THE JOURNAL OF COMPARA-
NEUROLOGY 304135-146 (1991)
Variation in Temninal Morphology and m p t i c Inhibition at Crustacean Neuromscular Junctions F.W. TSE, L. MARIN, S.S. JAHROMI, AND H.L. ATWOOD Department of Physiology, University of Toronto, Toronto, Ontario, Canada M5S 1A8
ABSTRACT Synaptic terminals of excitatory and inhibitory neurons supplying muscle fibers in leg muscles of crabs (Pachygrapsus crassipes and Hyas areneus) were investigated with light and electron microscopy. Terminals responsible for large excitatory postsynaptic potentials (EPSPs) at low frequencies of activation had a compact configuration with clusters of terminal boutons radiating from the main axon branch. Terminals responsible for small EPSPs had a more diffuse organization, with boutons often arranged in series along thin axon branches. Inhibitory neurons, when activated, produced both presynaptic and postsynaptic inhibitory effects, with the former being more potent at low frequencies of activation. Presynaptic inhibition was variable in magnitude but was generally strong in fibers with large EPSPs. Representative terminals from regions of strong and weak presynaptic inhibition were identified by activity-dependent uptake of horseradish peroxidase, serially sectioned, and reconstructed from electron micrographs. Both regions were found to contain axo-axonal synapses from inhibitory to excitatory terminals, with a larger number in the region of strong presynaptic inhibition. In addition, axo-axonal synapses were more uniformly distributed in the latter region. The number of inhibitory presynaptic dense bars (active zones) was somewhat higher in the region of weak inhibition, but larger individual dense bars occurred in the region of strong inhibition. Possible factors contributing to the differences in strength of inhibition include: (1) morphology and electrical properties of terminals; and (2) high probability of transmission at a relatively small number of inhibitory synapses during low frequency activation in the region of strong inhibition. Key words: vesicle, axo-axonal,bouton, synapse,postsynapticpotential
Presynaptic inhibition at crustacean neuromuscular junctions, first described by Dudel and Kufler (’61))is believed to be mediated by axo-axonal synapses made by peripheral inhibitory axons on terminals of the excitatory axons (Atwood and Jones, ’67; Atwood and Marin, ’70; Jahromi and Atwood, ’74). The strength of presynaptic inhibition demonstrated by electrophysiologicalrecording varies considerably from one muscle to another, and within a muscle, the various muscle fibers show different degrees of presynaptic inhibition (Atwood and Bittner, ’71).Focal recordings from individual neuromuscular junctions on a single muscle fiber have also shown nonhomogeneity of inhibitory effects (Dudel and Kuffler, ’61;Wiens and Atwood, ’75). Nonuniformity of inhibition implies nonuniform distribution andlor varying physiological potency of inhibitory axo-axonal synapses. Evidence for local “clustering” of axo-axonal synapses has been obtained in a few samples examined by electron microscopy (Atwood and Kwan, ’76). However, previous studies have not attempted to correlate
o 1991 WILEY-LISS, INC.
the samples obtained for electron microscopy with physiological expressions of inhibition. Recently, a method was developed for marking terminals situated under a macropatch electrode for electron microscopy using activity-dependent uptake of horseradish peroxidase (Tse et al., ’87). The same electrode can be used to record quanta1 currents during synaptic transmission (Dudel, 1981; Dudel et al., ’83) and to study the effectivenessof presynaptic inhibition (Tse and Atwood, ’86; Atwood and Tse, ’88). Thus, it is now possible to carry out correlated electrophysiologicaland morphological studies of presynaptic inhibition on specific nerve terminals. Several questions can be investigated with this approach: 1)Are axo-axonal synapses found at the site of physiological Accepted September 11,1990. F.W. Tse’s present address is Department of Physiology and Biophysics, University of Washington, Seattle, WA 98195. S.S. Jahromi’s present address is Playfair Neuroscience Unit, Toronto Western Hospital, 399 Bathurst Street, Toronto, Canada M5T 258.
F.W. TSE E'r AL.
136 effect, or are they remote from it? 2) Are axo-axonal synapses more common a t sites where strong presynaptic inhibition occurs? 3) Does the morphology of the excitatory nerve terminals, which is thought to be one of the factors determining the strength of presynaptic inhibition (Atwood et al., '841, differ in sites with strong and weak inhibition? In the present study, comparison of sites from fibers showing strong and weak presynaptic inhibition was made in crab muscles, with the aim of answering these questions. In another report (Atwood and Tse, '88), an analysis of changes in quantal currents during presynaptic inhibition has been presented.
MATJ3RIALSANDMETHODS Preparations of the "stretcher" muscle in legs of two species of crab, Pachygrapsus crassipes (Randall) and Hyas areneus (Linnaeus) were made as described in previous papers (Wiens and Atwood, '75; Stephens and Atwood, '82; Atwood et al., '84; Tse et al., '87; Atwood and Tse, '88). The isolated preparations were maintained at 15°C in appropriate physiological solutions for recording of electrical events. The physiological solution used for Hyas contained 450 mM NaC1, 8 mM KCl, 15 mM CaCl,, 20 mM MgCl,, and 1 mM Hepes, adjusted to pH 7.4. For Pachygrapsus, the physiological solution contained 470 mM NaC1, 8 mM KC1, 20 mM CaCl,, 10 mM MgCI,, and 1 mM Hepes, adjusted to pH 7.4. For physiological work, the single excitatory (E) axon was stimulated (Stephens and Atwood, '82) to evoke excitatory postsynaptic potentials (EPSPs) in muscle fibers, and the specific inhibitory (SI)axon, isolated in the meropodite, was stimulated at various times relative to the excitatory stimulus to test for presynaptic inhibition (Wiens and Atwood, '75). Frequencies of 1-10 Hz were used in most experiments. Records were made from muscle fibers with conventional intracellular electrodes, and from single synaptic terminals with macropatch electrodes filled with physiological solution (Dudel, '81; Tse et al., '87; Atwood and Tse, '88).
Morphological observations were made at the light microscopic level using membrane-permeant fluorescent dyes to stain nerve terminals. The method was adapted from Johnson et al. ('81) and Yoshikami and Okun ('84). Dyes that have been reported to stain mitochondria were used. The dyes, either Rhodamine 123 (Rh123; from Sigma, 5 &ml) or 3,3'-diethyloxadicarbocyanineiodide (DiOC,; from Aldrich, 0.5 pg/ml) were dissolved in crab saline. The dissected preparation was stained in the dye solution for 10 minutes, in a refrigerator (5°C). Then the preparation was rinsed three times with crab saline, and was observed or photographed with a compound microscope (Nikon Optiphot) equipped with an episcopic-fluorescenceattachment and appropriate filters. Blue excitation and green emission filters were employed for Rh123; for DiOC,, green excitation and red emission filters were used. The presynaptic terminals have a high mitochondria1 content (Jahromi and Atwood, '74) and the fluorescent dyes used in this study are selectively accumulated by mitochondria (Johnson et al., '80; Bunting et al., '89). These features account for the differential labeling of the presynaptic terminals. Since the dyes did not cause significant changes in the electrophysiology of the preparation, staining could be done before, or after, the collection of electrophysiological data. Both dyes were tested on neuromuscular systems of a few crustacean species (shore crab, crayfish, and lobster). Rh123
consistently gave better contrast and definition than DiOC,. In general, the best results were obtained with crab preparations. For ultrastructural work, samples were obtained from fibers whose EPSPs and degree of presynaptic inhibition had been observed with intracellular and extracellular macropatch recording. Activity-dependent uptake of horseradish peroxidase (HRP) was employed to mark terminals of the E axon, and also those of the SI axon (Holtzman et al., '71; Thompson and Atwood, '84). Terminals of the smaller common inhibitory (CI) axon could then be distinguished by their lack of HRP-labeled synaptic vesicles. The technique was further refined to label terminals from which recordings had been made with a macropatch electrode. The technique has been described in detail by Tse et al. ('87). In brief, the preparation was maintained in HRP-containing, Ca-free, high Mg solution (to eliminate synaptic transmission). Recordings were made with a macropatch electrode filled with a Ca-containing solution. Activity-dependent HRP uptake occurred only under the macropatch electrode, and not elsewhere on the surface of the preparation. HRP-marked terminals were located subsequently in sections made for electron microscopy. The fixation and sectioning procedures for electron microscopy have been described in detail in previous papers (Atwood et al., '84; Tse et al., '87). Serial sections were made from several terminals taken from physiologically identified muscle fibers, and reconstructions were made in a few cases to investigate the number and location of axo-axonal synapses. Three-dimensional reconstructions were made using a software program (Boulder HVEM 3-D Reconstruction: Dr. John Kinnamon, University of Colorado) that ran on an IBM PC-compatible microcomputer equipped with a digitizing tablet (GTCO DIG1 PAD 5).Two-dimensional drawings were then prepared to illustrate the complex networks of terminals and synapses that were found; these could not be seen to advantage in the initial three-dimensional reconstructions. The two-dimensional drawings took account of the distribution and dimensions of the nerve terminals and their synapses located in the samples.
RESULTS PhysiologicalvariationinEPSPs Previous studies have shown that considerable variation in amplitude and facilitation of EPSPs occurs in the muscle fiber populations of crustacean limb muscles, including the stretcher muscles of Pachygrapsus and Grapsus (Atwood, '67; Atwood and Bittner, '71) and Hyas (Sherman and Atwood, '72; Wiens and Atwood, '75). In general, EPSPs of relatively large amplitude show modest facilitation when the frequency of stimulation is raised from 1 to 5 Hz or higher. In contrast, small amplitude EPSPs generally increase greatly in amplitude as the frequency of stimulation is raised above 1Hz. The quantal output of transmitter at individual nerve terminals shows corresponding variation. At low frequencies of stimulation, individual excitatory terminals on the muscle fibers are known to release several quanta per impulse for largeEPSP fibers, but very few quanta per impulse (sometimes, on average, less than 1)for small-EPSP fibers (Atwood, '67; Bittner, '68; Sherman, '77; reviews, Atwood, '76, '82; Atwood and Wojtowicz, '86). Thus, the size of the EPSP is determined by the transmitter-releasing properties of the
CRUSTACEAN NEUROMUSCULAR INHIBITION
137
terminals (high-output or low-output type) and to a lesser mV when stimulated at less than 4 Hz) did not show extent by the membrane properties of the muscle fibers significant facilitation, and sometimes showed depression (Sherman and Atwood, '72; Atwood, '82). instead. These observations are in agreement with previous When the E axon was stimulated at low frequencies (1-4 work. Hz), the amplitude of the EPSPs ranged from less than 1 mV to over 30 mV in Pachygrupsus (Fig. 11, and from less Preynapticinhibition than 1mV to over 10 mV in Hyus. Small EPSPs exhibited considerable facilitation even at Crab stretcher muscles are innervated by two inhibitory moderate frequencies of stimulation; large EPSPs (over 15 neurons: a specific inhibitor (SI), which supplies only the
B JA z * i
(
E axon +
CI a x o n
0
4
8
12
16
20
24
28
32
36
EPSP amplitude (mV)
361 1
0
(ii)
32
0 0
0
0
n
:.
