zy zy SYNAF'SE 24:l-11 (1996) Low Calcium-InducedDisruption of Active Zone Structure and Function at the Frog Neuromuscular Junction zyxwv zyxwv zyx zyxwv STEPHEN D. MERINEY, BIRGIT WOLOWSKE, ELHAM EZZATI, AND ALAN D. GRINNELL Department of Neuroscience, University of Pittsburgh, Pittsburgh, Pennsylvania 15260 (S.D.M.); Jerry Lewis Neuromuscular Research Center, UCLA School of Medicine, Los Angeles, California 90024 (B.W, E.E., A.D.G.) KEY WORDS Freeze fracture, Transmitter release, Tetanic potentiation, Pairedpulse facilitation ABSTRACT Transmitter release from frog motor nerve terminals occurs at specialized sites on the nerve terminal called active zones (AZs). We have used a low calcium (0.1 nM) saline treatment to disrupt AZ structure and correlated these changes with alterations in transmitter release from the nerve terminal. Exposure to 0.1 nM free calcium saline for 3 h caused many individual AZs to break into two or three pieces, apparently unorganized particles drifted free of the AZ array, and the normally ordered alignment of AZ particles was loosened. Despite these forms of disruption in AZ organization, physiological function remained remarkably normal. Although the size of the endplate potential recorded in response to a single nerve stimulus was little affected, paired-pulse facilitation and tetanic potentiation were significantly increased. Synaptic depression was not apparent during the tetanus, but was revealed following the cessation of the stimulation. The results are consistent with the hypothesis that 0.1 nM calcium treatment detached AZ segments from the anchoring molecules that normally hold these proteins in alignment with other synapse-specific molecules. We propose that the ordered AZ organization serves to bring the calcium channels that regulate transmitter release in close proximity to other proteins that are critical to the modulation of release, especially during periods of high frequency stimulation. We hypothesize that the drifting AZ segments, although capable of apparently normal transmitter release, may not be tightly coupled with the intracellular calcium handling proteins that normally restrict the time that calcium ions have to act on the transmitter release apparatus following each action potential. 0 1996 Wiley-Liss, Inc. zyxwvuts INTRODUCTION The tightly regulated release of neurotransmitter is known to occur at active zones (AZs).Active zones are particularly conspicuous in freeze-fractured presynaptic membranes of frog motor nerve terminals, where there are regularly spaced double rows of large (approximately 10 nm) intramembranous proteins running across the width of the terminal on either side of a n elevated membrane ridge (Heuser et al., 1974, 1979; Peper et al., 1974; see Fig. 1). Comparison of these observations with those made using transmission electron microscopy reveals that each such AZ unit is associated with several "docked" synaptic vesicles and a localized cluster of additional vesicles in the terminal cytoplasm (Couteaux, 1974). AZs are also precisely aligned opposite postsynaptic junctional folds, where acetylcholine receptors are densely packed (Heuser and 0 1996 WILEY-LISS, INC. Reese, 1981). Direct evidence for physical intercellular connections of some type, particularly strong a t AZ sites, have been demonstrated in mammalian junctions where hypertonic saline is used to shrink the terminal. The pre- and postsynaptic membranes tend to pull away from each other except at AZ sites, where something continues to hold them together (Robbins and Polak, 1989). Whatever constitutes this connection very likely also contributes to the localization and anchoring of the AZ structures in the presynaptic membrane. Candidate filamentous strands in the extracellular matrix (ECM) that appear to attach selectively at the AZ in the presynaptic membrane have been visualized by Hirokawa and Heuser (1982) in deep-etch freeze-fracture cross sections of frog neuromuscular junctions. zyxw Received J u n e 25, 1995; accepted in revised form August 27, 1995. 2 S.D. MERINEY ET AL. zyxwvutsrqp zyxwvutsrqp zyxwv zyxwvut Fig. 1. Freeze-fracture through the “p” face of a control frog nerve terminal. High magnification (printed a t 90,OOOX for analysis) montage of a representative series of AZ segments. The active zones are defined as the two double rows of large (10nm) particles. Most of the parameters used for quantification of AZ integrity are labeled. Brackets identifying “AZ length” indicate a single AZ segment, DP = dispersed particles, calibration = 200 nM. It is widely accepted that quantal release occurs near the large intramembranous particles (Heuser et al., 1979), which are thought to represent both voltage sensitive calcium (Ca++)channels (Cohen et al., 1991; Pumplin e t al., 1981; Robitaille et al., 1990) and calciumactivated K+ channels (Robitaille and Charlton, 1992; Robitaille et al., 1993). The details of events inside the terminal that couple Ca+ influx to vesicle fusion and release are poorly understood, but are likely to involve interactions between immediately available Cat +-semitive molecules and one or more of the complex of membrane and cytoplasmic proteins implicated in vesicle docking and release, and depend critically on their proximity and organization (Jahn and Sudhof, 1994; Mastrogiacomo e t al., 1994; Sollner et al., 1993). I n addition, it may be essential that Cat+channels be in close proximity to other proteins in the active zone (Cat ’ activated potassium channels, local Ca++buffers, etc.) to account for cooperativity in release and the normal properties of paired-pulse facilitation, tetanic potentiation, tetanic and post-tetanic depression, and post-tetanic potentiation. Thus one might postulate that the high degree of organization of AZ structures in the presynaptic membrane is reflective of a similar organization inside the terminal, and that both are of importance. In fact, the importance of AZ integrity has been demonstrated in the study of the disease Lambert-Eaton Syndrome (LES). This autoantibody-mediated disease is characterized by a disruption of active zone organization a t the motor nerve terminal that results in both a breakdown in the alignment of