ABSTRACT
The patterned burst activity of cardiac pacemaker ganglion cells in Homarus americanus has been studied by means of intracellular recording electrodes.
Burst activity, highly similar to that seen in cells of intact ganglia, has been demonstrated in ganglion sections containing as few as two large-cell bodies.
Studies of the sectioned preparations have shown that potential deflexions during the burst period are mainly endogenous activity of the respective cells and not post-synaptic potentials.
The behaviour of the cells in the period between bursts suggests the action of an inhibitory conductance change in each of the cells during this period.
INTRODUCTION
The pacemaker ganglion of the lobster heart is a functionally complete neural structure of nine cells which is capable of spontaneously producing bursts of action-potential discharges separated by periods of relative inactivity. The burst discharge is maintained even when the ganglion is completely removed from the heart. The somata of five of the nine cells are quite large, approximately 60 µ in diameter, making them favourable for study by means of intracellular electrodes. A number of reports have been made, dealing with intracellular studies on Panulirus interruptus and P. japomcus, the California and Japanese spiny lobsters (Bullock & Terzuolo, 1957; Watanabe, 1958; Hagiwara & Bullock, 1957; Hagiwara, Watanabe & Saito, 1959; Watanabe & Bullock, 1960). Studies on Homaros americanus, the New England lobster, have been carried out by Welsh & Maynard (1951), Maynard (1955), and Hartline (1967), these investigators using extracellular recording techniques. Cooke (1966) has reported on intracellular activity in this preparation.
Electrical activity in the somata of the large ganglion cells of Panulirus, during a period of burst discharge, is typified by the record of Fig. 1. Both slow, summating potentials and low-amplitude spike potentials are present in the wave-forms. Hagiwara & Bullock (1957) concluded that the slow potentials were EPSP’s and the spikes were the electrotonic effects of action potentials arising in the penetrated neuron at loci distant from the soma. Axons of certain of the small cells were considered to be presynaptic to the large cells and hence the pacemakers of the ganglion. For a detailed description of the experimental evidence, the reader is referred to a review article by Hagiwara (1961).
In Panulirus it was found that impairing the function of the four small cells by one means or another resulted in extensive changes in the wave-forms of the large cells. Hagiwara & Bullock (1957) found that crushing the caudal portion of the ganglion, where the somata of the four small cells are located, abolished the normal burstpotential pattern in the large cell somata. Watanabe (1958), in experiments on Panulirus japonicus, used procaine to anaesthetize the caudal portion of the ganglion, also with the result that the normal burst pattern in the wave-forms of the large cell somata disappeared. In some of these anaesthetized preparations a smooth oscillation of the resting potential of the large cells occurred at a frequency of approximately I c/sec. Frequently, groups of spikes were superimposed around the high point of the oscillations, but the type of potential fluctuations which in the intact ganglion were considered to be EPSP’s never occurred. Watanabe considered the slow oscillations and the spikes to be the intracellular events probably underlying Maynard’s (1955) observations of burst activity in preparations containing only somata of large cells.
In the experiments discussed in this communication intracellular activity of the large cells of the cardiac pacemaker ganglion in Homarus has been studied. The conclusions cited above necessarily constitute a hypothesis for interpreting burst phenomena in the Homarus ganglion because of the similarity of the preparations (see Alexandrowicz, 1932) and of the extracellularly recorded activity (Maynard, 1955). The findings of the current experiments indicate, however, that this hypothesis is not valid for Homarus for the following reasons. First, burst activity is present in preparations from which the small-cell bodies and their immediate processes have been removed, and the intracellularly recorded activity pattern in these sectioned preparations is very similar to that recorded from cells of intact ganglia. This burst activity is present in preparations which contain as few as two large-cell somata. Secondly, in the cases where sectioning the preparation had removed or damaged the normal pacemaker, the spontaneous burst rate was low, and the cells responded to electrical stimulation, internal or external, with burst activity. The manner in which the bursts arose in these circumstances was indicative of activation endogenous to the large cells. Thirdly, the behaviour of the large cells in the period between spontaneous bursts is largely governed by a long-term inhibitory conductance change occurring in each of the large cells.