O.
n
I
: 0
0
v)
ti
0
0 0
0
12
o
I
0
r
. 0
30
40
50
60
% reduction in
70
80
90
100
EPSP amplitude
during optimal inhibition
0
0
30
40
50
60
70
80
90 over 97
% reduction in EPSP amplitude during optimal inhibition
Fig. 1. A Example of differential reduction of excitatory postsynaptic potentials (EPSPs) in the same muscle fiber by optimally timed stimulation of the specific inhibitor (SI) axon (i) or the common inhibitor (CI) axon (ii) in Pachygrapsus. In each case, EPSPs were recorded when the excitatory (E) axon was stimulated alone at 4 Hz, and when the SI or CI axon was stimulated a few milliseconds before the E axon to produce maximal inhibition. The effect of the SI axon was typically much greater than that of the CI axon. Calibration: 5 mV (vertical); 5 ms (horizontal). B. Range and distribution of EPSP amplitudes recorded from a sample of 50 stretcher muscle fibers in Pachygrapsus during 1 Hz stimulation (i), and of their reduction during optimal inhibition by the SI axon (ii). The two histograms were
obtained from a sample that included only muscle fibers with clearly demonstrable reduction in the amplitude or decay time constant of the EPSPs during optimal inhibition indicating clearly functional SI synapses; because of this, the sampling was biased towards selection of muscle fibers with strong inhibition. C: Reduction in EPSP amplitude during optimal inhibition produced by the SI axon at 1Hz in 50 muscle fibers in Pachygrupsus (same sample as in B). The best-fitted linear regression to these points (not shown) has a coefficient of correlation (r) of 0.435, which indicates that there was weak positive correlation (P < 0.01 for no correlation) between EPSP amplitudes and their maximal reduction by inhibition from the SI axon.
F.W. TSE ET AL.
138 stretcher muscle; and a common inhibitor (CI), which supplies all of the limb muscles (Wiens and Rathmayer, ’85; Wiens, ’89). In Grapsus and Hyas, the SI axon provides stronger inhibitory effects than the CI axon (Atwood and Bittner, ’71; Wiens and Atwood, ’75). The same is true in Pachygrapsus (Fig. 1A). Both SI and CI axons are known to produce presynaptic inhibition in stretcher muscles of several crab species (Wiens and Atwood, ‘75). The presynaptic mechanism accounts for most of the reduction of the EPSP at low frequencies of stimulation, and the SI axon is more effective in producing presynaptic inhibition than the CI axon in most muscle fibers. Responses sampled in Pachygrapsus showed a wide range in the effectiveness of the SI axon at low frequencies of stimulation (Fig. lB, C). When the SI axon was stimulated with optimal timing for inhibition, it always produced strong (overall) inhibition in muscle fibers with large EPSPs, but it sometimes produced only moderate inhibition in muscle fibers with intermediate or small EPSPs. A weak positive correlation was found between the EPSP amplitude and effectiveness of SI’s inhibition; this is in agreement with previous work on Grapsus (Atwood and Bittner, ’71). An assessment was made of the relative contribution of presynaptic and postsynaptic components to the overall inhibitory effect of the SI axon. The change in muscle fiber membrane conductance associated with postsynaptic inhibition was estimated by the “constant current method” (reviewed by Ginsborg, ’671, which assumes that all resistances associated with the postsynaptic muscle membrane are ohmic, and that the postsynaptic muscle membrane is isopotential (Fig. 2A). Representative results for this method are presented in Fig. 2B. The “constant current method” was applied to ten muscle fibers (six from Pachygrapsus and four from Hyas; Fig. 2) that had their resting membrane input resistances in the range typical of this preparation (Sherman and Atwood, ’72). In these ten muscle fibers, optimally timed stimulation of the SI axon reduced individual EPSPs by over 90%. The reversal potential of inhibitory postsynaptic potentials (IPSPs) in these muscle fibers was within a few millivolts of the resting membrane potential; therefore, the IPSP clamped the membrane potential of the muscle fiber close to the resting membrane potential. In order to reduce the amplitude of an EPSP by over 90%,the parallel “shunt resistance” of the IPSP (r in Fig. 2A) had to be less than l/lOth that of the resting membrane’s input resistance (R in Fig. 2A). The estimatedvalues of the “shunt resistances” for the IPSPs in these ten muscle fibers were in the range of 1to 11times that of the resting membrane input resistance (Fig. 2C). A “shunt resistance” in this range could only reduce the amplitude of an EPSP by 10 to 40%. Therefore, the additional mechanisms giving rise to the remaining 50 to 80% of the observed (90%)inhibition must be presynaptic.
Variationinterminal morphology Treatment of preparations with fluorescent dyes was undertaken to determine whether morphological differences accompany physiological ones. Previously, methylene blue staining of crayfish opener muscles did not yield evidence of morphological differences between low-output and high-output terminals (Bittner, ’68). In crab muscle fibers, morphological variation was found (Fig. 3). Two extreme morphologies with a range of intermediates were encountered.
At one extreme, the junctional region consisted of a cluster of varicose synaptic terminals, with branches mostly in parallel (in palmate fashion) from a preterminal branch of the axon (Fig. 3A). Such synaptic terminals resembled in many respects an amphibian neuromuscular junction (see Fig. 8a of McMahan et al., ’72). At the other end of the spectrum, the junctional region contained a long string of varicose synaptic terminals connected in series to a preterminal branch of the axon (Fig. 3G). Although the second variety of synaptic morphology has been described as “typical” for crustacean neuromuscular systems (primarily from observations on crayfish preparations: Van Harreveld, ’39; Wiersma, ’61; Onodera and Takeuchi, ’80; Florey and Cahill, ’82), synaptic terminals of this variety were found infrequently in stretcher preparations of the present study. Most synaptic terminals observed in the present study were intermediate between the two extremes. Electrophysiological measurements and conjoint fluorescent staining revealed good correlation between the output characteristics of synaptic terminals and their gross morphology. Neuromuscular junctions of fibers with small EPSPs that facilitated a t higher frequencies of stimulation always contained varicose synaptic terminals linked in a long series by “bottlenecks” (Fig. 3G; see also Florey and Cahill, ’82). Neuromuscular junctions of fibers with large EPSPs that did not facilitate at higher frequencies of stimulation always had dense clusters of synaptic terminals (Fig. 3A). In general, the number of discrete junctional areas was limited (three to six) in the latter type of fiber, whereas more diffuse innervation was common in the former.
Localization of terminals and synapws A macropatch electrode was used to locate spots where evoked quantal synaptic currents were present; then the SI axon was stimulated to see whether reduction in quantal output occurred, as an indication of presynaptic inhibition (Atwood and Tse, ’88).Subsequently, synaptic terminals of the E and SI axons in the recorded spot were labeled by activity-dependent uptake of HRP (see Materials and Methods). Sectioning was then undertaken to locate terminals containing HRP-labeled vesicles, and if possible serial sections and reconstructions were undertaken in terminal regions that contained the label. Three generd criteria for identification of excitatory and inhibitory terminals in electron micrographs were employed (Jahromi and Atwood, ’74; Atwood and Marin, ’83). The first was the polarity of axo-axonal synapses. Almost always, terminals of the inhibitory axon are presynaptic to those of the excitatory axon (but see Nakajima et al., ’73; Atwood and Kwan, ’79). The second was that synaptic vesicles of the excitatory axon were more regular in size and more spherical. The third was that the specialized postsynaptic membrane of excitatory synapses was more prominent and more densely stained. Since the SI axon was stimulated in HRP while the CI axon was not, unlabeled inhibitory terminals were assigned to the CI axon. Fig. 4 shows an example of a section containing both labeled and unlabeled inhibitory terminals. The unlabeled terminal was followed through a series of 85 sections; no labeled vesicles were found. This terminal was always located some distance away from the labeled inhibitory and excitatory terminals, and formed neuromuscular but not axo-axonal synapses. In other series of sections, we also did not locate any axo-axonal synapses for the unla-
CRUSTACEAN NEUROMUSCULAR INHIBITION
A
outside
139
4-----.
,,
i
)
R : resting input resistance of t h e muscle f i b r e
I
constant current source
muscle fibre
r :
hunt ” resistance of t h e IPSP
E : resting membrane potential
e : ‘reversal potential” of the
IPSP
i : injected current
inside
B
(i)
+-----‘ (ii)
C
(iii)
0
4 p‘ ’
mi (iv)
9
-\/---
(V)
0
A
J ‘A,
J
A
’1
0
0.3
0.4
0.5
0.6
0.7
Muscle input resistance (MR) Fig. 2. A Estimation of postsynaptic inhibitory conductance by the constant current method. B: Example of an inhibitory postsynaptic potential (IPSP) recorded while long pulses of current were injected into the muscle fiber. Upper traces: membrane potential; lower traces (not calibrated): injected current. Depolarizing current was injected in (i) and (ii); no current pulse was injected in (iii); hyperpolarizing
current was injected in (iv), (v), and (vi). Note that in (vi) the IPSP was reversed. Calibration: 10 mV (vertical); 400 ms (horizontal).C: Ratio of IPSP “shunt” resistance to the (resting) muscle input resistance plotted against the value of (resting) muscle input resistance. Filled circles are data from Pachygrapsus; filled triangles are data for Hyas.
beled (putative CI) axon. However, samples for this study were selected on the basis of demonstrable presynaptic inhibition mediated by the SI axon, rather than the CI axon. For location of axo-axonal synapses for the CI axon, samples would have to have been selected from recording locations at which the CI axon produced demonstrable presynaptic inhibition, as shown, for example, by Wiens and Atwood (’75). In all samples in which labeled terminals could be found and in which presynaptic inhibition was known to be produced by the SI axon, axo-axonal synapses of the appropriate type were found. Furthermore, presynaptic inhibition could be detected under the Ca-containingmacropatch electrode in a Ca-free external solution (Atwood and Tse, ’88). These two observations indicate that at least
some of the axo-axonal synapses responsible for presynaptic inhibition occur close to the points at which transmitter is released from the E axon, in confirmation of morphological data of Smith (’78) and Atwood et al. (’84).
Morphologyof terminals exhibitingstrong and weak inhibition Intracellular microelectrodes were employed to assess the strength of presynaptic inhibition by the SI axon in selected muscle fibers. A macropatch electrode was then used to locate terminals on the surface of the muscle fiber and to label them using local activity-dependent HRP uptake for subsequent examination by electron microscopy. Serial sections were made to locate the labeled terminals.
