ABSTRACT
When certain vertebrate and invertebrate interneurones are repeatedly stimulated through the sense organs, or by electric shocks delivered to sensory nerves, their response gradually fails. Typically the stimulus is delivered once every few seconds and the response consists of action potentials (e.g. Bell, Buendia, Sierra & Segundo, 1964; Horn & Hill, 1964, 1966; Horridge, Scholes, Shaw & Tunstall, 1965; Horn & Rowell, 1968) or excitatory postsynaptic potentials (Bruner & Tauc, 1966; Spencer, Thompson & Neilson, 1966; Segundo, Takenaka & Encabo, 1967 a). The physiological basis of this response decrement is not known though synaptic depression, desensitization of the postsynaptic membrane, accumulation of potassium ions in the extracellular clefts and build up of pre-and postsynaptic inhibition have all at one time or another been invoked.
The analysis of factors responsible for response decrement is likely to be more successful in a relatively simple system than in a complex one. A monosynaptic junction would be ideal for such a study, especially if the presynaptic and postsynaptic components of the junction could be seen and penetrated with micro-electrodes for stimulating and recording, and where there are no internuncial neurones. The giant synapse of the squid stellate ganglion fulfils all of these requirements (Young, 1939; Bullock & Hagiwara, 1957; Bryant, 1959; Takeuchi & Takeuchi, 1962; Miledi & Slater, 1966; Miledi, 1967). This junction rapidly ceases to transmit if the presynaptic fibres are stimulated continuously with shocks at high frequency (Bullock, 1948). In the present study discontinuous stimulation was used in order to determine whether response decrement curves analogous to those observed in more complex systems could be generated, and whether the mechanisms which operate to bring about response decrement during continuous stimulation also operate during discontinuous stimulation.
The anatomy of the squid stellate ganglion has been described by Young (1939). The mantle connective contains a number of giant axons (preganglionic giants) which end in the stellate ganglion. One of these fibres, the largest, gives off a series of branches each of which enters into synaptic contact with one of the postganglionic giant fibres to form a giant synapse. There are between eight and eleven postganglionic or stellar nerves, each of which contains a giant fibre. The largest postganglionic giant ten trains. However, by varying the parameters of the train, it was invariably possible to reduce the likelihood of transmission of spikes across the ganglion to less than 100% for repeated trains of stimuli applied to the pre-nerve.
Effects of varying the parameters of a train
After the experiment which gave the results plotted in curve 1 (Text-fig. 1) the number of shocks within a train was increased to twenty by increasing the frequency of stimulating pulses. Other stimulus parameters remained unchanged. A group of ten trains was delivered. The first two trains (Textfig. 1, curve 2) each evoked the maximum of twenty spikes in the post-nerve. Thereafter the response declined.
The rate at which the response to successive trains declines is, within wide limits, independent of the frequency of pulses within a train, provided the number of them is held constant. Thus the responses plotted in Text-fig. 2 declined when trains containing twenty shocks were delivered to the pre-nerve at 10 sec. intervals. The way in which the responses declined was similar whether the twenty shocks were delivered at a rate of 30/sec. (curve 1) or 50/sec. (curve 2). Before each of these groups of trains was delivered the preparation was allowed a 600 sec. rest period. The post-nerve response to the first train of each group contained fewer spikes than the expected twenty, suggesting that 600 sec. was not adequate to restore transmission after a period of prior activity (see below).
Effect of varying the interval between trains
It was usually possible to change the probability of transmission across the ganglion, by varying the interval (i.t.i.) between the beginning of successive trains of shocks. Thus in the four experiments on one preparation which gave the results plotted in Text-fig. 3 the i.t.i. was different in each experiment. The stimulus to the pre-nerve was otherwise the same, consisting of a group of ten trains each having a duration of 0·5 sec. and containing twenty shocks delivered at twice threshold intensity. There was no response decrement for i.t.i. of 40 sec. and 25 sec. (curves 1 and 2 respectively). When the interval was reduced to 15 sec. the response was maintained for the first seven presentations and then declined. At i.t.i. of 10 sec. the response declined rapidly. Before each of these groups of trains was presented, the preparation had been unstimulated for 600 sec.