MATERIALS AND METHODS
Most of the experiments described in this report were performed on New England lobsters (Homarus americanus). A few specimens of the California spiny lobster (Panulirus interruptus) were also studied. The cardiac pacemaker ganglion (Fig. 2) was isolated from the myocardium in much the same manner as that described by Hagiwara & Bullock (1957), except that fine forceps and a pair of small ophthalmic scissors were found more suitable for the fine dissection. Small bits of muscle were left attached to the ends of the ganglion for mounting purposes, but there were no connectives between the rostral and caudal ends other than through the ganglion trunk. The excised ganglion was mounted either on a flat glass plate, in which case chilled perfusion solution flowed over the preparation, or on a glass box through which chilled water was circulated. Cole’s lobster fluid (Cole, 1941) was used for perfusion, and in both cases the bath temperature was maintained at 10 °C. In favourable preparations burst discharge was maintained for 6-8 h.
With the use of a darkfield condenser and polarized light source, the five large cells in the rostral portion of the ganglion were generally visible, allowing penetration of the soma membranes to be made under visual control. Figure 2 shows the usual positions of the cells in the ganglion and their numeral designation.
Glass capillaries pulled to a tip diameter of 0·5 µ or less, and filled with 2·5 M-KCL were used for recording intracellular potentials. Electrode resistance ranged from 20 to 60 MΩ. In experiments where the membrane potential of a cell was altered by internally applied current, two separate electrodes were inserted into the cell, usually from different directions.
The recording amplifiers were electrometer-tube input types with input capacitance neutralization. A Precision Instrument Type 6104 tape-recorder was used for storing data. The 10 to 1 or 100 to 1 record-playback speed ratio of this machine made it possible to write out data with a pen recorder (Brush Inst., Mark 280) while maintaining an effective high-frequency response of at least 8 kc. Except for the cases where action potentials or the brief electronic potential spread therefrom were observed, the actual requirements on system bandwidth were much less than this 8 kc.
TERMINOLOGY
‘Burst’ (internally recorded) : a partial depolarization of the cell membrane lasting several hundred milliseconds, with a number of more rapid positive-going deflexions superimposed.
‘Burst period’: the time between the onset of the initial potential deflexion of a burst and the time at which the intracellular potential first returns to the value from which the initial deflexion started.
‘Interburst period’ : the time between the end of one burst period and the beginning of the next.
RESULTS
A. Activity of cells in the intact ganglion
Figure 3 a and the upper trace of Fig. 3 b show typical potential waveforms taken from the large-cell somata of the cardiac pacemaker ganglion in Homarus. Over 150 different soma records were examined in this series of experiments. The maximum value of soma membrane depolarization during the burst period, over all experimental records, was 35 mV. Generally the peak values of the burst potential ranged from 20 to 25mV. The membrane resting potential of cells in good condition was between − 50 and −60 mV. These experimental values are in good agreement with the burstpotential amplitude and resting-potential magnitude recorded from the large-cell somata of Panulirus (Hagiwara & Bullock, 1957, and personal observations). The pattern of the individual potential deflexions occurring during a burst period was very similar from burst to burst in a given cell, with a great deal of similarity in the features of simultaneous bursts recorded from different cells (Fig. 3 a). Burst onset in cell 5 generally occurred 3-4 msec, earlier than in the other large cells.
In contrast to the soma wave-forms recorded from the Panuliris ganglion, attenuated spike potentials were never seen superimposed on the slow burst potentials in ganglion-cell somata of Homarus. Spike potentials, however, became quite prominent in non-soma intracellular recordings. There was considerable uncertainty as to the exact nature of the penetrated region in these instances as only the elliptical shape of the cell bodies was discernible in the darkfield illumination. In these situations it was necessary to rely upon the d.c. resting-potential shift and large-amplitude potential deflexions during bursts as evidence that a neural membrane had been penetrated. Numerous instances of overshooting action potentials were observed in non-soma (presumably axon) records. The recording electrode tip in most of these cases was 200 µ or more distant from the nearest cell body. The amplitude of these action potentials ranged from 60 to 70 mV. The number of spikes fired by an axon during a burst seldom varied by more than 20% about the mean value, but between different fibres this number varied greatly, the observed range being 1−30. The burst of spikes of Fig. 3 b (lower trace) was recorded from a location near the cell-3 soma. A cell-4 soma record, taken simultaneously, is also shown in the figure. Other intracellular recordings of the large-amplitude spikes were taken from all regions in the rostral part of the ganglion.
The presence of the small-amplitude potential fluctuations at the end of the bursts in the lower trace of Fig. 3b indicates that the recording electrode was within a process of a large cell since this was the nature of wave-forms recorded from the somata of these cells. Further evidence of this is that the spike amplitude remained constant for approximately 40 min., which suggests that the recording electrode had penetrated a large process, such as an axon of a large cell. Owing to the size of the small cells and to the lack of success in recording from the somata of these cells, it is unlikely that any of the fibres from which records were taken were processes of the small cells.