140
F.W. TSE ET AL.
CRUSTACEAN NEUROMUSCULAR INHIBITION
141
For selected terminal regions, reconstructions were completed from serial sections. Information from the reconstructions was reduced to two-dimensionaldiagrams to illustrate synaptic interactions between E and SI terminals and synapticsize and location. Representativediagrams illustrating the general findings are presented in Figs. 5-10. In general, the local activity-dependent HRP labeling method worked well for regions of strong inhibition involving high-output E synapses, but was less successful for regions of weak inhibition involving intermediate or lowoutput E synapses. A possible reason for less HRP uptake in the latter situation is that the percentage of synaptic vesicles labeled with HRP is considerably higher for terminals with high quantal content at low frequencies of stimulation (Thompson and Atwood, ’84).The need to maintain mechanical stability during local labeling precluded use of high frequencies of stimulation since they produced contraction even when the muscle was superfused with low Ca solution. A region of weak inhibition is presented in Figs. 5-7 and a region of strong inhibition in Figs. 8-10. Physiological measurements of EPSPs gave values of 1and 15 mV (at 1 Hz) for the fibers showing weak and strong inhibition, respectively. Optimal timing of inhibitory impulses (at 1 Hz) produced 5 1 0 % and 90% reduction of EPSPs, respectively, in the two fibers. For each reconstructed synaptic region, three diagrams were made: one to show the interaction between excitatory and inhibitory terminals, one to show the relative dimensions of the excitatory terminals and their associated synapses, and one to show the relative dimensions of the inhibitory terminals and their associated synapses. The diagrams are two-dimensional representations of intricately intertwined three-dimensional structures. In the region of we& inhibition (Fig. 5), the E axon courses through the series and gives rise to an enlargement and to two side branches with synapse-bearing end-bulbs. The inhibitory axon gives rise to several very thin sidebranches bearing terminal end-bulbs. All axo-axonal synapses (five in this series) occurred on these terminal end-bulbs. The output synapses of both E axon and SI axon are concentrated in the enlarged portions of the terminals (Figs. 6 and 7). The number of individual synapses and the number of individual presynapticdense bars (“activezones”) associated with them are greater for the E axon than for the SI axon, as indicated in the figures. This observation confirms those made in previous studies (Jahromi and Atwood, ’74;Atwood and Kwan, ’76). Axo-axonal synapses possess one to five dense bars, and are formed between the enlarged portions of the two axons rather than on slender bottlenecks. In the region of strong inhibition (Figs. 8-10), E and SI axons are closely intertwined (Fig. 81, with a total of nine
axo-axonal synapses occurring between the two axons at several locations. The E axon (Fig. 9) has a larger number of output synapses with associated dense bodies than the SI axon (Fig. lo), and a higher proportion of terminal area devoted to synaptic contacts. A summary and comparison of data for inhibitory synapses (both neuromuscular and axo-axonall are provided in Fig. 11. The percentages of synapses with multiple dense bars is greater for the region of weak inhibition (Fig. IlA), that is, there are more synapses with two or more dense bars in the sample from that region. However, the proportion of dense bars of large size (i.e., more than three sections long) is marginally greater in the region of strong inhibition (Fig. 11B). In addition, there are more axoaxonal synapses in the region of strong inhibition. Average length of individual dense bars for inhibitory synapses is 0.19 Fm for the region of weak inhibition, and 0.21 Fm for the region of strong inhibition. Thus, no difference in average size of dense bars is apparent, even though size distributions (Fig. 11B)are different. Morphologyof individual synapses taken alone is not sufficient to account for the large difference in effectiveness of inhibition at low frequencies in the two regions.
Fig. 3. Morphology of preterminal axonal branches and varicose synaptic terminals on muscle fibers innervated by synaptic terminals with different output characteristics. Negative images of fluorescing terminals were obtained by enlarging positive color slides directly onto black-and-white photographic paper: the fluorescent preterminal axonal branches and synaptic terminals appear dark. A string of synaptic terminals in G stretched between the two arrows; however, the long thin bottlenecks connecting each varicose synaptic terminal are not well illustrated. The images in B to F were reconstructed from
montages because different regions of a synaptic terminal were at different planes of focus. The varicose synaptic terminals in A and B were located on different muscle fibers, from which EPSPs of over 15 mV were recorded at 1Hz stimulation; those in C to F were located on four different muscle fibers, from which EPSPs of 5 to 10 mV were recorded at 1Hz stimulation; those in G were located on a muscle fiber, from which EPSPs of under 2 mV were recorded at 1 Hz stimulation. Scale bar = 250 pm.
DISCUSSION The present study confirms the diversity of synaptic properties in limb muscles of crabs, and the nonuniformity of inhibition in different muscle fibers at low frequencies of stimulation. The major component of inhibitory reduction of EPSPs at low frequencies is presynaptic, as in other crustacean limb muscles studied previously (Atwood and Marin, ’70; Atwood and Bittner, ’71). The basic question addressed in the present study concerns the relationship between morphological features of nerve terminals and effectiveness of inhibition, particularly its presynaptic component. An earlier study by Atwood et al. (’84)demonstrated that inhibitory axo-axonal synapses can in principle produce a large attenuation of transmitter release through local action of axo-axonal synapses on terminal varicosities. The varicose configuration of the terminals constitutes a favorable substrate for presynaptic inhibition; during inhibitory transmitter action, a large attenuation of terminal depolarizations occurs, and this is more pronounced for small, high resistance axon processes. This general principle has now been confirmed and extended in a detailed theoretical study recently completed by Segev (’90). The ultrastructural work of the present study provides additional clear examples of varicose excitatory terminals joined to an axonal branch with thin processes and endowed with axo-axonal synapses (Figs. 5 and 8). Given that the structure of the terminals and the location of axo-axonal synapses provide favorable conditions for presynaptic inhibition in both high-output and low-output
Fig. 4. Identification of terminals of SI, CI, and E axons by differential labeling and morphology. All parts of the figure were from the same section of a Hyus muscle fiber. A Electron micrograph of a group of synaptic terminals obtained from a Hyas muscle fiber. HRP-labeled synaptic vesicles are identifiable (a few are circled). In this micrograph there are two synaptic terminals of the E axon ( E l and E2); four of the SI axon (Sl, S2,S3, and 54); and one of the CI axon (C). B: A higher magnification electron micrograph to show a presynaptic dense
bar associated with an axo-axonal (aa) synapse. Labeled synaptic vesicles are circled. E l is one of the excitatory terminals, and S2 one of the terminals of the SI axon from A. The axo-axonal synapse is on a thin bottleneck of the excitatory axon (E2). C: A high magnification electron micrograph of the synaptic terminal of the CI axon. The synaptic vesicles (sv) in this synaptic terminal were not labeled, although there were two neuromuscular synapses (nmj). Magnification: A, x 10,800; B, ~ 2 2 , 0 0 0C,D, ; ~66,000. Scale bars = 1pm.
143
CRUSTACEAN NEUROMUSCULAR INHIBITION
Axo-axonal synapses
A
Excitatory
A
Inhibitory
Fig. 5. Two-dimensional diagram to show synaptic interactions between E and SI terminals in a region of weak inhibition in Pachygrapsus. Excitatory branches (stippled) and inhibitory branches (clear) are shown in two-dimensional scale drawing. Position of axo-axonal synapses is indicated. The width of the terminals is determined from their cross sections and is proportional to linear circumference. Scale: vertical = 1 pm, horizontal = 5 pm.
cases, what factor or factors can account for the difference in effectiveness of inhibition at low frequencies of stimulation? Factors to be evaluated include: 1) differences in morphology of the terminals and number or placement of axo-axonal synapses; 2) differences in synaptic strength determined by ultrastructural features or by calcium channel availability at the synapse; and 3) differences in electrical properties-of the terminals, including density of mem-
Fig. 6. Two-dimensionaldiagram of the excitatory terminals in the region of Fig. 5. The width of each process is proportional to its linear circumference. The synaptice release zones are combined for each section and the width of the band representing them is proportional to combined linear dimension. Individual presynaptic dense bars (solid lines) are shown to indicate the locations and positions of those found in the series. Those of axo-axonal synapses are shown on the surface of the terminals, and identified as A, B, C, and D. Scale: vertical = 1 pm, horizontal = 1.5 um.
brane channels responsible for propagation of the action potential.
Terminal morphology In crab muscles, we found that high-output (and highly inhibited) terminals often have a compact, palmate configuration with many closely adjacent varicosities, whereas low-output, less strongly inhibited terminals are more diffuseli arranged, with-serial varicosities (Fig. 3). This
144
F.W. TSE ET' AL.
'=)
&
0
Axo-axonal synapses
A
A
Excitatory
Inhibitory
Fig. 7. Two-dimensional diagram of the inhibitory terminals in the region of Fig. 5. Conventions as in Fig. 6. Individual axo-axonal synapses are indicated (A, B, C, D). Scale: vertical = 1 pm, horizontal = 1.5 pm.
Fig. 8. Two-dimensional diagram to show synaptic interactions between E and SI terminals in a region of strong inhibition in Puchygrupsus. Conventions as in Fig. 5. Scale: vertical = 1 pm, horizontal = 5 pm.
difference could be significant, if coupled with a difference in extent of active invasion of the terminals by action potentials, as originally proposed by Atwood ('67) and Sherman and Atwood ('72), and again more recently by Dude1 ('82, '83).The compact terminal morphology of Fig. 4 would favor more uniform activation of terminal boutons by a subterminal action potential, and hence more effective release of transmitter by a single impulse in both excitatory and inhibitory terminals [see also Segev ('90) for a theoretical treatment of such differences]. At the ultrastructural level, the main differences between strongly and weakly inhibited regions are the larger number and more uniform distribution of axo-axonal synapses over the excitatory terminal processes in the former case, and the location of inhibitory axo-axonal synapses on varicosities supplied by very thin branches in the latter case. If the inhibitory action potential actively invades the main axon branches but not the thin side processes, the voltage change experienced by axo-axonal synapses may be reduced, on account of the thin, high-resistance side processes, and the probability of release of inhibitory transmitter for a single impulse may be relatively low for synapses on the terminal varicosities of the inhibitory axon. Con-
versely, in the region of strong inhibition, inhibitory transmitter release may be favored by a larger electrical event in the inhibitory terminal. The more widespread distribution of axo-axonalsynapses on the excitatory terminal in the region of strong inhibition, coupled with higher quantal content of inhibitory transmission, could serve to attenuate the electrical events in the excitatory terminal more strongly in this region.
synaptic strength The differences in synaptic size and dense bar (active zone) size and number do not, in themselves, provide an explanation for differences in transmitter release of excitatory or inhibitory axons in the two regions. Excitatory synapses formed a larger proportion of the surface area of high-output terminals, but this was not true in the case of the inhibitory terminals (Figs. 7 and 10). Individual active zones were not greatly different in size and number for excitatory terminals (Figs. 6 and 9). Some active zones of inhibitory synapses are relatively large in the region of strong inhibition; synapses possessing them could be relatively potent (Atwood and Wojtowicz, '86). The striking difference in number and surface area of inhibitory and
CRUSTACEAN NEUROMUSCULAR INHIBITION
145
A
0
go
Weak
Strong
50
Axo-axonal
3
40
n
n
30
0
8 20 10
0
Number of dense bars Per synapse
B “1 40
E
2
30
0
25
20
0
8 10
Fig. 9. Two-dimensional diagram of the excitatory terminals in the region of Fig. 8. Conventions as in Fig. 6. Individual axo-axonal synapses are indicated by the numbers 1-9) Scale: vertical = 1 pm, horizontal = 3 pm.
