, MD: Williams & Wilkins Co. FATT, P. & KATZ, B. (1952). Spontaneous subthreshold activity at motor nerve endings. Journal of Physiology 117, 109-128. FESCE, R., SEGAL, J. R., CECCARELLI, B. & HURLBUT, W. P. (1986b). Effects of black widow spider venom and Ca2+ on quantal secretion at the frog neuromuscular junction. Journal of General Physiology 88, 59-81. FESCE, R., SEGAL, J. R. & HURLBUT, W. P. (1986a). Fluctuation analysis of nonideal shot noise. Application to the neuromuscular junction. Journal of General Physiology 88, 25-57. FINKELSTEIN, A., RUBIN, L. L. & TZENa, M.-C. (1976). Black widow spider venom: effect of purified toxin on lipid bilayer membranes. Science 193, 1009-1011. FLORY, E. & KRIEBEL, M. E. (1983). Changes in acetylcholine concentration, miniature end-plate potentials and synaptic vesicles in frog neuromuscular preparations during lanthanum treatment. Comparative Biochemistry and Physiology 75C, 285-294. FRITZ, L. C. (1980). Black widow spider venom as a probe at the crustacean neuromuscular junction. Appendix B. Thesis, Rockefeller University. FRITZ, L. C. & MAURO, A. (1982). The ionic dependence of black widow spider venom action at the stretch receptor neuron and neuromuscular junction of crustaceans. Journal ofNeurobiology 13, 385-401. FRONTALI, N., CECCARELLI, B., GORIO, A., MAURO, A., SIEKEVITZ, P., TZENG, M. C. & HURLBUT, W. P. (1976). Purification from black widow spider venom of a protein factor causing the depletion of synaptic vesicles of neuromuscular junctions. Journal of Cell Biology 68, 462-479. GORIO, A., HURLBUT, W. P. & CECCARELLI, B. (1978). Acetylcholine compartments in mouse diaphragm: a comparison of the effects of black widow spider venom, electrical stimulation and high concentrations of potassium. Journal of Cell Biology 78, 716-733. HAIMANN, C., ToRRI-TARELLI, F., FESCE, R. & CECCARELLI, B. (1985). Measurement of quantal secretion induced by ouabain and its correlation with depletion of synaptic vesicles. Journal of Cell Biology 101, 1953-1965. HERRERA, A. A., GRINNELL, A. D. & WOLOWSKE, B. (1985). Ultrastructural correlates of experimentally altered transmitter release efficacy in frog motor nerve terminals. Neuroscience 16, 491-500. HEUSER, J. & MILEDI, R. (1971). Effect of lanthanum ions on function and structure of frog neuromuscular junctions. Proceedings of the Royal Society B 179, 247-260. 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., DENNIS, M. J., JAN, Y., JAN, L. & EVANS, L. (1979). Synaptic vesicle exocytosis captured by quick freezing and correlated with quantal transmitter release. Journal of Cell Biology 81, 275-300. ISRAEL, M. & LESBATS, B. (1981). Continuous determination by a chemiluminescent method of acetylcholine release and compartmentation in Torpedo electric organ synaptosomes. Journal of Neurochemistry 36, 1475-1483. ISRAEL, M. & MANARANCHE, R. (1985). The release of acetylcholine: from a cellular towards a molecular mechanism. Biology of the Cell 55, 1-14. KARNOVSKY, M. J. (1967). The ultrastructural basis of capillary permeability studied with peroxidase as a tracer. Journal of Cell Biology 35, 213-236. KATZ, B. & MILEDI, R. (1977). Transmitter leakage from motor nerve endings. Proceedings of the Royal Society B 196, 59-72. KRIEBEL, M. E. & FLOREY, E. (1983). Effect of lanthanum ions on the amplitude distributions of miniature endplate potentials and on synaptic vesicles in frog neuromuscular junctions. Neuroscience 9, 535-547. 