AZ particles and a decrease in the total number of particles (Fukunaga et al., 1982, 1983). Associated with this disruption is a decrement in transmitter release characterized by a reduction in quantal content (Elmqvist and Lambert, 1968; see Vincent e t al., 1989 for review). Despite extensive study of the structure of the frog AZ, the functional significance of the high degree of organization is still poorly understood. In a n effort to better understand the relationship between AZ organization and physiological function, experimental protocols that cause a disruption of this organization have been developed. It has been reported that a 2-3-h treatment of neuromuscular preparations with a zero Ca++ Ringer causes severe disruption of active zones and dispersion of the 10 nm intramembranous particles (Ceccarelli e t al., 1979; however, see Pumplin, 1983). We confirm that 3 h of exposure to a n EGTA-buffered saline containing 0.1 nM free Ca++causes significant disruption of active zones, and have used such preparations to ask whether changes in AZ structure result in alterations in the physiology of synaptic transmission. Some of this work has been presented in preliminary form (Meriney et al., 1989). zyxwv zyxwvu zyxwvut MATERIALS AND METHODS Freeze-fracture electron microscopy Adult Rana pipiens frogs were anesthetized with 0.1% tricaine methanesulfonate (Sigma Chemical Co., St. Louis, MO) and pithed. Cutaneous pectoris nervemuscle preparations were dissected out and pinned to a thin layer of Sylgard in a bath of normal frog Ringer ACTIVE ZONE STRUCTURE AND FUNCTION zyxwv 3 zyxwvutsrq (NFR; 116 mM NaCl, 1mM NaHC03, 2 mM KC1, 1.8 CaC12,1mM MgCl2,5 mM Hepes, pH 7.4). Experimental preparations were either fixed immediately, incubated in control saline for 3 h, incubated in 0.1 nM free Ca++ saline for 3 h at room temperature, or incubated in 0.1 nM free Ca++for 3 h and then returned to control saline for 2 h before fixation. The 0.1 nM free C a t +saline had the following composition (in mM): 94 NaCl, 1NaHC03, 2 KC1, 10 Hepes, 0.02 CaC12,2.2 MgC12, 20 EGTA, 2.5 glucose, pH 7.2. Control saline was buffered to 1.8 mM free Ca++using the following composition (in mM): 64 NaC1, 1 NaHC03, 2 KC1, 10 Hepes, 21.8 CaC12, 1.0 MgC12,ZO EGTA, 2.5 glucose, pH 7.2. Free Ca++concentrations were calculated by a custom software package. All salines had a free Mg++ concentration of 1 mM. Preparations were fixed for 1h with 2% glutaraldehyde in NFR and washed in NFR. Small (0.5-1 mm) nerve terminal-containing pieces were processed by standard freeze-fracture methodology (Heuser et al., 1974; KO, 1981), and viewed in a Zeiss EM 109 transmission electron microscope at 80 kV. To analyze the organization of this structure, the cytoplasmic halves (“p”faces) of nerve terminal membrane leaflets were photographed and printed at low (9,000~) and high (90,000 x ) magnification for construction of montages. Low magnification montages were used to determine nerve terminal position relative to the nerve trunk and innervating axon and to verify that fractured nerve terminal segments were all on a single muscle fiber. High magnification montages were used to quantify details of AZ structure. For terminals in which a significant portion (greater than 50 Fm) of the total length had been fractured, we characterized the disruption of this structure by the parameters shown in Figures 1and 3. Based on a n average nerve terminal length in this preparation of about 675 pm (Propst and KO, 19871, 50 pm is roughly 8% of the terminal length. We studied eight terminals from six control untreated muscles with fractured terminal lengths that ranged between 58-102 p,m (mean i SEM = 75 i 15 pm); eight terminals from three muscles treated with a 3-h exposure to 0.1 nM free calcium with lengths ranging between 55-199 pm (mean = 87 ? 48); four terminals from two muscles treated with 0.1 nM free calcium followed by control saline with lengths ranging between 56-269 pm (mean = 127 i 86); three terminals from one muscle treated for 3 h with control saline with lengths ranging between 73-209 pm (mean = 155 i 59). AZs were defined as the complex of double rows of large (about 10 nm) intramembranous particles on either side of a membrane ridge remaining in the cytoplasmic half (“p” face) of the presynaptic lipid bilayer after it has been fractured. Distinct AZ pieces were quantified if they contained a continuous double row of 10 nm particles at least 200 nm in length, and were positioned at least 50 nm from any other AZ piece. AZ pieces of smaller length were grouped into a category of short AZ segments and their length was not quantified. The number of single rows ofAZ particles per ridge, and the number of ridges that contained apparently unorganized, or free, 10 nm intramembranous particles were also quantified. Electrophysiology Intracellular recordings of endplate potentials were performed a s previously described (Meriney and Grinnel, 1991). Briefly, cutaneous pectoris nerve-muscle preparations were incubated for 10 min in 2 mg/ml FITC-labeled peanut agglutinin (Sigma Chemical Co., St. Louis, MO) to stain the extracellular matrix covering nerve terminals (KO,1987). The preparation was then rinsed several times in NFR and pinned at 1.1times resting in situ length to the bottom of a Sylgard coated recording chamber (5 ml capacity). The muscle nerve was sucked into a stimulating electrode and stimulated with 100-200 psec pulses at 2-3 times the level necessary to evoke a maximal contraction. To prevent nerveevoked muscle contractions during experiments, the preparation was superfused continually ( 1ml/min) with 3-5 FM d-tubocurarine chloride (dTC) in NFR. Intracellular recordings of endplate potentials (EPPs) were made a t 15 2 1°C with glass microelectrodes (20-40 m a ) filled either with 3 M KCl or with 0.6 potassium acetate, 5 mM KC1, and 50 mM EGTA. Muscle cells were penetrated near endplates under visual control, alternatively using Hoffman amplitude modulation contrast optics and fluorescence microscopy at 200 x with a n Olympus 20x long-working distance waterimmersion objective. Endplate potentials were recorded on magnetic tape (RACAL) and analyzed off-line using a custom modification