- Email: [email protected]
Ultrastructural correlates of experimentally altered transmitter release efficacy in frog motor nerve terminals
0306-4522/85 $3.00 + 0.00
Netuo~&nce Vol. 16, No. 3, pp. 491-500, 1982 Printed in Great Britain
Pergamon Press Ltd 0 1985IBRO
ULTRASTRUCTURAL CORRELATES OF EXPERIMENTALLY ALTERED TRANSMITTER RELEASE EFFICACY IN FROG MOTOR NERVE TERMINALS A. A. H-A, A. D. GRINNELL* and B. WOLOWSKE* Neurobiology Section, Department of Biological Sciences, and Program in Neural, Informational and Behavioral Sciences, University of Southern California, Los Angeles, CA 90089; *Department of Physiology and Jerry Lewis Neuromuscular Research Center, University of California, Los Angeles, CA 90024, U.S.A. Abstract-After experimentally inducing long term changes in transmitter release, a series of frog neuromuscular junctions were studied with intracellular recording and then semi-serially sectioned and examined in the electron microscope. Transmitter release per unit length of motor nerve terminal was well correlated with several measures of the length of individual presynaptic active zones and with the number of mitochondria per terminal. Total release from each terminal correlated with estimates of the total amount of active zone. This study of neuromuscular junctions in sartorius muscles of the frog Rnna pipiens was undertaken to search for ultrastructural correlates of the increase in transmitter release efficacy that follows denervation of the contralateral sartorius. This treatment typically results in greatly enhanced release at some synapses while others appear unaffected. In the present study, nine identified junctions with known physiological properties were sectioned every 6pm throughout much of their length to yield 40-105 cross-sectional profiles per junction. Overall, these 9 synapses showed a 33-fold range in quanta1 transmitter release and an ll-fold range in release per unit nerve terminal length. Release correlated with estimates of active zone size. No correlations were found between release and the density of synaptic vesicles adjacent to active zones. Our results suggest that active zones in motor nerve terminals are plastic structures, and that changes in active zone size may be the structural basis of long term changes in transmitter release and synaptic efficacy.
It is likely that cellular mechanisms of learning and memory in the central nervous system (CNS) involve long term changes in the effectiveness of synapses.% Because of the complexity and inaccessibility of CNS connections, however, it is useful to search for comparable forms of plasticity occurring in peripheral synapses, where underlying structural and physiological mechanisms can be more easily studied. Many investigators have found that the neuromuscular junction is a suitable preparation for this purpose.‘6 One component that would be expected to play a central role in the regulation of synaptic effectiveness is the presynaptic active zone, the presumed site of synaptic vesicle exocytosis and transmitter release9~25*4’ and probably also the site of voltagedependent Car+ influx .45Indeed, in an earlier study,23 we showed that differences in synaptic safety margin at frog neuromuscular junctions are associated with differences in the average size of individual active zones. In the present study, we ask whether similar ultrastructural changes underlie experimentally-induced alterations in synaptic effectiveness. We examine Corresponding author: A. A. Herrera, Department of Biological Sciences, University of Southern California, Los Angeles, CA 90089-0371, U.S.A. Abbreviation: EFT, endplate potential. 491
structural correlates of the large increase in transmitter release that occurs at certain nerve terminals in
a frog sartorius muscle when the contralateral sartorius is denervated.*’ A preliminary report has been published.** EXPERIMENTAL PROCEDURES Adult Rana pipiem (body length 6-7 cm) were used. To obtain muscles with enhanced transmitter release, 7 frogs were anesthetized by immersion in 0.2% tricaine methanesulfonate and the left sartorius nerve was crushed near its entrance to the muscle. Fifty-two to 74 days later, the frogs were pithed, and the right sartorius nerve-muscle preparations were dissected and immersed in Ringer’s solution (116 mM NaCl, 2 mM KCl, 1.8 mM CaCI,, 1 mM MaCl,. and 1 mM NaHCO, at pH 7.2 and 1YC). Using previ&siy described techniaues.‘$3031 these. muscles were then tested for average synaptic safety margin to identify those in which effectiveness was enhanced to the greatest degree because it was anticipated that ultrastructural changes would be most apparent in such preparations. Although differences in safety margin could be due to pre- or postsynaptic factors, previous studies “**’ showed that such changes were due mainly to differences in transmitter release. The 5 muscles with the highest average safety margin were next immersed in Ringer’s solution containing 0.3 mM Ca2+ for intracellular recording while viewina the iunctions through a compound mic&cope (Zeiss G/O.75 water immersion objective, 400X) modified for Hoffman modulation contrast. In the event that ultrastructural details may depend on previous activity, nerves were stimulated at 0.5 Hz continuously for 75 min after equilibration to the low Ca2+ solution. During this time, recordings were made from one
492
A. A. Hnaanaa ef al,
or two areas in each muscle and from 4 to 10 adjacent junctions in each area. A microcomputer was used to measure filmed records of 30-200 spontaneous miniature endplate potentials (EPPs) and 128 EPPs from each junction. Resting potentials varied between -88 and -99mV at different junctions. Quanta1 content was calculated using the direct method (mean EPP/mean miniature EPP), after correction of all potentials for non-linear smnmation.34 Bach junction was marked by inserting a beveled micropipette into the muscle fiber about SOOpm away and pressure injecting a 10-20 nm bolus of 4”/, aqueous Chicago Blue 6B. After 2250 nerve stimuli were delivered at 0.5 Hz, the muscles were pinned at rest length onto a thin layer of Sylgard and stained with the nitroblue tetrazolium nerve terminal stains3 and with cholinesterase stain.27 This light microscopic nerve terminal/cholinesterase method is compatible with subsequent electron microscopy when performed as described below. The following procedure was used: (a) fix for 5 min in 2y glutaraldehyde in Ringer’s solution made hyperosmottc with 3.75g/I ogalactose. Hyperosmotic solutions were found to be essential for adequate tissue preservation after nitroblue tetrazolium staining; (b) immerse in 10 ml of the same solution to which 10mg nitroblue tetrazolium and 1 mg phenazine methosulfate (Sigma) were added 10min previously; (c) monitor terminal staining at 400X with water immersion objective and terminate staining with a 1 min rinse in galactose-Ringer’s solution when optimal contrast between nerve terminal and muscle fiber has developed (4-6 min); (d) fix for 15 min in 2% glutaraldehyde; (e) w&r with galactose-Ringer’s solution and draw identified nerve terminals with camera lucida. The length of 20 sarcomeres was also recorded so that the degree of stretch of each junction could be precisely determined. (f) Stain for cholinesterase at 4°C; tj~) terminate stain by washing with ~lacto~-~nger’s solution when visual observation at 400X fhst reveals cholinesterase reaction product (1.5-2 mm); (h) draw the cholinesterase staining pattern superimposed on the nerve terminal drawings. These drawings were used to measure the total length of all nerve terminal branches in each junction; (i) 6x 1 h in 2% glmaraldehyde at 4°C; (j) cut out small pieces (about 1 x 3 mm) of muscle containing identified junctions; (k) use the camera1 lucida to make detailed drawings of each piece showing junctions, dye injection sites and other landmarks; (1) fix 3 h in 2% ahrtaraldehvde at 4°C: (m) wash 12 h in galactose-Ringer solution at 4T; (n) postix in 2% OsO, in gsactose-Ringer’s solution for 1 h; (0) wash for 30min then dehydrate in acetone and embed in Medcast (Pelco). Each piece was then cross-sectioned starting at one end. Sections were initially 4 pm thick. The local damage at the dye injection site was easily identified in these sections and served to identify the fibers of interest. The known distance between the dye injection and the junction was traversed by 1pm sections. The distalmost branch of the first junction encountered was thin-sectioned within l-2 /Irn of its tip and sequential thin sections were taken every 6brn throughout most of the extent of each identified junction. Thin sections were stained with saturated uranyl acetate and photographed in a Zeiss EMIOC microscope at 80 kV. Prints at a final magnification of 47,400X were analyzed with a digitizing tablet (resolution 0.1 mm on the tablet, corresponding to about 21 A) and microcomputer. Previous studies using freeze-fracture electron microscopy have established that in most cases active zone particles in the presynaptic membrane of frog neuromuscular junctions occur in rows aligned ~~c~~ly to the long axis of the terminal width.25*26**1 Active zones that are obliquely oriented or substantially shorter than the total terminal width seem relatively infrequent, Thus, we assume that the length of individual active zones would be proportional to the length of presynaptic membrane in close unobstructed contact with the postsynaptic membrane
where cross-sections passed through active zones. ‘fo estimate active zone length, we multiplied mean contact width at active zones by 0.76, that being the mean fraction of the apparent terminal width occupied by active zone that we measured from freeze-fracture electron micrographs published by others. ~wA mean active zone spacing of 1.3 active zones per pm of terminal length was also measured from these freeze-fracture views. This corresponds closely to the values of synaptic fold spacing reported earlier for the frog sartorius.” Other measurements were also made from each cross section as previously described.zr Briefly, these are: (a) the perimeter, width, height and cross-sectional area of the &mind; (b) the number and cross-sectional area of mitochondria; (c) synaptic vesicle density; (d) potential contact width, or the length of the terminal perimeter within 0.2 pm of the muscle fiber membrane, and (e) Schwamr cell interposition, or the extent of the potential contact width obstructed by Schwann cell processes interposed between pre- and postsynaptic membranes. RESULTS Selection of specimens