The ganglion cells of Homarus displayed syncytial characteristics similar to the cells of Panulirus (Hagiwara et al. 1959; Watanabe & Bullock, 1960). That is, a certain percentage of current of either polarity, passed into one of the cells, flowed out to the external medium via the remaining cells. Measurements were made by inserting two electrodes into one cell and a third electrode into a second cell. For adjacent cells in Homarus, the potential shift in the unstimulated cell was generally between 5 and 10 % of the shift in the stimulated cell for direct or slowly alternating current. Soma-to-soma transmission of a.c. dropped to a quarter of the d.c. value at approximately 30 c/sec.
In many, but not all, of the ganglion cells studied the membrane potential underwent a slow positive-going shift during the interburst interval (see Fig. 3 a). The total excursion of this potential shift was never over 6 mV. in any of the cells studied and in most cases was under 2 mV. This potential variation was found to have the following characteristics.
The shift had the highest positive slope in the initial part of the interburst interval (Fig. 4 a). In preparations which displayed relatively long interburst intervals (4−6 times the burst duration), the slope became vanishingly small towards the end of the interval.
Hyperpolarizing the soma membrane caused the shift to disappear from the soma wave-form (Fig. 4 c, upper trace). The exact value of hyperpolarization necessary varied from cell to cell but was generally less than 20 mV. Greater values of hyperpolarization caused a reversal to occur; that is, the soma membrane potential was more positive in the initial portion of the interburst interval than at the terminal portion. This situation is illustrated in Fig. 4b, where the end portions of three bursts and the following interburst intervals are superimposed. The membrane potential towards the end of the interval has been made common in the figure and the initial points are all at the same potential difference from this reference. Curve a of Fig. 4b was taken with the cell at normal resting potential. For curve b the cell soma was hyperpolarized by 20 mV and in c by 40 mV.
Abolishing the interburst shift in the soma record of one of the cells by the application of a mild hyperpolarizing current had little or no observable effect on a simultaneously observed shift in unstimulated cells (Fig. 4c). Strong hyperpolarizing currents tended to reduce the interburst shift in the unstimulated cells, but this was presumably due to hyperpolarization of these cells by current flowing from the stimulated cell through the syncytial pathways.
Depolarizing the soma membrane by 10 or 15mV. caused a noticeable inter-burst shift to appear in cells which displayed no shift at the normal resting potential.
Replacing the normal bathing solution for a short period by one deficient in K+ caused the shift to be enhanced by 1 or 2 mV.
At a given point of observation in a cell the most negative value which the inter-burst potential attained was equal to the ‘reset’ potential following spontaneous inter-burst spikes. This is demonstrated by the axon record of Fig. 4d. The fibre from which this record was taken tended to fire two or three spikes during the interburst period in addition to six spikes during the burst period. The peak amplitude of the spikes, 55 mV., indicated that the recording electrode tip was near to a locus of spike generation but that action potentials probably did not invade the axon segment which contained the recording electrode. The potential level following each interburst spike was identical to the most negative potential level following the burst period, although this level was much more slowly attained in the latter case. It can also be seen that the positive-going shift proceeded much more slowly after the burst than after the single spikes.
The reversal potential for the interburst potential shift, its enhancement by depolarization and low external potassium, and the equality of the reset potentials following single spikes and the burst period suggest a potassium conductance change, at some cellular region, from high to low during the interburst period. Further discussion will be given at a later point.
B. Activity in sectioned preparations
Figure 5 a shows the soma waveforms of two large cells (numbers 3 and 4) in an intact Homaros ganglion. Figure 5b shows the activity pattern in these same two cells after the caudal portion of the ganglion containing the four small-cell bodies and their immediate processes had been cut away. The records of Fig. 5 c show activity, again, in these same two cells, after a cut had been made between the somata of cells 4 and 5, isolating a smaller rostral portion of the ganglion. Due to the nearness of the second cut to cell 4, the electrode placed in this cell body was dislodged during the cutting procedure and had to be re-inserted, which caused some loss in amplitude of the observed potentials. The results shown in Fig. 5 are typical of ten experiments in which the primary aim was to determine the exact nature of the intracellular activity remaining after the four small-cell bodies and their immediate processes had been cut away. In none of these cases did the preparation fail to give repetitive burst activity similar to that shown. Stable burst activity was maintained by these sectioned preparations for several hours; indeed, the sectioned preparations containing all of the large-cell somata were nearly as viable as intact ones.
