- Email: [email protected]
Fatigue and recovery of transmission at the Mauthner fiber-giant fiber synapse of the hatchetfish
Brain Research, 98 (1975) 229-242 © Elsevier Scientific Publishing Company, Amsterdam - Printed in The Netherlands
229
FATIGUE AND RECOVERY OF TRANSMISSION AT THE MAUTHNER FIBER-GIANT FIBER SYNAPSE OF THE HATCHETFISH
S. M. H1GHSTEIN AND M. V. L. BENNETT
Division of Cellular Neurobiology, Department of Neuroscience, Albert Einstein College of Medicine, Bronx, N. Y. 10461 (U.S.A,) (Accepted May 7th, 1975)
SUMMARY
When the Mauthner fiber-giant fiber synapse of the hatchetfish is activated at gradually increasing frequencies, postsynaptic potentials (PSPs) in the giant fiber become progressively smaller, but complete failures of transmission are not observed even when PSP size is as small or smaller than miniature PSPs (mPSPs) simultaneously recorded. On the assumption of a Poisson distribution of amplitudes, calculations from the absence of failures and from variance suggest that quantum number remains at least as high as 5-10 and that quantal size is greatly reduced. During tetanic stimulation the frequency of mPSPs first increases and then decreases again, sometimes to a very low frequency. However, mPSP amplitude is reduced by no more than about 50 ~, which indicates that quanta giving rise to mPSPs come from a different population of vesicles than those comprising evoked PSPs. During rest following a tetanus, calculated quantal size in evoked PSPs recovers within several hundred milliseconds to mPSP size simultaneouly recorded. Most of this recovery time represents time for filling, since vesicles can be supplied at much higher rates during tetanic stimulation. After one second rest PSP amplitude exceeds threshold but recovery for later PSPs in a short train requires many minutes. The slowness of this recovery is consistent with the morphological demonstration of slow recovery of the vesicle population after depletion. These data are interpreted in terms of vesicle release, depletion and membrane recycling. Following depletion new vesicles are released after only partial filling which accounts for small quanta. Very small mPSPs are not seen because filling time is short compared to time for release as mPSPs. Since quantal size can be gradually reduced, release can interrupt filling, and filling and release sites are likely to be the same. The data in combination with the morphological observations support the hypothesis of vesicular release of transmitter and provide new evidence as to rates and sites for filling of vesicles.
230 1N T R O D U C T I O N
The synapse between Mauthner and giant fibers of the hatchetfish is unique among chemically transmitting synapses in vertebrates in that both pre- and postsynaptic elements can be simultaneously penetrated with microelectrodes close to the synaptic regionL Studies of the ultrastructure is and physiology'~ of this synapse demonstrate that the Mauthner fiber is presynaptic and the giant fiber is postsynaptic. An action potential in either Mauthner fiber is followed by an excitatory postsynaptic potential (PSP) in each giant fiber after a delay of about 0.4 msec. At repetition rates of less than a few per second each PSP is suprathreshold for spike initiation in the giant fiber and impulse transmission is one to one. If the Mauthner fiber is repetitively stimulated at a gradually increasing frequency, the amplitude of the PSP decreases and impulse transmission fails. As the frequency of stimulation is increased still further, the PSP can be reduced to an amplitude which is comparable to that of the spontaneous miniature PSPs, but there are no complete failures of the PSPs'~,I:L This phenomenon and subsequent recovery of PSP amplitude during rest are the subjects of the present paper. Our new data in conjunction with morphological findings presented in the accompanying paper is indicate that the total vesicular store of transmitter is quite rapidly depleted by stimulation of the Mauthner fibers and that later stimuli release the contents of partially filled vesicles. The resulting decrease of quantal size while quantum number remains high accounts for the small size of the PSP and lack of failures. Furthermore, from observations of recovery of PSP amplitude after a prolonged tetanus one can obtain an estimate of the time required to fill a synaptic vesicle. METHODS
Hatchetfish, 3-5 cm in length, were paralyzed with an intramuscular injection of o-tubocurarine chloride, pinned in their normal position to a block o f Takiwax (Cenco) and perfused through the mouth with oxygenated Ringer. Composition of the Ringer was 167 m M NaCI, 5 m M KCI, 3 m M CaCI2; 1.5 m M MgCI2, 4 m M Tris at pH 7.4. Mauthner fibers were stimulated with bipolar electrodes placed on the caudal spinal cord or stimulated directly through an intracellular microelectrode in a bridge circuit. The posterior cerebellum, medulla and anterior spinal cord were exposed from above. A silver wire was placed on the lateral line lobe to record the overall Mauthner fiber and giant fiber activity with respect to an indifferent electrode on the lateral surface of the skull. This recording served as a monitor of the adequacy of spinal stimulation and the general level of transmission. The glass microelectrodes employed were filled with 3 M KC1 and were 5-10 M ~ in resistance. They were inserted through the floor of the fourth ventricle by way of its posterior opening to penetrate Mauthner fibers and giant fibers just below the ventricular surface. The ventricle could be perfused with test solutions; as the synapses lie only 40-60 #m below the ependyma, appropriate agents could reach the synapse and affect transmission. Intracellular potentials were recorded using conventional circuitry. Filmed
231
A:! • i' !
B!
2msec
i
It,-.-"
150nA '
Fig. 1. Synaptic relation between Mauthner and giant fibers. A: a directly evoked spike in a Mauthner fiber (upper trace, current on lower trace) is followed by a spike in a giant fiber (middle trace). B: paired stimulation of the Mauthner fiber with a 12-msec interval between stimuli. The PSP is greatly reduced and fails to initiate an impulse in the giant fiber.
results were analyzed with the aid of Hewlett Packard digitizer and model 9830 calculator. RESULTS
Evoked PSPs and spontaneous mPSPs Direct stimulation of either Mauthner fiber is followed by an action potential in each giant fiber (Fig. 1A). This action potential is initiated by a PSP which can be revealed by paired stimulation (Fig. 1B). The PSP shows marked depression following even a single stimulus. It requires up to 0.7 sec to recover its maximum rate of rise, although it reaches threshold for spike initiation earlier. The depression is not accompanied by any change in the presynaptic impulse. Spontaneous mPSPs are generally recorded in the giant fibers (Figs. 2A, 4A and 5). Origin of the mPSPs from the Mauthner fiber is indicated by the fact that subthreshold depolarization of the Mauthner fiber increases miniature frequencyl,L Another indication that mPSPs are of Mauthner fiber origin is that tetanic stimulation of the Mauthner fibers increases miniature frequency (see below). In addition, topical curare blocks transmission from Mauthner fiber to giant fiber and also blocks miniatures 2z. As shown in Fig. 2, a few microliters of 10-4 g/1 of o-tubocurarine applied to the floor of the IVth ventricle blocks mPSPs within 60 sec. The evoked PSP is greatly reduced in size (Fig. 2Ba) while the giant fiber action potential elicited by depolarizing pulses remains unchanged in amplitude as does the giant fiber resting potential. After washing with Ringer the PSPs recover to near their control values. Finally morphological investigation has failed to reveal any synapses onto the giant fibers in the medulla other than those arising from the Mauthner fibers 17. Although there are synapses on the somata of the giant fibers, these are probably too distant to produce detectable mPSPs, and are at least in part non-cholinergic2~. Effects of repetitive stimulation on PSPs As the frequency of Mauthner fiber stimulation is gradually increased, the PSPs become steadily reduced in amplitude (Fig. 3). In the extreme the PSPs can be
232 A1
B}
.L.._J,,_~
~d.,
~1 . . . .
L~t .... I --
.!.
20 mse¢
!
l
5 0 mV
2 mV
:
Fig. 2. Block of miniature PSPs and reduction in PSP amplitude caused by topical application of curare. All traces are intracellularly recorded from a giant fiber. A: prior to curare application. AI: spontaneous miniature PSPs (3 records at high gain). As: a synaptically evoked spike following activation of the Mauthner fibers by spinal stimulation. Aa: a spike and two PSPs evoked by a train of 3 spinal stimuli. B: records one minute after topical application to the IVth ventricle of a few microliters of curare 10-4 g/1. BI: spontaneous miniatures are absent, recording as in Aa. B2: a directly evoked giant fiber spike is similar in amplitude to the synaptically evoked spike in A2. B~: 3 PSPs evoked by spinal stimulation are markedly reduced in amplitude compared to those in A3. High and low gain records are shown in Aa and B3.
m a d e to d i s a p p e a r into the noise level. Even when PSP a m p l i t u d e is as small or smaller t h a n that o f the miniatures recorded simultaneously, there are no c o m p l e t e failures o f PSPs (Fig. 3E-I). A t the higher frequencies when the P S P a m p l i t u d e is reduced, its time course is significantly longer lasting t h a n the miniatures riding u p o n it (Figs. 3 E - I a n d 4B a n d C). I f the size o f the miniatures c o m p r i s i n g small PSPs is the same as t h a t o f the s p o n t a n e o u s miniatures, there should be no (non-zero) PSPs smaller t h a n a miniature, a n d failures should be c o m m o n for a m p l i t u d e s where the m e a n n u m b e r o f miniatures c o n t r i b u t i n g to the P S P is small. W e believe t h a t q u a n t u m n u m b e r in the small PSPs is high, b u t t h a t q u a n t u m size has become m u c h smaller. The reduced q u a n t u m size can be a c c o u n t e d for by depletion o f the vesicular store o f t r a n s m i t t e r a n d subsequent release o f the contents o f incompletely filled vesicles. I f the a m p l i t u d e s o f the PSPs are Poisson distributed, which we will consider further in the discussion, we can estimate q u a n t u m n u m b e r f r o m variance a n d mean a m p l i t u d e o f the PSP 16. In the PSP o f Fig. 4C c o m p u t e d q u a n t u m n u m b e r is 13.6. M o r e o v e r , the absence o f failures provides an estimate o f a lower limit on q u a n t u m number. O n the a s s u m p t i o n o f a Poisson process, q u a n t u m n u m b e r s o f 5, 7.5 a n d 10 w o u l d result in a m e a n o f one failure p e r 6 sec, 72 sec a n d 15 min o f stimulation at 25/sec. W e conclude from m a n y seconds o f watching the oscilloscope t h a t a lower
233
A I ~ B
C
.