The results illustrated in Text-figs. 1 and 3 suggest that transmission across the ganglion represents a balance between the extent of recovery from the previous train and the number of shocks contained in the following one. The interplay of these two factors is illustrated in Text-fig. 4. Nine shocks were delivered to the pre-nerve in 450 msec., the train being repeated every 10 sec. Fifty-seven trains were applied and the response of the postsynaptic giant axon was constant at nine spikes. The duration of the train was then doubled (Text-fig. 4B) so that each train contained eighteen shocks. Instead of increasing, the modal number of spikes dropped to eight. After the seventieth train (Text-fig. 4C) the number of pulses per train was reduced to nine and again the modal response contained eight spikes. When (Text-fig. 4D) the shock frequency within a train was doubled (eighteen shocks in 450 msec.) the first response contained ten spikes and then fell to between seven and nine. These results suggested that an i.t.i. of 10 sec. was adequate to maintain transmission to between seven and nine shocks per train, but no more. If so, the response should rapidly fall if the interval between trains were shortened, an effect that was demonstrated. After D in Text-fig. 4, each of the 450 msec, trains contained eighteen shocks. From D to E (Text-fig. 4) the i.t.i. was 10 sec. At E it was reduced to 5 sec. The response (train 94) immediately declined and continued to do so until transmission completely failed. During this state, the i.t.i. was increased to 10 sec. (Text-fig. 4F). Transmission was immediately restored and maintained, though at a lower level than before E.
Recovery following a rest
Transmission across the ganglion was also affected by the duration of the rest period which intervened between the presentation of groups of trains. To study this effect groups of stimuli were delivered to the pre-nerve. Each group contained ten trains of shocks. Stimulus parameters within a train and the interval between trains was constant in each group, but the interval (i.g.i.) between groups was varied. The results of such an experiment are plotted in Text-fig. 5. Each train applied to the pre-nerve contained twenty shocks and the i.t.i. was 10 sec. The five curves plotted in Text-fig. 5 were obtained from a continuous series of experiments on the same preparation. Following an i.t.i. of 600 sec. (curve 1) the first and second trains of shocks each evoked twenty spikes in the giant axon. Thereafter the responses declined. When the i.g.i. was 300 sec. only the first train evoked the maximum discharge in the giant axon (curve 2). A rest period of 60 sec. was not adequate to restore maximum transmission across the ganglia (curve 3), and an i.g.i. of 30 sec. even less effective (curve 4). It may also be seen from these curves that as the i.g.i. is reduced the level to which the postsynaptic responses fall gradually declines. These effects— a decline in the postsynaptic response to the first train within a group and a failure to sustain the tail of the response curve as the i.g.i. was shortened—characterized transmission across the ganglion. A second i.g.i. of 600 sec. (curve 5) was sufficient to restore transmission to approximately the original level (curve 1).
B. Bioelectric changes accompanying response decrement
When the giant axon failed to respond to every pulse in a train, it was always to the last shock or shocks. The effect is illustrated in Text-fig. 6 which is a series of records taken during an experiment in which the pre-nerve was stimulated by trains of shocks. The duration of each train was 0·5 sec. and contained 20 shocks. The i.t.i. was 10 sec. Each shock of the first and second trains evoked an impulse in the giant axon (Text-fig. 6, records 1 and 2 respectively). The remaining stimuli failed to elicit the maximum response and as the response declined more and more spikes dropped out from the end of each train. This pattern of response decrement is quite different from that which is seen on stimulating the pre-nerve with shocks at high frequency, suggesting the operation of different mechanisms. The effects of high-frequency stimulation are illustrated in Text-fig. 7. In the experiments from which these records were taken the pre-nerve was stimulated with trains of ten shocks. The frequency of the shocks was varied. Ten minutes were allowed to elapse between successive trains to ensure recovery from previous stimulation. When the pulse frequency was 120/sec. (Text-fig. 7 A) each shock evoked a spike in the post-nerve, but when the shock frequency was increased beyond this the action potentials began to drop out (Text-fig. 7 B, C, and D), some shocks evoking an impulse in the giant axon, others failing to do so.