From all the available material, one group of adjacent junctions was selected from each of two muscles (muscles 134 and 141) for semi-serial sectioning. Camera lucida drawings of these 9 junctions indicating the location of the 50 thin sections are shown in Fig. 1 along with physiological and terminal length data. These 50 sections yielded 40-105 cross-sectional profiles of nerve terminal branches per junction. Of these, 28 to 65% (mean 42%) passed through active zones. Nerve terminal lengths were normalized to a sarcomere spacing of 2.2pm to correct for slight differences in the degree to which muscle fibers were stretched at the time of fixation. In our previous study?’ it was apparent that contralateral denervation did not alRxt all junctions equally. Some terminals underwent large increases in release while many were apparently unaf%kered, i.e. they had values of release per unit terminal length that overlapped completely with values obtainexl from muscles of unoperated frogs. Thus, within a single muscle from an operated animal, junctions of both abnormally high and normal release efficacy could be found side by side. The nine junctions shown in Fig. 1 represent such a group. Two of them (134-l 134-3) showed very high release, higher than any normal junction tested to date under the same conditions. Quanta1 content at the weakest of the remaining junctions was not differeut from that of junctions in control muscles. Overall, these 9 identified junctions represented a greater than 33-fold range of EPP quanta1 content (W-30.2 qu@WWP at this Ca2+ concentration), which when correct& for the approximately 3-fold range of terminal length (307-970pmm), translated to a nearly l&fold range of upsetter release per unit terminal length (0.18-3.19q~~~l~~rn). This last parameter is a particularly usefiul measure of the inherent transmitter releasing ability of a nerve terminal. Figure 2a shows the typical appearance of a cross section passing through an active zone. Active zones
Fig. 2. Typical electron micrographs of cross-sections passing through an active zone (A) and near but not through an active zone (B). Criteria for recognizing active zones and morphometric methods given in text. Electron-dense crystals in synaptic cleft are cholinesterase reaction product. 493
llllllllllllllllHlllllllllllllllllllllllllllllIII
EP
M
L
M/L
N
134-l29.4924 3.19 80 134-Z 5.6 605 0.92
52
134-3 302 970 3.11 52 134-4 1.7437 0.39 56 134-5 8.6900 0.96 94
741-5 2.1307 0.68 40 141-6 5.68580.65 105 141-7 2.0 414 0.48 57 141-8 0.9 517 0.18 81
Fig. 1. Camera lueida drawings showing 5 adjacent endplates in one sartorius muscle (no. 134) and 4 adjacent endplates in another sartorius muscle (no. 141) from a different frog. Contralateral sartorius muscles were denervated 60 (no. 134) and 59 (no. 141)days earlier. Thin eross sections were taken at the indicated locations, mostly at 6pm intervals. Also shown for each endplate is the measured value of quantal content (M) in Ringer’s solution with lowered [W+], the total kngtb in pm of all the nerve terminal branches (L), the number of transmitter quanta released per I00 pm nerve terminal kngtb per nerve impulse&f/L), and the number of terminalbranch cross-seetionsobtained@).
were identified by the presence of a presynaptic membrane thickening apposed to the opening of a postsynaptic fold and by the absence of Schwann cell processes between pre- and postsynaptic membranes. Figure 2B shows a section passing near, but not through, an active zone. Even though there is a suggestion of a postsynaptic fold in Fig. 2B and there are many vesicles, the interposed Schwann cell process and the lack of a presynaptic density caused us to exclude such images from our active zone data pool. We felt this approach would eliminate some unnecessary variability.
vesicles/O.045pm* at active zones only. The index of cross-sectional shape obtained by dividing mean terminal width by mean terminal height varied to a greater extent, from 1.2 for the “tallest” terminal to 2.9 for the “flattest” terminal (1.2 and 3.0 for active zone sections only). However, both vesicle density and terminal shape were not correlated with release per unit length (P > 0.05). The morphological parameters that did correlate signScantly with transmitter release per unit terminal length arc plotted in Fig. 3. Closed symbols represent measurements taken from sections through active zones while open symbols show measurements made Ultrastructural observations for portions of terminals between active zones. As in the previous ultrastructural comparison of Figures 3A and B show that nerve terminal perinormal cutaneous pectoris and sartorius neurometer and cross-sectional area correlated significantly muscular junctions,” synaptic vesicle density and with release per unit length and that, in 7 of 9 nerve terminal shape showed no significant cor- junctions studied, these measurements were larger at relation with transmitter release levels. Vesicle density active zones than between active zones. Figure 3C varied only slightly between 4.3 and 5.9 vesicles/ shows that for these 9 identified contralateral junc0.045 pm2 (mean values) at different junctions when tions there is a positive correlation between release all sections were considered, and between 5.2 and 9.7 per unit length and the number of mitochondria per
496
A. A. Hw
section. However, as Fig. 3D shows, the number of ~t~hon~a per section is a linear function of the cross-sectional area of the terminal and more likely related to axoplasmic volume than to release per unit length. A closer examination of Fig. 3C reinforces the conclusion that the number of mitochondria per section may not be an important determinant of release since a v?eak junction 141-2 (release per unit length 0.92) had the same average number of mitochondria as the two strong junctions 134-I and 134-3 (release per unit length 3.19 and 3.11). No difference in mitochondrial size was apparent in different terminals. Figure 3 also shows that release per unit length is correlated with the length of presynaptic membrane in unobstructed close contact with postsynaptic membrane whether expressed absolutely (Fig. 3E) or as the proportion of the terminal perimeter in close contact (Fig. 3F). In nearly every case, these last two measurements were larger at active zones than between active zones. Figure 3G shows the relation between release per unit length and another approximation of active zone length, the width of presynaptic membrane within 0.2 pm of postsynaptic membrane. This potential contact width is largest for the most strongly releasing terminals, and is greater at active zones than between active zones for all 9 terminals. In Fig. 3H, release per unit length is plotted against the fraction of the potential contact width obstructed by Schwann cell processes. A significant correlation exists only for sections between active zones. Figure 3H also illustrates the well known fact that Schwann cell wrapping of terminals occurs principally in regions between active zones. Correlations between release and active zone length
Since several of these measurements suggested that the average length of individual active zones was an important determinant of transmitter release per unit length, it was of interest to see how estimates of total active zone length in individual terminals correlated with total transmitter release from those terminals. To estimate total active zone length from our data 3 factors must be known: total terminal length, the mean length of individual active zones, and the spacing between active zones. Terminal length can be accurately determined from light microscopic camera lucida drawings. We estimated the mean length of individual active zones by multiplying mean contact width at active zones by the fraction of the apparent terminal width occupied by active zone as seen in freez+fracture views of frog terminals avaihtble in the literature (see Experimental Procedures). Active zone spacing was assumed to be a constant 1.3 active zones per pm of terminal length, a figure likewise dctctmined from data pub&shed by others2’Jb and in good agreement with our previous estimates based on synaptic fold spacing. I5 The results are shown in Fig. 4, which depicts the relation between total trans-
el al.
mitter release per nerve impulse and total active zone length for the 9 identified contralateral junctions. DISCUSSION
~e~at~o~hip between release and active zone sire
These results, together with those previously published,23 lead to the conclusion that the amount of presynaptic active zone is an important ultrastructural determinant of the level of transmitter release from a motor nerve terminal, The quantitative relationship apparently has a steep slope, e.g. comparing the 4 weakest junctions with the 2 strongest, a S-fold difference in total active zone length correlates with an approximately 1%fold increase in total release (Fig. 4). This implies that the relation between active zone size and release is non-linear and/or that there are correlated physiological differences, not readily discerned in electron micrographs. Indeed, it has been shown that there are differences in resting and impulse-evoked Ca2+ influx correlated with release per unit length in frog cutaneous pectoris junctions. 3pHowever, for these 9 physiologically identified sartorius jtmctions, the amount of transmitter release per unit length of terminal was proportional to the size of individual active zones, and total transmitter output was well correlated with estimates of the total amount of active zone in all the terminal branches. Others have also found correlations between transmitter release levels and active zones in normal,25 degenera~ng,~ and regenerating3u frog junctions, in human junctions affected by Lambert-Eaton myasthenic syndrome,12 in crustacean neuromuscular junctions,‘3.‘4.35*36,49and in Aplysia3 and squid” synapses. Nerve terminals on fast and slow muscle fibers in the frog, which are known to differ in transmitter release, show corresponding differences in total active zone size.53 One obvious way for terminals to increase the total amount of active zone is to grow larger. In fact, a positive correlation between terminal size and transmitter release has been noted by a number of authors.2*6*1’*32 In our results the two junctions with the highest release levels are indeed the two largest junctions (134-1, 134-3 in Fig. 1). However, the overall correlation is quite poor, since two other terminals with nearly the same size (134-5, 141-6) release much less transmitter. Other studies show that when subpopulations of muscle fibers with the same input resistance are considered, there is actually an inverse relationship between terminal size and release.S7*” Clearly there are other regulatory mechanisms that also govern release. Our data do not, of course, exclude the possibility that a change in the spacing of active zones or the density of intramembrane particles within active zones also contribute to the increase in release. Neither of these parameters can be determined from our cross-sectional views. A recent freeze-fracture study showed that active zone spacing and the num-
497
Ultrastruetural and physiological plasticity length vary considerably at different junctions within the same muscle. 40 Similar ultrastructural studies comparing junctions in two muscles whose terminals differ greatly in release’9 confirm that active zone
1
30
her of active zone particles per unit terminal
n ii
0
I 1000
0
TOTAL
0
2 4 5 6 10 + YITOCHOWDRIA
:ON:ACT2Wl:TH
0
:,m,
1 2 3 4 POTENTIAL CONTACT WIDTH (pm)
clto**
iECT. k*
&I?)