IOOmV D
~ ~
2msec ,o,,~c
": i :ec -.. ~
II
--
,,tk
E
33j~ec
H
ll~sec
F
~o,~
J !1mY imsec
143/seC
.... c
u-
.
.
.
.. . .. . . .. . .. .
I p
Fig. 3. Reduction in evoked PSPs to the size of miniature PSPs by tetanic stimulation. A: spinal stimulation at a low rate (1/sec) showing the giant fiber spike and the extracellular field of the Mauthner and giant fibers recorded on the lateral line lobe: B: after some seconds of stimulation at 10/sec the rate of rise of the PSP is reduced, but initiation of impulses in the giant fiber persists; recorded as in A. C: at 20/sec stimulation, PSP amplitude is reduced below threshold. The upper trace is a high gain intracellular recording. The middle trace is a just extracellular recording obtained subsequently at the same gain and stimulation frequency. D-I: progressively increasing frequencies of stimulation (display as in C, faster sweep in E and F). The PSP becomes progressively and gradually smaller in amplitude until it is smaller than the mPSPs simultaneously recorded. The evoked PSPs are longer in time course than the mPSPs. The mPSPs are generated at any time during the stimulation cycle, and show little preferential occurrence during the evoked PSPs. limit for q u a n t u m n u m b e r is between 5 and 10 in experiments such as those illustrated in Figs. 3 and 4.
Effects of tetanic stimulation on miniatures The frequency o f miniatures increases markedly just after the onset o f a tetanus (Fig. 4B and C, top row and Fig. 6A), and then decreases again to a value near or below the resting level (Fig. 4B and C, b o t t o m row). In occasional experiments miniature frequency becomes very small during a tetanus. The mean amplitudes o f the mPSPs decrease somewhat, but even when their frequency is markedly reduced they are not greatly reduced in size. In the experiment o f Fig. 4 during the tetanus there was a roughly 5 0 ~ decline in m P S P amplitude from a mean o f 0.73 ± 0.25 to a mean o f 0.38 ~ 0.14 mV (Fig. 5). In no experiment were the mPSPs reduced in amplitude by more than this amount. The observed reduction in m P S P amplitude is certainly not adequate to explain the reduction in PSP amplitude given the q u a n t u m n u m b e r implied by the calculations from the variance and absence o f failures that are given above. In the experiment o f Fig. 4 the mean m P S P size increases slightly to 0.51 ~ 0.17 mV within a few seconds after the tetanus, but time for complete recovery to control
234
A
,
•'
C
D
20~lec
I mV
2 msec
Fig. 4. Changes in mPSP frequency and amplitude during tetanic stimulation. Three records in each column. A : spontaneous mPSPs prior to tetanic stimulation (recorded at low sweep speed). B: mPSPs and evoked PSPs during spinal stimulation at 25/sec. PSP amplitude is reduced to the level of the mPSPs superimposed on it. The top traces were recorded near the beginning of the tetanus and there is an increase in the frequency of mPSPs. After about one minute and several minutes (next two records below) the mPSP frequency is progressively reduced, but evoked PSP amplitude is unchanged. Several superimposed sweeps in each record. C: as in B, but single sweeps. D: the amplitude of mPSPs is somewhat reduced immediately after cessation of the tetanus.
amplitude was not measured. In no experiment did we observe an increase in the number of small mPSPs during recovery from a tetanus.
Short-term recovery of PSPs from tetanic stimulation When a tetanus is terminated, PSP amplitude recovers rapidly. Generally after a tetanus a one second rest allows recovery of the PSP to an amplitude at which it initiates an impulse. To explore the recovery process, trains of stimuli were given at varying rates that were selected to allow partial recovery between trains. The PSP
t/) a. t/) o. E --I
.4
o .3
O
CONTROL
o
DURING TETANUS
® POST TETANUS
o
I--
•
®
14.
O .2 o®
Z
o
o•® ®
U
.1
•
o
o
oe ®
•
n~
LL •
®
•
o
o0
I
I
I
I
a
t
.2
.4
.6
.8
1
t2
oo
oo
oo i
a
I
n
1.4 1.6 1.8 2
mPSP AMPLITUDE mV Fig. 5. Amplitude distributions of the miniature PSPs before, during and after a tetanus. Same tetanus as in Fig. 4. The ordinate is the fraction of the total number of miniature PSPs and the abscissa is the mPSP amplitude. The means and standard deviations in the 3 cases are 0.73 ± 0.25 mV, 0.38 ± 0.14 mV and 0.51 ~ 0.17 mV, respectively.
235
A
l
l
I
t
1
2my 2 0 reset
.5
mV
msec
~"
~
.2
Z
_o
i~.---
~' '
0
iii
PSP AMPLITUDE mV
Fig. 6. Recovery of PSPs between trains of stimuli. A: PSPs and mPSPs during a train of 15 stimuli at 25-msec intervals. The train was repeated at l/sec until a steady level of responding was obtained. Continuous record from upper to lower trace, mPSP frequency increases during the early part of the train. B1-3: the first PSP in the train at faster sweep speed. C: amplitude distributions of the first PSP and of miniature PSPs during repetition of the train at 1/sec. Mean mPSP amplitude is 0.38 mV and at this interval between trains quantal size calculated from variance is 0.32 mV. Noise level indicated by shading along the ordinate.
evoked by the first stimulus in the train was then studied with respect to amplitude variations in order to make inferences about quantal size. Such an experiment is illustrated in Fig. 6. When the train was repeated once per second (725 msec between trains), the amplitude of the first PSP in the train showed considerable variability. On the assumption that the amplitudes are distributed according to Poisson statistics, we calculated quantal size and content from the coefficient of variation and mean amplitude of the PSP 16. The calculated quantal size was similar to the size of the spontaneous mPSPs observed simultaneously. This agreement suggests that Poisson statistics were applicable and that at this interval between stimuli the quantal size had largely recovered. When 400 msec separated successive trains, PSP amplitude decreased, but the size of the spontaneous mPSPs remained unchanged. The quantal size for the PSP calculated as before was now half the mPSP amplitude. In two other similar experiments calculated quantal size equalled the size of the spontaneous mPSPs with 200 msec recovery time. Many other experiments gave qualitatively similar results; PSP amplitude always recovered to threshold in a second or less, although PSP amplitude had been greatly reduced during a preceding tetanus. Determined from the variance of recovering PSPs 0.2-0.4 msec seems to be the minimum recovery time for the depleted preparation to generate PSPs consisting of
236 lO 9 E
.8
t~
.7
~)
.6
~'-- .5
.2 9--15
A
TIME (sets)
.o c5
0 OCt
~o
oo o
E
6Z
o
o
o_
o
•
••
o •
o
le
ee
.
.
.
.
.
.
.
.
.
j
1
,
i
i
,
i
i
5
i
i
t
j
,
i
I
i
h
9 ' 1o','1
f
1,
TIME (sees) Fig. 7. Effects of a single spinal stimulus on amplitude and frequency of mPSPs in a giant fiber. The upper graph shows the time course of amplitude changes for a preparation that had been stimulated extensively (filled circles) and for a preparation that had been stimulated very little (open circles). Mean and standard deviations were computed for 1 sec intervals. In the two examples the amplitudes are depressed by about 25 % with a significance at 0.001 level for 3 sec (open circles) and 4 sec (closed circles). The lower graph shows the frequency changes in the same examples measured as mPSPs per half sec. The rate of occurrence was depressed in the extensively stimulated fiber (filled circles) and augmented in the relatively fresh preparation (open circles).
'normal size' quanta of transmitter. Since an appreciable number o f small quanta can be released with much shorter interstimulus intervals, most of this period should represent time for filling the vesicles.