Length
of donee bar (number of sectlone)
Fig. 11. Comparison of features of inhibitory (SI) synapses for Puchygrupsus in regions of strong and weak inhibition. Percentages of inhibitory synapses with different numbers of dense bars (A) and distribution of dense bar sizes (B) are illustrated. Axo-axonal synapse (shaded) and neuromuscular synapses are included. Asterisk marks an axo-axonal synapse in the region of strong inhibition.
excitatory synapses (consistently found in crustacean limb muscles), together with indications of a higher quantal content for inhibitory transmission (Atwood and Bittner, ’71), suggest that effectivetransmission at low frequencies of stimulation is determined largely by transmitter release from relatively few “potent” synapses. This conclusion is supported by statistical analyses of quantal release at excitatory terminals (Atwood and Tse, ’88). At present, data on calcium channel number and density at these synapses is lacking. Such data, which could eventually be supplied by freeze-fracture techniques (see, for example, Govind and Walrond, ’891, could address in part the question of “potent” and “impotent” synapses for both excitatory and inhibitory terminals. 9
Differencesin electrical properties
No direct evidence on nerve terminal electrical properties was provided by the present study. However, the differences in strength of inhibition in the two regions are not readily explained by the ultrastructural findings on differences in synaptic size and number of active zones. As noted in Terminal morphology, above, if action potential propagation were limited to main axons, the more common occurrence of thin, inexcitable side processes in the region of Fig. 10. Two-dimensional diagram of the inhibitory terminals in the region of Fig, 8, Conventions as in Fig, 6 , Individual axo-axonal weak inhibition could account for part of the difference in synapses are indicated by the numbers 1-91 scale: vertical = 1 wm, effectiveness of presynaptic inhibition. Such branches could attenuate an electrotonically conducted action potential, horizontal = 3 pm.
F.W. TSE ET AL.
146 and a smaller electrical signal would in turn be less effective in releasing quantal units of transmitter from inhibitory synapses. In conclusion, the morphological differences discerned for individual synapses of excitatory and inhibitory terminals do not in themselves lead to a clear resolution of the question of differential physiological effectiveness, but the number and distribution of axo-axonal synapses coupled with differences in terminal morphology and putative terminal electrical properties provides a working hypothesis. A more complete explanation must include more data on: 1) extent of action potential propagation in the terminals; and 2) physiological potency of individual synapses, including information on the number of active calcium channels.
ACKNOWLEDGMENTS The work was supported by a grant from NSERC, Canada, and by an NSERC Visiting Scientist award to Dr. S.S. Jahromi, who was on sabbatical leave from the University of Shiraz, Iran. Marianne Hegstrom-Wojtowicz assisted in preparing the illustrations.
XxrFBAmcm Atwood, H.L. (1967) Variation in physiological properties of crustacean motor synapses. Nature 215:57-58. Atwood, H.L. (1976) Organization and synaptic physiology of crustacean neuromuscular systems. Progr. Neurobiol. 7:291-391. Atwood, H.L. (1982) Synapses and neurotransmitters. In H.L. Atwood and D.C. Sandeman (eds): The Biology of Crustacea, Vol. 3, Neurobiology Structure and Function. New York: Academic Press, pp. 105-150. Atwood, H.L., and A. Jones (1967) Presynaptic inhibition in crustacean muscle: Axo-axonal synapse. Experientia 23:103&1038. Atwood, H.L., and W.A. Morin (1970) Neuromuscular and axoaxonal synapses of crayfish opener muscle. J. Ultrastruct. Res. 32:351-369. Atwood, H.L., and G.D. Bittner (1971) Matching of excitatory and inhibitory inputs to crustacean muscle fibers. J. Neurophysiol. 34:157-170. Atwood, H.L., and I. Kwan (1976) Synaptic development in the crayfish opener muscle. J. Neurobiol. 7:289-312. Atwood, H.L., and I. Kwan (1979) Reciprocal axo-axonal synapses between excitatory and inhibitory neurons in crustaceans. Brain Res. 174:324328. Atwood, H.L., and L. Marin (1983) Ultrastructure of synapses with different transmitter-releasing characteristics on motor axon terminals of a crab, Hyas areneus. Cell Tissue Res. 231:103-115. Atwood, H.L., and J.M. Wojtowicz (1986) Short-term and long-term plasticity and physiological differentiation of crustacean motor synapses. Int. Rev. Neurobiol. 28275-362. Atwood, H.L., and F. Tse (1988) Changes in binomial parameters of quantal release at crustacean motor axon terminals during presynaptic inhibition. J. Physiol. (Lond.)402:177-193. Atwood, H.L., J.K. Stevens, and L. Marin (1984) Axoaxonal synapse location and consequences for presynaptic inhibition in crustacean motor axon terminals. J. Comp. Neurol. 22564-74. Bittner, G.D. (1968) Differentiation of nerve terminals in the crayfish opener muscle and its functional significance. J. Gen. Physiol. 51t731-758. Bunting, J.R., T.V. Phan, E. Kamali, and R.M. Dowben (1989) Fluorescent cationic probes of mitrochondria: Metrics and mechanism of interaction. Biophys. J. 56t979-993. Dudel, J. (1981) The effect of reduced calcium on quantal unit current and release at the crayfish neuromuscular junction. Pflugers Arch. 391:3540. Dudel, J. (1982) Transmitter release by graded local depolarization of presynaptic nerve terminals at the crayfish neuromuscular junction. Neurosci. Lett. 32:181-186. Dudel, J. (1983) Graded or all-or-nothing release of transmitter quanta by
local depolarization of nerve terminals on crayfish muscle. Pflugers Arch. 398: 155-164. Dudel, J., and S.W. KuRler (1961) Presynaptic inhibition a t the crayfish neuromuscular junction. J. Physiol. (Lond.)271r369-390. Dudel, J., I. Parnas, and H. Parnas (1983) Neurotransmitter release and its facilitation in crayfish. VI. Release determined by both intracellular calcium concentration and depolarization of the nerve terminal. Pflugers Arch. 399:l-10. Florey, E., and M.A. Cahill (1982) The innervation patterns of crustacean skeletal muscle. Cell Tissue Res. 224t527-541. Ginsborg, B.L. (1967) Ionic movement in junctional transmission. Pharmacol. Rev. 19:289-316. Govind, C.K., and J.P. Walrond (1989) Structural plasticity at crustacean neuromuscular synapses. J. Neurobiol. 20:409421. Holtzman, E., A.R. Freeman, and L.A. Kashner (1971) Stimulationdependent alterations in peroxidase uptake at lobster neuromuscular junctions. Science 173t733-736. Jahromi, S.S., and Atwood, H.L. (1974) Three-dimensionalultrastructure of the crayfish neuromuscular apparatus. J. Cell Biol. 63:599413. Johnson, L.V., M.L. Walsh, and L.B. Chen (1980) Localization of mitochondria in living cells with rhodamine. Proc. Natl. Acad. Sci. U.S.A. 77:990-999. Johnson, L.V., M.L. Walsh, B.J. Bockus, and L.B. Chen (1981) Monitoringof relative mitochondrial membrane potential in living cells by fluorescence microscopy. J. Cell Biol. 88:526-535. McMahan, U.J., N.C. Spitzer, and K. Peper (1972) Visual identification of nerve terminals in living isolated skeletal muscles. Proc. R. SOC.(Lond) B181t421-430. Nakajima, Y., A.D. Tisdale, and M.P. Henkart (1973) Presynaptic inhibition at inhibitory nerve terminals. A new synapse in the crayfish stretch receptor. Proc. Natl. Acad. Sci. U S A . 702462-2466. Onodera, K., and A. Takeuchi (1980) Distribution and pharmacological properties of synaptic and extrasynaptic glutamate receptors on crayfish muscles. J. Physiol. (Lond.)306.233-250. Segev, I. (1990) Computer study of presynaptic inhibition controlling the spread of action potentials into axonal terminals. J. Neurophysiol. 63:987-998. Sherman, R.G. (1977) Comparative analysis of an excitatory motor unit in crustaceans. J. Comp. Physiol. 114t91-101. Sherman, R.G., and H.L. Atwood (1972) Correlated electrophysiological and ultrastructural studies of a crustacean motor unit. J. Gen. Physiol. 59:586-615. Smith, D.O. (1978) Ultrastructural specificity of synaptic sites in nerve terminals mediating both presynaptic and postsynaptic inhibition. J. Comp. Neurol. 182:839-850. Stephens. P.J., and H.L. Atwood (1982) Thermal acclimation in acrustacean neuromuscular system. J. Exp. Biol. 98t3947. Thompson, C.S., and H.L. Atwood (1984) Synaptic strength and horseradish peroxidase uptake in crayfish nerve terminals. J. Neurocytol. 13267280. Tse, F.W., and H.L. Atwood (1986) Presynaptic inhibition at the crustacean neuromuscular junction. News Physiol. Sci. It47-50. Tse, F.W., L. Marin, and H.L. Atwood (1987) Focal labeling of axonal terminals with active synapses recorded by an extracellular macro-patch electrode. J. Neurosci. Methods 21: 17-29. Van Harreveld, A. (1939) The nerve supply of doubly and triply innervated crayfish muscle related to their function. J. Comp. Neurol. 70:285-296. Wiens, T.J. (1989) Common and specific inhibition in leg muscles of decapods: Sharpened distinctions. J. Neurobiol. 20:45%469. Wiens, T.J., and H.L. Atwood (1975) Dual inhibitory control in crab leg muscles. J. Comp. Physiol. 99.211-230. Wiens, T.J., and Rathmayer, W. (1985) The distribution of the common inhibitory neuron in brachyuran limb musculature. I . Target muscles. J. Comp. Physiol. 156t305-313. Wiersma, C.A.G. (1961) The neuromuscular system. In T.H. Waterman (ed): The Physiology of Crustacea, Vol. 2. New York: Academic Press, pp. 241-279. Yoshikami, D., and L.M. Okun (1984) Staining of living presynaptic nerve terminals with selective fluorescent dyes. Nature 310:53-56.