0 0.1 0.2 0.a 0.4 0.5 CONTACT WIDTH/PCRlMETER
0
0.1 0.2 0.3 0.4 0.5 SCI /POTENTIAL CONTACT WIDTH
Fig. 3. Scatter plots relating aspects of nerve terminal ultrastructure to transmitter release per unit nerve terminal length (M/L). Each closed symbol is the mean value for a given endplate for all cross-sections passing through active zones. Each horizontally adjacent open symbol is the mean value for the same endplate for cross-sections passing between active zones. Data at and between active zones were combined and averaged for each endplate to obtain the single set of points in 3D. Circles are data from endplates
in muscle no. 134, triangles are data from muscle 141. For each graph the linear correlation eoetlkient is 8iven below for closed points (r,) and open points (rO).Significance levels for 7 degrees of freedom are r = 0.67 (P < 0.05) and r = 0.80 (P < 0.01). 3A: M/L vs nerve terminal perimeter, r, = 0.70, r, = 0.81. 3B: M/L vs nerve terminal cross seetional area, r, = 0.74, r, =0.86. 3C: M/L vs number of mitochondria per section, r, = 0.75, r,, = 0.56. 3D: Numhcr of mitochondria for all sections vs cross-sectional area for all sections, r = 0.97. 3E: M/L vs contact width, or the let@ of presynaptic membrane in close unobstructed con-
tact with the postsynaptic membrane, rc = 0.89, r,, = 0.93. 3F: M/L vs the ratio of contact width to nerve tmninal perimeter, r, = 0.97, r, = 0.90.3G: M/L vs potential contact width, or the length of the terminal perimeter within $2 Am of the muscle fiber membrane, r, = 0.84, r, = 0.90.3H: M/L vs the ratio of Schwann cell interposition (SCI) to potential contact width, or the fraction of the potential contact width where actual synaptic contact is occluded by Schwann cell processes, rs = 0.53, r, = 0.55.
ACTIVE ZONE
I
2000 LENGTH
1
3000 (Mm)
Fig. 4. The relation between total quanta1 transmitter release per nerve stimulus and the estimated total length of all the active zone in each of the 9 identified contralateral sartorius endplates. Line fit by linear regression, r = 0.96, P
A, A. HERRERAet al.
498
differences in the depth of invagination of terminals into their gutters. Finally, differences in contact width may reflect differences in the extent to which Schwann cell processes are interposed between pre- and postsynaptic membranes. Studies on the neurohypophysis have revealed inverse correlations between levels of hormone release and the extent to which glial cells wrap the neurosecretory endings.42*s’*52It is not clear whether a similar correlation holds for our data. At active zones, the percentage of the potential contact width (portion of terminal perimeter membrane within 0.2 pm of the muscle membrane) obstructed by Schwann cell processes was not significantly correlated with release. There was, however, a significant correlation for sections passing between active zones (Fig. 3H). ~eiat~o~hip ameters
between release and some other par-
Our finding that the density of synaptic vesicles near the presynaptic membrane does not correlate with transmitter release IeveIs is in agreement with a previous ultrastructural study of physiologically identified frog neuromuscular junctions4’ This earlier study found that vesicle density was an extremely variable feature and showed little if any correlation with the amount of release just before fixation. In addition, nerve terminals in muscles of aged mice, which release as much or more transmitter as terminals in younger mice, contain reduced numbers of vesicles.5~“*28 On the other hand, a positive correlation has been found in some studies. Developing motor nerve terminals in the avian iris gradually become able to follow repetitive nerve activity as terminals enlarge and vesicle density increases.43 In long term habituated Aplysiu, the number of vesicles associated with active zones in the synapses involved decreases, while sensitized animals show increased vesicle numbers3 It is tempting to equate the vesicles immediately adjacent to active zones or presynaptic densities with the immediately releasable transmitter pool, but this has not been directly confirmed. Given the fact that only vesicles with the most intimate contact may be released, and given the uncertainties regarding both the rate at which new vesicles are moved into this immediately releasable pool and the effects of tissue preparation on vesicle position, it would seem prudent to be cautious in interpreting all such studies. Similar cautions may apply when one attempts to correlate mitochondrial numbers with release. Release may be affected by the Ca*+-sequestering ability of mitochondria’ but other Ca*+ buffers which are not so easily quantified from micrographs may be more important, such as smooth endoplasmic reticulum,* soluble buffers:’ other organelles,‘*,~ and active
pumping to the eel1 exterior.‘*1° Our results imply that the Ca*+-sequestering activity of mit~hon~a is less important than their role in providing adenosine 5’-triphosphate for the metabolic activity associated with release. More strongly releasing nerve terminals were generally found to contain more mitochondria. This may reflect the greater energy demand caused by high release levels or, since mitochondrial number was directly proportional to cross-sectional area (Fig. 3D), the number may simply reflect cytoplasmic volume. It should be noted, however, that in an earlier study comparing cutaneous pectoris and sartorius junctions, the more strongly releasing cutaneous pectoris terminals contained more ~t~hond~a without a difference in terminal voIume.” Mechanisms of contralateral increase in transmitter release
The present results also allow some speculation on the mechanism by which unilateral denervation of a sartorius muscle causes an increase in transmitter release from nerve terminals in the contralateral mu&e. Our sample was chosen such that 2 of the 9 identified contralateral terminals released more transmitter per unit terminal length than any control terminal we have ever encountered. Thus it is highly likely that these 2 junctions, 134-l and 134-3, would best demonstrate the morphological basis of long term plasticity in transmitter release. Examination of Figs 3 and 4 shows that these 2 strongly releasing terminals uniquely differed from the remaining 7 weaker terminals in 4 ways: the actual width of close synaptic contact, the potential contact width, the fraction of the terminal perimeter in synaptic contact, and total active zone length. It is unlikely that enhanced release could be explained by sprouting and new synapse formation, of the type described by RotshenkeP and Ring et aI.* follo~ngcontr~ateral denervation of the frog cutaneous pectoris and sartorius muscles, respectively. A recent study found no evidence that contralateral denervation enhances the sprouting that normally occurs in the sartorius as part of the ongoing process of synaptic remodelling.24 In addition, there was no significant correlation in the present results between release and terminal length. We conclude that the active zone in adult motor nerve terminafs may be a plastic structure, and that long term changes in transmitter reiease from these terminals can be achieved by increasing the length of individual active zones. Aeknowledaements-We thank L. Batter, H. Kabe, F. Knight ana D. Scott for technical a&taace This work was supported by USPHS grants MB6232 to A.D.G. and NS18186 to A.A.H. and by the Muscular Dystrophy Association.
REFERENCES 1, Alnaes E. and Rahamimoff
R. (1975) On the role of mitochondria terminals. J. Physfoi., Land. 248, 285-306.
in transmitter
release from frog motor nerve
Ultrastructural
and physiological plasticity
499
2. Angaut-Petit D. and Mallart A. (1979) Dual innervation of endplate sites and its consequences for newmmh transmission in muscles of adult Xenopus haeuis. J. Physiol., Lond. 289, 203-218. 3. Bailey C. H. and Chen M. (1983) Morphological basis of long-term habituation and sensitization in Aplysia. Science, N.Y. 2243, 91-93. 4. Baker P. F. and Schlaepfer W. W. (1978) Uptake and binding of calcium by axoplasm isolated from giant axons of Loligo and Myxicola. J. Physiol., Land. 276, 102125. 5. Banker B. Q., Kelly S. S. and N. Robbins (1983) Neuromuscular transmission and correlative morphology in young and old mice. J. Physiol., Land. 339, 355-375. 6. Bennett M. R. and Raftos J. (1977) The formation and regression of synapses during the re-innervation of axolotl striated muscles. J. Physiol., Lond. 265, 261-295. 7. Blaustein M. P. and Hodgkin A. L. (1969) The effect of cyanide on the efflux of calcium from squid axons. J. Physiol., Lond. 280, 497-527. 8. Blaustein M. P., Ratzlaff R. W. and Schweitzer E. S. (1978) Calcium buffering in presynaptic nerve texminals. II-Kinetic properties of the nonmitochondrial Ca sequestration mechanism. J. gen. Physiol. 72, 43-66. 9. Couteaux R. and Pecot-Dechavassine M. (1970) Vesicules synaptiques et poches au niveau de les “zones actives” de la jonction neuromusculaire. C.r. hebd. S&nc. Acad. Sci., Paris 271, 2346-2349.