of the pClamp suite of programs (Axon Instruments, Foster City, CA). Recordings were made in NFR before, and 0.5-1 h following a 3-h treatment with 0.1 nM free Ca+’ saline. zyxw zyxwvuts zyxwvut zyxwvutsrq zyxwvutsrqpo RESULTS The normal structure of the AZ Control frog nerve terminals typically have well organized AZ structure a s visualized by freeze-fracture techniques. Figure 1 shows a freeze-fracture replica of a control nerve terminal a t high magnification (90,OOOX). Normally, the 10 nm intramembranous AZ particles are arranged in neatly organized rows that usually span the width of the nerve terminal (see “AZ l e n g t h in Fig. 1).In control terminals, a double row of AZ particles is conspicuous on either side of the crests of AZ ridges in p-face replicas. These particle arrays are arranged roughly perpendicular to the long axis of the terminal (see “angle” in Fig. 1) and spaced roughly 1pm apart (see “interzone distance” in Fig. 1). Structural disruption of active zone integrity Although control nerve terminals occasionally show some disruption in organization, treatment for 3 h with 4 zyxwvutsrqpon S.D. MERINEY ET AL zyxwvu zyxw zyxw zyxwvuts Fig. 2. Freeze-fracture through the "p" face of a 3 h 0.1 nM Ca"-treated frog nerve terminal. Most AZs show many of the characteristics of disrupted structure outlined in the text and Table I (SR = single rows; DP = dispersed particles; AZ segments are bracketed for one sample AZ ridge). Calibration = 200 nM. 0.1 nM free Ca++resulted in a clear-cut increase in the frequency with which a disruption of the normally ordered structure could be observed. Figure 2 shows a freeze-fracture replica taken from a 0.1 nM free Ca++treated nerve terminal. To quantify the disruption in AZ structure observed, we measured the number of AZ segments (>200 nm in length) per ridge, the percentage of ridges that contained very short ((200 nm) AZ pieces, the percentage of ridges that contained single rows of 10 nm particles (defined as either an unopposed double row of particles on one side of the AZ ridge, or a true single row of particles), the percentage of ridges that contained clusters of apparently unorganized particles, and the total amount of AZ per unit length. Representative examples of the characteristics of disruption that were quantified are indicated in Figures 1 and 2. In control nerve terminals, 91% of AZ ridges con- zyxw zyxwvu TABLE I. Disruption of AZ integrity is revealed as treatment-induced changes in the number of active zone pieces or segments per presynaptic ridee exuressed as a uercent of total' zyxwvuts zyxwvu zyxwvu ~~ Number ofAZ uieces uer ridge (% of total) No. AZ Dieces Control 3 h control 3 h 0.1 nM Ca-3 h 0.1 nM Ca' '; 2 h NFR 1 2 3 4 90.9 92.9 76.9 88.2 9.1 6.1 18.4 11.4 0 0.5 0 0 0.4 0 4 0.4 'For each of t h e four experimental conditions, t h e percentage of active zone ridges that contained one or more active zone pieces is shown. tained a single, long AZ segment (see Table I). Occasionally (9% of the time), control ridges had two AZ segments. In contrast, a 3-h treatment in 0.1 nM free Ca++ resulted in a significant increase in the percentage of ridges that showed more than one AZ segment (about zy zyxwvu zyx zyxw 5 ACTIVE ZONE STRUCTURE AND FUNCTION A AZ segments/ridge * T B Ridges with A2 fragments 20 16 4 9 12 0 k a 8 4 0 c Ridges with single rows n m * zyxw zyxwv zy D E dges with T free particles Control 3 h r . NFR 3 hr. 0.1 nM Calcium 3 hr. 0.1 nM Calcium then 2 hr. NFR Fig. 3. Quantification of the disruption of active zone integrity. A Control active zones have a n average of 1.08 active zone segments per ridge (open bar). A 3-h exposure to control saline (hatched bar) does not affect this structural arrangement, however, a 3-h treatment with 0.1 nM free Ca" significantly increases that number to 1.27 segments per ridge (dense hatched bar). This effect can be reversed with a recovery exposure to control saline for a n additional 2 h (solid bar). B Only 4.2% of control active zone ridges (open bar) have very small active zone fragments (less than 200 nm in length), and this number does not change with a 3-h exposure to control saline (hatched bar). However, a 3-h exposure to 0.1 nM free Ca-- significantly increases this percentage to 16 (dense hatched bar), and this disruption shows only slight recovery following a subsequent 2-h exposure to control saline (solid bar). C: Only 5% of control active zone ridges have single rows of particles, but this number can be significantly increased by a 3-h incubation in control saline (hatched bar) or 0.1 nM free Ca" (densely hatched bar), and this form of disruption is not significantly different following a further 2-h exposure to control saline (solid bar). D. Seventeen percent of control active zone ridges have clusters of free 10 nm particles that do not appear to be organized (open bar). Although this number has a tendency to increase following incubation in control saline (hatched bar), the disruption is not significantly different from control unless the incubation is performed in 0.1 nM free Ca" (dense hatched bar). This form of disruption does not recover following a subsequent exposure to control saline for 2 h (solid bar). * = significantly different from control ( P < 0.05) using a one-way analysis of variance. 