As a check for possible differences in effect of crushing, anaesthetizing, and removing the small cells, several specimens of Panulirus interruptus from California were obtained and studied. It was found that sectioning the ganglion between the fifth large cell and the first small cell completely stopped normal burst discharge in the large cells of this preparation. Only tonic firing like that of Fig. 6 occurred after the cut. This discharge often persisted for 30−60 min. after the cut was made. The resting potential of these isolated cells remained at approximately its normal value of − 50 to − 60 mV. for several hours after the cut. The oscillatory resting potential and superimposed bursts of spikes reported by Watanabe (1958) did not occur in the cells studied. The effects of anaesthetizing the small cells rather than cutting them off may therefore have been slightly different. However, this difference is not significant in the present investigation as the intent was to determine whether or not normal burst activity remained after sectioning, which it did not. All data described hereafter in this report were taken from the Homarus cardiac ganglion.
A partial cut across the ganglion trunk between two large-cell bodies, which left only a shred of sheath connecting the two segments, caused the ganglion to exhibit independent spontaneous rhythms in the two segments. The extent of isolation of the two segments was checked whenever possible by fixing the end of one of the segments and completing the cut. If wave-form changes were not observed in the cells of one of the segments, isolation was considered to have been effective. In all cases observed the partial cut described above produced effective isolation. Figure 7 a shows activity recorded from cell bodies on either side of such a cut between cell bodies 3 and 4 (see inset). Cell 5, still attached to the small-cell group, displayed irregular bursts, some of which were unusually long (Fig.7a, upper trace). In the rostral section of this preparation, which contained only the three most rostral cell bodies, very uniform burst activity was present (Fig. 7a and b, lower traces); but the burst repetition rate was low. The record of Fig. 7b, upper trace, was taken from the soma of cell 4 after a second cut had been made across the trunk, in this case just caudal to the cell-5 soma. This record, then, demonstrates the spontaneous activity in a ganglion section containing two large-cell somata. The bursts, although short, occured very regularly and were quite uniform.
Figure 8 and 9 a show spontaneous burst activity of two different preparations which consisted of only the bifurcation area and one of the upper branches, and which contained only two cell bodies. Burst activity was routinely observed in these preparations for 3−4 hrs.
The cells suffered varying losses in peak amplitude of the burst wave-forms after sectioning (Fig. 5, for example). It is not possible to set exact figures for the amplitude loss because in making the cuts the tissue surrounding the penetrated cells was unavoidably moved, slightly re-positioning the electrodes in their respective cells. Occasionally the electrode in the cell near the cut was dislodged, and re-insertion introduced indeterminate variations. It also appeared that there was loss in resting potential magnitude due to sectioning, as typical values observed lay between −40 and −55 mV.
Generally the burst rate increased, often as much as doubled, and the burst duration decreased with the removal of the length of tissue containing the small-cell bodies. As cell bodies were taken away beyond this point, the burst rate decreased and the burst duration showed further decrease. A tabulation of burst durations and interburst durations of preparations observed both in the intact and sectioned states is given in Table 1.
The brief low-amplitude potential deflexions which occurred in the interburst interval in some sectioned preparations (see Fig. 5 b, c) were most probably due to injury from sectioning, and in many cases tended to die out after a short time. In others, especially in preparations of only two or three cell bodies, these interburst deflexions persisted for hours. Although in a number of instances this activity immediately preceded the bursts, no definite relationship between the two will be inferred.
A number of observations were made on sections of the ganglion containing only one cell body. A steady resting potential of − 50 to − 60 mV. was maintained, but there was no spontaneous burst activity. Any low-amplitude tonic firing in these cells tended to die out within a few minutes after the cut was made.
The rostral sections of the Homarus ganglion, because of their simpler nature, proved to be most useful for the study of burst activity. Furthermore, it can be expected that the findings concerning burst activity in these sectioned preparations carry over to burst activity in the intact preparations for the following reasons. First, burst activity was maintained in these cells for a number of hours. Secondly, the activity during these bursts lasted from 100 to 400 msec, and consisted of a number of events. Thirdly, the amplitude of potential deflexions in the cell bodies was nearly the same as that in the cell bodies of intact ganglia. Fourthly, the temporal characteristics of the initial rise of the burst and of the potential variations during the burst period were the same in cells of intact ganglia and of sectioned ganglia. Fifthly, the high degree of cell-to-cell synchrony between potential deflexions was still present. Sixthly, a positivegoing potential shift was still present in the soma wave-forms during the interburst period. In other words, no new burst-formation mechanism, such as the slow oscillatory waveform described by Watanabe, became evident when the caudal portion of the ganglion was removed.