10. Dip010 R. and Beauge L. (1979) Physiological role of ATP-driven calcium pump in squid axon. Narure, Load. 278, 271-273. 11. Fahim M. A. and Robbins N. (1982) Ultrastructural studies of young and old mouse neuromuscular junctions. J. Neurocyfol. 11, 641-656. 12. Fukunaga H., Engel A. G., Gsame M. and Lambert E. H. (1982) Paucity and disorganization of presynaptic membrane active zones in the Lambert-Eaton myasthenic syndrome. Muscle Neroe 5,686-697. 13. Govind C. K. and Chiang R. G. (1979) Correlation between presynaptic dense bodies and transmitter output at lobster neuromuscular terminals by serial section eleotron microscopy. Brain Res. 161, 377-388. 14. Govind C. K. and Meiss D. E. (1979) Quantitative comparison of low and high-output neuromuscular synapses from a motomeuron of the lobster (Homarus americanus). Cell T&s. Res. M&455-463. 15. Grinnell A. D. and Herrera A. A. (1980) Physiological regulation of synaptic effectiveness at frog neuromuscular junctions. J. Physiol., Lond. 307, 301-317. 16. Grinnell A. D. and Herrera A. A. (1981) Specificity and plasticity of neuromuscular connections: long term regulation of motoneuron function. Prog. Neurobiol. 17, 203-282. 17. Harris J. B. and Ribchester R. R. (1979) The relationship between endplate size and transmitter release in normal and dystrophic muscles of the mouse. J. Physiol., Lond. 296, 245-265. 18. Henkart M. P., Reese T. S. and Brinley F. J. (1978) Endoplasmic reticulum sequesters calcium in the squid giant axon. Science, N. Y. 202, 1300-l 302. 19. Herrera A. A., Banner L. R. and Mason P. (1985) Differences in synaptic efficacy at frog neuromuscular junctions. Sot. Neurosci. Absr 11, 302. 20. Herrera A. A. and Grinnell A. D. (1980) Transmitter release from frog motor nerve terminals depends on motor unit size. Nature, Lond. 287, 649-651. 21. Herrera A. A. and Grinnell A. D. (1981) Contralateral denervation causes enhanced transmitter release from frog motor nerve terminals. Nature, Lond. 291, 495-497. 22. Herrera A. A. and Grinnell A. D. (1982) Synaptic apposition, glial obstruction, and synaptic strength at frog neuromuscular junctions. Sot. Neurosci. Abs 8, 492. 23. Herrera A. A., Grinnell A. D. and Wolowske B. (1985) Ultrastructural correlates of naturally occurring differences in transmitter release efficacy in frog motor nerve tetminals. J. Neurocytol. 14, 193-202. 24. Herrera A. A. and Scott D. R. (1985) Motor axon sprouting in frog sartorius muscles is not altered by contralateral axotomy. J. Neurocyiol. 14, 145-156. 25. Heuser J. E., Reese T. S., Dennis M. J., Jan Y., Jan L. and Evans L. (1979) Synaptic vesicle exocytosis captured by quick freezing and correlated with quanta1 transmitter release. J. Cell Biol. 81, 275-300. 26. Heuser J. E., Reese T. S. and Landis D. M. D. (1974) Functional changes in frog neuromuscular junctions studied with freeze-fracture. J. Neurocytol. 3, 109-131. 27. Kamovsky M. J. (1964) The localization of cholinesterase activity in rat cardiac muscle by electron microscopy. J. Cell Biol. 23, 217-232. 28. Kelly S. S. and Robbins N. (1983) Progression of age changes in synaptic transmission at mouse neuromuscular junctions. J. Physiol., Lond. 343, 375-383. 29. Ko C.-P. (1981) Electrophysiological and freeze-fracture studies of changes following denervation at frog neuromuscular junctions. J. Physiol., Lond. 321, 627-639. 30. Ko C.-P.-(1984) Regeneration of the active zone at the frog neuromuscular junction. J. Cell Viol. 9?I, 16854695.
31. Ko C-P., Probst J. W. and Herrera A. A. (1985) A comparison of active zone ultrastructure in two muscles with markedly different synaptic efficacy. Sot. Neurosci. Abst 11, 302. 32. Kuno M., Turkanis S. A. and Weakly J. N. (1971) Correlation between nerve terminal size and transmitter release at the neuromuscular junction of the frog. J. Physiol., Lond. 213, 545-556. 33. Letinsky M. S. and DeCino P. (1980) Histological staining of pre- and postsynaptic components of amphibian neuromuscular junctions. J. Neurocytol. 9, 303-320. 34. Martin A. R. (1955) A further study of the statistical composition of the end-plate potential. J. Physiol., Lond. 130, 114-122. 35. Meiss D. E. and Govind C. K. (1979) Multiterminal innervation: non-uniform density along single lobster muscle fibers. Brain Res. 160, 163-169. 36. Meiss D. E. and Govind C. K. (1980) Heterogeneity of existing synapses at the ends of single muscle fibers in the lobster, Homarus americanus. J. Neurobiol. 11, 381-395. 37. Nude11 B. M. and Grinnell A. D. (1982) Inverse relationship between transmitter release and terminal length in synapses on frog muscle fibers of uniform input resistance. J. Neurosci. 2, 216224. 38. Nude11 B. M. and Grinnell A. D. (1983) Regulation of synaptic position, size, and strength in anuran skeletal muscle. J. Neurosci. 3, 161-176.
500
A. A. HERRERAet al.
39. Pawson P. A. and Grinnell A. D. (1984) Posttetanic potentiation in strong and weak neuromuscular junctions: physiological differences caused by a differential Ca2+ influx. Brain Res. 323, 31 t-315. 40. Pawaon P. A. and Grinnell A. D. (1985) Physiological and possible structural bases for the dif%rences in synaptic efficacy at the frog neuromuscular junction. Can. I Physiol. Pharmac. 63, Axxiii. 41. Peper K., Dreyer F., Sandri C., Akert K. and Moor H. (1974) Structure and ultrastructure of frog motor endplate. Cell Tissue Res. 149, 437-455. 42. Perfmutter L. S., Hatton G. 1. and Tweedle C. D. (1984) Plasticity in the in c~itroueuroh~ophysis: effects of osmotic changes on pituicytes. Neuroscience 12, 503-S 11. 43. Piiar G., Tuttle J. and Vaca K. (1981) Functional maturation of motor nerve terminals in the avian iris: ultra-structure, transmitter metabolism and synaptic reliability. J. Physiol., Lond. 321, 175-193. 44. Politoff A. L., Rose S. and Pappas G. D. (1974) Calcium binding sites of synaptic vesicles of frog sartorius neuromuscular junction. J. Cell Biul. 61, 81%823. 45. Pumplin D. W., Reese T. S. and Llinh R. (I 981) Are presynaptic membrane particles the calcium channels? Pruc. mtn. Acad. Sci., U.S.A. 78, 7210-7213. 46. Ring G., Reichert F. and Rotshenker S. (1983) Sprouting in intact sartorius muscles of the frog following contralateral axotomy. Brain Res. 260, 313-316. 47. Rose S. J., Pappas G. D. and Kriebel M. E. (1978) The fine structure of identified frog ne~omu~ular junctions in relation to synaptic activity. Brain Res. 144,213-239. 48. Rotshenker S. (1979) Synapse formation in intact innervated cutaneous pectoris muscles of the frog fohowing denervation of the opposite-muscle. J. Physiol., Lond. 292, 535-547. 49. Sherman R. G. and Atwood H. L. (1972) Correlated electroohvsioloaical and ultrastructural studies of a crustacean .motor unit. J. gem Physiof. 59, S&615. 50. Thompson R. F., Rerger 1. W. and Madden J. (1983) Cellular processes of learning and memory in the mammalian CNS. A. Rev. Neurosci. 6, 447491. 51. Tweedle C. D. (1983) Ultrastructural manifestations of increased hormone release in neurohypophysis. Prog. Bruin Res. 60, 259-272. 52. Tweedle C. D. and Hatton G. I. (1982) Magnocellular neuropeptidergic terminals in ne~ohy~physis: rapid gbal release of enclosed axons during pa~u~~on. Brain Res. Bull. 8, 205-209. 53. Verma V. and Reese T. S. (1984) Structure and distribution of neuromuscular junctions on slow muscle fibers in the frog. Neuroscience 12, 647-662. (Accepted 30 May 1985)
Netuo~&nce Vol. 16, No. 3, pp. 491-500, 1982 Printed in Great Britain
Pergamon Press Ltd 0 1985IBRO
ULTRASTRUCTURAL CORRELATES OF EXPERIMENTALLY ALTERED TRANSMITTER RELEASE EFFICACY IN FROG MOTOR NERVE TERMINALS A. A. H-A, A. D. GRINNELL* and B. WOLOWSKE* Neurobiology Section, Department of Biological Sciences, and Program in Neural, Informational and Behavioral Sciences, University of Southern California, Los Angeles, CA 90089; *Department of Physiology and Jerry Lewis Neuromuscular Research Center, University of California, Los Angeles, CA 90024, U.S.A. Abstract-After experimentally inducing long term changes in transmitter release, a series of frog neuromuscular junctions were studied with intracellular recording and then semi-serially sectioned and examined in the electron microscope. Transmitter release per unit length of motor nerve terminal was well correlated with several measures of the length of individual presynaptic active zones and with the number of mitochondria per terminal. Total release from each terminal correlated with estimates of the total amount of active zone. This study of neuromuscular junctions in sartorius muscles of the frog Rnna pipiens was undertaken to search for ultrastructural correlates of the increase in transmitter release efficacy that follows denervation of the contralateral sartorius. This treatment typically results in greatly enhanced release at some synapses while others appear unaffected. In the present study, nine identified junctions with known physiological properties were sectioned every 6pm throughout much of their length to yield 40-105 cross-sectional profiles per junction. Overall, these 9 synapses showed a 33-fold range in quanta1 transmitter release and an ll-fold range in release per unit nerve terminal length. Release correlated with estimates of active zone size. No correlations were found between release and the density of synaptic vesicles adjacent to active zones. Our results suggest that active zones in motor nerve terminals are plastic structures, and that changes in active zone size may be the structural basis of long term changes in transmitter release and synaptic efficacy.