Long-term effects of tetanic stimulation T h e m o r p h o l o g i c a l d a t a 18 i n d i c a t e that over one hour is required for full recovery of the vesicle population. Although after a tetanus the response to low frequency stimulation (l/sec) recovers within a second to the level required to initiate impulses, there is a much more persistent decrement to prolonged trains o f stimuli. Following a train that greatly reduces PSP amplitude and a brief rest that is long enough to allow quantum size to recover fully, responses to a second train decrement
237 again within a few seconds, which is more much rapid than during the first train. To investigate the time course of recovery after a prolonged train, experiments were carried out using a train of three stimuli separated by 15-20 msec. The short train was repeated once per second and mean amplitudes for the 2nd and 3rd PSPs in the train were obtained. The first PSP was suprathreshold under these conditions. Then the train was repeated at 10/sec for 10 min which reduced PSP amplitude to a low level and presumably depleted the synapse of vesicles18. Recovery was tested by applying the train 100 times at one/sec after 15 min, 30 min, etc. In three experiments recovery of the 2nd and 3rd PSPs was complete or nearly so after 15 min, 45 min, and 1 h. At all rest intervals the 1st PSP initiated an impulse and its amplitude was not determined. In several other experiments partial recovery was observed over 15-30 min, but the penetration was lost before complete recovery was obtained.
Loss of large miniature potentials after single stimuli Even a single stimulus has a small effect on miniature amplitude. Fig. 7 graphs the results of two experiments in which a single stimulus to the Mauthner fiber was observed to transiently decrease the amplitude of the spontaneous mPSPs. Open circles illustrate the case of an 'unstimulated' preparation which had been stimulated briefly at 1/sec while penetrating the giant fiber. A single Mauthner fiber stimulus increased the frequency of mPSPs for several seconds, but decreased the amplitude of these potentials. The filled circles illustrate a second preparation which had been extensively stimulated prior to the single shock test. In this case both the mPSP frequency and amplitude transiently declined. A single train of 2 or 3 stimuli to the Mauthner fiber produced a similar transient reduction in the mPSP amplitude. This result suggests that if the population of vesicles is somewhat variable in terms of contents of transmitter, the fuller ones are preferentially released, perhaps because they are the most recently filled, and filling and release sites are close together. DISCUSSION
A primary goal of the present study was to explain the small size of PSPs obtained by repetitive stimulation at gradually increasing frequencies. The evoked PSPs become of a size similar to or smaller than that of the miniature PSPs simultaneously observed, without there being the failures that would be predicted from simple application of the quantal hypothesis of transmitter release. We can put forward 3 hypotheses consistent with the quantal release of transmitter from vesicles to explain the absence of failures when PSP amplitude becomes small. (1) Quantum number is high, but the number of molecules of transmitter per quantum (or vesicle) is reduced (the partial filling hypothesis). (2) Quantum number is high and the quanta involve the normal amount of transmitter, but the postsynaptic effect is reduced by desensitization (the desensitization hypothesis). (3) Quantum number is small, but the probability of release is high. These hypotheses will be considered in order. The evidence favors the partial filling hypothesis and that partially filled vesicles arise through membrane recycling. The preceding paper demonstrates depletion of
238 vesicles by tetani that reduce PSP amplitudes to low levels and indicates that new vesicles are formed from intermediate membranous compartments 1~. If the contents of these new vesicles were released prior to complete filling with transmitter, quantal size would be reduced. In the depleted synapse vesicles are undergoing exocytosis as fast as they are being made, and it is reasonable that they have time for only partial filling. A possible objection to this explanation is that the spontaneous miniature PSPs do not undergo a change in size comparable to that inferred for quanta comprising the evoked responses. The evidence indicates that the mPSPs arise at the Mauthner fiber synapse, but the synaptic vesicles that are responsible for the mPSPs often appear to be tess accessible for depletion than those comprising the evoked response. The persistence of relatively full sized mPSPs can be explained in several ways. Our initial view was that these represented ordinary full vesicles late in diffusing to the release site. This explanation seems unlikely however since these mPSPs are not preferentially released during the PSP but instead occur more or less at random during the tetanic stimulation cycle. Another possibility is that there are release sites that have their own supply of vesicles and that are only slightly affected by impulse activity. In support of this explanation electron microscopy reveals occasional normal appearing clusters of vesicles close to the surface membrane in otherwise depleted synapses is. These islands of vesicles might be the origin of the relatively full sized mPSPs that occur while quantal size is reduced in the evoked responses. The poor synchronization of the mPSPs with evoked PSPs and the long latency of onset of the tetanic increase in mPSP frequency suggest that there are different release mechanisms or separate release sites. The slow onset of the tetanic increase in frequency allows the site of mPSP generation to be more distant from the site at which the action potential acts on the presynaptic membrane. A similar situation may obtain at the neuromuscular junction, for the increase in mPSP frequency caused by tetanic stimulation is slow in onset as well as independent of extracellular calcium (see ref. 14). The desensitization hypothesis would seem to be contradicted by the observation that nearly normal sized mPSPs persist even when the evoked PSP is greatly reduced. If, however, the persistent mPSPs were generated at different regions of the synapse from where the PSPs were generated, they would not reveal desensitization. Although the marked changes in mPSP frequency and the smaller changes in amplitude suggest that the sites for evoked release and for at least some of the mPSPs are not far apart, it is uncertain whether the same postsynaptic receptors are affected. Nevertheless the correlation between depletion of vesicles and depression of transmission leads us to favor partial filling over desensitization as the origin of small quanta. Furthermore, the effects of repeated trains are inconsistent with desensitization. I f desensitization were responsible for the reduction in quantum size, one would expect that once quantum size had recovered after a prolonged train, the time course of reduction during a second train would be as slow as that in the first. Instead the second reduction is much more rapid which is of course consistent with the morphological observations of slow recovery from depletion of vesicles 18. Initially we expected that small mPSPs would be observed during recovery from
239 a tetanus whether depletion or desensitization was occurring. However, the rapid filling of vesicles inferred from the rapid recovery of the PSP after a tetanus makes the failure to observe small mPSPs understandable. It is likely that the filling of a vesicle is so rapid that the time it spends partially full is small compared to its lifetime before release as a spontaneous miniature. Another explanation of the lack of small mPSPs is that only more or less full vesicles can be released as miniatures. This is not unreasonable, but would require an important difference between release of vesicles that was spontaneous and release that was evoked by tetanic stimulation. The hypothesis that the small PSPs consist of a small number of normal quanta released with high probability is the simplest to reject since PSPs of smoothly graded amplitudes smaller than the normal miniatures are easily obtained (Fig. 3). Moreover, the different synaptic areas contributing to the PSPs are separated by distances such that they would be expected to act independently of each other, particularly on the ipsilateral side where several distinct synapses occur between the Mauthner and giant fibers. If reduction in PSP amplitude were due to a sudden drop in the probability of release from large to small at these individual sites, a stepwise reduction in the PSP would occur, whereas a gradual drop in probability would lead to a binomial or Poisson process, neither of which fits the evidence. The recovery time of PSP amplitude appears to provide a measure of the filling time of the vesicles. The experiments involving repeated trains with recovery between them indicates that quantal size in the evoked PSP has recovered to the spontaneous mPSP size within a few tenths of a second. Certainly at this time the total vesicular store is replaced to an insignificant extent, although there are enough vesicles observed to explain the persistence of transmission during the tetanus as well as the recovery 18. Since the PSPs can be evoked at intervals less than 20 msec without failures, most of the recovery time of quantal size appears to be the time for filling the vesicles rather than the time for their formation and movement to the release site. The longer time for full recovery of PSP amplitude presumably involves increase in the number of vesicles available for release by an impulse as well as filling of more recently formed vesicles. Because quantum size appears to decrease gradually as the interval between stimuli decreases, it is likely that release can interrupt filling. This observation suggests that the filling and release sites are essentially the same. The most attractive alternative is that vesicles take up ACh from the cytoplasm as they move from formation to release site. In this case the transit time for essentially all the vesicles would during a tetanus have to be uniformly short compared to the filling time which as already noted is a few tenths of a second. The ultrastructural data suggest vesicles are formed within a micron of the presynaptic membrane and not in a particularly uniform way 18 (see also ref. 12). Thus, filling at the release site itself seems more plausible. It is likely that filled vesicles move away from the filling (and release) sites to comprise the mobilizable store of transmitter corresponding to the large pool of vesicles that are depleted by tetanic stimulation. Although the data suggest that during a tetanus the time required to supply
240 vesicles is short compared to the filling time, they do not specify whether new vesicles are supplied for each release event or each vesicle is used for a number of release events 19,24. An alternative to loss of a vesicle's identity with each exocytotic event is that a vesicle releases its transmitter by merely transient contact with the surface membrane and maintenance of its vesicular form. If it went through subsequent release events prior to complete refilling, quantal size would be reduced. The morphological observations of depletion clearly indicate that exocytosis occurs, although they do not establish that it occurs with every release event. However, we think it likely that once a vesicle fuses with the external surface that its membrane can only be reclaimed by coated vesicles 18. One may ask what specific characteristics of the hatchetfish synapse are important in the observations of reduced quantal size and of filling time. Obviously vesicles must be releasable faster than they can be made and it must be possible to deplete the mobilizable store of vesicles in a reasonable period. A factor that is particularly important in measuring filling time is the high probability of release of the immediately available store by a single impulse. The probability of release at the hatchetfish synapse approaches unity as described earlier 2, the most important evidence being that a second stimulus closely following a test impulse produces only a greatly reduced PSP (Fig. 1) and that PSP size is independent of amplitude of the Mauthner fiber impulse over a wide range. Since a single impulse releases virtually all the quanta immediately available for release, the vesicles that are released by the next stimulus have recently arrived at the release site. If the probability of release is near unity, we should justify the use of Poisson statistics to infer quantal size during vesicle filling, since these statistics apply where a small probability operates on a large population of vesicles. However, if the population is Poisson distributed and probability of release is unity, clearly the number released is Poisson distributed. The population of vesicles would vary according to Poisson statistics if vesicles occupying release sites were the relevant population, vesicles freely diffused in the terminal, and only a small fraction of available release sites were occupied. In the depleted synapse the vesicles are sufficiently sparsely distributed that they should not interfere with each other and quantum number is reduced well below that for PSPs in the unstimulated preparation; thus it is reasonable that Poisson statistics apply to the vesicles available for release. Further investigation of the statistics of release is required in the unstimulated preparation in which vesicles are quite densely packed and a larger number of vesicles are released per impulse. Another factor in the statistics for calculation of quantal size is variability of the quanta 16. Presumably vesicles arrive all during the interval between stimuli and are filled to differing degrees. Thus, a correction for increased variance of quantal amplitude should be included in the calculations, a correction that would tend to decrease calculated quantal size and increase calculate quantum number. We have not made this correction in the absence of direct measures of quantal variability. The correction would prolong the inferred time to fill vesicles to a moderate degree, but at this stage in the investigation, a change of even a factor or two would not be of any great importance.
