J. Physiol. (1975), 251, pp. 409-426 With 11 text-figure8 Printed in Great Britain
409
SYNAPTIC TRANSFER AT A VERTEBRATE CENTRAL NERVOUS SYSTEM SYNAPSE
BY A. R. MARTIN AND G. L. RINGHAM* From the Department of Physiology, University of Colorado Medical Center, Denver, Colorado 80220 U.S.A.
(Received 17 January 1975) SUMMARY
1. The relation between presynaptic depolarization and transmitter release was examined at a synapse between a Mifller axon and a lateral interneurone in the spinal cord of the lamprey. Two micro-electrodes, one for passing current and the other for recording the resulting voltage change, were placed in the presynaptic axon; a single electrode for recording the post-synaptic potential produced by release of transmitter was placed in the post-synaptic cell. 2. When action potentials were blocked with tetrodotoxin, brief depolarizing pulses in the presynaptic fibre were as effective as the action potential had been in producing transmitter release. 3. The release process had an apparent threshold depolarization of 40-50 mV and saturated- at presynaptic depolarizations of the order of 100 mV. Increasing the duration of the presynaptic pulse increased the maximum level of release. 4. Displacing the presynaptic voltage recording electrode from the position of synaptic contact toward the current passing electrode increased the apparent depolarization required to produce a given level of transmitter release. This shift in the input-output relation was consistent in magnitude with the voltage attenuation between the presynaptic recording electrode and the synapse expected from the space constant of the fibre. 5. The effect of conditioning hyperpolarization and depolarization of the presynaptic fibre on subsequent transmitter release by brief depolarizing pulses was examined. No effect was observed when the presynaptic recording electrode was in the region of synaptic contact. When the presynaptic electrode was not so positioned, conditioning effects were observed which depended on electrode position and could be attributed to changes * Present address: Department of Physiology, University of Utah College of Medicine, Salt Lake City, Utah 84132, U.S.A.
410 A. R. MARTIN AND G. L. RINGHAM in the space constant of the presynaptic fibre. No conditioning effects were observed on transmitter release by the action potential. INTRODUCTION
During the normal process of chemical synaptic transmission an action potential in the presynaptic nerve causes release of transmitter on to the post-synaptic cell, producing a conductance change in the post-synaptic membrane. The release of transmitter is mediated by an influx of Ca ions into the presynaptic nerve as a result of the depolarization associated with the arrival of the action potential (see Katz, 1969). Studies of the relation between presynaptic depolarization and transmitter release are normally limited by the presence of an all-or-nothing action potential in the presynaptic nerve. However, the action potential may be blocked with tetrodotoxin without interfering with the transmitter release process (Furukawa, Sasaoka & Hosoya, 1959), thus enabling graded depolarizations of the presynaptic nerve. Studies of this kind have been carried out on the neuromuscular junction of the frog (Katz & Miledi, 1967 a, b) and on the giant synapse of the squid (Bloedel, Gage, Llinas & Quastel, 1966; Katz & Miledi, 1967 c; Kusano, 1968), but only at the squid synapse was it possible to record the magnitude of the presynaptic potential changes. In the experiments to be reported here, studies of the relation between presynaptic polarization and transmitter release have been carried out at the synapse described in the preceding paper (Ringham, 1975), occurring in the lamprey spinal cord between a Mifller axon and a lateral interneurone. Apart from extending previous observations on the squid to a vertebrate central nervous system synapse, the preparation has the advantage that the presynaptic element is a continuous cylindrical axon which forms a discrete synaptic contact less than 100 I=m in extent, probably with a single short dendrite extending from the post-synaptic cell. This is in contrast with the somewhat more complicated geometry of the squid synapse (Hama, 1962) in which the presynaptic terminal is of the order of 1 mm in extent, and consequently, may be subject to non-uniform depolarization because of the electrotonic decrement along its length. A preliminary report of some of the present observations has been published previously (Martin & Ringham, 1974). METHODS The method of preparing and mounting the lamprey spinal cord was as described in the previous paper (Ringham, 1975). Two micro-electrodes were placed in the presynaptic fibre, one for passing depolarizing currents and the other for recording the presynaptic potential changes. A third electrode was placed in the lateral interneurone to record the excitatory post-synaptic potential (e.p.s.p.). In the initial experiments the presynaptic recording electrode was placed as close as possible to