It is likely that cellular mechanisms of learning and memory in the central nervous system (CNS) involve long term changes in the effectiveness of synapses.% Because of the complexity and inaccessibility of CNS connections, however, it is useful to search for comparable forms of plasticity occurring in peripheral synapses, where underlying structural and physiological mechanisms can be more easily studied. Many investigators have found that the neuromuscular junction is a suitable preparation for this purpose.‘6 One component that would be expected to play a central role in the regulation of synaptic effectiveness is the presynaptic active zone, the presumed site of synaptic vesicle exocytosis and transmitter release9~25*4’ and probably also the site of voltagedependent Car+ influx .45Indeed, in an earlier study,23 we showed that differences in synaptic safety margin at frog neuromuscular junctions are associated with differences in the average size of individual active zones. In the present study, we ask whether similar ultrastructural changes underlie experimentally-induced alterations in synaptic effectiveness. We examine Corresponding author: A. A. Herrera, Department of Biological Sciences, University of Southern California, Los Angeles, CA 90089-0371, U.S.A. Abbreviation: EFT, endplate potential. 491
structural correlates of the large increase in transmitter release that occurs at certain nerve terminals in
a frog sartorius muscle when the contralateral sartorius is denervated.*’ A preliminary report has been published.** EXPERIMENTAL PROCEDURES Adult Rana pipiem (body length 6-7 cm) were used. To obtain muscles with enhanced transmitter release, 7 frogs were anesthetized by immersion in 0.2% tricaine methanesulfonate and the left sartorius nerve was crushed near its entrance to the muscle. Fifty-two to 74 days later, the frogs were pithed, and the right sartorius nerve-muscle preparations were dissected and immersed in Ringer’s solution (116 mM NaCl, 2 mM KCl, 1.8 mM CaCI,, 1 mM MaCl,. and 1 mM NaHCO, at pH 7.2 and 1YC). Using previ&siy described techniaues.‘$3031 these. muscles were then tested for average synaptic safety margin to identify those in which effectiveness was enhanced to the greatest degree because it was anticipated that ultrastructural changes would be most apparent in such preparations. Although differences in safety margin could be due to pre- or postsynaptic factors, previous studies “**’ showed that such changes were due mainly to differences in transmitter release. The 5 muscles with the highest average safety margin were next immersed in Ringer’s solution containing 0.3 mM Ca2+ for intracellular recording while viewina the iunctions through a compound mic&cope (Zeiss G/O.75 water immersion objective, 400X) modified for Hoffman modulation contrast. In the event that ultrastructural details may depend on previous activity, nerves were stimulated at 0.5 Hz continuously for 75 min after equilibration to the low Ca2+ solution. During this time, recordings were made from one
492
A. A. Hnaanaa ef al,
or two areas in each muscle and from 4 to 10 adjacent junctions in each area. A microcomputer was used to measure filmed records of 30-200 spontaneous miniature endplate potentials (EPPs) and 128 EPPs from each junction. Resting potentials varied between -88 and -99mV at different junctions. Quanta1 content was calculated using the direct method (mean EPP/mean miniature EPP), after correction of all potentials for non-linear smnmation.34 Bach junction was marked by inserting a beveled micropipette into the muscle fiber about SOOpm away and pressure injecting a 10-20 nm bolus of 4”/, aqueous Chicago Blue 6B. After 2250 nerve stimuli were delivered at 0.5 Hz, the muscles were pinned at rest length onto a thin layer of Sylgard and stained with the nitroblue tetrazolium nerve terminal stains3 and with cholinesterase stain.27 This light microscopic nerve terminal/cholinesterase method is compatible with subsequent electron microscopy when performed as described below. The following procedure was used: (a) fix for 5 min in 2y glutaraldehyde in Ringer’s solution made hyperosmottc with 3.75g/I ogalactose. Hyperosmotic solutions were found to be essential for adequate tissue preservation after nitroblue tetrazolium staining; (b) immerse in 10 ml of the same solution to which 10mg nitroblue tetrazolium and 1 mg phenazine methosulfate (Sigma) were added 10min previously; (c) monitor terminal staining at 400X with water immersion objective and terminate staining with a 1 min rinse in galactose-Ringer’s solution when optimal contrast between nerve terminal and muscle fiber has developed (4-6 min); (d) fix for 15 min in 2% glutaraldehyde; (e) w&r with galactose-Ringer’s solution and draw identified nerve terminals with camera lucida. The length of 20 sarcomeres was also recorded so that the degree of stretch of each junction could be precisely determined. (f) Stain for cholinesterase at 4°C; tj~) terminate stain by washing with ~lacto~-~nger’s solution when visual observation at 400X fhst reveals cholinesterase reaction product (1.5-2 mm); (h) draw the cholinesterase staining pattern superimposed on the nerve terminal drawings. These drawings were used to measure the total length of all nerve terminal branches in each junction; (i) 6x 1 h in 2% glmaraldehyde at 4°C; (j) cut out small pieces (about 1 x 3 mm) of muscle containing identified junctions; (k) use the camera1 lucida to make detailed drawings of each piece showing junctions, dye injection sites and other landmarks; (1) fix 3 h in 2% ahrtaraldehvde at 4°C: (m) wash 12 h in galactose-Ringer solution at 4T; (n) postix in 2% OsO, in gsactose-Ringer’s solution for 1 h; (0) wash for 30min then dehydrate in acetone and embed in Medcast (Pelco). Each piece was then cross-sectioned starting at one end. Sections were initially 4 pm thick. The local damage at the dye injection site was easily identified in these sections and served to identify the fibers of interest. The known distance between the dye injection and the junction was traversed by 1pm sections. The distalmost branch of the first junction encountered was thin-sectioned within l-2 /Irn of its tip and sequential thin sections were taken every 6brn throughout most of the extent of each identified junction. Thin sections were stained with saturated uranyl acetate and photographed in a Zeiss EMIOC microscope at 80 kV. Prints at a final magnification of 47,400X were analyzed with a digitizing tablet (resolution 0.1 mm on the tablet, corresponding to about 21 A) and microcomputer. Previous studies using freeze-fracture electron microscopy have established that in most cases active zone particles in the presynaptic membrane of frog neuromuscular junctions occur in rows aligned ~~c~~ly to the long axis of the terminal width.25*26**1 Active zones that are obliquely oriented or substantially shorter than the total terminal width seem relatively infrequent, Thus, we assume that the length of individual active zones would be proportional to the length of presynaptic membrane in close unobstructed contact with the postsynaptic membrane
where cross-sections passed through active zones. ‘fo estimate active zone length, we multiplied mean contact width at active zones by 0.76, that being the mean fraction of the apparent terminal width occupied by active zone that we measured from freeze-fracture electron micrographs published by others. ~wA mean active zone spacing of 1.3 active zones per pm of terminal length was also measured from these freeze-fracture views. This corresponds closely to the values of synaptic fold spacing reported earlier for the frog sartorius.” Other measurements were also made from each cross section as previously described.zr Briefly, these are: (a) the perimeter, width, height and cross-sectional area of the &mind; (b) the number and cross-sectional area of mitochondria; (c) synaptic vesicle density; (d) potential contact width, or the length of the terminal perimeter within 0.2 pm of the muscle fiber membrane, and (e) Schwamr cell interposition, or the extent of the potential contact width obstructed by Schwann cell processes interposed between pre- and postsynaptic membranes. RESULTS Selection of specimens
From all the available material, one group of adjacent junctions was selected from each of two muscles (muscles 134 and 141) for semi-serial sectioning. Camera lucida drawings of these 9 junctions indicating the location of the 50 thin sections are shown in Fig. 1 along with physiological and terminal length data. These 50 sections yielded 40-105 cross-sectional profiles of nerve terminal branches per junction. Of these, 28 to 65% (mean 42%) passed through active zones. Nerve terminal lengths were normalized to a sarcomere spacing of 2.2pm to correct for slight differences in the degree to which muscle fibers were stretched at the time of fixation. In our previous study?’ it was apparent that contralateral denervation did not alRxt all junctions equally. Some terminals underwent large increases in release while many were apparently unaf%kered, i.e. they had values of release per unit terminal length that overlapped completely with values obtainexl from muscles of unoperated frogs. Thus, within a single muscle from an operated animal, junctions of both abnormally high and normal release efficacy could be found side by side. The nine junctions shown in Fig. 1 represent such a group. Two of them (134-l 134-3) showed very high release, higher than any normal junction tested to date under the same conditions. Quanta1 content at the weakest of the remaining junctions was not differeut from that of junctions in control muscles. Overall, these 9 identified junctions represented a greater than 33-fold range of EPP quanta1 content (W-30.2 qu@WWP at this Ca2+ concentration), which when correct& for the approximately 3-fold range of terminal length (307-970pmm), translated to a nearly l&fold range of upsetter release per unit terminal length (0.18-3.19q~~~l~~rn). This last parameter is a particularly usefiul measure of the inherent transmitter releasing ability of a nerve terminal. Figure 2a shows the typical appearance of a cross section passing through an active zone. Active zones