The progressive failure of response in the post-nerve, such as shown in Text-fig. 6, is not due to impedance changes at the stimulating electrodes because (a) the current drawn is unchanged, although the response fails (Pl. 1), (b) transmission cannot be restored by increasing the shock intensity up to ten times threshold and (c) when recording electrodes are placed on the pre-nerve, response decrement is not observed.
When the action potentials have dropped out a small graded potential becomes visible, its amplitude declining as stimulation is continued (Text-fig. 6, e.g. record 10; Pl. 1, C). This local potential, which from its waveform appears to be a unitary one, is thought to be the postsynaptic potential of the giant synapse (Bullock, 1948; Bullock & Hagiwara, 1957). Further details of the changes in transmission may be obtained by superimposing the response to each shock within a train. Sample records are shown in Pl. 2. The responses of the giant axon to the first of a group of ten trains of shocks to the pre-nerve are shown in Pl. 2, A. Each of the ten shocks in a train evoked a single spike, but the latency of the spike gradually increased up to approximately 0·7 msec. The trains were delivered at intervals of 10 sec. The response to the fourth train (Pl. 2, B) contained ten spikes but the response to the sixth train (Pl. 2, C) contained only six spikes spread over 1−2 msec. The tenth train of shocks evoked only four spikes (Pl. 2, D). The latency to the origin of the synaptic potential was constant both within a train and between trains. Within a train the slope of the leading edge of the local response decreased and finally fell to below the threshold at which an action potential was generated, which was constant. The effects are more clearly shown in Pl. 2, E. This record was taken from the same preparation as the other records in the figure, but the recording electrode positions were changed slightly, to emphasize the synaptic potential. After a number of trains, with identical parameters to those described above, had been delivered the response began to fail and the ten pre-nerve shocks evoked only three spikes (Pl. 2, E). The point of origin of the local response was constant but the slope of its leading edge and the peak response amplitude gradually declined.
Another feature of the change in responsiveness may be seen in Pl. 2, A−D. In the first train, the latency to the first spike (Pl. 2, A) is 3·8 msec, while it is 4·3 msec, for the fourth train (Pl. 2, B), 4·6 msec, for the sixth (Pl. 2, C) and 4·7 msec, for the tenth (Pl. 2, D). Thus each burst of spikes has an after-effect which is not dissipated completely in the interval between trains. As a consequence the slope of the leading edge of the synaptic potential evoked by the first shock of a train is less than that evoked by the first shock of the previous train. This effect can also be seen in Textfig. 8. These records were obtained from an experiment in which the pre-nerve was repeatedly stimulated with trains of shocks similar to those described above. Responses are shown to the first shock of the first (A), third (B) and fourth (C) trains. The first train evoked the maximum number of spikes (ten) in the giant axon, the third only five and the fourth only four. The time of origin of the synaptic potential is unchanged, but its slope declines.
Once transmission had failed it was not possible to reestablish it by stimulating the pre-nerve with shocks at high frequency. Either there is no obvious effect of such stimulation on the rate of decline of responsiveness (Text-fig. 9 A) or there is a transient depression of transmission affecting the train following the high-frequency shocks (Text-fig. 9B). No other effects were observed even when the high-frequency stimulation was prolonged (e.g. 100/sec. for 20 sec.). This failure, which contrasts with the success of this procedure in Aplysia (Bruner & Tauc, 1966) is consistent with Miledi’s (personal communication) findings, using intracellular electrodes, that post-activation potentiation is not observed at the giant synapse.
DISCUSSION
When the mantle connective is repeatedly stimulated by trains of shocks the number of impulses evoked in the giant axon of the last stellar nerve can, by appropriate selection of stimulus parameters, be made to decline. The evidence presented suggests strongly that this failure of impulse activity, both within a train and to successive trains, is secondary to a decline in amplitude of the synaptic potential. The progressive delay in the time of spike initiation is accounted for by the progressively slower rate of rise of the synaptic potential, and the ultimate failure to generate a spike is accounted for by the failure of the synaptic potential to reach the threshold for spike initiation. The time of origin of the synaptic potential following the shock artifact is not affected by any of these changes.