Burst activity observed in the two-cell, bifurcation upper arm (or simply upper arm) preparations (Figs. 8,9) was consistent from preparation to preparation in two respects, although there were wide differences in interburst intervals (Table 1), and the cells displayed differing tendencies toward firing in the interburst interval. First, there was always a small deflexion (pre-potential) in the potential wave forms of at least one of the two somata, which immediately preceded the initial burst deflexion. The prepotential took one of several shapes: (1) fast onset and slow decay (Fig. 8, lower arrow); (2) fast onset and a maintained plateau for a few tens of milliseconds (not shown here); and (3) a ramp-type rise (Fig. 8, upper arrow). But in all cases the membrane potential remained lowered between the pre-potential onset and the initial rise of the burst. A pre-potential was frequently, but not always, observed in cell wave-forms in sectioned preparations which contained more than two cell bodies.
Secondly, the membrane potential during the interburst interval displayed the following characteristic. The slow positive-going potential shift occurred in the initial portion of the interburst interval. During the remainder of the interval the potential formed a plateau (interburst potential plateau in the following discussion). This pattern is most easily seen in Fig. 9 a, where an arrow marks the transition region between the shift and plateau. This behaviour also occurred in the waveforms of Fig. 10, but the transition is not seen in these cases owing to the expanded time-scale. In many of the preparations studied the transition was not well demarcated, or the interburst potential shift was not smooth. Consequently, the localization of the transition region became a matter of interpretation. The time-course of the potential shift period for a given cell was nearly identical from burst to burst and was consistent at around 1 sec. in nearly all cells of the upper-arm preparations studied. The large variations in total interburst interval observed in these preparations (see Table 1) were accounted for by variations in the duration of the plateau period, and not in the duration of the shift period. Also, interburst firing by the cells took place principally during the plateau period, but in some cases during both the plateau period and the latter portion of the potential shift period. Such firing never occurred during the first 500-600 msec, of the shift period. The plateau phase of the interburst potential was never observed in cells of intact ganglia, although in some instances (Fig. 4 a, for example), it seemed that the membrane potential might have reached a plateau had not the next burst occurred.
C. Stimulus-induced activity
Response to depolarizing current pulses
Depolarizing current pulses passed into the somata of the large cells during the period of the interburst potential shift in upper-arm preparations, or correspondingly during the total interburst period in intact preparations, caused only minor alteration of the spontaneous pattern unless the depolarization produced by the pulse was above 30mV · and the pulse duration was greater than 25−30 msec. However, if a pulse was applied to a cell of an upper-arm preparation during the plateau period, a burst followed readily, and the build-up of the burst-period depolarization was indicative of activity endogenous to the large cells and not of synaptic excitation. The following examples are typical of twenty-five experiments carried out on upperarm preparations.
Figure 9b shows the differing effects of two identical depolarizing current pulses to the soma of cell 1 of upper-arm preparation. The spontaneous activity pattern is given in Fig. 9 a. The maximum depolarization caused by the stimulus current was 10 mV · and the pulse duration was 2 msec. The first pulse was applied before the interburst interval had run its time-course and no deflexions followed it. The small foreshortening of the first interburst interval over the displayed spontaneous intervals cannot be ascribed with certainty to the injection of the current pulse, as this interval fell just within the range of variation of the spontaneous interburst intervals. The second pulse was applied approximately 150 msec, after the interburst potential had attained its plateau state (arrow), and was followed by events which culminated in a burst. In this case the interval between the bursts was considerably shorter than the spontaneous interburst intervals. Also, the cell remained partially depolarized during the total time between stimulus and burst onset. Therefore, it can safely be said that the stimulus triggered the burst.