Fig. 2. Typical electron micrographs of cross-sections passing through an active zone (A) and near but not through an active zone (B). Criteria for recognizing active zones and morphometric methods given in text. Electron-dense crystals in synaptic cleft are cholinesterase reaction product. 493
llllllllllllllllHlllllllllllllllllllllllllllllIII
EP
M
L
M/L
N
134-l29.4924 3.19 80 134-Z 5.6 605 0.92
52
134-3 302 970 3.11 52 134-4 1.7437 0.39 56 134-5 8.6900 0.96 94
741-5 2.1307 0.68 40 141-6 5.68580.65 105 141-7 2.0 414 0.48 57 141-8 0.9 517 0.18 81
Fig. 1. Camera lueida drawings showing 5 adjacent endplates in one sartorius muscle (no. 134) and 4 adjacent endplates in another sartorius muscle (no. 141) from a different frog. Contralateral sartorius muscles were denervated 60 (no. 134) and 59 (no. 141)days earlier. Thin eross sections were taken at the indicated locations, mostly at 6pm intervals. Also shown for each endplate is the measured value of quantal content (M) in Ringer’s solution with lowered [W+], the total kngtb in pm of all the nerve terminal branches (L), the number of transmitter quanta released per I00 pm nerve terminal kngtb per nerve impulse&f/L), and the number of terminalbranch cross-seetionsobtained@).
were identified by the presence of a presynaptic membrane thickening apposed to the opening of a postsynaptic fold and by the absence of Schwann cell processes between pre- and postsynaptic membranes. Figure 2B shows a section passing near, but not through, an active zone. Even though there is a suggestion of a postsynaptic fold in Fig. 2B and there are many vesicles, the interposed Schwann cell process and the lack of a presynaptic density caused us to exclude such images from our active zone data pool. We felt this approach would eliminate some unnecessary variability.
vesicles/O.045pm* at active zones only. The index of cross-sectional shape obtained by dividing mean terminal width by mean terminal height varied to a greater extent, from 1.2 for the “tallest” terminal to 2.9 for the “flattest” terminal (1.2 and 3.0 for active zone sections only). However, both vesicle density and terminal shape were not correlated with release per unit length (P > 0.05). The morphological parameters that did correlate signScantly with transmitter release per unit terminal length arc plotted in Fig. 3. Closed symbols represent measurements taken from sections through active zones while open symbols show measurements made Ultrastructural observations for portions of terminals between active zones. As in the previous ultrastructural comparison of Figures 3A and B show that nerve terminal perinormal cutaneous pectoris and sartorius neurometer and cross-sectional area correlated significantly muscular junctions,” synaptic vesicle density and with release per unit length and that, in 7 of 9 nerve terminal shape showed no significant cor- junctions studied, these measurements were larger at relation with transmitter release levels. Vesicle density active zones than between active zones. Figure 3C varied only slightly between 4.3 and 5.9 vesicles/ shows that for these 9 identified contralateral junc0.045 pm2 (mean values) at different junctions when tions there is a positive correlation between release all sections were considered, and between 5.2 and 9.7 per unit length and the number of mitochondria per
496
A. A. Hw
section. However, as Fig. 3D shows, the number of ~t~hon~a per section is a linear function of the cross-sectional area of the terminal and more likely related to axoplasmic volume than to release per unit length. A closer examination of Fig. 3C reinforces the conclusion that the number of mitochondria per section may not be an important determinant of release since a v?eak junction 141-2 (release per unit length 0.92) had the same average number of mitochondria as the two strong junctions 134-I and 134-3 (release per unit length 3.19 and 3.11). No difference in mitochondrial size was apparent in different terminals. Figure 3 also shows that release per unit length is correlated with the length of presynaptic membrane in unobstructed close contact with postsynaptic membrane whether expressed absolutely (Fig. 3E) or as the proportion of the terminal perimeter in close contact (Fig. 3F). In nearly every case, these last two measurements were larger at active zones than between active zones. Figure 3G shows the relation between release per unit length and another approximation of active zone length, the width of presynaptic membrane within 0.2 pm of postsynaptic membrane. This potential contact width is largest for the most strongly releasing terminals, and is greater at active zones than between active zones for all 9 terminals. In Fig. 3H, release per unit length is plotted against the fraction of the potential contact width obstructed by Schwann cell processes. A significant correlation exists only for sections between active zones. Figure 3H also illustrates the well known fact that Schwann cell wrapping of terminals occurs principally in regions between active zones. Correlations between release and active zone length
Since several of these measurements suggested that the average length of individual active zones was an important determinant of transmitter release per unit length, it was of interest to see how estimates of total active zone length in individual terminals correlated with total transmitter release from those terminals. To estimate total active zone length from our data 3 factors must be known: total terminal length, the mean length of individual active zones, and the spacing between active zones. Terminal length can be accurately determined from light microscopic camera lucida drawings. We estimated the mean length of individual active zones by multiplying mean contact width at active zones by the fraction of the apparent terminal width occupied by active zone as seen in freez+fracture views of frog terminals avaihtble in the literature (see Experimental Procedures). Active zone spacing was assumed to be a constant 1.3 active zones per pm of terminal length, a figure likewise dctctmined from data pub&shed by others2’Jb and in good agreement with our previous estimates based on synaptic fold spacing. I5 The results are shown in Fig. 4, which depicts the relation between total trans-
el al.
mitter release per nerve impulse and total active zone length for the 9 identified contralateral junctions. DISCUSSION
~e~at~o~hip between release and active zone sire
These results, together with those previously published,23 lead to the conclusion that the amount of presynaptic active zone is an important ultrastructural determinant of the level of transmitter release from a motor nerve terminal, The quantitative relationship apparently has a steep slope, e.g. comparing the 4 weakest junctions with the 2 strongest, a S-fold difference in total active zone length correlates with an approximately 1%fold increase in total release (Fig. 4). This implies that the relation between active zone size and release is non-linear and/or that there are correlated physiological differences, not readily discerned in electron micrographs. Indeed, it has been shown that there are differences in resting and impulse-evoked Ca2+ influx correlated with release per unit length in frog cutaneous pectoris junctions. 3pHowever, for these 9 physiologically identified sartorius jtmctions, the amount of transmitter release per unit length of terminal was proportional to the size of individual active zones, and total transmitter output was well correlated with estimates of the total amount of active zone in all the terminal branches. Others have also found correlations between transmitter release levels and active zones in normal,25 degenera~ng,~ and regenerating3u frog junctions, in human junctions affected by Lambert-Eaton myasthenic syndrome,12 in crustacean neuromuscular junctions,‘3.‘4.35*36,49and in Aplysia3 and squid” synapses. Nerve terminals on fast and slow muscle fibers in the frog, which are known to differ in transmitter release, show corresponding differences in total active zone size.53 One obvious way for terminals to increase the total amount of active zone is to grow larger. In fact, a positive correlation between terminal size and transmitter release has been noted by a number of authors.2*6*1’*32 In our results the two junctions with the highest release levels are indeed the two largest junctions (134-1, 134-3 in Fig. 1). However, the overall correlation is quite poor, since two other terminals with nearly the same size (134-5, 141-6) release much less transmitter. Other studies show that when subpopulations of muscle fibers with the same input resistance are considered, there is actually an inverse relationship between terminal size and release.S7*” Clearly there are other regulatory mechanisms that also govern release. Our data do not, of course, exclude the possibility that a change in the spacing of active zones or the density of intramembrane particles within active zones also contribute to the increase in release. Neither of these parameters can be determined from our cross-sectional views. A recent freeze-fracture study showed that active zone spacing and the num-
497
Ultrastruetural and physiological plasticity length vary considerably at different junctions within the same muscle. 40 Similar ultrastructural studies comparing junctions in two muscles whose terminals differ greatly in release’9 confirm that active zone
1
30
her of active zone particles per unit terminal
n ii
0
I 1000
0
TOTAL
0
2 4 5 6 10 + YITOCHOWDRIA
:ON:ACT2Wl:TH
0
:,m,
1 2 3 4 POTENTIAL CONTACT WIDTH (pm)
clto**
iECT. k*
&I?)