SYNAPTIC TRANSFER
411
the region of synaptic contact on the fibre, which was medial and usually within 150 ,um rostral to the lateral cell. The presynaptic stimulating electrode was then placed a short distance (usually within 300 ,um) away. However, the voltage electrode sometimes produced damage in the presynaptic region, resulting in partial or complete loss of the e.p.s.p. In later experiments the electrodes were arranged so that the synapse was on one side of the presynaptic current electrode and the voltage recording electrode an equal distance on the other side. As the decrement in presynaptic depolarization along the cylindrical fibre should have been the same in either direction, the voltage electrode was assumed to record the same potential changes as were occurring at the synapse. For the experiments in which the effects of conditioning pulses were examined, the conditioning pulses and test depolarizations were applied to the same presynaptic current electrode. Tetrodotoxin (Sankyo) was added to all bathing solutions at a concentration of 0-2 fig/ml. RESULTS
After a lateral interneurone had been impaled and current and voltage electrodes placed in the appropriate Miller fibre, tetrodotoxin was added to the perfusing fluid to block initiation of propagated action potentials. Action potentials in the fibre, evoked by brief depolarizing pulses, were not blocked immediately and their amplitudes, together with those of the corresponding e.p.s.p.s in the cell, could be monitored as block progressed. After block was complete, e.p.s.p.s could then be produced by presynaptic depolarization without the intervention of action potentials. One such sequence of records is shown in Fig. 1. In each frame the upper record is from the presynaptic fibre, the lower from the interneurone. The left column shows responses obtained during the development of tetrodotoxin block. In the first frame a presynaptic action potential of about 85 mV amplitude produced an e.p.s.p. in the cell of about 2 mV. As the action potential failed (second and third frames) the e.p.s.p. also diminished in size. In the right column, obtained after the block was complete, depolarizing pulses 2 msec in duration and similar in amplitude to the failing action potentials on the left produced similar e.p.s.p.s. Thus the artificial pulses were similar in efficacy to the action potentials in producing transmitter release at the synapse. The relation between e.p.s.p. amplitude and presynaptic depolarization, obtained from the experiment described above, is shown in Fig. 2. With both the action potentials (curve A) and the electrotonic pulses (P) there was an apparent 'threshold' for production of a detectable e.p.s.p. of about 50 mV, and after about 90 mV presynaptic depolarization no further increase in e.p.s.p. amplitude occurred. The early part of each curve is linear on a semilogarithmic plot (not shown) with a tenfold increase in e.p.s.p. amplitude being produced by a 24 mV increment in presynaptic depolarization. The fact that the electrotonic pulses were slightly less effective in producing transmitter release than were the failing action
412 A. R. MARTIN AND G. L. RINGHAM potentials could suggest that the presynaptic recording electrodes were positioned a short distance from the synapse, toward the current passing electrode. The recorded presynaptic depolarization would then be somewhat larger than that actually occurring at the synapse. However, if this were the case, the slope of the curve (P) should have been less than that obtained with the action potentials and the plateau level the same. A more likely explanation can be obtained by examining Fig. 1 more closely and noting that the depolarizing pulses were slightly shorter in duration than the action potentials. Thus they would be expected to release less
transmitter. I
i
U'
VpreI
4
2
S 'I~~
__
__
a
3
6 %ftam _wrq~_
V"~
, r
]2mV
~10msec Fig. 1. Intracellular records from presynaptic axon (upper trace) and postsynaptic cell (lower trace) in lamprey spinal cord. First column: failure of presynaptic action potential after addition of tetrodotoxin 02 ,ug/ml. to the bathing solution. First action potential is normal. Failure during second and third frames is accompanied by reduction in amplitude of post-synaptic potential (e.p.s.p.). Second column: e.p.s.p.s produced by artificial presynaptic depolarizing pulses 2 msec in duration after tetrodotoxin block. Responses are similar to those produced by failing action potentials. Tetrodotoxin was present in experiments illustrated by all subsequent Figures except Fig. 11.