241 The suggestion that filling and release sites are the same is supported by biochemical observations that acetylcholine released by physiological stimuli is richer in recently synthesized material than is the total storeS,6, 8,9. Also, effects of a brief rest during prolonged stimulation of the neuromuscular junction suggest that a subpopulation of vesicles ready for release is filled preferentially11. To be sure the time scale for these experiments is much longer than for those with the hatchetfish. We do not know whether the transmitter going into vesicles comes from the cytoplasm or is synthesized in the filling process. Moreover, it is possible that in the unstimulated condition vesicles are continually visiting the filling site and exchanging or 'topping up' their contents. The loss of large quanta for several seconds after a single stimulus could be explained by this mechanism. Some synapses can transmit at much higher frequencies than the hatchetfish synapse. The maximum is the nerve-electrocyte junction in the weakly electric gymnotid fish Eigenmannia which transmits indefinitely at 300-600/sec (see ref. 4). For the mechanisms of filling to be the same as in the hatchetfish, one would require a much larger number of sites for filling and release with a much lower probability of release at a particular site. Synapses that do not fatigue and at which vesicles cannot be depleted are explicable in terms of vesicle formation and filling mechanisms that can process vesicles at least as rapidly than the release mechanism. Whether the formation mechanism is faster in non-depleting synapses or whether the cellular machinery is just present in larger amount is a question that may be open to morphological investigation. At the neuromuscular junction reduction in quantal size has been observed after relatively high frequency stimulation10,15, and after prolonged low frequency stimulation that depletes the ending of transmitterT, 11. In the sympathetic ganglion, however, quantal size was not altered by stimulation to depletion 21. It is possible that in the preganglionic fibers the rate limiting step is the formation rather than the filling of vesicles, and in this tissue as well as in the neuromuscular juntion, depletion of vesicles can be produced by repetitive activationS,z0. Small mPSPs can be recorded at the neuromuscular junction after exhaustive stimulation and it has been suggested that the usual large mPSPs result from multiple release of the small components 15. The hatchetfish results suggest that partially filled vesicles could give rise to the small mPSPs, although in the hatchetfish the contents of partially filled vesicles appear to be at most rarely released spontaneously. The combined physiological and morphological data support the view that vesicles are lost in the release process and their membrane is recycled to form new vesicles. In addition one can obtain a measure of the time required to fill newly formed vesicles. The filling process appears accessible for study because new vesicles can be so rapidly released that adding transmitter is the rate limiting step. Although full recovery after depletion requires prolonged rest, transmission at low rates recovers very rapidly. The survival value is obvious, and the mechanism may be of some generality. The observations reported here allow a further elaboration on the hypotheses of quantal release and membrane recycling in specifying the time and site of addition of transmitter.
242 ACKNOWLEDGEMENTS A n u m b e r o f t h e early e x p e r i m e n t s w e r e d o n e in c o l l a b o r a t i o n w i t h M i c h a E. Spira, w h o also c o n t r i b u t e d m u c h h e l p f u l criticism. T h e w o r k was s u p p o r t e d in p a r t by N I H
Grants NB-07512 and HD-04248,
a n d g r a n t s f r o m the A l f r e d P. S l o a n F o u n d a t i o n . REFERENCES 1 AUERBACH,A. A., Spontaneous and evoked quantal transmitter release at a vertebrate control synapse, Nature (Lond.), 234 (1971) 181-183. 2 AUEREACH,A. A., AND BENNETT, M. V. L., Chemically mediated transmission at a giant fiber synapse in the central nervous system of a vertebrate, J. gen. Physiol., 53 (1969) 183-210. 3 BARKER, L. A., DOWDALL, M.J., AND WHn"rAKER, 'V. P., Choline metabolism in the cerebral cortex of guinea pigs, Biochem. J., 130 (1972) 1063-1080. 4 BENNETT, M. V. L., Electric organs. In W. S. HOAR AND D. J. RANDALL (Eds.), Fish Physiology, Vol. 5, Academic Press, New York, 1971, pp. 347~,91. 5 BIRKS, R. I., The relationship of transmitter release and storage to fine structure in a sympathetic ganglion, J. Neurocytol., 3 (1974) 133-160. 6 BIRKS, R. I., AND FITCH, S. J. G., Storage and release of acetylcholine in a sympathetic ganglion, J. Physiol. (Lond.), 240 (1974) 125-134. 7 CECCARELLI,B., HURLBUT,W. P., AND MAURO, A., Turnover of transmitter and synaptic vesicles at the frog neuromuscular juntion, J. Cell Biol., 57 (1973) 499-524. 8 CHAKRIN,L. W., MARCHBANKS,R. M., MITCHELL,J. F., AND WHITTAKER,V. P., The origin of the acetylcholine released from the surface of the cortex, J. Neurochem., 19 (1972) 2727-2736. 9 COLLIER,B., The preferential release of newly synthesized transmitter by a sympathetic ganglion, J. Physiol. (Lond.), 205 (1969) 341-352. 10 COOKE,J. D., AND QUASTEL,D. M. J., Transmitter release by mammalian motor nerve terminals in response to focal polarization, J. Physiol. (Lond.), 228 (1973) 377-405. 11 ELMQVIST,D., AND QUASTEL,D. M. J., Presynaptic action of hemicholinium at the neuromuscular junction, J. Physiol. (Lond.), 177 (1965) 463--482. 12 HEUSER, J. E., AND REESE, T. S., Evidence for recycling of synaptic vesicle membrane during transmitter release at the frog neuromuscular junction, J. Cell Biol., 57 (1973) 315-344. 13 HIGHSTEIN,S. M., AND BENNETT, M. V. L., Fatigue at Mauthner fiber-giant fiber synapse of the hatchetfish, Fed. Proc., 32 (1973) 443 abstract. 14 HURLBUT, W. P., LONGENECKER,H. B., JR., AND MAURO, A., Effects of calcium and magnesium on the frequency of miniature end plate potentials during prolonged tetanization, J. Physiol. (Lond.), 219 (1971) 17-38. 15 KREIBEL, M. E., AND GROSS,C., Multimodal distribution of frog miniature end plate potentials in adult, denervated, and tadpole leg muscles, J. gen. Physiol., 64 (1974) 85-104. 16 MARTIN, A. R., Quantal nature of synaptic transmission, Physiol. Rev., 46 (1966) 51~i6. 17 MODEL, P. G., SPIRA, M. E., AND BENNETT,M. V. L., Synaptic inputs to the cell bodies of the giant fibers of the hatchetfish, Brain Research, 45 (1972) 288-295. 18 MODEL, P. G., HIGHSTEIN, S. M., AND BENNETT, M. V. L., Depletion of vesicles and fatigue of transmission at a vertebrate central synapse, Brain Research, 98 (1975) 209-228. 19 PFENNINGER, g . H . , AND ROVAINEN, C. M., Stimulation- and calcium-dependence of vesicle attachment sites in the presynaptic membrane; a freeze-cleave study on the lamprey spinal cord, Brain Research, 72 (1974) 1-23. 20 PYSH, J. J., AND WILEY, R. G., Synaptic vesicle depletion and recovery in cat sympathetic ganglia electrically stimulated in vivo, J. Cell BioL, 60 (1974) 365-374. 21 SACCHI,O., ANDI~RRI, V., Quantal mechanism of transmitter release during progressive depletion of the presynaptic stores at a ganglionic synapse, J. geE. Physiol., 61 (1973) 342-360. 22 SPIRA, M. E., MODEL, P. G., AND BENNETT,M. V. L., Cholinergic transmission at a central synapse, J. Cell BioL, 47 (1970) 199a-200a. 23 SPIRA, M. E., AND BENNETr, M. V. L., Penicillin induced seizure activity in the hatchetfish, Brain Research, 43 (1972) 235-241. 24 STREIT,P., AKERT, K., SANDRt, C., LIVINGSTON,R. B., AND MOOR, H., Dynamic ultrastructure of presynaptic membranes at nerve terminals in the spinal cord of rats. Anesthetized and unanesthetized preparations compared, Brain Research, 48 (1972) 11-26.