NEUROLOGY 304135-146 (1991)
Variation in Temninal Morphology and m p t i c Inhibition at Crustacean Neuromscular Junctions F.W. TSE, L. MARIN, S.S. JAHROMI, AND H.L. ATWOOD Department of Physiology, University of Toronto, Toronto, Ontario, Canada M5S 1A8
ABSTRACT Synaptic terminals of excitatory and inhibitory neurons supplying muscle fibers in leg muscles of crabs (Pachygrapsus crassipes and Hyas areneus) were investigated with light and electron microscopy. Terminals responsible for large excitatory postsynaptic potentials (EPSPs) at low frequencies of activation had a compact configuration with clusters of terminal boutons radiating from the main axon branch. Terminals responsible for small EPSPs had a more diffuse organization, with boutons often arranged in series along thin axon branches. Inhibitory neurons, when activated, produced both presynaptic and postsynaptic inhibitory effects, with the former being more potent at low frequencies of activation. Presynaptic inhibition was variable in magnitude but was generally strong in fibers with large EPSPs. Representative terminals from regions of strong and weak presynaptic inhibition were identified by activity-dependent uptake of horseradish peroxidase, serially sectioned, and reconstructed from electron micrographs. Both regions were found to contain axo-axonal synapses from inhibitory to excitatory terminals, with a larger number in the region of strong presynaptic inhibition. In addition, axo-axonal synapses were more uniformly distributed in the latter region. The number of inhibitory presynaptic dense bars (active zones) was somewhat higher in the region of weak inhibition, but larger individual dense bars occurred in the region of strong inhibition. Possible factors contributing to the differences in strength of inhibition include: (1) morphology and electrical properties of terminals; and (2) high probability of transmission at a relatively small number of inhibitory synapses during low frequency activation in the region of strong inhibition. Key words: vesicle, axo-axonal,bouton, synapse,postsynapticpotential
Presynaptic inhibition at crustacean neuromuscular junctions, first described by Dudel and Kufler (’61))is believed to be mediated by axo-axonal synapses made by peripheral inhibitory axons on terminals of the excitatory axons (Atwood and Jones, ’67; Atwood and Marin, ’70; Jahromi and Atwood, ’74). The strength of presynaptic inhibition demonstrated by electrophysiologicalrecording varies considerably from one muscle to another, and within a muscle, the various muscle fibers show different degrees of presynaptic inhibition (Atwood and Bittner, ’71).Focal recordings from individual neuromuscular junctions on a single muscle fiber have also shown nonhomogeneity of inhibitory effects (Dudel and Kuffler, ’61;Wiens and Atwood, ’75). Nonuniformity of inhibition implies nonuniform distribution andlor varying physiological potency of inhibitory axo-axonal synapses. Evidence for local “clustering” of axo-axonal synapses has been obtained in a few samples examined by electron microscopy (Atwood and Kwan, ’76). However, previous studies have not attempted to correlate
o 1991 WILEY-LISS, INC.
the samples obtained for electron microscopy with physiological expressions of inhibition. Recently, a method was developed for marking terminals situated under a macropatch electrode for electron microscopy using activity-dependent uptake of horseradish peroxidase (Tse et al., ’87). The same electrode can be used to record quanta1 currents during synaptic transmission (Dudel, 1981; Dudel et al., ’83) and to study the effectivenessof presynaptic inhibition (Tse and Atwood, ’86; Atwood and Tse, ’88). Thus, it is now possible to carry out correlated electrophysiologicaland morphological studies of presynaptic inhibition on specific nerve terminals. Several questions can be investigated with this approach: 1)Are axo-axonal synapses found at the site of physiological Accepted September 11,1990. F.W. Tse’s present address is Department of Physiology and Biophysics, University of Washington, Seattle, WA 98195. S.S. Jahromi’s present address is Playfair Neuroscience Unit, Toronto Western Hospital, 399 Bathurst Street, Toronto, Canada M5T 258.
F.W. TSE E'r AL.
136 effect, or are they remote from it? 2) Are axo-axonal synapses more common a t sites where strong presynaptic inhibition occurs? 3) Does the morphology of the excitatory nerve terminals, which is thought to be one of the factors determining the strength of presynaptic inhibition (Atwood et al., '841, differ in sites with strong and weak inhibition? In the present study, comparison of sites from fibers showing strong and weak presynaptic inhibition was made in crab muscles, with the aim of answering these questions. In another report (Atwood and Tse, '88), an analysis of changes in quantal currents during presynaptic inhibition has been presented.
MATJ3RIALSANDMETHODS Preparations of the "stretcher" muscle in legs of two species of crab, Pachygrapsus crassipes (Randall) and Hyas areneus (Linnaeus) were made as described in previous papers (Wiens and Atwood, '75; Stephens and Atwood, '82; Atwood et al., '84; Tse et al., '87; Atwood and Tse, '88). The isolated preparations were maintained at 15°C in appropriate physiological solutions for recording of electrical events. The physiological solution used for Hyas contained 450 mM NaC1, 8 mM KCl, 15 mM CaCl,, 20 mM MgCl,, and 1 mM Hepes, adjusted to pH 7.4. For Pachygrapsus, the physiological solution contained 470 mM NaC1, 8 mM KC1, 20 mM CaCl,, 10 mM MgCI,, and 1 mM Hepes, adjusted to pH 7.4. For physiological work, the single excitatory (E) axon was stimulated (Stephens and Atwood, '82) to evoke excitatory postsynaptic potentials (EPSPs) in muscle fibers, and the specific inhibitory (SI)axon, isolated in the meropodite, was stimulated at various times relative to the excitatory stimulus to test for presynaptic inhibition (Wiens and Atwood, '75). Frequencies of 1-10 Hz were used in most experiments. Records were made from muscle fibers with conventional intracellular electrodes, and from single synaptic terminals with macropatch electrodes filled with physiological solution (Dudel, '81; Tse et al., '87; Atwood and Tse, '88).
Morphological observations were made at the light microscopic level using membrane-permeant fluorescent dyes to stain nerve terminals. The method was adapted from Johnson et al. ('81) and Yoshikami and Okun ('84). Dyes that have been reported to stain mitochondria were used. The dyes, either Rhodamine 123 (Rh123; from Sigma, 5 &ml) or 3,3'-diethyloxadicarbocyanineiodide (DiOC,; from Aldrich, 0.5 pg/ml) were dissolved in crab saline. The dissected preparation was stained in the dye solution for 10 minutes, in a refrigerator (5°C). Then the preparation was rinsed three times with crab saline, and was observed or photographed with a compound microscope (Nikon Optiphot) equipped with an episcopic-fluorescenceattachment and appropriate filters. Blue excitation and green emission filters were employed for Rh123; for DiOC,, green excitation and red emission filters were used. The presynaptic terminals have a high mitochondria1 content (Jahromi and Atwood, '74) and the fluorescent dyes used in this study are selectively accumulated by mitochondria (Johnson et al., '80; Bunting et al., '89). These features account for the differential labeling of the presynaptic terminals. Since the dyes did not cause significant changes in the electrophysiology of the preparation, staining could be done before, or after, the collection of electrophysiological data. Both dyes were tested on neuromuscular systems of a few crustacean species (shore crab, crayfish, and lobster). Rh123
consistently gave better contrast and definition than DiOC,. In general, the best results were obtained with crab preparations. For ultrastructural work, samples were obtained from fibers whose EPSPs and degree of presynaptic inhibition had been observed with intracellular and extracellular macropatch recording. Activity-dependent uptake of horseradish peroxidase (HRP) was employed to mark terminals of the E axon, and also those of the SI axon (Holtzman et al., '71; Thompson and Atwood, '84). Terminals of the smaller common inhibitory (CI) axon could then be distinguished by their lack of HRP-labeled synaptic vesicles. The technique was further refined to label terminals from which recordings had been made with a macropatch electrode. The technique has been described in detail by Tse et al. ('87). In brief, the preparation was maintained in HRP-containing, Ca-free, high Mg solution (to eliminate synaptic transmission). Recordings were made with a macropatch electrode filled with a Ca-containing solution. Activity-dependent HRP uptake occurred only under the macropatch electrode, and not elsewhere on the surface of the preparation. HRP-marked terminals were located subsequently in sections made for electron microscopy. The fixation and sectioning procedures for electron microscopy have been described in detail in previous papers (Atwood et al., '84; Tse et al., '87). Serial sections were made from several terminals taken from physiologically identified muscle fibers, and reconstructions were made in a few cases to investigate the number and location of axo-axonal synapses. Three-dimensional reconstructions were made using a software program (Boulder HVEM 3-D Reconstruction: Dr. John Kinnamon, University of Colorado) that ran on an IBM PC-compatible microcomputer equipped with a digitizing tablet (GTCO DIG1 PAD 5).Two-dimensional drawings were then prepared to illustrate the complex networks of terminals and synapses that were found; these could not be seen to advantage in the initial three-dimensional reconstructions. The two-dimensional drawings took account of the distribution and dimensions of the nerve terminals and their synapses located in the samples.
RESULTS PhysiologicalvariationinEPSPs Previous studies have shown that considerable variation in amplitude and facilitation of EPSPs occurs in the muscle fiber populations of crustacean limb muscles, including the stretcher muscles of Pachygrapsus and Grapsus (Atwood, '67; Atwood and Bittner, '71) and Hyas (Sherman and Atwood, '72; Wiens and Atwood, '75). In general, EPSPs of relatively large amplitude show modest facilitation when the frequency of stimulation is raised from 1 to 5 Hz or higher. In contrast, small amplitude EPSPs generally increase greatly in amplitude as the frequency of stimulation is raised above 1Hz. The quantal output of transmitter at individual nerve terminals shows corresponding variation. At low frequencies of stimulation, individual excitatory terminals on the muscle fibers are known to release several quanta per impulse for largeEPSP fibers, but very few quanta per impulse (sometimes, on average, less than 1)for small-EPSP fibers (Atwood, '67; Bittner, '68; Sherman, '77; reviews, Atwood, '76, '82; Atwood and Wojtowicz, '86). Thus, the size of the EPSP is determined by the transmitter-releasing properties of the
CRUSTACEAN NEUROMUSCULAR INHIBITION
137
terminals (high-output or low-output type) and to a lesser mV when stimulated at less than 4 Hz) did not show extent by the membrane properties of the muscle fibers significant facilitation, and sometimes showed depression (Sherman and Atwood, '72; Atwood, '82). instead. These observations are in agreement with previous When the E axon was stimulated at low frequencies (1-4 work. Hz), the amplitude of the EPSPs ranged from less than 1 mV to over 30 mV in Pachygrupsus (Fig. 11, and from less Preynapticinhibition than 1mV to over 10 mV in Hyus. Small EPSPs exhibited considerable facilitation even at Crab stretcher muscles are innervated by two inhibitory moderate frequencies of stimulation; large EPSPs (over 15 neurons: a specific inhibitor (SI), which supplies only the
B JA z * i
(
E axon +
CI a x o n
0
4
8
12
16
20
24
28
32
36
EPSP amplitude (mV)
361 1
0
(ii)
32
0 0
0
0
n
:.