23%; see Table I). This disruption in 0.1 nM free Cat+ showed some reversal following a 2-h return to control saline (see Table I). When the mean numbers of AZ segments per ridge were compared, 0.1 nM free Ca+' treatment significantly increased the number from 1.08 in control terminals, to 1.27 (see Fig. 3A). Furthermore, the percentage of AZ ridges that contained very small ( ~ 2 0 nm) 0 AZ segments also increased from about 4% to greater than 16%,and this effect only partially reversed following a 2-h return to control saline (Fig. 3B). Therefore, it is clear that the normally continuous array of 10 nm intramembranous particles spanning most of the width of the nerve terminal along the crest of the AZ ridge, was fragmented into smaller pieces by exposure to a very low free Ca++saline. Control AZ ridges occasionally (about 5% of the time) had single rows of 10 nm particles that did not appear to have parallel counterparts. The percentage of AZ ridges with single rows increased to about 25% in nerve terminals that had been incubated for 3 h in either control saline or 0.1 nM free Ca++saline, and this form of disruption persisted following a further 2-h exposure to control saline (see Fig. 3C). The appearance of clusters of apparently unorganized particles also had a tendency to increase in terminals bathed in control saline for 3 h (Fig. 3D), but this increase was not statistically significant. This form of disruption did increase significantly when terminals were exposed to 0.1 nM free Ca++,and became even more prevalent when these treated terminals were returned to control saline for a n additional 2 h (Fig. 3D). Even though the individual significant differences zyxw zyxwvutsrqpo 6 zyxwvutsrqp zyxwvutsrqp zyxwvu S.D. MERINEY ET AL described above are modest when examined in isolation (see Fig. 3), taken together, these disruptions of AZ structure are a n obvious departure from normal. Furthermore, the data presented in Figure 3 represent only the quantification of specific AZ structural characteristics and do not convey every aspect of the disruption in organization. The clear disruption of AZ structure is prominent when one compares Figures 1 and 2. For example, although this parameter was not quantified, nerve terminals treated with 0.1 nM free Cat+ showed a n apparent decrease in the precision with which 10 nm particles were aligned in their ordered arrays. Control nerve terminal AZ segments were constructed of ordered arrays of particles that appeared precisely aligned with one another. A 3-h treatment in low Ca++ appeared to loosen that alignment, so that the AZ segment often appeared wavy and disordered, but with 10 nm particles still within 50 nm of each other, and as such, not classified as disrupted by our definitions outlined above. Although this type of disruption did not appear in our quantification, it is probable that this loosening is reflective of a significant disruption of organization. Despite these forms of disorganization, measurements of the total amount of AZ/km of terminal length, i.e., the summed length of the long and short fragments showing a t least one row of 10 nm particles on either side of a n intramembranous ridge, was not statistically different in 0.1 nM Ca++-treated(1.27 t- 0.22pm AZ/ pm terminal length) vs. control (1.09 5 0.22) muscles. Thus the low Ca++treatment does not appear to cause a n absolute loss or internalization ofAZ structures, only their dispersion or disorganization in the membrane. 1.0 I 0 zyxwz I I I I I I I 5 10 15 20 25 30 35 1 1 1 40 45 50 Interval (msec) Fig. 4. Paired-pulse facilitation measured i n control (filled circles) and experimental neuromuscular junctions treated for 3 h in 0.1 nM free Ca", and then returned for 1 h to NFR with 5 pM dTC (open circles). A t all intervals tested (5-50 msec.) a 3-h treatment with 0.1 nM free Ca+- significantly increased the magnitude of pairedpulse facilitation. zyxwvu zyxwvut zyxwvut zyxwvutsrq zyxwv Physiological consequences of active zone disruption Despite significant disruption of AZ structure, alterations in physiological function were surprisingly limited. In fact, EPPs recorded in curarized preparations and compared before and after the 0.1 nM free Ca++ treatment did not differ in latency or time course, and were quite similar in amplitude, although the mean was slightly reduced (4.95 2 3.5 mV after 0.1 nM C a t + vs. 5.2 ?I 3.3 mV in controls; n = 27 neuromuscular junctions in four muscles). Although transmission did not appear affected when examined following a single stimulus, we used paired pulses and trains of stimuli to examine more complex features of release. The response of endplate potentials to paired-pulse stimuli of varying inter-pulse intervals is plotted in Figure 4. Control endplates (filled circles, Fig. 4)showed significant paired-pulse facilitation at intervals between 5 and 50 msec. Preparations treated with 0.1 nM free Ca++ saline displayed significantly larger paired-pulse facilitation than controls at all intervals tested (open circles, Fig. 4). This difference in potentiation was even more con- spicuous in recordings of EPPs during trains of stimuli. Figure 5 shows EPPs recorded in response to 20 Hz nerve stimulation from a representative endplate before (Fig. 5A) and following a 3-h exposure to 0.1 nM free Ca++saline (Fig. 5B). In control nerve terminals, EPP size usually increased during the first few stimuli (tetanic potentiation), and subsequently decreased in size (tetanic depression). Strikingly, following a 3-h exposure to 0.1 nM free Ca++saline and return to control saline for 0.5 to 1h, the tetanic potentiation that developed in the first few stimuli persisted throughout the stimulation period with little or no apparent depression. The mean response measured in 10 endplates (before and after treatment) from one experiment is plotted in Figure 6. In addition, Figure 6 plots recovery after the 20 Hz tetanus tested with stimulation a t 0.5 Hz. Control endplates showed no significant change in the depressed EPP in the first post-tetanic test stimulus (5 seconds after the tetanic stimulus was terminated), but a rapid recovery of EPP size that, within 5-10 seconds, was within control ranges. I n contrast, following treatment with 0.1 nM free Ca++saline, the first post-tetanic stimulus pulse resulted in a n EPP that was significantly depressed to near control depression levels (see open circles in Fig. 6). This depressed EPP rapidly potentiated over the next 5-10 seconds to a amplitude that was significantly larger than the control EPP size. Therefore, despite apparently normal release measured following a single nerve stimulus, endplates treated with 0.1 nM free Ca++saline showed increased paired- ACTIVE ZONE STRUCTURE AND FUNCTION zyxwv zy zyx zy 7 5 0 0 ms ‘ 7 I 3mV zyxwvut Fig. 5. Endplate potentials recorded from a muscle fiber during 20 Hz nerve stimulation in NFR containing 5 )LMdTC. A: A control neuromuscular junction displays a characteristic facilitation following the first few stimuli, followed by a depression of release. B Following a 3-h treatment in 0.1 nM free Ca*’, and 1 