229
FATIGUE AND RECOVERY OF TRANSMISSION AT THE MAUTHNER FIBER-GIANT FIBER SYNAPSE OF THE HATCHETFISH
S. M. H1GHSTEIN AND M. V. L. BENNETT
Division of Cellular Neurobiology, Department of Neuroscience, Albert Einstein College of Medicine, Bronx, N. Y. 10461 (U.S.A,) (Accepted May 7th, 1975)
SUMMARY
When the Mauthner fiber-giant fiber synapse of the hatchetfish is activated at gradually increasing frequencies, postsynaptic potentials (PSPs) in the giant fiber become progressively smaller, but complete failures of transmission are not observed even when PSP size is as small or smaller than miniature PSPs (mPSPs) simultaneously recorded. On the assumption of a Poisson distribution of amplitudes, calculations from the absence of failures and from variance suggest that quantum number remains at least as high as 5-10 and that quantal size is greatly reduced. During tetanic stimulation the frequency of mPSPs first increases and then decreases again, sometimes to a very low frequency. However, mPSP amplitude is reduced by no more than about 50 ~, which indicates that quanta giving rise to mPSPs come from a different population of vesicles than those comprising evoked PSPs. During rest following a tetanus, calculated quantal size in evoked PSPs recovers within several hundred milliseconds to mPSP size simultaneouly recorded. Most of this recovery time represents time for filling, since vesicles can be supplied at much higher rates during tetanic stimulation. After one second rest PSP amplitude exceeds threshold but recovery for later PSPs in a short train requires many minutes. The slowness of this recovery is consistent with the morphological demonstration of slow recovery of the vesicle population after depletion. These data are interpreted in terms of vesicle release, depletion and membrane recycling. Following depletion new vesicles are released after only partial filling which accounts for small quanta. Very small mPSPs are not seen because filling time is short compared to time for release as mPSPs. Since quantal size can be gradually reduced, release can interrupt filling, and filling and release sites are likely to be the same. The data in combination with the morphological observations support the hypothesis of vesicular release of transmitter and provide new evidence as to rates and sites for filling of vesicles.
230 1N T R O D U C T I O N
The synapse between Mauthner and giant fibers of the hatchetfish is unique among chemically transmitting synapses in vertebrates in that both pre- and postsynaptic elements can be simultaneously penetrated with microelectrodes close to the synaptic regionL Studies of the ultrastructure is and physiology'~ of this synapse demonstrate that the Mauthner fiber is presynaptic and the giant fiber is postsynaptic. An action potential in either Mauthner fiber is followed by an excitatory postsynaptic potential (PSP) in each giant fiber after a delay of about 0.4 msec. At repetition rates of less than a few per second each PSP is suprathreshold for spike initiation in the giant fiber and impulse transmission is one to one. If the Mauthner fiber is repetitively stimulated at a gradually increasing frequency, the amplitude of the PSP decreases and impulse transmission fails. As the frequency of stimulation is increased still further, the PSP can be reduced to an amplitude which is comparable to that of the spontaneous miniature PSPs, but there are no complete failures of the PSPs'~,I:L This phenomenon and subsequent recovery of PSP amplitude during rest are the subjects of the present paper. Our new data in conjunction with morphological findings presented in the accompanying paper is indicate that the total vesicular store of transmitter is quite rapidly depleted by stimulation of the Mauthner fibers and that later stimuli release the contents of partially filled vesicles. The resulting decrease of quantal size while quantum number remains high accounts for the small size of the PSP and lack of failures. Furthermore, from observations of recovery of PSP amplitude after a prolonged tetanus one can obtain an estimate of the time required to fill a synaptic vesicle. METHODS
Hatchetfish, 3-5 cm in length, were paralyzed with an intramuscular injection of o-tubocurarine chloride, pinned in their normal position to a block o f Takiwax (Cenco) and perfused through the mouth with oxygenated Ringer. Composition of the Ringer was 167 m M NaCI, 5 m M KCI, 3 m M CaCI2; 1.5 m M MgCI2, 4 m M Tris at pH 7.4. Mauthner fibers were stimulated with bipolar electrodes placed on the caudal spinal cord or stimulated directly through an intracellular microelectrode in a bridge circuit. The posterior cerebellum, medulla and anterior spinal cord were exposed from above. A silver wire was placed on the lateral line lobe to record the overall Mauthner fiber and giant fiber activity with respect to an indifferent electrode on the lateral surface of the skull. This recording served as a monitor of the adequacy of spinal stimulation and the general level of transmission. The glass microelectrodes employed were filled with 3 M KC1 and were 5-10 M ~ in resistance. They were inserted through the floor of the fourth ventricle by way of its posterior opening to penetrate Mauthner fibers and giant fibers just below the ventricular surface. The ventricle could be perfused with test solutions; as the synapses lie only 40-60 #m below the ependyma, appropriate agents could reach the synapse and affect transmission. Intracellular potentials were recorded using conventional circuitry. Filmed
231
A:! • i' !
B!
2msec
i
It,-.-"
150nA '
Fig. 1. Synaptic relation between Mauthner and giant fibers. A: a directly evoked spike in a Mauthner fiber (upper trace, current on lower trace) is followed by a spike in a giant fiber (middle trace). B: paired stimulation of the Mauthner fiber with a 12-msec interval between stimuli. The PSP is greatly reduced and fails to initiate an impulse in the giant fiber.
results were analyzed with the aid of Hewlett Packard digitizer and model 9830 calculator. RESULTS
Evoked PSPs and spontaneous mPSPs Direct stimulation of either Mauthner fiber is followed by an action potential in each giant fiber (Fig. 1A). This action potential is initiated by a PSP which can be revealed by paired stimulation (Fig. 1B). The PSP shows marked depression following even a single stimulus. It requires up to 0.7 sec to recover its maximum rate of rise, although it reaches threshold for spike initiation earlier. The depression is not accompanied by any change in the presynaptic impulse. Spontaneous mPSPs are generally recorded in the giant fibers (Figs. 2A, 4A and 5). Origin of the mPSPs from the Mauthner fiber is indicated by the fact that subthreshold depolarization of the Mauthner fiber increases miniature frequencyl,L Another indication that mPSPs are of Mauthner fiber origin is that tetanic stimulation of the Mauthner fibers increases miniature frequency (see below). In addition, topical curare blocks transmission from Mauthner fiber to giant fiber and also blocks miniatures 2z. As shown in Fig. 2, a few microliters of 10-4 g/1 of o-tubocurarine applied to the floor of the IVth ventricle blocks mPSPs within 60 sec. The evoked PSP is greatly reduced in size (Fig. 2Ba) while the giant fiber action potential elicited by depolarizing pulses remains unchanged in amplitude as does the giant fiber resting potential. After washing with Ringer the PSPs recover to near their control values. Finally morphological investigation has failed to reveal any synapses onto the giant fibers in the medulla other than those arising from the Mauthner fibers 17. Although there are synapses on the somata of the giant fibers, these are probably too distant to produce detectable mPSPs, and are at least in part non-cholinergic2~. Effects of repetitive stimulation on PSPs As the frequency of Mauthner fiber stimulation is gradually increased, the PSPs become steadily reduced in amplitude (Fig. 3). In the extreme the PSPs can be
232 A1
B}
.L.._J,,_~
~d.,
~1 . . . .
L~t .... I --
.!.
20 mse¢
!
l
5 0 mV
2 mV
:
Fig. 2. Block of miniature PSPs and reduction in PSP amplitude caused by topical application of curare. All traces are intracellularly recorded from a giant fiber. A: prior to curare application. AI: spontaneous miniature PSPs (3 records at high gain). As: a synaptically evoked spike following activation of the Mauthner fibers by spinal stimulation. Aa: a spike and two PSPs evoked by a train of 3 spinal stimuli. B: records one minute after topical application to the IVth ventricle of a few microliters of curare 10-4 g/1. BI: spontaneous miniatures are absent, recording as in Aa. B2: a directly evoked giant fiber spike is similar in amplitude to the synaptically evoked spike in A2. B~: 3 PSPs evoked by spinal stimulation are markedly reduced in amplitude compared to those in A3. High and low gain records are shown in Aa and B3.
m a d e to d i s a p p e a r into the noise level. Even when PSP a m p l i t u d e is as small or smaller t h a n that o f the miniatures recorded simultaneously, there are no c o m p l e t e failures o f PSPs (Fig. 3E-I). A t the higher frequencies when the P S P a m p l i t u d e is reduced, its time course is significantly longer lasting t h a n the miniatures riding u p o n it (Figs. 3 E - I a n d 4B a n d C). I f the size o f the miniatures c o m p r i s i n g small PSPs is the same as t h a t o f the s p o n t a n e o u s miniatures, there should be no (non-zero) PSPs smaller t h a n a miniature, a n d failures should be c o m m o n for a m p l i t u d e s where the m e a n n u m b e r o f miniatures c o n t r i b u t i n g to the P S P is small. W e believe t h a t q u a n t u m n u m b e r in the small PSPs is high, b u t t h a t q u a n t u m size has become m u c h smaller. The reduced q u a n t u m size can be a c c o u n t e d for by depletion o f the vesicular store o f t r a n s m i t t e r a n d subsequent release o f the contents o f incompletely filled vesicles. I f the a m p l i t u d e s o f the PSPs are Poisson distributed, which we will consider further in the discussion, we can estimate q u a n t u m n u m b e r f r o m variance a n d mean a m p l i t u d e o f the PSP 16. In the PSP o f Fig. 4C c o m p u t e d q u a n t u m n u m b e r is 13.6. M o r e o v e r , the absence o f failures provides an estimate o f a lower limit on q u a n t u m number. O n the a s s u m p t i o n o f a Poisson process, q u a n t u m n u m b e r s o f 5, 7.5 a n d 10 w o u l d result in a m e a n o f one failure p e r 6 sec, 72 sec a n d 15 min o f stimulation at 25/sec. W e conclude from m a n y seconds o f watching the oscilloscope t h a t a lower
233
A I ~ B
C
.