O.
n
I
: 0
0
v)
ti
0
0 0
0
12
o
I
0
r
. 0
30
40
50
60
% reduction in
70
80
90
100
EPSP amplitude
during optimal inhibition
0
0
30
40
50
60
70
80
90 over 97
% reduction in EPSP amplitude during optimal inhibition
Fig. 1. A Example of differential reduction of excitatory postsynaptic potentials (EPSPs) in the same muscle fiber by optimally timed stimulation of the specific inhibitor (SI) axon (i) or the common inhibitor (CI) axon (ii) in Pachygrapsus. In each case, EPSPs were recorded when the excitatory (E) axon was stimulated alone at 4 Hz, and when the SI or CI axon was stimulated a few milliseconds before the E axon to produce maximal inhibition. The effect of the SI axon was typically much greater than that of the CI axon. Calibration: 5 mV (vertical); 5 ms (horizontal). B. Range and distribution of EPSP amplitudes recorded from a sample of 50 stretcher muscle fibers in Pachygrapsus during 1 Hz stimulation (i), and of their reduction during optimal inhibition by the SI axon (ii). The two histograms were
obtained from a sample that included only muscle fibers with clearly demonstrable reduction in the amplitude or decay time constant of the EPSPs during optimal inhibition indicating clearly functional SI synapses; because of this, the sampling was biased towards selection of muscle fibers with strong inhibition. C: Reduction in EPSP amplitude during optimal inhibition produced by the SI axon at 1Hz in 50 muscle fibers in Pachygrupsus (same sample as in B). The best-fitted linear regression to these points (not shown) has a coefficient of correlation (r) of 0.435, which indicates that there was weak positive correlation (P < 0.01 for no correlation) between EPSP amplitudes and their maximal reduction by inhibition from the SI axon.
F.W. TSE ET AL.
138 stretcher muscle; and a common inhibitor (CI), which supplies all of the limb muscles (Wiens and Rathmayer, ’85; Wiens, ’89). In Grapsus and Hyas, the SI axon provides stronger inhibitory effects than the CI axon (Atwood and Bittner, ’71; Wiens and Atwood, ’75). The same is true in Pachygrapsus (Fig. 1A). Both SI and CI axons are known to produce presynaptic inhibition in stretcher muscles of several crab species (Wiens and Atwood, ‘75). The presynaptic mechanism accounts for most of the reduction of the EPSP at low frequencies of stimulation, and the SI axon is more effective in producing presynaptic inhibition than the CI axon in most muscle fibers. Responses sampled in Pachygrapsus showed a wide range in the effectiveness of the SI axon at low frequencies of stimulation (Fig. lB, C). When the SI axon was stimulated with optimal timing for inhibition, it always produced strong (overall) inhibition in muscle fibers with large EPSPs, but it sometimes produced only moderate inhibition in muscle fibers with intermediate or small EPSPs. A weak positive correlation was found between the EPSP amplitude and effectiveness of SI’s inhibition; this is in agreement with previous work on Grapsus (Atwood and Bittner, ’71). An assessment was made of the relative contribution of presynaptic and postsynaptic components to the overall inhibitory effect of the SI axon. The change in muscle fiber membrane conductance associated with postsynaptic inhibition was estimated by the “constant current method” (reviewed by Ginsborg, ’671, which assumes that all resistances associated with the postsynaptic muscle membrane are ohmic, and that the postsynaptic muscle membrane is isopotential (Fig. 2A). Representative results for this method are presented in Fig. 2B. The “constant current method” was applied to ten muscle fibers (six from Pachygrapsus and four from Hyas; Fig. 2) that had their resting membrane input resistances in the range typical of this preparation (Sherman and Atwood, ’72). In these ten muscle fibers, optimally timed stimulation of the SI axon reduced individual EPSPs by over 90%. The reversal potential of inhibitory postsynaptic potentials (IPSPs) in these muscle fibers was within a few millivolts of the resting membrane potential; therefore, the IPSP clamped the membrane potential of the muscle fiber close to the resting membrane potential. In order to reduce the amplitude of an EPSP by over 90%,the parallel “shunt resistance” of the IPSP (r in Fig. 2A) had to be less than l/lOth that of the resting membrane’s input resistance (R in Fig. 2A). The estimatedvalues of the “shunt resistances” for the IPSPs in these ten muscle fibers were in the range of 1to 11times that of the resting membrane input resistance (Fig. 2C). A “shunt resistance” in this range could only reduce the amplitude of an EPSP by 10 to 40%. Therefore, the additional mechanisms giving rise to the remaining 50 to 80% of the observed (90%)inhibition must be presynaptic.
Variationinterminal morphology Treatment of preparations with fluorescent dyes was undertaken to determine whether morphological differences accompany physiological ones. Previously, methylene blue staining of crayfish opener muscles did not yield evidence of morphological differences between low-output and high-output terminals (Bittner, ’68). In crab muscle fibers, morphological variation was found (Fig. 3). Two extreme morphologies with a range of intermediates were encountered.
At one extreme, the junctional region consisted of a cluster of varicose synaptic terminals, with branches mostly in parallel (in palmate fashion) from a preterminal branch of the axon (Fig. 3A). Such synaptic terminals resembled in many respects an amphibian neuromuscular junction (see Fig. 8a of McMahan et al., ’72). At the other end of the spectrum, the junctional region contained a long string of varicose synaptic terminals connected in series to a preterminal branch of the axon (Fig. 3G). Although the second variety of synaptic morphology has been described as “typical” for crustacean neuromuscular systems (primarily from observations on crayfish preparations: Van Harreveld, ’39; Wiersma, ’61; Onodera and Takeuchi, ’80; Florey and Cahill, ’82), synaptic terminals of this variety were found infrequently in stretcher preparations of the present study. Most synaptic terminals observed in the present study were intermediate between the two extremes. Electrophysiological measurements and conjoint fluorescent staining revealed good correlation between the output characteristics of synaptic terminals and their gross morphology. Neuromuscular junctions of fibers with small EPSPs that facilitated a t higher frequencies of stimulation always contained varicose synaptic terminals linked in a long series by “bottlenecks” (Fig. 3G; see also Florey and Cahill, ’82). Neuromuscular junctions of fibers with large EPSPs that did not facilitate at higher frequencies of stimulation always had dense clusters of synaptic terminals (Fig. 3A). In general, the number of discrete junctional areas was limited (three to six) in the latter type of fiber, whereas more diffuse innervation was common in the former.
Localization of terminals and synapws A macropatch electrode was used to locate spots where evoked quantal synaptic currents were present; then the SI axon was stimulated to see whether reduction in quantal output occurred, as an indication of presynaptic inhibition (Atwood and Tse, ’88).Subsequently, synaptic terminals of the E and SI axons in the recorded spot were labeled by activity-dependent uptake of HRP (see Materials and Methods). Sectioning was then undertaken to locate terminals containing HRP-labeled vesicles, and if possible serial sections and reconstructions were undertaken in terminal regions that contained the label. Three generd criteria for identification of excitatory and inhibitory terminals in electron micrographs were employed (Jahromi and Atwood, ’74; Atwood and Marin, ’83). The first was the polarity of axo-axonal synapses. Almost always, terminals of the inhibitory axon are presynaptic to those of the excitatory axon (but see Nakajima et al., ’73; Atwood and Kwan, ’79). The second was that synaptic vesicles of the excitatory axon were more regular in size and more spherical. The third was that the specialized postsynaptic membrane of excitatory synapses was more prominent and more densely stained. Since the SI axon was stimulated in HRP while the CI axon was not, unlabeled inhibitory terminals were assigned to the CI axon. Fig. 4 shows an example of a section containing both labeled and unlabeled inhibitory terminals. The unlabeled terminal was followed through a series of 85 sections; no labeled vesicles were found. This terminal was always located some distance away from the labeled inhibitory and excitatory terminals, and formed neuromuscular but not axo-axonal synapses. In other series of sections, we also did not locate any axo-axonal synapses for the unla-
CRUSTACEAN NEUROMUSCULAR INHIBITION
A
outside
139
4-----.
,,
i
)
R : resting input resistance of t h e muscle f i b r e
I
constant current source
muscle fibre
r :
hunt ” resistance of t h e IPSP
E : resting membrane potential
e : ‘reversal potential” of the
IPSP
i : injected current
inside
B
(i)
+-----‘ (ii)
C
(iii)
0
4 p‘ ’
mi (iv)
9
-\/---
(V)
0
A
J ‘A,
J
A
’1
0
0.3
0.4
0.5
0.6
0.7
Muscle input resistance (MR) Fig. 2. A Estimation of postsynaptic inhibitory conductance by the constant current method. B: Example of an inhibitory postsynaptic potential (IPSP) recorded while long pulses of current were injected into the muscle fiber. Upper traces: membrane potential; lower traces (not calibrated): injected current. Depolarizing current was injected in (i) and (ii); no current pulse was injected in (iii); hyperpolarizing
current was injected in (iv), (v), and (vi). Note that in (vi) the IPSP was reversed. Calibration: 10 mV (vertical); 400 ms (horizontal).C: Ratio of IPSP “shunt” resistance to the (resting) muscle input resistance plotted against the value of (resting) muscle input resistance. Filled circles are data from Pachygrapsus; filled triangles are data for Hyas.
beled (putative CI) axon. However, samples for this study were selected on the basis of demonstrable presynaptic inhibition mediated by the SI axon, rather than the CI axon. For location of axo-axonal synapses for the CI axon, samples would have to have been selected from recording locations at which the CI axon produced demonstrable presynaptic inhibition, as shown, for example, by Wiens and Atwood (’75). In all samples in which labeled terminals could be found and in which presynaptic inhibition was known to be produced by the SI axon, axo-axonal synapses of the appropriate type were found. Furthermore, presynaptic inhibition could be detected under the Ca-containingmacropatch electrode in a Ca-free external solution (Atwood and Tse, ’88). These two observations indicate that at least
some of the axo-axonal synapses responsible for presynaptic inhibition occur close to the points at which transmitter is released from the E axon, in confirmation of morphological data of Smith (’78) and Atwood et al. (’84).
Morphologyof terminals exhibitingstrong and weak inhibition Intracellular microelectrodes were employed to assess the strength of presynaptic inhibition by the SI axon in selected muscle fibers. A macropatch electrode was then used to locate terminals on the surface of the muscle fiber and to label them using local activity-dependent HRP uptake for subsequent examination by electron microscopy. Serial sections were made to locate the labeled terminals.
140
F.W. TSE ET AL.