h after return to NFR plus 5 )LM dTC, 20 Hz stimulation results in facilitation that is maintained over the course of the stimulation. o 0 zyxwv Low calcium treatment Control zy zyxwvu zy 1 3 ’ ’ 6 8 10 12 14 1 2 3 5 6 7 8 91011 91-100 I / 4 20 Hz Stimulus Number / I 2 4 I I f I Recovery at 0.5 Hz (sec) Fig. 6. Plot of the mean response of 10 nerve terminals to 20 Hz nerve stimulation. Although endplate potential (EPP) size (expressed as a percentage of control EPP size) of control nerve terminals (filled circles) shows facilitation over the first 11 stimuli, pretreatment with 0.1 nM free Ca++shows significantly greater facilitation. Furthermore, this facilitation is maintained only in the terminals pretreated for 3 h with 0.1 nM free Ca*’ (see average EPP size for the last 10 stimuli in the train: stimuli #91-100). Interestingly, although treated terminals do not show significant depression of EPP size during the 20 Hz 5 sec train, there is a significant depression of transmitter release during the first few seconds of recovery immediately following the train. This post-tetanic depression is not significant in control terminals, and is quickly overcome by a post-tetanic potentiation that is significantly greater in terminals treated with 0.1 nM free Ca“ for 3 h. pulse facilitation, tetanic potentiation, and post-tetanic potentiation. The presence of normal tetanic depression was revealed in low Ca++-treatedpreparations only following the removal of tetanic nerve stimulation. control preparations, there are instances in which this organization has been disrupted, especially near the ends of terminal branches (see Pumplin, 1983; Pawson and Grinnell, unpublished observations). Presumably this disruption in AZ organization in control preparations, particularly at the ends of nerve terminal branches, reflects remodeling associated with ongoing extension and retraction of terminal branches. This type of remodeling has been described during alterations in AZ morphology associated with changes in activity of animals in “summer”vs. “winter” frogs (Dor- DISCUSSION Frog motor nerve terminal active zones are normally highly organized linear arrays of intramembranous proteins precisely aligned both with respect to acetylcholine receptor accumulations in the muscle membrane and with intraterminal release-associated proteins. In 8 zyxwvu zyxwvu zyxwvutsrq S.D. MERINEY ET AL lochter et al., 1993). Similar disorganizations has been described a t regenerating frog neuromuscularjunctions (KO,1984). Surprisingly, it is possible to disrupt AZ organization without grossly affecting transmitter release. Ceccarelli and his colleagues (1979) reported that when a frog neuromuscular preparation was bathed in zero Ca++ saline for 2-3 h, the AZs became highly disorganized, breaking into fragments that drifted apart in the presynaptic membrane and lost their normal orientation. This disruption persisted following a 1h return to Ca++containing saline a t which time recordings of EPPs showed normal quantal contents and mEPP frequency (Haimann e t al., 1980). On the other hand, they reported a small drop in mean quantal size, which was interpreted as reflecting the fact that some release sites were further from acetylcholine receptor concentrations (Haimann e t al., 1980). Pumplin (1983) has disputed the morphological finding, reporting no greater disruption of AZs in zero Ca++saline than in control saline. Here, we quantify clear-cut disruption of AZ organization following treatment with a saline that buffers Ca++ to 0.1 nM. The number of AZ segments per ridge increased as AZs broke into smaller pieces (Fig. 3A,B). The fragmentation of large pieces showed essentially complete reversal following a 2-h return to control saline (Fig. 3A). In contrast, when small AZ pieces (<200 nm) broke free, they were much less likely to re-connect during 2-h return to control saline (Fig. 3B). The presence of single rows of AZ particles cannot be attributed to the low Ca++treatment, since prolonged exposure to control saline was almost equally effective at creating this form of disruption (Fig. 3C). The presence of dispersed AZ particles, apparently free-floating in the membrane, was significantly increased by exposure to low-Ca++saline, although partial disorganization of this type increased with time in control saline, and a return to control saline after 0.1 nM Ca++treatment did not prevent further dispersion (Fig. 3D). Presumably these AZ particles were completely dissociated from their original organized arrangement, and despite a 2-h return to control saline, they were unable to be reorganized. The low Ca++(0.1 nM) treatment may be able to disrupt connections that extend between AZ proteins and the postsynaptic membrane near the tops of the junctional folds (Hirokawa and Heuser, 1982; Robbins and Polak, 1989), or, more likely, from the corresponding part of the basal lamina in both directions. I t seems reasonable to suggest that this connection is composed of one or more Ca++-sensitiveadhesion molecules t h a t serve to direct the alignment of pre- and postsynaptic elements. Despite significant disruption of AZ organization, physiological function was remarkably robust. The principal physiological differences between 0.1 nM Ca++-treatedand control preparations were a slightly reduced mean EPP amplitude, increased paired pulse facilitation, and sustained tetanic potentiation. The slight reduction in mean EPP amplitude might be a reflection of the smaller mean mEPP size reported in similarly treated preparations by Haimann et al. (1980). There is no evidence for significantly reduced quantal content following treatment with low Ca++salines. If the quantal contents were sharply decreased, one would have expected enhanced facilitation and reduced depression. However, Haimann et al. (1980) found normal quantal contents in such preparations, and we observed enhanced facilitation in identified junctions studied both before and following treatment in which the EPP size did not change. Additionally, 0.1 nM Ca++-treatedjunctions revealed close to control levels of depression following the cessation of tetanic stimulation (see Fig. 5). These data show that the increased tetanic