0 0.1 0.2 0.a 0.4 0.5 CONTACT WIDTH/PCRlMETER
0
0.1 0.2 0.3 0.4 0.5 SCI /POTENTIAL CONTACT WIDTH
Fig. 3. Scatter plots relating aspects of nerve terminal ultrastructure to transmitter release per unit nerve terminal length (M/L). Each closed symbol is the mean value for a given endplate for all cross-sections passing through active zones. Each horizontally adjacent open symbol is the mean value for the same endplate for cross-sections passing between active zones. Data at and between active zones were combined and averaged for each endplate to obtain the single set of points in 3D. Circles are data from endplates
in muscle no. 134, triangles are data from muscle 141. For each graph the linear correlation eoetlkient is 8iven below for closed points (r,) and open points (rO).Significance levels for 7 degrees of freedom are r = 0.67 (P < 0.05) and r = 0.80 (P < 0.01). 3A: M/L vs nerve terminal perimeter, r, = 0.70, r, = 0.81. 3B: M/L vs nerve terminal cross seetional area, r, = 0.74, r, =0.86. 3C: M/L vs number of mitochondria per section, r, = 0.75, r,, = 0.56. 3D: Numhcr of mitochondria for all sections vs cross-sectional area for all sections, r = 0.97. 3E: M/L vs contact width, or the let@ of presynaptic membrane in close unobstructed con-
tact with the postsynaptic membrane, rc = 0.89, r,, = 0.93. 3F: M/L vs the ratio of contact width to nerve tmninal perimeter, r, = 0.97, r, = 0.90.3G: M/L vs potential contact width, or the length of the terminal perimeter within $2 Am of the muscle fiber membrane, r, = 0.84, r, = 0.90.3H: M/L vs the ratio of Schwann cell interposition (SCI) to potential contact width, or the fraction of the potential contact width where actual synaptic contact is occluded by Schwann cell processes, rs = 0.53, r, = 0.55.
ACTIVE ZONE
I
2000 LENGTH
1
3000 (Mm)
Fig. 4. The relation between total quanta1 transmitter release per nerve stimulus and the estimated total length of all the active zone in each of the 9 identified contralateral sartorius endplates. Line fit by linear regression, r = 0.96, P
A, A. HERRERAet al.
498
differences in the depth of invagination of terminals into their gutters. Finally, differences in contact width may reflect differences in the extent to which Schwann cell processes are interposed between pre- and postsynaptic membranes. Studies on the neurohypophysis have revealed inverse correlations between levels of hormone release and the extent to which glial cells wrap the neurosecretory endings.42*s’*52It is not clear whether a similar correlation holds for our data. At active zones, the percentage of the potential contact width (portion of terminal perimeter membrane within 0.2 pm of the muscle membrane) obstructed by Schwann cell processes was not significantly correlated with release. There was, however, a significant correlation for sections passing between active zones (Fig. 3H). ~eiat~o~hip ameters
between release and some other par-
Our finding that the density of synaptic vesicles near the presynaptic membrane does not correlate with transmitter release IeveIs is in agreement with a previous ultrastructural study of physiologically identified frog neuromuscular junctions4’ This earlier study found that vesicle density was an extremely variable feature and showed little if any correlation with the amount of release just before fixation. In addition, nerve terminals in muscles of aged mice, which release as much or more transmitter as terminals in younger mice, contain reduced numbers of vesicles.5~“*28 On the other hand, a positive correlation has been found in some studies. Developing motor nerve terminals in the avian iris gradually become able to follow repetitive nerve activity as terminals enlarge and vesicle density increases.43 In long term habituated Aplysiu, the number of vesicles associated with active zones in the synapses involved decreases, while sensitized animals show increased vesicle numbers3 It is tempting to equate the vesicles immediately adjacent to active zones or presynaptic densities with the immediately releasable transmitter pool, but this has not been directly confirmed. Given the fact that only vesicles with the most intimate contact may be released, and given the uncertainties regarding both the rate at which new vesicles are moved into this immediately releasable pool and the effects of tissue preparation on vesicle position, it would seem prudent to be cautious in interpreting all such studies. Similar cautions may apply when one attempts to correlate mitochondrial numbers with release. Release may be affected by the Ca*+-sequestering ability of mitochondria’ but other Ca*+ buffers which are not so easily quantified from micrographs may be more important, such as smooth endoplasmic reticulum,* soluble buffers:’ other organelles,‘*,~ and active
pumping to the eel1 exterior.‘*1° Our results imply that the Ca*+-sequestering activity of mit~hon~a is less important than their role in providing adenosine 5’-triphosphate for the metabolic activity associated with release. More strongly releasing nerve terminals were generally found to contain more mitochondria. This may reflect the greater energy demand caused by high release levels or, since mitochondrial number was directly proportional to cross-sectional area (Fig. 3D), the number may simply reflect cytoplasmic volume. It should be noted, however, that in an earlier study comparing cutaneous pectoris and sartorius junctions, the more strongly releasing cutaneous pectoris terminals contained more ~t~hond~a without a difference in terminal voIume.” Mechanisms of contralateral increase in transmitter release
The present results also allow some speculation on the mechanism by which unilateral denervation of a sartorius muscle causes an increase in transmitter release from nerve terminals in the contralateral mu&e. Our sample was chosen such that 2 of the 9 identified contralateral terminals released more transmitter per unit terminal length than any control terminal we have ever encountered. Thus it is highly likely that these 2 junctions, 134-l and 134-3, would best demonstrate the morphological basis of long term plasticity in transmitter release. Examination of Figs 3 and 4 shows that these 2 strongly releasing terminals uniquely differed from the remaining 7 weaker terminals in 4 ways: the actual width of close synaptic contact, the potential contact width, the fraction of the terminal perimeter in synaptic contact, and total active zone length. It is unlikely that enhanced release could be explained by sprouting and new synapse formation, of the type described by RotshenkeP and Ring et aI.* follo~ngcontr~ateral denervation of the frog cutaneous pectoris and sartorius muscles, respectively. A recent study found no evidence that contralateral denervation enhances the sprouting that normally occurs in the sartorius as part of the ongoing process of synaptic remodelling.24 In addition, there was no significant correlation in the present results between release and terminal length. We conclude that the active zone in adult motor nerve terminafs may be a plastic structure, and that long term changes in transmitter reiease from these terminals can be achieved by increasing the length of individual active zones. Aeknowledaements-We thank L. Batter, H. Kabe, F. Knight ana D. Scott for technical a&taace This work was supported by USPHS grants MB6232 to A.D.G. and NS18186 to A.A.H. and by the Muscular Dystrophy Association.
REFERENCES 1, Alnaes E. and Rahamimoff
R. (1975) On the role of mitochondria terminals. J. Physfoi., Land. 248, 285-306.
in transmitter
release from frog motor nerve
Ultrastructural
and physiological plasticity
499
2. Angaut-Petit D. and Mallart A. (1979) Dual innervation of endplate sites and its consequences for newmmh transmission in muscles of adult Xenopus haeuis. J. Physiol., Lond. 289, 203-218. 3. Bailey C. H. and Chen M. (1983) Morphological basis of long-term habituation and sensitization in Aplysia. Science, N.Y. 2243, 91-93. 4. Baker P. F. and Schlaepfer W. W. (1978) Uptake and binding of calcium by axoplasm isolated from giant axons of Loligo and Myxicola. J. Physiol., Land. 276, 102125. 5. Banker B. Q., Kelly S. S. and N. Robbins (1983) Neuromuscular transmission and correlative morphology in young and old mice. J. Physiol., Land. 339, 355-375. 6. Bennett M. R. and Raftos J. (1977) The formation and regression of synapses during the re-innervation of axolotl striated muscles. J. Physiol., Lond. 265, 261-295. 7. Blaustein M. P. and Hodgkin A. L. (1969) The effect of cyanide on the efflux of calcium from squid axons. J. Physiol., Lond. 280, 497-527. 8. Blaustein M. P., Ratzlaff R. W. and Schweitzer E. S. (1978) Calcium buffering in presynaptic nerve texminals. II-Kinetic properties of the nonmitochondrial Ca sequestration mechanism. J. gen. Physiol. 72, 43-66. 9. Couteaux R. and Pecot-Dechavassine M. (1970) Vesicules synaptiques et poches au niveau de les “zones actives” de la jonction neuromusculaire. C.r. hebd. S&nc. Acad. Sci., Paris 271, 2346-2349.