IOOmV D
~ ~
2msec ,o,,~c
": i :ec -.. ~
II
--
,,tk
E
33j~ec
H
ll~sec
F
~o,~
J !1mY imsec
143/seC
.... c
u-
.
.
.
.. . .. . . .. . .. .
I p
Fig. 3. Reduction in evoked PSPs to the size of miniature PSPs by tetanic stimulation. A: spinal stimulation at a low rate (1/sec) showing the giant fiber spike and the extracellular field of the Mauthner and giant fibers recorded on the lateral line lobe: B: after some seconds of stimulation at 10/sec the rate of rise of the PSP is reduced, but initiation of impulses in the giant fiber persists; recorded as in A. C: at 20/sec stimulation, PSP amplitude is reduced below threshold. The upper trace is a high gain intracellular recording. The middle trace is a just extracellular recording obtained subsequently at the same gain and stimulation frequency. D-I: progressively increasing frequencies of stimulation (display as in C, faster sweep in E and F). The PSP becomes progressively and gradually smaller in amplitude until it is smaller than the mPSPs simultaneously recorded. The evoked PSPs are longer in time course than the mPSPs. The mPSPs are generated at any time during the stimulation cycle, and show little preferential occurrence during the evoked PSPs. limit for q u a n t u m n u m b e r is between 5 and 10 in experiments such as those illustrated in Figs. 3 and 4.
Effects of tetanic stimulation on miniatures The frequency o f miniatures increases markedly just after the onset o f a tetanus (Fig. 4B and C, top row and Fig. 6A), and then decreases again to a value near or below the resting level (Fig. 4B and C, b o t t o m row). In occasional experiments miniature frequency becomes very small during a tetanus. The mean amplitudes o f the mPSPs decrease somewhat, but even when their frequency is markedly reduced they are not greatly reduced in size. In the experiment o f Fig. 4 during the tetanus there was a roughly 5 0 ~ decline in m P S P amplitude from a mean o f 0.73 ± 0.25 to a mean o f 0.38 ~ 0.14 mV (Fig. 5). In no experiment were the mPSPs reduced in amplitude by more than this amount. The observed reduction in m P S P amplitude is certainly not adequate to explain the reduction in PSP amplitude given the q u a n t u m n u m b e r implied by the calculations from the variance and absence o f failures that are given above. In the experiment o f Fig. 4 the mean m P S P size increases slightly to 0.51 ~ 0.17 mV within a few seconds after the tetanus, but time for complete recovery to control
234
A
,
•'
C
D
20~lec
I mV
2 msec
Fig. 4. Changes in mPSP frequency and amplitude during tetanic stimulation. Three records in each column. A : spontaneous mPSPs prior to tetanic stimulation (recorded at low sweep speed). B: mPSPs and evoked PSPs during spinal stimulation at 25/sec. PSP amplitude is reduced to the level of the mPSPs superimposed on it. The top traces were recorded near the beginning of the tetanus and there is an increase in the frequency of mPSPs. After about one minute and several minutes (next two records below) the mPSP frequency is progressively reduced, but evoked PSP amplitude is unchanged. Several superimposed sweeps in each record. C: as in B, but single sweeps. D: the amplitude of mPSPs is somewhat reduced immediately after cessation of the tetanus.
amplitude was not measured. In no experiment did we observe an increase in the number of small mPSPs during recovery from a tetanus.
Short-term recovery of PSPs from tetanic stimulation When a tetanus is terminated, PSP amplitude recovers rapidly. Generally after a tetanus a one second rest allows recovery of the PSP to an amplitude at which it initiates an impulse. To explore the recovery process, trains of stimuli were given at varying rates that were selected to allow partial recovery between trains. The PSP
t/) a. t/) o. E --I
.4
o .3
O
CONTROL
o
DURING TETANUS
® POST TETANUS
o
I--
•
®
14.
O .2 o®
Z
o
o•® ®
U
.1
•
o
o
oe ®
•
n~
LL •
®
•
o
o0
I
I
I
I
a
t
.2
.4
.6
.8
1
t2
oo
oo
oo i
a
I
n
1.4 1.6 1.8 2
mPSP AMPLITUDE mV Fig. 5. Amplitude distributions of the miniature PSPs before, during and after a tetanus. Same tetanus as in Fig. 4. The ordinate is the fraction of the total number of miniature PSPs and the abscissa is the mPSP amplitude. The means and standard deviations in the 3 cases are 0.73 ± 0.25 mV, 0.38 ± 0.14 mV and 0.51 ~ 0.17 mV, respectively.
235
A
l
l
I
t
1
2my 2 0 reset
.5
mV
msec
~"
~
.2
Z
_o
i~.---
~' '
0
iii
PSP AMPLITUDE mV
Fig. 6. Recovery of PSPs between trains of stimuli. A: PSPs and mPSPs during a train of 15 stimuli at 25-msec intervals. The train was repeated at l/sec until a steady level of responding was obtained. Continuous record from upper to lower trace, mPSP frequency increases during the early part of the train. B1-3: the first PSP in the train at faster sweep speed. C: amplitude distributions of the first PSP and of miniature PSPs during repetition of the train at 1/sec. Mean mPSP amplitude is 0.38 mV and at this interval between trains quantal size calculated from variance is 0.32 mV. Noise level indicated by shading along the ordinate.
evoked by the first stimulus in the train was then studied with respect to amplitude variations in order to make inferences about quantal size. Such an experiment is illustrated in Fig. 6. When the train was repeated once per second (725 msec between trains), the amplitude of the first PSP in the train showed considerable variability. On the assumption that the amplitudes are distributed according to Poisson statistics, we calculated quantal size and content from the coefficient of variation and mean amplitude of the PSP 16. The calculated quantal size was similar to the size of the spontaneous mPSPs observed simultaneously. This agreement suggests that Poisson statistics were applicable and that at this interval between stimuli the quantal size had largely recovered. When 400 msec separated successive trains, PSP amplitude decreased, but the size of the spontaneous mPSPs remained unchanged. The quantal size for the PSP calculated as before was now half the mPSP amplitude. In two other similar experiments calculated quantal size equalled the size of the spontaneous mPSPs with 200 msec recovery time. Many other experiments gave qualitatively similar results; PSP amplitude always recovered to threshold in a second or less, although PSP amplitude had been greatly reduced during a preceding tetanus. Determined from the variance of recovering PSPs 0.2-0.4 msec seems to be the minimum recovery time for the depleted preparation to generate PSPs consisting of
236 lO 9 E
.8
t~
.7
~)
.6
~'-- .5
.2 9--15
A
TIME (sets)
.o c5
0 OCt
~o
oo o
E
6Z
o
o
o_
o
•
••
o •
o
le
ee
.
.
.
.
.
.
.
.
.
j
1
,
i
i
,
i
i
5
i
i
t
j
,
i
I
i
h
9 ' 1o','1
f
1,
TIME (sees) Fig. 7. Effects of a single spinal stimulus on amplitude and frequency of mPSPs in a giant fiber. The upper graph shows the time course of amplitude changes for a preparation that had been stimulated extensively (filled circles) and for a preparation that had been stimulated very little (open circles). Mean and standard deviations were computed for 1 sec intervals. In the two examples the amplitudes are depressed by about 25 % with a significance at 0.001 level for 3 sec (open circles) and 4 sec (closed circles). The lower graph shows the frequency changes in the same examples measured as mPSPs per half sec. The rate of occurrence was depressed in the extensively stimulated fiber (filled circles) and augmented in the relatively fresh preparation (open circles).
'normal size' quanta of transmitter. Since an appreciable number o f small quanta can be released with much shorter interstimulus intervals, most of this period should represent time for filling the vesicles.
Long-term effects of tetanic stimulation T h e m o r p h o l o g i c a l d a t a 18 i n d i c a t e that over one hour is required for full recovery of the vesicle population. Although after a tetanus the response to low frequency stimulation (l/sec) recovers within a second to the level required to initiate impulses, there is a much more persistent decrement to prolonged trains o f stimuli. Following a train that greatly reduces PSP amplitude and a brief rest that is long enough to allow quantum size to recover fully, responses to a second train decrement
237 again within a few seconds, which is more much rapid than during the first train. To investigate the time course of recovery after a prolonged train, experiments were carried out using a train of three stimuli separated by 15-20 msec. The short train was repeated once per second and mean amplitudes for the 2nd and 3rd PSPs in the train were obtained. The first PSP was suprathreshold under these conditions. Then the train was repeated at 10/sec for 10 min which reduced PSP amplitude to a low level and presumably depleted the synapse of vesicles18. Recovery was tested by applying the train 100 times at one/sec after 15 min, 30 min, etc. In three experiments recovery of the 2nd and 3rd PSPs was complete or nearly so after 15 min, 45 min, and 1 h. At all rest intervals the 1st PSP initiated an impulse and its amplitude was not determined. In several other experiments partial recovery was observed over 15-30 min, but the penetration was lost before complete recovery was obtained.