216 B. CECCARELLI, W. P. HURLBUT AND N. IEZZI KRIEBEL, M. E. & GROSS, C. E. (1974). Multimodal distribution of frog miniature end plate potentials in adult, denervated, and tadpole leg muscle. Journal of General Physiology 64, 85-103. KUNO, M., TURKANIS, S. A. & WEAKLY, J. N. (1971). Correlation between nerve terminal size and transmitter release at the neuromuscular junction of the frog. Journal of Physiology 213, 545-556. LETINSKY, M. S., FISHBECK, K. H. & MCMAHAN, U. J. (1976). Precision of reinnervation of original postsynaptic sites in frog muscle after a nerve crush. Journal of Neurocytology 5, 691-718. MELDOLESI, J. (1982). Studies on a-latrotoxin receptors in rat brain synaptosomes: correlation between toxin binding and stimulation of transmitter release. Journal of Neurochemistry 38, 1559-1569. MELDOLESI, J., HUTTNER, W. B., TSIEN, R. Y. & POZZAN, T. (1984). Free cytoplasmic Ca2+ and transmitter release. Studies on PC 12 cells and synaptosomes exposed to a-latrotoxin. Proceedings of the National Academy of Sciences of the U.S.A. 81, 620-624. MELDOLESI, J., SCHEER, H., MADEDDU, L. & WANKE, E. (1986). Mechanism of action of a-LT: The presynaptic stimulatory toxin of black widow spider venom. Trends in Pharmacological Sciences 35, 213-236. MISLER, S. & FALKE, L. C. (1987). Dependence upon multivalent cations of quantal release of transmitter induced by black widow spider venom at the frog neuromuscular junction. American Journal of Physiology 253, C469-476. MISLER, S. & HURLBUT, W. P. (1979). Action of black widow spider venom on quantized release of acetylcholine at the frog neuromuscular junction: Dependence upon external Mg2+. Proceedings of the National Academy of Sciences of the U.S.A. 76, 991-995. NIciiOLLS, D. G., RuGOLO, M., SCOTT, I. G. & MELDOLESI, J. (1982). a-Latrotoxin of black widow spider venom depolarizes the plasma membrane, induces massive calcium influx, and stimulates transmitter release in guinea pig synaptosomes. Proceedings of the National Academy of Sciences of the U.S.A. 79, 7924-7928. PUMPLIN, D. W. & REESE, T. (1977). Action of brown widow spider venom and botulinum toxin on the frog neuromuscular junction examined with the freeze-fracture technique. Journal of Physiology 273, 443-457. RICE, S. 0. (1944). Mathematical analysis of random noise. Bell Telephone System Journal 23, 282-332. RUBIN, L. L. (1976). The neuromuscular junction in tissue culture: Studies on mechanisms of transmitter release. Thesis, Rockefeller University. SEGAL, J. R., CECCARELLI, B., FESCE, R. & HURLBUT, W. P. (1985). Miniature endplate potential frequency and amplitude determined by an extension of Campbell's theorem. Biophysical Journal 47, 183-202. SIGWORTH, F. J. (1981). Interpreting power spectra from nonstationary membrane current fluctuations. Biophysical Journal 35, 289-300. SMITH, J. E., CLARK, A. W. & KUSTER, T. A. (1977). Suppression by elevated calcium of black widow spider venom activity at frog neuromuscular junctions. Journal of Neurocytology 6, 519-539. STEINMAN, R. M., MELLMAN, I. S., MULLER, W. A. & COHN, Z. A. (1983). Endocytosis and the recycling of plasma membrane. Journal of Cell Biology 96, 1-27. TAUC, L. (1982). Nonvesicular release of neurotransmitter. Physiological Reviews 62, 857-893. TORRI-TARELLI, F., GROHOVAZ, F., FESCE, R. & CECCARELLI, B. (1985). Temporal coincidence between synaptic vesicle fusion and quantal secretion of acetylcholine. Journal of Cell Biology 101, 1386-1399. VALTORTA, F., JAHN, R., FESCE, R., GREENGARD, P. & CECARELLI, B. (1988). Synaptophysin (p38) at the frog neuromuscular junction: its incorporation into the axolemma and recycling after intense quantal secretion of neurotransmitter. Journal of Cell Biology (in the Press). VALTORTA, F., MADEDDU, L., MELDOLESI, J. & CECCARELLI, B. (1984). Specific localization of the a-latrotoxin receptor in the nerve terminal plasma membrane. Journal of Cell Biology 99, 124-132. VERVEEN, A. A. & DE FELICE, L. J. (1974). Membrane noise. Progress in Biophysics and Molecular Biology 28, 189-268. VYSKOCIL, F. & ILLES, P. (1979). Non-quantal release of transmitter at mouse neuromuscular junction and its dependence on the activity of Na+-K+-ATP-ase. Pflugers Archiv 370, 295-297. The Journal of Physiology, Vol. 402 Plate 1