The effect of pulse duration on transmitter release is illustrated in Fig. 3. E.p.s.p.s produced by pulses of 4 msec duration are shown in Fig. 3A. In Fig. 3B, synaptic transfer curves are shown for a different experiment in which pulse durations of 1, 2 and 4 msec were used. There was little difference between the responses to 1 msec and 2 msec pulses except at
SYNAPTIC TRANSFER 413 large depolarizations; with 4 msec pulses the transfer curve was shifted to the left by about 15 mV and the maximum response increased by 2-3 mV. However, the maximum rate of change of e.p.s.p. amplitude with depolarization was unaltered, again being of the order of tenfold for a 25 mV increment in peak presynaptic depolarization. The dependence of release on duration is somewhat different than that reported for the neuromuscular junction of the frog (Katz & Miledi, 1967b), where the maximum effect of increasing pulse duration occurred between durations of 1 and 2 msec.
3 A
P
0
Aoe I
0
I
20
I
I
40
I
I
60
I
I
80
I
I
100
Vpre(mV) Fig. 2. Synaptic transfer curves from experiment shown in Fig. 1. Abscissa: amplitude of presynaptic depolarization; ordinate: e.p.s.p. amplitude. Curve A (@) was obtained with failing action potentials during development of tetrodotoxin block, curve P (A) with 2 msec depolarizing pulses after block was complete.
Attempts to depolarize the presynaptic axon for more than 4 msec were frustrated by the onset of delayed rectification, after which the micropipette could not pass enough current to depolarize the fibre by more than about 40 mV. However, as noted in the previous paper (Ringham, 1975), the synapse was electrically as well as chemically coupled, and prolonging the depolarization resulted in an increase in the post-synaptic response because of summation of the electrical coupling potential with the e.p.s.p. One such experiment is shown in Fig. 4, in which pulses of 50 msec duration were used. Electrical coupling to the cell can be seen with both hyperpolarizing and depolarizing pulses, and whea the initial depolarization was sufficient, e.p.s.p.s. appeared on top of the coupling potentials. The synaptic transfer curve for this experiment is shown in Fig. 5. The
A. R. MARTIN AND G. L. RINGHAM post-synaptic responses in the hyperpolarizing region and the responses in the early part of the depolarizing region were due to electrical coupling. At about 35-40 mV depolarization the e.p.s.p-. appeared and reached its maximum amplitude at about 80 mV peak depolarization. Beyond this, the response continued to increase at a rate consistent with that expected for the electrical coupling potential alone (dashed line). The contribution of the 414
A
VPre
f0
vPost
49 8>~~~~~~~
4-
50 mV
2-
-g I2 mV 5 msec
0
100
50
150
V~re(MV) Fig. 3. Effect of pulse duration on transmitter release. A, e.p.s.p.s (V,,,.) produced by presynaptic depolarizing pulses (Ipr) 4 msec in duration. B, synaptic transfer curves obtained from one preparation with presynaptic pulses, 1, (0), 2 (A) and 4 (N) msec duration. Increasing duration increases synaptic efficacy.
coupling potential to the over-all response was estimated by measuring the break in the post-synaptic potential at the end of the depolarizing pulse (Fig. 4). This was then plotted against peak depolarization in Fig. 5 (open circles) to obtain the dashed line. It should be pointed out that the coupling potential is produced by the steady-state depolarization, not the peak. Because the points are plotted against peak depolarization, the dashed line does not represent the transfer curve for the electrical coupling, but rather the contribution of the coupling to the total response.For example, the point at 94 mV peak depolarization corresponded to a steady-state presynaptic depolarization of only 37 mV, and the coupling potential contributed 2-3 mV to the total post-synaptic response of 7-7 mV. When
SYNAPTIC TRANSFER 415 the coupling potential amplitudes are plotted against steady-state depolarization (points represented by x 's, Fig. 5), it can be seen that the electrical coupling is highly rectifying, as reported previously (Ringham, 1975). 1 Vpre
4
vpost
2
5
36
- j'
5omV
_mewt---_ i_ ]2 mV 10 msec Fig. 4. Responses to presynaptic depolarizing pulses 50 msec in duration, showing both e.p.s.p. and electrical coupling between presynaptic axon and post-synaptic cell. In last three frames e.p.s.p.s are superimposed on coupling potentials. Breaks at beginning and end of pulses are due to capacitative coupling between current-passing and recording electrodes. Note delayed rectification in presynaptic fibre.