It is probable that transmission failure occurs at all junctions between the giant fibres of the pre-nerve and the giant axon of the last stellar nerve. The reasons for supposing this to be so are (a) each of these fibres is capable of evoking an excitatory synaptic potential large enough to generate an action potential in the postsynaptic giant (Bryant, 1959; Miledi & Slater, 1966) and (b) once the responses to successive trains of stimuli have fallen to low levels, increasing the intensity of the stimulus up to ten times threshold fails to increase the response. These results suggest that at the intensity normally used to stimulate the pre-nerve (twice threshold) all the large axons in the mantle connective synapsing with the giant axon were excited, though the unitary synaptic potential recorded with the extracellular electrodes appears to originate at the giant synapse (Bullock, 1948; Bullock & Hagiwara, 1957).
When high frequency (e.g. ⪕ 50/sec.) trains of shocks are repeatedly applied to the pre-nerve, the pattern of spike drop-out (Text-fig. 6) is quite different from that which occurs during very high-frequency stimulation (e.g. ⪖ 150/sec., Text-fig. 7); but it is identical, as is the pattern of change of the synaptic potential, to the changes which occur during continuous stimulation of the pre-nerve with brief shocks (∼3o/sec.) which have been studied by Bullock (1948), Bullock & Hagiwara (1957), Bryant (1959), Miledi & Slater (1966), Miledi (1967) and Katz & Miledi (1967). These similarities, together with the evidence presented above (pp. 225−7 and Pls.1 and 2) suggest that the processes underlying transmission failure during continuous stimulation are the same as those which underlie failure during intermittent stimulation with trains of shocks, if allowance is made for partial recovery in the interval between successive trains.
In studies of the giant synapse in which transmission failure was brought about by continuous stimulation with brief shocks applied to the pre-nerve it has been shown (Bullock & Hagiwara, 1957; Bryant, 1959), using intracellular electrodes, that an action potential continues to invade the presynaptic terminal even when the postsynaptic response is declining. Bryant (1959) has shown that once transmission across the giant (distal) synapse has failed transmission continues at the proximal synapse and conversely. That is, the response decrement is synapse-specific. When the amplitude of the synaptic potential at the giant synapse has waned as a result of prolonged stimulation the amplitude may be restored by hyperpolarizing the presynaptic terminal (Hagiwara & Tasaki, 1958; Takeuchi & Takeuchi, 1962; Miledi & Slater, 1966), an effect which is thought to be brought about by increasing the amount of available transmitter. These studies suggest that stimulus repetition brings about an uncoupling between the action potential present in the presynaptic terminal and the expression of transmitter release manifest as the postsynaptic potential. The basis of this uncoupling is not known though certain possibilities can be excluded. The amount of transmitter substance released by the presynaptic terminal would fall if stimulus repetition brought about a reduction of its membrane potential (Hagiwara & Tasaki, 1958; Takeuchi & Takeuchi, 1962). This does not appear to happen (Miledi & Slater, 1966), so that explanations for a presynaptic origin of the failure brought about wholly by an accumulation of potassium ions in the extracellular spaces and/or by presynaptic inhibition are untenable. The decline of the response could be accounted for by postsynaptic inhibition, building up progressively in the course of stimulation. Such an explanation is improbable because stimulation of the second-order giant fibres has been shown to evoke only excitatory synaptic potentials (Bryant, 1959; Miledi & Slater, 1966) and although hyperpolarizing potentials have been recorded from cell bodies in the giant-fibre lobe they are not thought to be inhibitory. These potentials probably arise from fields created by large excitatory synaptic potentials in synapses near-by (Miledi, 1967). It seems improbable, therefore, that transmission failure is brought about through the activity of an internuncial neurone. Other possibilities to be considered include an imbalance between the mobilization and utilization of transmitter substance, changes involving the concentration, movement or reactions of calcium ions, and de-sensitization of the postsynaptic membrane. While the latter cannot at present be ruled out, the clear dependence of the synaptic potential on events which can be controlled presynaptically (e.g. recovery of transmission brought about by hyperpolarizing the presynaptic terminal) reduces the likelihood of the last possibility being of primary importance.