CRUSTACEAN NEUROMUSCULAR INHIBITION
141
For selected terminal regions, reconstructions were completed from serial sections. Information from the reconstructions was reduced to two-dimensionaldiagrams to illustrate synaptic interactions between E and SI terminals and synapticsize and location. Representativediagrams illustrating the general findings are presented in Figs. 5-10. In general, the local activity-dependent HRP labeling method worked well for regions of strong inhibition involving high-output E synapses, but was less successful for regions of weak inhibition involving intermediate or lowoutput E synapses. A possible reason for less HRP uptake in the latter situation is that the percentage of synaptic vesicles labeled with HRP is considerably higher for terminals with high quantal content at low frequencies of stimulation (Thompson and Atwood, ’84).The need to maintain mechanical stability during local labeling precluded use of high frequencies of stimulation since they produced contraction even when the muscle was superfused with low Ca solution. A region of weak inhibition is presented in Figs. 5-7 and a region of strong inhibition in Figs. 8-10. Physiological measurements of EPSPs gave values of 1and 15 mV (at 1 Hz) for the fibers showing weak and strong inhibition, respectively. Optimal timing of inhibitory impulses (at 1 Hz) produced 5 1 0 % and 90% reduction of EPSPs, respectively, in the two fibers. For each reconstructed synaptic region, three diagrams were made: one to show the interaction between excitatory and inhibitory terminals, one to show the relative dimensions of the excitatory terminals and their associated synapses, and one to show the relative dimensions of the inhibitory terminals and their associated synapses. The diagrams are two-dimensional representations of intricately intertwined three-dimensional structures. In the region of we& inhibition (Fig. 5), the E axon courses through the series and gives rise to an enlargement and to two side branches with synapse-bearing end-bulbs. The inhibitory axon gives rise to several very thin sidebranches bearing terminal end-bulbs. All axo-axonal synapses (five in this series) occurred on these terminal end-bulbs. The output synapses of both E axon and SI axon are concentrated in the enlarged portions of the terminals (Figs. 6 and 7). The number of individual synapses and the number of individual presynapticdense bars (“activezones”) associated with them are greater for the E axon than for the SI axon, as indicated in the figures. This observation confirms those made in previous studies (Jahromi and Atwood, ’74;Atwood and Kwan, ’76). Axo-axonal synapses possess one to five dense bars, and are formed between the enlarged portions of the two axons rather than on slender bottlenecks. In the region of strong inhibition (Figs. 8-10), E and SI axons are closely intertwined (Fig. 81, with a total of nine
axo-axonal synapses occurring between the two axons at several locations. The E axon (Fig. 9) has a larger number of output synapses with associated dense bodies than the SI axon (Fig. lo), and a higher proportion of terminal area devoted to synaptic contacts. A summary and comparison of data for inhibitory synapses (both neuromuscular and axo-axonall are provided in Fig. 11. The percentages of synapses with multiple dense bars is greater for the region of weak inhibition (Fig. IlA), that is, there are more synapses with two or more dense bars in the sample from that region. However, the proportion of dense bars of large size (i.e., more than three sections long) is marginally greater in the region of strong inhibition (Fig. 11B). In addition, there are more axoaxonal synapses in the region of strong inhibition. Average length of individual dense bars for inhibitory synapses is 0.19 Fm for the region of weak inhibition, and 0.21 Fm for the region of strong inhibition. Thus, no difference in average size of dense bars is apparent, even though size distributions (Fig. 11B)are different. Morphologyof individual synapses taken alone is not sufficient to account for the large difference in effectiveness of inhibition at low frequencies in the two regions.
Fig. 3. Morphology of preterminal axonal branches and varicose synaptic terminals on muscle fibers innervated by synaptic terminals with different output characteristics. Negative images of fluorescing terminals were obtained by enlarging positive color slides directly onto black-and-white photographic paper: the fluorescent preterminal axonal branches and synaptic terminals appear dark. A string of synaptic terminals in G stretched between the two arrows; however, the long thin bottlenecks connecting each varicose synaptic terminal are not well illustrated. The images in B to F were reconstructed from
montages because different regions of a synaptic terminal were at different planes of focus. The varicose synaptic terminals in A and B were located on different muscle fibers, from which EPSPs of over 15 mV were recorded at 1Hz stimulation; those in C to F were located on four different muscle fibers, from which EPSPs of 5 to 10 mV were recorded at 1Hz stimulation; those in G were located on a muscle fiber, from which EPSPs of under 2 mV were recorded at 1 Hz stimulation. Scale bar = 250 pm.
DISCUSSION The present study confirms the diversity of synaptic properties in limb muscles of crabs, and the nonuniformity of inhibition in different muscle fibers at low frequencies of stimulation. The major component of inhibitory reduction of EPSPs at low frequencies is presynaptic, as in other crustacean limb muscles studied previously (Atwood and Marin, ’70; Atwood and Bittner, ’71). The basic question addressed in the present study concerns the relationship between morphological features of nerve terminals and effectiveness of inhibition, particularly its presynaptic component. An earlier study by Atwood et al. (’84)demonstrated that inhibitory axo-axonal synapses can in principle produce a large attenuation of transmitter release through local action of axo-axonal synapses on terminal varicosities. The varicose configuration of the terminals constitutes a favorable substrate for presynaptic inhibition; during inhibitory transmitter action, a large attenuation of terminal depolarizations occurs, and this is more pronounced for small, high resistance axon processes. This general principle has now been confirmed and extended in a detailed theoretical study recently completed by Segev (’90). The ultrastructural work of the present study provides additional clear examples of varicose excitatory terminals joined to an axonal branch with thin processes and endowed with axo-axonal synapses (Figs. 5 and 8). Given that the structure of the terminals and the location of axo-axonal synapses provide favorable conditions for presynaptic inhibition in both high-output and low-output
Fig. 4. Identification of terminals of SI, CI, and E axons by differential labeling and morphology. All parts of the figure were from the same section of a Hyus muscle fiber. A Electron micrograph of a group of synaptic terminals obtained from a Hyas muscle fiber. HRP-labeled synaptic vesicles are identifiable (a few are circled). In this micrograph there are two synaptic terminals of the E axon ( E l and E2); four of the SI axon (Sl, S2,S3, and 54); and one of the CI axon (C). B: A higher magnification electron micrograph to show a presynaptic dense
bar associated with an axo-axonal (aa) synapse. Labeled synaptic vesicles are circled. E l is one of the excitatory terminals, and S2 one of the terminals of the SI axon from A. The axo-axonal synapse is on a thin bottleneck of the excitatory axon (E2). C: A high magnification electron micrograph of the synaptic terminal of the CI axon. The synaptic vesicles (sv) in this synaptic terminal were not labeled, although there were two neuromuscular synapses (nmj). Magnification: A, x 10,800; B, ~ 2 2 , 0 0 0C,D, ; ~66,000. Scale bars = 1pm.
143
CRUSTACEAN NEUROMUSCULAR INHIBITION
Axo-axonal synapses
A
Excitatory
A
Inhibitory
Fig. 5. Two-dimensional diagram to show synaptic interactions between E and SI terminals in a region of weak inhibition in Pachygrapsus. Excitatory branches (stippled) and inhibitory branches (clear) are shown in two-dimensional scale drawing. Position of axo-axonal synapses is indicated. The width of the terminals is determined from their cross sections and is proportional to linear circumference. Scale: vertical = 1 pm, horizontal = 5 pm.
cases, what factor or factors can account for the difference in effectiveness of inhibition at low frequencies of stimulation? Factors to be evaluated include: 1) differences in morphology of the terminals and number or placement of axo-axonal synapses; 2) differences in synaptic strength determined by ultrastructural features or by calcium channel availability at the synapse; and 3) differences in electrical properties-of the terminals, including density of mem-
Fig. 6. Two-dimensionaldiagram of the excitatory terminals in the region of Fig. 5. The width of each process is proportional to its linear circumference. The synaptice release zones are combined for each section and the width of the band representing them is proportional to combined linear dimension. Individual presynaptic dense bars (solid lines) are shown to indicate the locations and positions of those found in the series. Those of axo-axonal synapses are shown on the surface of the terminals, and identified as A, B, C, and D. Scale: vertical = 1 pm, horizontal = 1.5 um.
brane channels responsible for propagation of the action potential.
Terminal morphology In crab muscles, we found that high-output (and highly inhibited) terminals often have a compact, palmate configuration with many closely adjacent varicosities, whereas low-output, less strongly inhibited terminals are more diffuseli arranged, with-serial varicosities (Fig. 3). This
144
F.W. TSE ET' AL.
'=)
&
0
Axo-axonal synapses
A
A
Excitatory
Inhibitory
Fig. 7. Two-dimensional diagram of the inhibitory terminals in the region of Fig. 5. Conventions as in Fig. 6. Individual axo-axonal synapses are indicated (A, B, C, D). Scale: vertical = 1 pm, horizontal = 1.5 pm.
Fig. 8. Two-dimensional diagram to show synaptic interactions between E and SI terminals in a region of strong inhibition in Puchygrupsus. Conventions as in Fig. 5. Scale: vertical = 1 pm, horizontal = 5 pm.
difference could be significant, if coupled with a difference in extent of active invasion of the terminals by action potentials, as originally proposed by Atwood ('67) and Sherman and Atwood ('72), and again more recently by Dude1 ('82, '83).The compact terminal morphology of Fig. 4 would favor more uniform activation of terminal boutons by a subterminal action potential, and hence more effective release of transmitter by a single impulse in both excitatory and inhibitory terminals [see also Segev ('90) for a theoretical treatment of such differences]. At the ultrastructural level, the main differences between strongly and weakly inhibited regions are the larger number and more uniform distribution of axo-axonal synapses over the excitatory terminal processes in the former case, and the location of inhibitory axo-axonal synapses on varicosities supplied by very thin branches in the latter case. If the inhibitory action potential actively invades the main axon branches but not the thin side processes, the voltage change experienced by axo-axonal synapses may be reduced, on account of the thin, high-resistance side processes, and the probability of release of inhibitory transmitter for a single impulse may be relatively low for synapses on the terminal varicosities of the inhibitory axon. Con-
versely, in the region of strong inhibition, inhibitory transmitter release may be favored by a larger electrical event in the inhibitory terminal. The more widespread distribution of axo-axonalsynapses on the excitatory terminal in the region of strong inhibition, coupled with higher quantal content of inhibitory transmission, could serve to attenuate the electrical events in the excitatory terminal more strongly in this region.
synaptic strength The differences in synaptic size and dense bar (active zone) size and number do not, in themselves, provide an explanation for differences in transmitter release of excitatory or inhibitory axons in the two regions. Excitatory synapses formed a larger proportion of the surface area of high-output terminals, but this was not true in the case of the inhibitory terminals (Figs. 7 and 10). Individual active zones were not greatly different in size and number for excitatory terminals (Figs. 6 and 9). Some active zones of inhibitory synapses are relatively large in the region of strong inhibition; synapses possessing them could be relatively potent (Atwood and Wojtowicz, '86). The striking difference in number and surface area of inhibitory and
CRUSTACEAN NEUROMUSCULAR INHIBITION
145
A
0
go
Weak
Strong
50
Axo-axonal
3
40
n
n
30
0
8 20 10
0
Number of dense bars Per synapse
B “1 40
E
2
30
0
25
20
0
8 10
Fig. 9. Two-dimensional diagram of the excitatory terminals in the region of Fig. 8. Conventions as in Fig. 6. Individual axo-axonal synapses are indicated by the numbers 1-9) Scale: vertical = 1 pm, horizontal = 3 pm.