The behaviour of the potential in the period immediately following the stimulus pulses, and the dependence of this behaviour upon the cellular state (as evidenced by the interburst potential) is shown in Figs. 10 and 11. For these records simultaneous electrical events in both somata of a second upper-arm preparation are shown. The duration of the stimulus in these figures was 14 msec., and the depolarization of cell 1 soma due to the stimulus current was i8mV. The pulse of Fig. 10 b was applied during an interburst plateau period approximately 200 msec, later than was the pulse of Fig. 10a, and the burst followed the stimulus more rapidly in Fig. 10b than in Fig. 10a. In both cases the cell I membrane potential began to decay towards the resting level immediately after the stimulus termination. Also, in both cases this decay was arrested approximately 10 mV. positive to resting potential. In Fig. 10a a 50 msec, period when cell I underwent a slight repolarization followed this arrest. The initial deflexion of the burst wave-form then set in. In Fig. 10b the burst depolarization set in immediately after the point of arrest, and the initial rise was much more gradual than that in Fig. 10 a. The elapsed time between the potential arrest at 10 mV. and the following burst peak was 65 msec, in Fig. 10a, and 40 msec, in Fig. 10b. The value of this elapsed time decreased with increase in the length of time for which the membrane potential had been in its plateau state. For this preparation, at the above stimulus strength and duration, the elapsed time ranged between 15 and 70 msec. In all cases, though, the next spontaneous burst would not have been expected for several hundred milliseconds.
During the time immediately following the stimulus cut-off, the membrane potential of the cell-3 soma (Fig. 10, lower traces) exhibited a relatively slow depolarization which led smoothly into the more rapid rise of the initial burst deflexion. The direct depolarization of the cell-3 soma, due to the stimulus current passed into the cell-i soma, was very small as expected from the low-pass filter characteristics of soma-to-soma transmission described earlier. Therefore, the depolarization observed in cell 3 may be considered to be due entirely to activity set off by the pulse, but not to electro Ionic effects from it. The rapidly rising phase of the depolarization set in approximately 20 msec, later in Fig. 10a than in Fig. 10b, and occurred at almost the same time as the intial post-stimulus rise in cell I.
The records of Fig. 11 a were taken from the same preparations as those of Fig. 10. The stimulus pulses, identical to those of Fig. 10, were applied while the cells were still undergoing the interburst shift. The responses to three pulses have been superimposed, with the stimulus onset taken as the time reference. No bursts followed the pulses, nor were there sizeable single deflexions; however, the events which did follow the pulses displayed a continuity of form with those of Fig. 10. The curves labelled a resulted when the stimulus pulse was applied 200 msec, after the termination of the preceding burst, or in other words, very early during the period of interburst potential shift. This was the smallest of all responses of these cells to pulses of this strength and duration. Curves b and c resulted when stimulus pulses were applied 500 and 700 msec, respectively after the preceding bursts. In b and c the membrane potential of cell 1 (upper traces) showed a marked change in slope at approximately 10 mV. above the resting potential and remained partially depolarized for over 100/msec. The depolarization in cell 3 was greatest in c (Fig. 11 a, lower), and least in a. Subtracting the cell-I smallest response (Fig. 11a, curve a) from each cell-I waveform in Figs. 10 and 11a, and reproducing the cell-3 wave-form directly gave the wave-form comparison of Fig. 11 b. This procedure was carried out in order to eliminate as far as possible the deflexion in the cell-I wave-form which resulted directly from the stimulus current. In Fig. 11b both the cell-3 and the cell-I wave-forms had the following common features. First, the depolarization of the cells increased rapidly after the initial rising phase in the cases where a burst arose (curves, b’, Fig. 11 b), but slowly increased or decreased in the others (c’, d’). The initial rising phase was nearly identical in all cases. Secondly, in the cases where the burst did not arise, the somata of both cells remained partially depolarized for a long period.
Response to hyperpolarization
Hyperpolarizing the soma membrane of one cell in the upper-arm preparations by 10 or 15 mV. generally interrupted ongoing sponataneous burst discharge. Potential shifts of these amounts imposed in the large cells of the intact ganglion produced only minor lowering of the burst repetition rate. Most notable in the upper-arm preparations was the tendency for a burst to arise at the termination of a pulse of hyperpolarizing current of sufficient strength and duration. Association of the hyperpolarization removal with the following burst was especially clear in cases where the period between spontaneous bursts was 3 to 4 sec. These situations allowed the stimulus pulse to be inserted well in advance of the next expected burst. Again it was necessary that the pulse be applied during the potential plateau period.
The average interval between spontaneous bursts for the preparation, from which the record of Fig. 12 was made, was 4 sec. The first pulse failed to trigger a burst; however, low-amplitude depolarization occurred in the unstimulated cell after the hyperpolarization release. This low-amplitude depolarization was identical to that preceding the burst triggered by the second pulse. The termination of a sufficiently strong hyperpolarizing pulse was occasionally followed by two separate bursts in quick succession.
Response to external stimulus
Burst activity in the sectioned, as well as in the intact preparations, was readily set off by current pulses from wire electrodes placed in contact with the ganglion sheath.