10. Dip010 R. and Beauge L. (1979) Physiological role of ATP-driven calcium pump in squid axon. Narure, Load. 278, 271-273. 11. Fahim M. A. and Robbins N. (1982) Ultrastructural studies of young and old mouse neuromuscular junctions. J. Neurocyfol. 11, 641-656. 12. Fukunaga H., Engel A. G., Gsame M. and Lambert E. H. (1982) Paucity and disorganization of presynaptic membrane active zones in the Lambert-Eaton myasthenic syndrome. Muscle Neroe 5,686-697. 13. Govind C. K. and Chiang R. G. (1979) Correlation between presynaptic dense bodies and transmitter output at lobster neuromuscular terminals by serial section eleotron microscopy. Brain Res. 161, 377-388. 14. Govind C. K. and Meiss D. E. (1979) Quantitative comparison of low and high-output neuromuscular synapses from a motomeuron of the lobster (Homarus americanus). Cell T&s. Res. M&455-463. 15. Grinnell A. D. and Herrera A. A. (1980) Physiological regulation of synaptic effectiveness at frog neuromuscular junctions. J. Physiol., Lond. 307, 301-317. 16. Grinnell A. D. and Herrera A. A. (1981) Specificity and plasticity of neuromuscular connections: long term regulation of motoneuron function. Prog. Neurobiol. 17, 203-282. 17. Harris J. B. and Ribchester R. R. (1979) The relationship between endplate size and transmitter release in normal and dystrophic muscles of the mouse. J. Physiol., Lond. 296, 245-265. 18. Henkart M. P., Reese T. S. and Brinley F. J. (1978) Endoplasmic reticulum sequesters calcium in the squid giant axon. Science, N. Y. 202, 1300-l 302. 19. Herrera A. A., Banner L. R. and Mason P. (1985) Differences in synaptic efficacy at frog neuromuscular junctions. Sot. Neurosci. Absr 11, 302. 20. Herrera A. A. and Grinnell A. D. (1980) Transmitter release from frog motor nerve terminals depends on motor unit size. Nature, Lond. 287, 649-651. 21. Herrera A. A. and Grinnell A. D. (1981) Contralateral denervation causes enhanced transmitter release from frog motor nerve terminals. Nature, Lond. 291, 495-497. 22. Herrera A. A. and Grinnell A. D. (1982) Synaptic apposition, glial obstruction, and synaptic strength at frog neuromuscular junctions. Sot. Neurosci. Abs 8, 492. 23. Herrera A. A., Grinnell A. D. and Wolowske B. (1985) Ultrastructural correlates of naturally occurring differences in transmitter release efficacy in frog motor nerve tetminals. J. Neurocytol. 14, 193-202. 24. Herrera A. A. and Scott D. R. (1985) Motor axon sprouting in frog sartorius muscles is not altered by contralateral axotomy. J. Neurocyiol. 14, 145-156. 25. Heuser J. E., Reese T. S., Dennis M. J., Jan Y., Jan L. and Evans L. (1979) Synaptic vesicle exocytosis captured by quick freezing and correlated with quanta1 transmitter release. J. Cell Biol. 81, 275-300. 26. Heuser J. E., Reese T. S. and Landis D. M. D. (1974) Functional changes in frog neuromuscular junctions studied with freeze-fracture. J. Neurocytol. 3, 109-131. 27. Kamovsky M. J. (1964) The localization of cholinesterase activity in rat cardiac muscle by electron microscopy. J. Cell Biol. 23, 217-232. 28. Kelly S. S. and Robbins N. (1983) Progression of age changes in synaptic transmission at mouse neuromuscular junctions. J. Physiol., Lond. 343, 375-383. 29. Ko C.-P. (1981) Electrophysiological and freeze-fracture studies of changes following denervation at frog neuromuscular junctions. J. Physiol., Lond. 321, 627-639. 30. Ko C.-P.-(1984) Regeneration of the active zone at the frog neuromuscular junction. J. Cell Viol. 9?I, 16854695.
31. Ko C-P., Probst J. W. and Herrera A. A. (1985) A comparison of active zone ultrastructure in two muscles with markedly different synaptic efficacy. Sot. Neurosci. Abst 11, 302. 32. Kuno M., Turkanis S. A. and Weakly J. N. (1971) Correlation between nerve terminal size and transmitter release at the neuromuscular junction of the frog. J. Physiol., Lond. 213, 545-556. 33. Letinsky M. S. and DeCino P. (1980) Histological staining of pre- and postsynaptic components of amphibian neuromuscular junctions. J. Neurocytol. 9, 303-320. 34. Martin A. R. (1955) A further study of the statistical composition of the end-plate potential. J. Physiol., Lond. 130, 114-122. 35. Meiss D. E. and Govind C. K. (1979) Multiterminal innervation: non-uniform density along single lobster muscle fibers. Brain Res. 160, 163-169. 36. Meiss D. E. and Govind C. K. (1980) Heterogeneity of existing synapses at the ends of single muscle fibers in the lobster, Homarus americanus. J. Neurobiol. 11, 381-395. 37. Nude11 B. M. and Grinnell A. D. (1982) Inverse relationship between transmitter release and terminal length in synapses on frog muscle fibers of uniform input resistance. J. Neurosci. 2, 216224. 38. Nude11 B. M. and Grinnell A. D. (1983) Regulation of synaptic position, size, and strength in anuran skeletal muscle. J. Neurosci. 3, 161-176.
500
A. A. HERRERAet al.
39. Pawson P. A. and Grinnell A. D. (1984) Posttetanic potentiation in strong and weak neuromuscular junctions: physiological differences caused by a differential Ca2+ influx. Brain Res. 323, 31 t-315. 40. Pawaon P. A. and Grinnell A. D. (1985) Physiological and possible structural bases for the dif%rences in synaptic efficacy at the frog neuromuscular junction. Can. I Physiol. Pharmac. 63, Axxiii. 41. Peper K., Dreyer F., Sandri C., Akert K. and Moor H. (1974) Structure and ultrastructure of frog motor endplate. Cell Tissue Res. 149, 437-455. 42. Perfmutter L. S., Hatton G. 1. and Tweedle C. D. (1984) Plasticity in the in c~itroueuroh~ophysis: effects of osmotic changes on pituicytes. Neuroscience 12, 503-S 11. 43. Piiar G., Tuttle J. and Vaca K. (1981) Functional maturation of motor nerve terminals in the avian iris: ultra-structure, transmitter metabolism and synaptic reliability. J. Physiol., Lond. 321, 175-193. 44. Politoff A. L., Rose S. and Pappas G. D. (1974) Calcium binding sites of synaptic vesicles of frog sartorius neuromuscular junction. J. Cell Biul. 61, 81%823. 45. Pumplin D. W., Reese T. S. and Llinh R. (I 981) Are presynaptic membrane particles the calcium channels? Pruc. mtn. Acad. Sci., U.S.A. 78, 7210-7213. 46. Ring G., Reichert F. and Rotshenker S. (1983) Sprouting in intact sartorius muscles of the frog following contralateral axotomy. Brain Res. 260, 313-316. 47. Rose S. J., Pappas G. D. and Kriebel M. E. (1978) The fine structure of identified frog ne~omu~ular junctions in relation to synaptic activity. Brain Res. 144,213-239. 48. Rotshenker S. (1979) Synapse formation in intact innervated cutaneous pectoris muscles of the frog fohowing denervation of the opposite-muscle. J. Physiol., Lond. 292, 535-547. 49. Sherman R. G. and Atwood H. L. (1972) Correlated electroohvsioloaical and ultrastructural studies of a crustacean .motor unit. J. gem Physiof. 59, S&615. 50. Thompson R. F., Rerger 1. W. and Madden J. (1983) Cellular processes of learning and memory in the mammalian CNS. A. Rev. Neurosci. 6, 447491. 51. Tweedle C. D. (1983) Ultrastructural manifestations of increased hormone release in neurohypophysis. Prog. Bruin Res. 60, 259-272. 52. Tweedle C. D. and Hatton G. I. (1982) Magnocellular neuropeptidergic terminals in ne~ohy~physis: rapid gbal release of enclosed axons during pa~u~~on. Brain Res. Bull. 8, 205-209. 53. Verma V. and Reese T. S. (1984) Structure and distribution of neuromuscular junctions on slow muscle fibers in the frog. Neuroscience 12, 647-662. (Accepted 30 May 1985)