Loss of large miniature potentials after single stimuli Even a single stimulus has a small effect on miniature amplitude. Fig. 7 graphs the results of two experiments in which a single stimulus to the Mauthner fiber was observed to transiently decrease the amplitude of the spontaneous mPSPs. Open circles illustrate the case of an 'unstimulated' preparation which had been stimulated briefly at 1/sec while penetrating the giant fiber. A single Mauthner fiber stimulus increased the frequency of mPSPs for several seconds, but decreased the amplitude of these potentials. The filled circles illustrate a second preparation which had been extensively stimulated prior to the single shock test. In this case both the mPSP frequency and amplitude transiently declined. A single train of 2 or 3 stimuli to the Mauthner fiber produced a similar transient reduction in the mPSP amplitude. This result suggests that if the population of vesicles is somewhat variable in terms of contents of transmitter, the fuller ones are preferentially released, perhaps because they are the most recently filled, and filling and release sites are close together. DISCUSSION
A primary goal of the present study was to explain the small size of PSPs obtained by repetitive stimulation at gradually increasing frequencies. The evoked PSPs become of a size similar to or smaller than that of the miniature PSPs simultaneously observed, without there being the failures that would be predicted from simple application of the quantal hypothesis of transmitter release. We can put forward 3 hypotheses consistent with the quantal release of transmitter from vesicles to explain the absence of failures when PSP amplitude becomes small. (1) Quantum number is high, but the number of molecules of transmitter per quantum (or vesicle) is reduced (the partial filling hypothesis). (2) Quantum number is high and the quanta involve the normal amount of transmitter, but the postsynaptic effect is reduced by desensitization (the desensitization hypothesis). (3) Quantum number is small, but the probability of release is high. These hypotheses will be considered in order. The evidence favors the partial filling hypothesis and that partially filled vesicles arise through membrane recycling. The preceding paper demonstrates depletion of
238 vesicles by tetani that reduce PSP amplitudes to low levels and indicates that new vesicles are formed from intermediate membranous compartments 1~. If the contents of these new vesicles were released prior to complete filling with transmitter, quantal size would be reduced. In the depleted synapse vesicles are undergoing exocytosis as fast as they are being made, and it is reasonable that they have time for only partial filling. A possible objection to this explanation is that the spontaneous miniature PSPs do not undergo a change in size comparable to that inferred for quanta comprising the evoked responses. The evidence indicates that the mPSPs arise at the Mauthner fiber synapse, but the synaptic vesicles that are responsible for the mPSPs often appear to be tess accessible for depletion than those comprising the evoked response. The persistence of relatively full sized mPSPs can be explained in several ways. Our initial view was that these represented ordinary full vesicles late in diffusing to the release site. This explanation seems unlikely however since these mPSPs are not preferentially released during the PSP but instead occur more or less at random during the tetanic stimulation cycle. Another possibility is that there are release sites that have their own supply of vesicles and that are only slightly affected by impulse activity. In support of this explanation electron microscopy reveals occasional normal appearing clusters of vesicles close to the surface membrane in otherwise depleted synapses is. These islands of vesicles might be the origin of the relatively full sized mPSPs that occur while quantal size is reduced in the evoked responses. The poor synchronization of the mPSPs with evoked PSPs and the long latency of onset of the tetanic increase in mPSP frequency suggest that there are different release mechanisms or separate release sites. The slow onset of the tetanic increase in frequency allows the site of mPSP generation to be more distant from the site at which the action potential acts on the presynaptic membrane. A similar situation may obtain at the neuromuscular junction, for the increase in mPSP frequency caused by tetanic stimulation is slow in onset as well as independent of extracellular calcium (see ref. 14). The desensitization hypothesis would seem to be contradicted by the observation that nearly normal sized mPSPs persist even when the evoked PSP is greatly reduced. If, however, the persistent mPSPs were generated at different regions of the synapse from where the PSPs were generated, they would not reveal desensitization. Although the marked changes in mPSP frequency and the smaller changes in amplitude suggest that the sites for evoked release and for at least some of the mPSPs are not far apart, it is uncertain whether the same postsynaptic receptors are affected. Nevertheless the correlation between depletion of vesicles and depression of transmission leads us to favor partial filling over desensitization as the origin of small quanta. Furthermore, the effects of repeated trains are inconsistent with desensitization. I f desensitization were responsible for the reduction in quantum size, one would expect that once quantum size had recovered after a prolonged train, the time course of reduction during a second train would be as slow as that in the first. Instead the second reduction is much more rapid which is of course consistent with the morphological observations of slow recovery from depletion of vesicles 18. Initially we expected that small mPSPs would be observed during recovery from
239 a tetanus whether depletion or desensitization was occurring. However, the rapid filling of vesicles inferred from the rapid recovery of the PSP after a tetanus makes the failure to observe small mPSPs understandable. It is likely that the filling of a vesicle is so rapid that the time it spends partially full is small compared to its lifetime before release as a spontaneous miniature. Another explanation of the lack of small mPSPs is that only more or less full vesicles can be released as miniatures. This is not unreasonable, but would require an important difference between release of vesicles that was spontaneous and release that was evoked by tetanic stimulation. The hypothesis that the small PSPs consist of a small number of normal quanta released with high probability is the simplest to reject since PSPs of smoothly graded amplitudes smaller than the normal miniatures are easily obtained (Fig. 3). Moreover, the different synaptic areas contributing to the PSPs are separated by distances such that they would be expected to act independently of each other, particularly on the ipsilateral side where several distinct synapses occur between the Mauthner and giant fibers. If reduction in PSP amplitude were due to a sudden drop in the probability of release from large to small at these individual sites, a stepwise reduction in the PSP would occur, whereas a gradual drop in probability would lead to a binomial or Poisson process, neither of which fits the evidence. The recovery time of PSP amplitude appears to provide a measure of the filling time of the vesicles. The experiments involving repeated trains with recovery between them indicates that quantal size in the evoked PSP has recovered to the spontaneous mPSP size within a few tenths of a second. Certainly at this time the total vesicular store is replaced to an insignificant extent, although there are enough vesicles observed to explain the persistence of transmission during the tetanus as well as the recovery 18. Since the PSPs can be evoked at intervals less than 20 msec without failures, most of the recovery time of quantal size appears to be the time for filling the vesicles rather than the time for their formation and movement to the release site. The longer time for full recovery of PSP amplitude presumably involves increase in the number of vesicles available for release by an impulse as well as filling of more recently formed vesicles. Because quantum size appears to decrease gradually as the interval between stimuli decreases, it is likely that release can interrupt filling. This observation suggests that the filling and release sites are essentially the same. The most attractive alternative is that vesicles take up ACh from the cytoplasm as they move from formation to release site. In this case the transit time for essentially all the vesicles would during a tetanus have to be uniformly short compared to the filling time which as already noted is a few tenths of a second. The ultrastructural data suggest vesicles are formed within a micron of the presynaptic membrane and not in a particularly uniform way 18 (see also ref. 12). Thus, filling at the release site itself seems more plausible. It is likely that filled vesicles move away from the filling (and release) sites to comprise the mobilizable store of transmitter corresponding to the large pool of vesicles that are depleted by tetanic stimulation. Although the data suggest that during a tetanus the time required to supply
240 vesicles is short compared to the filling time, they do not specify whether new vesicles are supplied for each release event or each vesicle is used for a number of release events 19,24. An alternative to loss of a vesicle's identity with each exocytotic event is that a vesicle releases its transmitter by merely transient contact with the surface membrane and maintenance of its vesicular form. If it went through subsequent release events prior to complete refilling, quantal size would be reduced. The morphological observations of depletion clearly indicate that exocytosis occurs, although they do not establish that it occurs with every release event. However, we think it likely that once a vesicle fuses with the external surface that its membrane can only be reclaimed by coated vesicles 18. One may ask what specific characteristics of the hatchetfish synapse are important in the observations of reduced quantal size and of filling time. Obviously vesicles must be releasable faster than they can be made and it must be possible to deplete the mobilizable store of vesicles in a reasonable period. A factor that is particularly important in measuring filling time is the high probability of release of the immediately available store by a single impulse. The probability of release at the hatchetfish synapse approaches unity as described earlier 2, the most important evidence being that a second stimulus closely following a test impulse produces only a greatly reduced PSP (Fig. 1) and that PSP size is independent of amplitude of the Mauthner fiber impulse over a wide range. Since a single impulse releases virtually all the quanta immediately available for release, the vesicles that are released by the next stimulus have recently arrived at the release site. If the probability of release is near unity, we should justify the use of Poisson statistics to infer quantal size during vesicle filling, since these statistics apply where a small probability operates on a large population of vesicles. However, if the population is Poisson distributed and probability of release is unity, clearly the number released is Poisson distributed. The population of vesicles would vary according to Poisson statistics if vesicles occupying release sites were the relevant population, vesicles freely diffused in the terminal, and only a small fraction of available release sites were occupied. In the depleted synapse the vesicles are sufficiently sparsely distributed that they should not interfere with each other and quantum number is reduced well below that for PSPs in the unstimulated preparation; thus it is reasonable that Poisson statistics apply to the vesicles available for release. Further investigation of the statistics of release is required in the unstimulated preparation in which vesicles are quite densely packed and a larger number of vesicles are released per impulse. Another factor in the statistics for calculation of quantal size is variability of the quanta 16. Presumably vesicles arrive all during the interval between stimuli and are filled to differing degrees. Thus, a correction for increased variance of quantal amplitude should be included in the calculations, a correction that would tend to decrease calculated quantal size and increase calculate quantum number. We have not made this correction in the absence of direct measures of quantal variability. The correction would prolong the inferred time to fill vesicles to a moderate degree, but at this stage in the investigation, a change of even a factor or two would not be of any great importance.