The over-all synaptic transfer curve shown in Fig. 5 represents, of course, a highly artificial situation. Normally the presynaptic depolarization is brief and the chemical and electrical responses do not sum because of synaptic delay. In Fig. 4, and in the experimental records shown in the previous figures, the rising and falling phases of the electrotonic potentials in both the cell and the fibre and the onset of the e.p.s.p.s were obscured by large capacitative artifacts due to coupling between the current passing electrode and the two recording electrodes. Because of such artifacts it was not possible to measure synaptic delay accurately. In other experiments in which e.p.s.p.s were produced by action potentials the delay was in the range of 1-3 msec at about 100 C.
Effects of electrode position The shape of the synaptic transfer curves and their position in relation to the abscissa (Vpre) is obviously dependent on the location of the
416 A. R. MARTIN AND G. L. RINGHAM presynaptic voltage recording electrode with respect to the synapse. Initially regions of synaptic contact were located by exploring with the presynaptic current electrode to find the position which gave a maximum post-synaptic response with moderate depolarization. The region of synaptic contact was almost invariably 0-150,um rostral to the position of the interneurone, and in later experiments the synapse was simply assumed to be located in the middle of this region. Thus, localization of the synapse was probably in error by less than + 1001tm. During several experiments a check was made to determine the probable error in the synaptic transfer curve due to such errors in electrode positioning. One such
+10 B E
6
0
X
2
4-
20
2
40
60
80
100
120 140 160+
~~~~Vpre(miV)
Fig. 5. Synaptic transfer curve for experiment of Fig. 4 with 50 msec presynaptic hyperpolarizing and depolarizing pulses. For points indicated by 0 and * abscissa represents peak presynaptic depolarization. *: total amplitude of post-synaptic response; 0, contribution of electrical coupling potential. Difference between these two curves is e.p.s.p. amplitude. x, coupling potential amplitude replotted with abscissa representing steadystate presynaptic depolarization. These points plus those in hyperpolarizing quadrant represent synaptic transfer curve for electrical coupling potential.
experiment is shown in Fig. 6, with the usual recording arrangement in the inset. Current was passed through electrode 2 and presynaptic potential changes recorded with electrode 1. With this arrangement, disruption of the synapse by the recording micropipette was avoided. The depolarization at the synapse, VA, is related to the recorded depolarization, V1, by VA = V1 exp - (x - d)/A, where A is the space constant of the fibre, x the distance from the current passing electrode to the synapse and d the inter-electrode distance. Thus when x = d, the recorded depolarization, V1, should equal
SYNAPTIC TRANSFER 417 the actual presynaptic depolarization, VA. Curve A of Fig. 6 was obtained in this way, using 30 msec presynaptic pulses. After curve A was obtained, the functions of the two electrodes were reversed, electrode 1 becoming the current electrode and electrode 2 the voltage electrode. This arrangement resulted in curve B. In this case the presynaptic depolarization, VB, should be related to the recorded depolarization, V2, by VB = V2 exp -x/A. If one now assumes that equal post-synaptic responses indicate equal presynaptic depolarizations, then when the post-synaptic responses are the same for the two curves, VA = VB and the displacement of the 12+ 10 8
~~~~~~Fibre
//
-80
60
40
20
~
_20
cow ~~~~~2
40
60
80
100 120 140 160+
Vpre(mV
4-
Fig. 6. Effect of electrode position. Location of presynaptic electrodes with respect to synapse shown in inset. Distance d was 257 jum; distance x assumed to be similar. Curve A obtained with electrode 2 as currentpassing electrode, electrode 1 as presynaptic recording electrode. Curve B obtained with electrode functions reversed. Shift in transfer curve for small responses (V2 - V1) is consistent with that expected with a presynaptic space constant of 1 mm (see text).