In the studies of the upper-arm preparation described here the stimulator wires, spaced approximately 0·5 mm. apart, were placed on the sheath close to midway between the two cell bodies. The cell bodies were at least 2·4 mm. apart in all preparations used. The effects of the stimulus current would then be felt primarily by the cell axons and other processes distal to the soma. Activity could be evoked by brief external pulses even when the cells were undergoing the potential shift following the preceding burst. During the latter portion of the shift period or during the plateau period, bursts could be triggered in this manner. The wave-forms were very similar to those of spontaneous bursts. A single, low-amplitude deflexion in the soma potential wave-form tended to follow a pulse administered during the earlier portions of the shift period.
It was possible to block propagation, for a time, to one side or the other of the stimulus electrodes; that is, increasing the stimulus pulse duration beyond a critical value caused propagation to be blocked at the anodal wire. This caused the burst depolarization in the cell nearer the anode to come later than in the cell nearer the cathode and to take on a considerably different shape. There was very little alteration in the activity of the cell on the cathodal side with pulse duration changes. Figure 13 demonstrates three stages of anodal blocking in an upper-arm preparation. All of the pulses shown were applied during periods of interburst potential plateau. Had all three pulses been of the same duration, the burst wave-forms in the respective cells would have been identical in a, b, and c; i.e. recovery effects were not responsible for waveform differences. The responses of the cells to a stimulus of 0·5 msec, duration is shown in part a of the figure. Identical wave-forms were triggered by suprathreshold stimuli of shorter duration. The wave-forms displayed a rapid initial depolarization occurring at nearly the same time in both cells, as was the case in the onset of spontaneous bursts (see Fig. 8). The initial depolarization in the cell-I waveform built up much more slowly following a 0·7 msec, stimulus pulse (Fig. 13b, lower trace). The cell-3 wave-form was relatively unchanged from a, however. Only a small long-lasting depolarization of cell I followed a 0·8 msec, pulse (Fig. 13 c, lower trace). The cell-3 wave-form was again relatively unchanged. The cell-I depolarization, although considerably smaller than that following the briefer pulses (a and b), was too large to be accounted for by current flow from the cell-3 soma via syncytial pathways. Switching the stimulator output so that the anode was nearer the cell-3 soma caused the initial deflexion of the cell-3 wave-form to be delayed or blocked while the cell-I wave-form remained consistent.
DISCUSSION
The characteristics of the interburst potential shift were quite constant whether the cell under observation was in an intact ganglion or in one or the upper-arm preparations (Fig. 6 a and Fig. 9a). In this latter case the potential-shift period became separated from the following burst period by several hundred milliseconds in many cases (see Fig. 9a), and by as much as several seconds in some of the more irregularly firing preparations. Therefore, this potential shift would not seem to be the result of an excitatory process as such (i.e. a sodium conductance increase), since the membrane potential remained roughly constant between the times when the interburst potential reached its most positive level and when the first major excitatory event occurred (the burst initiation). Also, in the upper-arm preparations, the excitability of the cells was shown to be dependent upon the state of the membrane potential in the interburst period; i.e. the cells were easily excited during the plateau period (Fig. 9b, Fig. 10), and decreasingly refractory during the potential-shift period (Fig. 11 a). The depolarization of the soma membrane necessary to trigger even a single deflexion during the potential-shift period was generally in excess of 30 mV. Comparable values of intracellular stimulus were also required to trigger single deflexions in cells of intact ganglia. This would be expected, since in intact preparations the potential shift normally occupied the total interburst interval. It is unlikely that the difference in excitability between the shift and plateau periods is due primarily to the slight elevation in membrane potential during the shift period. Therefore, I consider the interburst potential shift to be the electrical correlate of an inhibitory process which follows the burst-period activity and which weakens as the interburst period progresses. This is in good agreement with the data presented at the end of section A, indicating a potassium-conductance mechanism. First, a high-potassium conductance somewhere in the cell would tend to hold the membrane potential near the potassium equilibrium potential and, with a conductance decrease, the membrane potential would shift in a positive direction. Secondly, during the time the conductance was high, the cell would be in a state of lowered excitability. Only when the conductance had decreased to a low value, the electrical correlate being the plateau period, could a burst be set off by weak transmembrane stimulus at the soma.