potentiation was not associated with reduced depression. I t seems probable, therefore, that these physiological abnormalities represent true departures from normal since they dissociate increases in paired-pulse facilitation and tetanic potentiation from changes in tetanic and post-tetanic depression. Disruption of these features of the release process might be expected if components of the release apparatus are not lost, but simply drift apart sufficiently that properties of release that depend on cooperativity would be affected. The disease LES is characterized by both disruption of AZ organization and a loss of AZ particles (Fukunaga et al., 1982,1983).Under these conditions, quantal content drops significantly (Vincent e t al., 1989). Similarly, comparisons of AZ structure in fast vs. slow frog neuromuscular junctions (Verma and Reese, 1984) or twitch vs. tonic lizard neuromuscular junctions (Walrond and Reese, 1985)have identified differences in AZ structure that correlate with physiological properties in these junctions. However, as with LES antibody-treated preparations, physiological differences here are most likely to be due to the large differences in the total number of AZ particles per terminal length rather than differences in the organization of those particles. Furthermore, comparisons made between nerve terminals in different species, or types of neuromuscular junctions within a single species, are complicated by potentially large differences in organization and function of the intraterminal proteins that buffer Cat+ and regulate exocytosis. In fact, single types of neuromuscular junctions can show significant differences in structure-function relationships when examined a t different developmental ages, or following regeneration. At newly regenerated neuromuscular junctions, AZs are small and poorly organized, and evoked release of transmitter recovers before reinnervation is complete such that release per unit length of the nerve terminal is initially higher than in control junctions and is occurring from AZ structures t h a t are smaller (KO,1981, 1984; Ding, 1982b). At endplates that reinnervate ectopic sites on muscle, nerve terminals are very small, yet quantal zyxwvutsr zyxwvut ACTIVE ZONE STRUCTURE AND FUNCTION content is often five times larger than in control reinnervated junctions (Ding, 1982a). Thus, AZ structure-function relationships can be complicated by intraterminal differences in biochemical organization and, as such, are difficult to interpret when comparisons are made between different types of nerve terminals or even between similar terminals of different developmental age. In contrast, we have studied acute changes in AZ structure and function within a single population of nerve terminals. We see only a disruption of AZ organization, and this is not associated with a significant drop in quanta1 content (see Ceccarelli et al., 1979; Haimann et al., 1980), but rather a relatively subtle alteration in physiological function-increases in paired-pulse facilitation and tetanic potentiation. Our observations, in comparison with acute effects in LES antibody-treated preparations, effectively dissociate physiological effects of AZ disruption with loss of AZ particles. Since disruption in the proximity of some AZ particles to one another does not significantly affect the magnitude of release following a single stimulus, it is possible that the high degree of AZ organization a t the frog neuromuscular junction is important not so much for bringing Ca++ channels in close proximity to one another, as in bringing individual transmitter release-associated Ca++ channels in close proximity to other important synaptic proteins. In fact, AZ regions of frog nerve terminals have been shown to contain calcium-activated K' channels, as well as Cat t channels (Cohen, 1991; Robitaille and Charlton, 1992; Robitaille et al., 1990; Robitaille et al., 1993; see also Roberts et al., 19901, and certainly include a variety of other proteins including those involved in vesicle docking and fusion (see Sollner et al., 1993).If AZs are organized primarily to bring a heterogenous group of proteins together for interaction, and not to bring multiple C a t t channels in close proximity to one another, transmitter release must be capable of being evoked by the flow of Ca++through a single Ca++ channel. Stanley (1993) has presented evidence that transmitter release is capable of being evoked by the opening of single Ca++channels. At a single frog neuromuscular junction, there are about 500 AZs that collectively release several hundred vesicles following each action potential (Katz and Miledi, 1979). Since release appears to be well distributed along the length of nerve terminals (DAlonzo and Grinnell, 1985), this implies that each AZ releases no more than one or two vesicles, and often may not release any transmitter following each action potential. Since there are about 300 10-nm particles within each AZ,perhaps one half or fewer of which may represent Ca++channels, several scenarios are possible. At one extreme, most of these Ca++channels open during each action potential and there is a low probability that Ca++ion entry leads to transmitter release. At the other extreme, only one or two percent of the Cat+channels in a n AZ open following each action potential, and following Cat+ entry, the probability that 9 a vesicle is released is high. Some combination of these two extreme hypotheses is probable. At the frog neuromuscular junction, N-type C a t + channels are thought to regulate evoked transmitter release (Kerr and Yoshikami, 1984). N-type Ca++channels have been shown to display several modes of gating, some with low open probability (Delcour and Tsien, 1993; Rittenhouse and Hess, 1994). I n frog motoneuron-muscle co-cultures, we have observed that Cat+ channels expressed in transmitter-releasing regions of neurites have a very low probability of opening during short voltage depolarizations (Meriney and Grinnell, unpublished observations). Taken together, the above information leads us to favor the following hypothesis: an action potential induces, on average, the opening of only a few of the Ca+ channels that exist at a n AZ and