B. CECCARELLI AND OTHERS (Facing p. 216) The Journal of Physiology, Vol. 402 Plate 2

kL'-i ' Cs<, E ~i

.t......

B. CECCARELLI AND OTHERS The Journal of Physiology, Vol. 402 Plate 3 A ~~~~~~B

7V%

'¾ 4

Cv JYt% NtU 2. t.j.,L< 3'

. !t ;t -.i

-. K) ¶ *2 4%

.t...)

:.~ J* ,kim .-.y -, I.^4,; .. , B. CECCARELLI ANqD OTHERS The Journal of Physiology, Vol. 402 Plate 4

gsvr rA:XtrXiL34> 8>,F~~~~~~i:>>j~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~Wjt3 ff*

B. CECCARELLI AND OTHERS a-LATROTOXIN AT LOW TEMPERATURE 217 WEIBEL, E. R. & PAUMGARTNER, D. (1978). Integrated stereological and biochemical studies on hepatic membranes. II. Correction of section thickness effect on volume and surface density measurements. Journal of Cell Biology 77, 584-597. WHITTAKER, V. P., MICHAELSON, I. A. & KIRKLAND, R. J. (1964). The separation of synaptic vesicles from nerve ending particles ('synaptosomes'). Biochemical Journal 90, 293-303. WOODWARD, M. P. & ROTH, T. F. (1979). Influence of buffer ions and divalent cations on coated vesicle disassembly and reassembly. Journal of Supramolecular Structure 11, 237-250.

EXPLANATION OF PLATES PLATE 1 Electron micrographs of longitudinal sections of portions of resting neuromuscular junctions bathed for 2 h at 2 °C in Ringer solution (A), Ca2 -free Ringer solution (B) or Ringer solution with Mg2+ (C). The general subcellular organization of the axonal endings (A) appears normal under the three conditions. The terminals are covered by Schwann cell processes (SC) and they contain mitochondria, neurofilaments, elements of smooth endoplasmic reticulum and normal complements of synaptic vesicles. Arrow-heads indicate interdigitations of Schwann cell processes that normally occur between the pre- and postsynaptic membranes. A-C, x 20000; scale marker 1 /im.

PLATE 2 Electron micrographs of longitudinal sections of portions of terminals treated with a-LTx (2 jug/ ml) for 2 h in Ringer solution (A), Ca2+-free Ringer solution (B) or Ringer solution with Mg2+ (C). The axon terminals contain swollen mitochondria, many large vesicular structures and some coated vesicles; their axolemmas are deeply infolded (arrow-heads), and they have lost most of their synaptic vesicles. Note that many of the infoldings in A and B traverse the axoplasm and are easy to recognize, whereas the numerous infoldings (arrow-heads) in C are less obvious because they are confined to the region of the axoplasm in the vicinity of the presynaptic membrane, which is marked by the thin line running along the tops of the post-junctional folds. A-C, x 20000; scale marker 1,um. PLATE 3 A and B, electron micrographs of transverse sections of portions of neuromuscular junctions from control muscles bathed for 2 h at 2 °C in Ringer solution (A) or in Ca2+-free Ringer solution (B). C, transverse section of a terminal from a preparation treated with a-LTx (2 #g/ml) for 2 h at 2 °C in Ca2+-free Ringer solution. The terminal and its mitochondria are swollen and the terminal is almost completely depleted of synaptic vesicles. D and E, transverse sections of nerve terminal branches from preparations treated at 2 °C with a-LTx in Ringer solution (A), or in Ca2+-free Ringer solution (B), with HRP. These terminals were fixed 60 min after the toxin had been applied. The junctional cleft and the extracellular space between the Schwann cell (SC) membrane and the axolemma contain rich deposits of reaction product. Only a few synaptic vesicles containing reaction product are present in these fields (circle). The dense large particles that occur in the clear region of the axoplasm in E are glycogen. A-E, x 20000; scale marker 1 /sm. PLATE 4 A, transverse section of two terminal branches from a preparation treated with a-LTx (2 ,ug/ml) for 2 h at 2 °C in Ringer solution with Mg2+. These branches have lost most oftheir synaptic vesicles and their axolemmas are infolded extensively (arrow-heads). B, high-power electron micrograph of a terminal from a preparation treated as the one in A. Many coated structures are evident in this field (arrows) and some pleomorphic small vesicular structures are seen near an active zone. C, transverse section of two terminal branches from a preparation that had been treated with a-LTx for 2 h at 2 °C in Ringer solution with Mg2+ and then warmed to 20 °C for 30 min. The terminals are swollen and totally depleted of synaptic vesicles, and no infoldings of the axolemma are evident (compare with A). The cytoplasmic matrix is less dense than in A or B. A and C, x 20000; scale marker 1 ,um. B, x 65000; scale marker 0 5 /tm.