The data of Figs. 10−12 have demonstrated the nature of the build-up of depolarization in the large cells following transmembrane stimulus at the soma. In the cases where a burst followed the stimulus, the depolarization proceeded smoothly to its peak value in a time course which could vary by several tens of milliseconds for a given cell. In all cases the nature of activity that followed the stimulus was dependent upon the state of the stimulated cell, as indicated by the interburst potential pattern. I am therefore led to conclude that the depolarization of the cells shown in these figures was due to endogenous activation and not to synaptic activity. The conversion of the initial burst depolarization from a rapid event (Fig. 13 a) to one of a much slower time-course (Fig. 13b), and subsequently to a low-amplitude disturbance (Fig. 13 c) by external pulses of slightly varying durations indicates that the intial burst-period deflexion is also due to endogenous activity since (1) there is no reason to suppose that a variation of 0· 1 or 0·2 msec, in the stimulus pulse duration could cause a change in the mode of activation (from synaptic to endogenous), and (2) the rise of the initial depolarization in Fig. 13 a and b would require the unsupported assumption of a large number of summating EPSP’s or that the rise time of the initial EPSP underwent a change from a few milliseconds to a few tens of milliseconds. The cell-I wave-forms of Fig. 10 also showed this variation of rise time. I am therefore led to reject the hypothesis that the deflexions observed in the soma wave-forms of the large cells in the cardiac ganglion of Homarus are EPSP’s, as was the conclusion drawn concerning the large cells of the Panulirus cardiac ganglion (Hagiwara & Bullock, 1957). My conclusion follows both from the data cited above, which indicate that activity of an endogenous nature can produce the depolarization of the burst period, and from the failure to observe in the Homarus preparation two of the characteristics which were basic to the development of the EPSP model for the Panulirus ganglion: (1) the small spike potentials superimposed upon the slower deflexions of the burst waveform (for example, cf. Figs, 1 and 3), and (2) the complete alteration of the potential pattern in the large cells after the removal of the small-cell group (see Fig. 5). Instead, I consider the rapid potential deflexions of the spontaneous burst wave-forms of the large-cell somata to be the effects of endogenous action potentials which are initiated in the large-cell axons at some distance from the soma and are propagated toward the soma (and also away from it), but do not invade it. A discussion of soma potentials produced by noninvading, antidromic action potentials in Aplysia neurones, which has some bearing on the occurrences described here, has been given by Tauc & Hughes (1963). Where a spontaneous burst arises, or when a burst is evoked by brief external stimulus (Fig. 13 a), the initial action potential (or possibly potentials) creates a rapid initial deflexion in the soma wave-form. Transmembrane current applied at the soma triggers a graded response which may or may not build into the large deflexions of the burst period (Figs. 9−12)
It is to be emphasized that the idea of a primary pacemaker which initiates and possibly maintains burst activity in the large cells in most circumstances is not rejected. It is the manner in which the large cells are driven and the events which comprise the soma wave-form during the burst period which are different from the model for Panulirus.
It is obvious from the studies of the rostral sections that the capability for pacemaking activity is not uniquely associated with the region in which the small-cell somata are located but rather exists in almost all parts of the Homarus cardiac ganglion, except possibly the isolated upper arms. In the intact preparation the pacemaking excitation may indeed come from this caudal region; but in the cases where this region is eliminated, another pacemaking region becomes effective. The data at hand are insufficient to describe the pacemaking excitation process or to postulate exactly where it arises, except that the large cells must be excited through their axons at locations distal to the somata. The tendency of depolarization of nearly equal amplitude to occur more or less simultaneously in the two cells of upper-arm preparations when only one of them was internally stimulated (see Figs. 10 and 11) would suggest that the excitation which produced the initial part of the depolarization in these instances occurred at a region which communicated electrically with both cells. Possibly this excitation occurred at one of the neuropiles formed by collaterals from the major cell processes (described by Alexandrowicz, 1932).
It is noted in this regard that Watanabe and his colleagues reported simultaneous build-up of excitation in the pacemaking neurones of the pacemaker ganglion of the shrimp heart (Watanabe, Obara & Akiyama, 1967). In this preparation the build-up could be shown in the case of spontaneous bursts, whereas in the lobster it was clearly evident only when the burst was triggered by electrical stimulus.
ACKNOWLEDGEMENT
I wish to express appreciation to Professor Richard W. Jones, Northwestern University, Evanston, Illinois, for advice and encouragement during the course of this work, and to Dr James R. Wennemark, University of Tennessee Medical School, Memphis, Tennessee, for initial help in developing experimental techniques.
This work was supported by Predoctoral Fellowship Award 4-F1-GM-24, 940-03 from the United States Public Health Service and by the Northwestern University Biomedical Engineering Center.