241 The suggestion that filling and release sites are the same is supported by biochemical observations that acetylcholine released by physiological stimuli is richer in recently synthesized material than is the total storeS,6, 8,9. Also, effects of a brief rest during prolonged stimulation of the neuromuscular junction suggest that a subpopulation of vesicles ready for release is filled preferentially11. To be sure the time scale for these experiments is much longer than for those with the hatchetfish. We do not know whether the transmitter going into vesicles comes from the cytoplasm or is synthesized in the filling process. Moreover, it is possible that in the unstimulated condition vesicles are continually visiting the filling site and exchanging or 'topping up' their contents. The loss of large quanta for several seconds after a single stimulus could be explained by this mechanism. Some synapses can transmit at much higher frequencies than the hatchetfish synapse. The maximum is the nerve-electrocyte junction in the weakly electric gymnotid fish Eigenmannia which transmits indefinitely at 300-600/sec (see ref. 4). For the mechanisms of filling to be the same as in the hatchetfish, one would require a much larger number of sites for filling and release with a much lower probability of release at a particular site. Synapses that do not fatigue and at which vesicles cannot be depleted are explicable in terms of vesicle formation and filling mechanisms that can process vesicles at least as rapidly than the release mechanism. Whether the formation mechanism is faster in non-depleting synapses or whether the cellular machinery is just present in larger amount is a question that may be open to morphological investigation. At the neuromuscular junction reduction in quantal size has been observed after relatively high frequency stimulation10,15, and after prolonged low frequency stimulation that depletes the ending of transmitterT, 11. In the sympathetic ganglion, however, quantal size was not altered by stimulation to depletion 21. It is possible that in the preganglionic fibers the rate limiting step is the formation rather than the filling of vesicles, and in this tissue as well as in the neuromuscular juntion, depletion of vesicles can be produced by repetitive activationS,z0. Small mPSPs can be recorded at the neuromuscular junction after exhaustive stimulation and it has been suggested that the usual large mPSPs result from multiple release of the small components 15. The hatchetfish results suggest that partially filled vesicles could give rise to the small mPSPs, although in the hatchetfish the contents of partially filled vesicles appear to be at most rarely released spontaneously. The combined physiological and morphological data support the view that vesicles are lost in the release process and their membrane is recycled to form new vesicles. In addition one can obtain a measure of the time required to fill newly formed vesicles. The filling process appears accessible for study because new vesicles can be so rapidly released that adding transmitter is the rate limiting step. Although full recovery after depletion requires prolonged rest, transmission at low rates recovers very rapidly. The survival value is obvious, and the mechanism may be of some generality. The observations reported here allow a further elaboration on the hypotheses of quantal release and membrane recycling in specifying the time and site of addition of transmitter.
242 ACKNOWLEDGEMENTS A n u m b e r o f t h e early e x p e r i m e n t s w e r e d o n e in c o l l a b o r a t i o n w i t h M i c h a E. Spira, w h o also c o n t r i b u t e d m u c h h e l p f u l criticism. T h e w o r k was s u p p o r t e d in p a r t by N I H
Grants NB-07512 and HD-04248,
a n d g r a n t s f r o m the A l f r e d P. S l o a n F o u n d a t i o n . REFERENCES 1 AUERBACH,A. A., Spontaneous and evoked quantal transmitter release at a vertebrate control synapse, Nature (Lond.), 234 (1971) 181-183. 2 AUEREACH,A. A., AND BENNETT, M. V. L., Chemically mediated transmission at a giant fiber synapse in the central nervous system of a vertebrate, J. gen. Physiol., 53 (1969) 183-210. 3 BARKER, L. A., DOWDALL, M.J., AND WHn"rAKER, 'V. P., Choline metabolism in the cerebral cortex of guinea pigs, Biochem. J., 130 (1972) 1063-1080. 4 BENNETT, M. V. L., Electric organs. In W. S. HOAR AND D. J. RANDALL (Eds.), Fish Physiology, Vol. 5, Academic Press, New York, 1971, pp. 347~,91. 5 BIRKS, R. I., The relationship of transmitter release and storage to fine structure in a sympathetic ganglion, J. Neurocytol., 3 (1974) 133-160. 6 BIRKS, R. I., AND FITCH, S. J. G., Storage and release of acetylcholine in a sympathetic ganglion, J. Physiol. (Lond.), 240 (1974) 125-134. 7 CECCARELLI,B., HURLBUT,W. P., AND MAURO, A., Turnover of transmitter and synaptic vesicles at the frog neuromuscular juntion, J. Cell Biol., 57 (1973) 499-524. 8 CHAKRIN,L. W., MARCHBANKS,R. M., MITCHELL,J. F., AND WHITTAKER,V. P., The origin of the acetylcholine released from the surface of the cortex, J. Neurochem., 19 (1972) 2727-2736. 9 COLLIER,B., The preferential release of newly synthesized transmitter by a sympathetic ganglion, J. Physiol. (Lond.), 205 (1969) 341-352. 10 COOKE,J. D., AND QUASTEL,D. M. J., Transmitter release by mammalian motor nerve terminals in response to focal polarization, J. Physiol. (Lond.), 228 (1973) 377-405. 11 ELMQVIST,D., AND QUASTEL,D. M. J., Presynaptic action of hemicholinium at the neuromuscular junction, J. Physiol. (Lond.), 177 (1965) 463--482. 12 HEUSER, J. E., AND REESE, T. S., Evidence for recycling of synaptic vesicle membrane during transmitter release at the frog neuromuscular junction, J. Cell Biol., 57 (1973) 315-344. 13 HIGHSTEIN,S. M., AND BENNETT, M. V. L., Fatigue at Mauthner fiber-giant fiber synapse of the hatchetfish, Fed. Proc., 32 (1973) 443 abstract. 14 HURLBUT, W. P., LONGENECKER,H. B., JR., AND MAURO, A., Effects of calcium and magnesium on the frequency of miniature end plate potentials during prolonged tetanization, J. Physiol. (Lond.), 219 (1971) 17-38. 15 KREIBEL, M. E., AND GROSS,C., Multimodal distribution of frog miniature end plate potentials in adult, denervated, and tadpole leg muscles, J. gen. Physiol., 64 (1974) 85-104. 16 MARTIN, A. R., Quantal nature of synaptic transmission, Physiol. Rev., 46 (1966) 51~i6. 17 MODEL, P. G., SPIRA, M. E., AND BENNETT,M. V. L., Synaptic inputs to the cell bodies of the giant fibers of the hatchetfish, Brain Research, 45 (1972) 288-295. 18 MODEL, P. G., HIGHSTEIN, S. M., AND BENNETT, M. V. L., Depletion of vesicles and fatigue of transmission at a vertebrate central synapse, Brain Research, 98 (1975) 209-228. 19 PFENNINGER, g . H . , AND ROVAINEN, C. M., Stimulation- and calcium-dependence of vesicle attachment sites in the presynaptic membrane; a freeze-cleave study on the lamprey spinal cord, Brain Research, 72 (1974) 1-23. 20 PYSH, J. J., AND WILEY, R. G., Synaptic vesicle depletion and recovery in cat sympathetic ganglia electrically stimulated in vivo, J. Cell BioL, 60 (1974) 365-374. 21 SACCHI,O., ANDI~RRI, V., Quantal mechanism of transmitter release during progressive depletion of the presynaptic stores at a ganglionic synapse, J. geE. Physiol., 61 (1973) 342-360. 22 SPIRA, M. E., MODEL, P. G., AND BENNETT,M. V. L., Cholinergic transmission at a central synapse, J. Cell BioL, 47 (1970) 199a-200a. 23 SPIRA, M. E., AND BENNETr, M. V. L., Penicillin induced seizure activity in the hatchetfish, Brain Research, 43 (1972) 235-241. 24 STREIT,P., AKERT, K., SANDRt, C., LIVINGSTON,R. B., AND MOOR, H., Dynamic ultrastructure of presynaptic membranes at nerve terminals in the spinal cord of rats. Anesthetized and unanesthetized preparations compared, Brain Research, 48 (1972) 11-26.