the flow of Cat+ through these individual Ca+' channels can, a significant fraction of the time, induce transmitter release. Other proteins, normally clustered near this releaseassociated Cat+ channel in the AZ, are positioned both to initiate vesicle fusion and to remove free Ca++,terminating the transmitter release process. It is possible that the low Ca+' treatment significantly reduces the intraterminal Ca++concentration such that i t does not recover in NFR before physiological recordings are made 0.5-1 h later. However, we feel that this is not a likely explanation for the physiological effects observed since a reduced intraterminal Ca++concentration would be expected to decrease, not increase, paired-pulse facilitation. On the other hand, our data are consistent with the hypothesis that the disruption in the anchoring of AZ pieces following exposure to low Ca+' saline results in a decrease in the normally tight association of AZ proteins with presumptive intraterminal Ca++ binding proteins. This physical disruption would be expected to alter the handling of C a t + ions following each action potential. A reduction in C a t + handling would result in the increase in paired-pulse facilitation and tetanic potentiation observed. Both paired-pulse facilitation and tetanic potentiation are short-term forms of synaptic plasticity that are thought to be dependent on residual Ca+t (see Zucker, 1993, for review). The relationship between these forms of plasticity and residual Ca++suggests that these effects are due to Cat+ acting at a different site than the lowaffinity site(s ) that are thought to trigger transmitter release (Zucker, 1993). Recently, several Cat+ binding proteins have been identified in neurons, and at synaptic sites, that may modulate the transmitter release process by acting a s local Ca+' buffers (Michaelis et aI., 1983; Roberts, 1993; Roberts, 1994; Tolosa de Talamoni et al., 1993). When appropriately positioned, these proteins may both restrict the region in which the internal Cat+ ion concentration exceeds 1 pM, and shorten the time that intracellular Cat+has to act on the transmitter release apparatus. Support for this interpretation appears strongest when one compares the size of the + zyxwv zyxwvutsrqp 10 S.D. MERINEY ET AL. zyxwv Ding, R. (198213) Lack of correlation between physiological and morphological features of regenerating frog neuromuscular junctions. Brain Res., 253:47-55. and Wernig, A. (1993) Acetylcholine recepDorlochter, M., Meurer, S., tor bars and transmitter release in frog neuromuscular junctions. Neuroscience, 52:987-999. Elmqvist, D., and Lambert, E.H. (1968)Detailed analysis ofneuromuscular transmission in a patient with the myasthenic syndrome sometimes associated with bronchogenic carcinoma. Mayo Clin. Proc., 43:689-713. Fukunaga, H., Engel, A,, Osame, M., and Lambert, H. (1982) Paucity and disorganization of presynaptic active zones in the LambertEaton myasthenic syndrome. Muscle Nerve, 5:686-697. Fukunaga, H., Engel, A,, Lang, B., Newsom-Davis, J., and Vincent, A. (1983) Passive transfer of Lambert-Eaton myasthenic syndrome with IgG from man to mouse depletes the presynaptic active zones. Proc. Natl. Acad. Sci. U.S.A., 80:7636-7640. Haimann, C., Grohovaz, F., Hurlbut, P., andceccarelli, B. (1980)Possible significance of changes in the structure of the active zones induced by Ca2+-freesolutions. In: Ontogenesis and Functional Mechanisms of Peripheral Synapses. J. Taxi, ed. Elsevier/North-Holland Biomedical Press, New York, pp. 157-170. Heuser, J.E., and Reese, T.S. (1981)Structural changes after transmitter release at the frog neuromuscular junction. J . Cell Biol., 88: 564-580. Heuser, J.E., Reese, T.S., and Landis, D.M.D. (1974) Functional changes in frog neuromuscular junctions studies with freeze-fracture. J . Neurocytol., 3:109-131. Heuser, J.E., Reese, T.S., Dennis, M.J., Jan, Y., Jan, L., and Evans, L. (1979) Synaptic vesicle exocvtosis captured bv auick freezinp.and correlated -with quantal transmitter' release: i. Cell Biol, 81: 275-300. Hirokawa, N., and Heuser, J.E. (1982)Internal and external differentiations of the postsynaptic membrane at the neuromuscular junction. J. Neurocytol., 11:487-510. Jahn, R., and Sudhof, T.C. (1994) Synaptic vesicles and exocytosis. . . Ann. Rev. Neurosci., 17:219-246. Katz, B., and Miledi, R. (1979) Estimates of quantal content during 'chemical potentiation' of transmitter release. Proc. R. SOC.Lond. [Biol.] 205:369-378. Kerr, L.M., and Yoshikami, D. (1984) A venom peptide with a novel presynaptic blocking action. Nature, 308:282-284. KO, C.-P. (1981) Electrophysiological and freeze fracture studies of changes following denervation a t the frog neuromuscular junctions. J. Physiol. (Lond.), 321:627-639. KO,C.-P. (1984) Regeneration of the active zone a t the frog neuromuscular junction. J. Cell Biol., 98:1685-1695. KO,C.-P. (1987)A lectin, peanut agglutinin, as a probe for the extracelACKNOWLEDGMENTS lular matrix in living neuromuscular junctions. J. Neurocytol., 16567-576. We thank Mike Kreman for technical assistance, and Mastrogiacomo, A., Parsons, S.M., Zampighi, G., Jenden, D.J., UmDrs. S. Sesack and G. Barrioneuvo for critical evaluabach, J.A., and Gundersen, C.B. (1994) Cysteine string proteins: a potential link between synaptic vesicles and presynaptic calcium tion of the manuscript. This work was supported by a channels. Science, 263:98 1-982. National Science Foundation Grant BNS 8719613 Meriney, S.D., and Grinnell, A.D. (1991)Endogenous adenosine modulates stimulation-induced depression at the frog neuromuscular (A.D.G.) and National Institutes of Health Grants junction. J . Physiol., 443:441-455. NS06232 (A.D.G.) and NS32345 (S.D.M.). Meriney, S.D., Pawson, P.A., and Grinnell, A.D. (1989) Presynaptic active zone integrity and ACh release from frog motor nerve terminals. SOC.Neurosci. Abst., 15258. REFERENCES Michaelis, E.K., Michaelis, M.L., Chang, H.H., and Kitos, T.E. (1983) Ceccarelli, B., Grohovaz, F., and Hurlbut, W.P. (1979) Freeze-fracture High affinity Ca2'-stimulated Mg2+-dependentATPase in rat brain studies of frog neuromuscular junctions during intense release of synaptosomes, synaptic membranes, and microsomes. J . Biol. neurotransmitter. I. Effects of black widow spider venom and Ca2+ Chem., 258:6101-6108. free solutions on the structure of the active zone. J . Cell Biol., Peper, K., Dreyer, F., Sandri, C., Akert, K., and Moor, H. (1974) Struc81:163-177. ture and ultrastructure of frog motor endplate. Cell Tissue Res., Cohen, M.W., Jones, O.T., and Angelides, K.J. (1991) Distribution of 149:437-455. Ca+ channels on frog motor nerve terminals revealed by fluorescent Propst, J.W., and KO, C.