The reduction in amplitude of the synaptic potential described in the present study resembles changes in the excitatory postsynaptic potentials recorded from the giant cell in the abdominal ganglion of Aplysia on repeated stimulation of the skin or of an afferent nerve (Hughes & Tauc, 1963; Bruner & Tauc, 1966) and from spinal motoneurones on repeatedly stimulating cutaneous afferent nerves (Spencer, Thompson & Neilson, 1966). The reduction in number of action potentials evoked by successive trains of spikes resembles the reduction in number of spikes evoked, by repeated stimulation of sense organs or afferent nerves, in neurones of the spinal cord (Buchwald, Halas & Schramm, 1965) and brain stem of mammals (Bell et al. 1964; Horn & Hill, 1964,1966 ; Scheibel & Scheibel, 1965) and in the brain of insects (Horridge et al. 1965 ; Horn & Rowell, 1968). The parametric similarities are sufficiently striking to raise the question of whether synaptic depression, of the kind described in the present study, may not be common to many forms of such response decrement. (Of course, the precise mechanism of synaptic depression is bound to vary according to the properties [e.g. diameter of presynaptic terminals, size of extracellular cleft] of the synapse being studied.) This view does not, of course, exclude the possibility that in some situations response decrement to repeated stimulation results from the activity of inhibitory interneurones, but there is as yet no direct evidence that this is so (Horn, 1970).
The pattern of spike drop-out has not been studied in detail for central neurones. A regular pattern similar to that described in the present study has been observed in the spinal cord (Wickelgren, 1967), but there are clear examples (e.g. Fig. 9, Buchwald et al. 1965; Fig. 5, Segundo et al. 1967 b) where the spikes fall out irregularly. Such a pattern might be seen if synaptic depression occurred at several junctions in the pathway leading from the site of stimulation to the recorded cell.
It seems unlikely that response failure, such as described in the present study, occurs at the giant synapse of the intact animal. The reason is that a single shock delivered to the mantle connective results in a contraction of the mantle musculature (Young, 1938), an effect which we have repeatedly confirmed. It seems unlikely that, even when the most vigorous efforts are made to move away from a situation, as many as twenty impulses travel along a giant fibre in the pre-nerve over a period of 0·5 sec. ; and even more improbable that the discharge will be made regularly every few seconds. In all probability therefore there is a large factor of safety for transmission across this junction in the living animal. What we have shown is that transmission fails if the junction is challenged sufficiently vigorously. It is probable, however, that other junctions fail when activated in normal circumstances, as appears to be the case for certain synapses in the central nervous system. Such a property would convert the postsynaptic cell into a novelty-detecting unit.
SUMMARY
Failure of transmission was studied in the squid stellate ganglion by delivering shocks to the mantle connective and recording extracellularly from the last stellar nerve.
When the mantle connective was repeatedly stimulated by trains of shocks, it was invariably possible to select stimulus parameters at which the number of impulses evoked in the giant axon of the last stellar nerve declined.
The course of transmission failure with repeated stimulation was studied over a wide range of stimulus parameters. It was closely similar to the course of habituation which has been described for interneurones in more complex systems.
When a train of shocks was delivered to the mantle connective the synaptic potential evoked by successive shocks in the train declined. The time of initiation of the synaptic potential remained constant, but the slope of its leading edge decreased so that the time of spike initiation became progressively more delayed. Ultimate failure to generate a spike is accounted for by the failure of the synaptic potential to reach the threshold for spike initiation which was unchanged.
Recovery occurred with the passage of time. If a second train of shocks was delivered before restoration was complete the number of impulses evoked was usually fewer than those evoked by the previous train. This is because the synaptic potential is depressed and fewer shocks are needed to depress it still further to and below the level necessary for spike initiation.
Possible mechanisms responsible for the progressive failure of the synaptic potential are discussed.
ACKNOWLEDGEMENTS
We are grateful to Professor R. Miledi for helpful discussions, to the Director and Staff of the Stazione Zoologica for providing laboratory facilities for this work and to the U.S. Public Health Service (Grant NB 04787) for financial support.