Length
of donee bar (number of sectlone)
Fig. 11. Comparison of features of inhibitory (SI) synapses for Puchygrupsus in regions of strong and weak inhibition. Percentages of inhibitory synapses with different numbers of dense bars (A) and distribution of dense bar sizes (B) are illustrated. Axo-axonal synapse (shaded) and neuromuscular synapses are included. Asterisk marks an axo-axonal synapse in the region of strong inhibition.
excitatory synapses (consistently found in crustacean limb muscles), together with indications of a higher quantal content for inhibitory transmission (Atwood and Bittner, ’71), suggest that effectivetransmission at low frequencies of stimulation is determined largely by transmitter release from relatively few “potent” synapses. This conclusion is supported by statistical analyses of quantal release at excitatory terminals (Atwood and Tse, ’88). At present, data on calcium channel number and density at these synapses is lacking. Such data, which could eventually be supplied by freeze-fracture techniques (see, for example, Govind and Walrond, ’891, could address in part the question of “potent” and “impotent” synapses for both excitatory and inhibitory terminals. 9
Differencesin electrical properties
No direct evidence on nerve terminal electrical properties was provided by the present study. However, the differences in strength of inhibition in the two regions are not readily explained by the ultrastructural findings on differences in synaptic size and number of active zones. As noted in Terminal morphology, above, if action potential propagation were limited to main axons, the more common occurrence of thin, inexcitable side processes in the region of Fig. 10. Two-dimensional diagram of the inhibitory terminals in the region of Fig, 8, Conventions as in Fig, 6 , Individual axo-axonal weak inhibition could account for part of the difference in synapses are indicated by the numbers 1-91 scale: vertical = 1 wm, effectiveness of presynaptic inhibition. Such branches could attenuate an electrotonically conducted action potential, horizontal = 3 pm.
F.W. TSE ET AL.
146 and a smaller electrical signal would in turn be less effective in releasing quantal units of transmitter from inhibitory synapses. In conclusion, the morphological differences discerned for individual synapses of excitatory and inhibitory terminals do not in themselves lead to a clear resolution of the question of differential physiological effectiveness, but the number and distribution of axo-axonal synapses coupled with differences in terminal morphology and putative terminal electrical properties provides a working hypothesis. A more complete explanation must include more data on: 1) extent of action potential propagation in the terminals; and 2) physiological potency of individual synapses, including information on the number of active calcium channels.
ACKNOWLEDGMENTS The work was supported by a grant from NSERC, Canada, and by an NSERC Visiting Scientist award to Dr. S.S. Jahromi, who was on sabbatical leave from the University of Shiraz, Iran. Marianne Hegstrom-Wojtowicz assisted in preparing the illustrations.
XxrFBAmcm Atwood, H.L. (1967) Variation in physiological properties of crustacean motor synapses. Nature 215:57-58. Atwood, H.L. (1976) Organization and synaptic physiology of crustacean neuromuscular systems. Progr. Neurobiol. 7:291-391. Atwood, H.L. (1982) Synapses and neurotransmitters. In H.L. Atwood and D.C. Sandeman (eds): The Biology of Crustacea, Vol. 3, Neurobiology Structure and Function. New York: Academic Press, pp. 105-150. Atwood, H.L., and A. Jones (1967) Presynaptic inhibition in crustacean muscle: Axo-axonal synapse. Experientia 23:103&1038. Atwood, H.L., and W.A. Morin (1970) Neuromuscular and axoaxonal synapses of crayfish opener muscle. J. Ultrastruct. Res. 32:351-369. Atwood, H.L., and G.D. Bittner (1971) Matching of excitatory and inhibitory inputs to crustacean muscle fibers. J. Neurophysiol. 34:157-170. Atwood, H.L., and I. Kwan (1976) Synaptic development in the crayfish opener muscle. J. Neurobiol. 7:289-312. Atwood, H.L., and I. Kwan (1979) Reciprocal axo-axonal synapses between excitatory and inhibitory neurons in crustaceans. Brain Res. 174:324328. Atwood, H.L., and L. Marin (1983) Ultrastructure of synapses with different transmitter-releasing characteristics on motor axon terminals of a crab, Hyas areneus. Cell Tissue Res. 231:103-115. Atwood, H.L., and J.M. Wojtowicz (1986) Short-term and long-term plasticity and physiological differentiation of crustacean motor synapses. Int. Rev. Neurobiol. 28275-362. Atwood, H.L., and F. Tse (1988) Changes in binomial parameters of quantal release at crustacean motor axon terminals during presynaptic inhibition. J. Physiol. (Lond.)402:177-193. Atwood, H.L., J.K. Stevens, and L. Marin (1984) Axoaxonal synapse location and consequences for presynaptic inhibition in crustacean motor axon terminals. J. Comp. Neurol. 22564-74. Bittner, G.D. (1968) Differentiation of nerve terminals in the crayfish opener muscle and its functional significance. J. Gen. Physiol. 51t731-758. Bunting, J.R., T.V. Phan, E. Kamali, and R.M. Dowben (1989) Fluorescent cationic probes of mitrochondria: Metrics and mechanism of interaction. Biophys. J. 56t979-993. Dudel, J. (1981) The effect of reduced calcium on quantal unit current and release at the crayfish neuromuscular junction. Pflugers Arch. 391:3540. Dudel, J. (1982) Transmitter release by graded local depolarization of presynaptic nerve terminals at the crayfish neuromuscular junction. Neurosci. Lett. 32:181-186. Dudel, J. (1983) Graded or all-or-nothing release of transmitter quanta by
local depolarization of nerve terminals on crayfish muscle. Pflugers Arch. 398: 155-164. Dudel, J., and S.W. KuRler (1961) Presynaptic inhibition a t the crayfish neuromuscular junction. J. Physiol. (Lond.)271r369-390. Dudel, J., I. Parnas, and H. Parnas (1983) Neurotransmitter release and its facilitation in crayfish. VI. Release determined by both intracellular calcium concentration and depolarization of the nerve terminal. Pflugers Arch. 399:l-10. Florey, E., and M.A. Cahill (1982) The innervation patterns of crustacean skeletal muscle. Cell Tissue Res. 224t527-541. Ginsborg, B.L. (1967) Ionic movement in junctional transmission. Pharmacol. Rev. 19:289-316. Govind, C.K., and J.P. Walrond (1989) Structural plasticity at crustacean neuromuscular synapses. J. Neurobiol. 20:409421. Holtzman, E., A.R. Freeman, and L.A. Kashner (1971) Stimulationdependent alterations in peroxidase uptake at lobster neuromuscular junctions. Science 173t733-736. Jahromi, S.S., and Atwood, H.L. (1974) Three-dimensionalultrastructure of the crayfish neuromuscular apparatus. J. Cell Biol. 63:599413. Johnson, L.V., M.L. Walsh, and L.B. Chen (1980) Localization of mitochondria in living cells with rhodamine. Proc. Natl. Acad. Sci. U.S.A. 77:990-999. Johnson, L.V., M.L. Walsh, B.J. Bockus, and L.B. Chen (1981) Monitoringof relative mitochondrial membrane potential in living cells by fluorescence microscopy. J. Cell Biol. 88:526-535. McMahan, U.J., N.C. Spitzer, and K. Peper (1972) Visual identification of nerve terminals in living isolated skeletal muscles. Proc. R. SOC.(Lond) B181t421-430. Nakajima, Y., A.D. Tisdale, and M.P. Henkart (1973) Presynaptic inhibition at inhibitory nerve terminals. A new synapse in the crayfish stretch receptor. Proc. Natl. Acad. Sci. U S A . 702462-2466. Onodera, K., and A. Takeuchi (1980) Distribution and pharmacological properties of synaptic and extrasynaptic glutamate receptors on crayfish muscles. J. Physiol. (Lond.)306.233-250. Segev, I. (1990) Computer study of presynaptic inhibition controlling the spread of action potentials into axonal terminals. J. Neurophysiol. 63:987-998. Sherman, R.G. (1977) Comparative analysis of an excitatory motor unit in crustaceans. J. Comp. Physiol. 114t91-101. Sherman, R.G., and H.L. Atwood (1972) Correlated electrophysiological and ultrastructural studies of a crustacean motor unit. J. Gen. Physiol. 59:586-615. Smith, D.O. (1978) Ultrastructural specificity of synaptic sites in nerve terminals mediating both presynaptic and postsynaptic inhibition. J. Comp. Neurol. 182:839-850. Stephens. P.J., and H.L. Atwood (1982) Thermal acclimation in acrustacean neuromuscular system. J. Exp. Biol. 98t3947. Thompson, C.S., and H.L. Atwood (1984) Synaptic strength and horseradish peroxidase uptake in crayfish nerve terminals. J. Neurocytol. 13267280. Tse, F.W., and H.L. Atwood (1986) Presynaptic inhibition at the crustacean neuromuscular junction. News Physiol. Sci. It47-50. Tse, F.W., L. Marin, and H.L. Atwood (1987) Focal labeling of axonal terminals with active synapses recorded by an extracellular macro-patch electrode. J. Neurosci. Methods 21: 17-29. Van Harreveld, A. (1939) The nerve supply of doubly and triply innervated crayfish muscle related to their function. J. Comp. Neurol. 70:285-296. Wiens, T.J. (1989) Common and specific inhibition in leg muscles of decapods: Sharpened distinctions. J. Neurobiol. 20:45%469. Wiens, T.J., and H.L. Atwood (1975) Dual inhibitory control in crab leg muscles. J. Comp. Physiol. 99.211-230. Wiens, T.J., and Rathmayer, W. (1985) The distribution of the common inhibitory neuron in brachyuran limb musculature. I . Target muscles. J. Comp. Physiol. 156t305-313. Wiersma, C.A.G. (1961) The neuromuscular system. In T.H. Waterman (ed): The Physiology of Crustacea, Vol. 2. New York: Academic Press, pp. 241-279. Yoshikami, D., and L.M. Okun (1984) Staining of living presynaptic nerve terminals with selective fluorescent dyes. Nature 310:53-56.