-P. (1987) Correlations between active zone w-conotoxin. J. Neurosci., 11:1032-1039. ultrastructure and synaptic function studied with freeze-fracture of Couteaux, R. (1974) Remarks on the organization of axon terminals physiologically identified neuromuscular junctions. J. Neurosci., in relation to secretory processes a t synapses. In: Advances in Cyto7:3654-3664. pharmacology, vol. 2. B. Ceccarelli, F. Clementi, and J . Meldolesi, Pumplin, D.W. (1983) Normal variations in presynaptic active zones eds. Raven Press, New York, pp. 369-379. of frog neuromuscular junctions. J . Neurocytol., 12:317-323. DAlonzo, A.J., and Grinnell, A.D. (1985) Profiles of evoked release Pumplin, D.W., Reese, T.S., and Llinas, R. (1981) Are the presynaptic along the length of frog motor nerve terminals. J. Physiol. membrane particles the calcium channels? Proc. Natl. Acad. Sci. (Lond.), 359:235-258. U.S.A., 78:7210-7213. Delcour, A.H., and Tsien, R.W. (1993) Multiple gating modes ofN-type Rittenhouse, A.R., and Hess, P. (1994) Microscopic heterogeneity in Ca2+channel activity distinguished by differences in gating kinetics. unitary N-type calcium currents in rat sympathetic neurons. J. J. Neurosci., 13:181-194. Physiol. (Lond.), 474:87-99. Ding, R. (1982a) Comparison of morphology and physiology ofsynapses Robbins, N., and Polak, J. (1989) Focal adhesions of nerve terminal formed a t ectopic and original endplates sites in frog muscle. Brain to synaptic matrix and schwann cell a t mouse neuromuscular juncRes., 253:57-63. tions. SOC. Neurosci. Abst., 15:258. EPPs at the end of, and immediately following, the cessation of a 20 Hz tetanic stimulus train in control and low Ca"-treated preparations (Fig. 6). Low (0.1 nM) Ca"-treated preparations showed maintained potentiation throughout the tetanus that rapidly gave way to depression immediately following the cessation of the tetanus. This observation appears most consistent with a decrease in the local Ca++buffering capacity of low Ca++-treatednerve terminals. Ifthis were to occur, Cat+ entering the nerve terminal during each action potential would be buffered at a slower rate than in control terminals. In this case, a n increase in facilitation and tetanic potentiation would be expected, while at the cessation of the tetanus, as the residual Ca++became buffered, depression would be unmasked. These data might also be expected if the nerve terminal calcium handling mechanisms are down-regulated in response to the 3-h exposure to very low extracellular calcium, independent of the relationship of AZ pieces to intracellular nerve terminal components. If this is the case, this compensatory response must require longer than 1h to recover following a return to normal calcium levels. In any event, the AZ disruption described here does not significantly alter release triggered by single action potentials a t low frequency. Departures from normal physiology observed with paired pulses and during high frequency trains are either a compensatory response of the nerve terminal to the low calcium exposure, or a consequence of the disruption in AZ structure. Both of these hypotheses explain physiological effects by alterations in the mechanisms of handling of calcium within the nerve terminal. zyxwvu zy zyxwvut ~ zyxwvutsrqpon ACTIVE ZONE STRUCTURE AND FUNCTION Roberts, W.M. (1993) Spatial calcium buffering in saccular hair cells. Nature, 363:74-76. Roberts, W.M. (1994) Localization of calcium signals by a mobile calcium buffer in frog saccular hair cells. J . Neurosci., 14:3246-3262. Roberts, W.M., Jacobs, R.A., and Hudspeth, A.J. (1990) Colocalization of ion channels involved in frequency selectivity and synaptic transmission at presynaptic active zones of hair cells. J . Neurosci., 10:366&3684. Robitaille, R., and Charlton, M.P. (1992) Presynaptic calcium signals and transmitter release are modulated by Caz'-activated potassium channels. J. Neurosci., 12:297-305. Robitaille, R., Adler, E.M., and Charlton, M.P. (1990)Strategic localization of calcium channels a t transmitter release sites of frog neuromuscular synapses. Neuron, 5:773-779. Robitaille, R., Garcia, M.L., Kaczorowski, G.J., and Charlton, M.P. (1993) Functional colocalization of calcium and calcium-gated potassium channels in control of transmitter release. Neuron, 11:645-655. Sollner, T., Bennett, M.K., Whiteheart, S.W., Scheller, R.H., and Rothman, J.E. (1993) A protein assembly-disassembly pathway in vitro that may correspond to sequential steps of synaptic vesicle docking, activation, and fusion. Cell, 75409418. 11 Stanley, E.F. (1993) Single calcium channels and acetylcholine release at a presynaptic nerve terminal. Neuron, 11:1007-1011. Tolosa de Talamoni, N., Smith, C.A., Wasserman, R.H., Beltramino, C., Fullmer, C.S., and Penniston, J.T. (1993) Immunocytochemical localization of the plasma membrane calcium pump, calbindin-Dz8K, and parvalbumin in purkinje cells of avian and mammalian cerebellum. Proc. Natl. Acad. Sci. U.S.A., 90:11949-11953. Verma, V., and Reese, T.S. (1984) Structure and distribution ofneuromuscular junctions on slow muscle fibers in the frog. Neurosci., 12:647-662. Vincent, A,, Lang, B., and Newsom-Davis, J . (1989) Autoimmunity to the voltage-gated calcium channels underlies the Lambert-Eaton myasthenic syndrome, a paraneoplastic disorder. Trends Neurosci., 12-496-502. Walrond, J.P., and Reese, T.S. (1985) Structure of axon terminals and active zones at synapses on lizard twitch and tonic muscle fibers. J.Neurosci., 5:1118-1131. Zampighi, G., and Kreman, M. (1985) Intercellular fibrillar skeleton in the basal interdigitations of kidney tubular cells. J. Membr. Biol., 88:33-43. Zucker, R.S. (1993) Calcium and transmitter release. J. Physiol. (Paris), 87:25-36.