transfer curve should be such that V1/V2 = exp - dIA. In the experiment shown, the inter-electrode distance, d, was about 260 /um, and for small post-synaptic responses the shift in the curves (indicated by dashed lines) was consistent with a space constant of about 1 mm. The shiftwas relatively larger for larger responses (e.g. VpO8t = 6 mV), the estimated value for A being about 0 7 mm. These values are consistent with direct measures of A (Ringham, 1975), the smaller value for the larger depolarizations being due to delayed rectification. If we assume that with the original arrangement the voltage electrode (1) was within + 70 um from the appropriate position, then the recorded presynaptic depolarizations should be within
418 A. R. MARTIN AND G. L. RINGHAM + 10 % of those actually occurring at the synapse, assuming a space constant of 0-7 mm. It seems likely that the synapse between the axon and the interneurone is formed on a single dendrite, although candidate dendrites occur morphologically along a region of axon about 150 ,um in extent (Ringham, 1975). If several synaptic contacts were made over this region their depolarizations would not be uniform. The information obtained above indicates that if such were the case the variation in depolarization from one end of the synaptic region to the other would be about + 10 % of the depolarization in the middle of the region.
Effects of presynaptic conditioning pulses In experiments on the squid giant synapse, the e.p.s.p. resulting from a given presynaptic depolarizing pulse was increased when the depolarization was preceded by a hyperpolarizing conditioning pulse and decreased when the conditioning pulse was in the depolarizing direction (Bloedel et al. 1967; Katz & Miledi, 1967c, 1970, 1971; Kusano, 1968). Thus, the synaptic transfer curves were shifted along the abscissa to the left by conditioning hyperpolarization and to the right by conditioning depolarization. These observations were initially attributed to direct effects on the transmitter release mechanism, possibly related to inactivation of the Ca conductance mechanism by depolarization and removal of resting inactivation by hyperpolarization. However, Katz & Miledi (1971) demonstrated that the effects could be abolished by intracellular injection of tetraethylammonium. Since such injection also abolishes delayed rectification, they suggested that the effects of the conditioning pulses were due simply to changes in presynaptic membrane resistance and the consequent alteration in the space constant of the presynaptic terminal. On hyperpolarization, for example, the space constant would increase and a subsequent depolarizing pulse would spread more completely into the distal portion of the nerve terminal, thus increasing transmitter release. At the lamprey synapse, where the region of synaptic contact is relatively restricted and the presynaptic axon fully accessible for penetration by micropipettes, this hypothesis could be tested directly. Fig. 7A shows the effect of 30 msec conditioning pulses on e.p.s.p.s produced by a subsequent 4 msec depolarization. In frame 1 responses to two successive stimuli are superimposed; in the first only the test depolarization was applied while in the second this was preceded by a hyperpolarizing conditioning pulse. After the conditioning pulse both the subsequent fibre depolarization and the corresponding e.p.s.p. were increased in size. In frame 2 the same procedure was followed with a depolarizi conditioning pulse. In this case both the subsequent fibre depolarization and the e.p.s.p. were decreased. The results obtained with the hyperpolarizing and depolarizing conditioning pulses are summarized in Fig. 7B, in
SYNAPTIC TRANSFER 419 which e.p.s.p. amplitude is plotted against presynaptic current. After hyperpolarization (triangles) a given current was more effective in producing transmitter release than after depolarization (squares). A I
2
VP,. V
-] 50 mV
e-
aPM ","I
Iw ~4 ] 2 mV 10 msec
7-
0
B 0
65-
E
I 321I
20
60
I
I
I
I
I
100 140 180 220 260
I
300 340
I
I
380 420
Ipr(nA) Fig. 7. Effects of hyperpolarizing and depolarizing conditioning pulses. A, presynaptic potential changes (Vpre) and e.p.s.p.s (VP) produced by 4 msec presynaptic current pulses (Ipre) with and without preceding conditioning pulses 30 msec in duration. 1: hyperpolarizing conditioning pulse increases both presynaptic depolarization and e.p.s.p. produced by subsequent test pulse. 2: depolarizing conditioning pulse produces reduction in both presynaptic depolarization and e.p.s.p. B, e.p.s.p. amplitude (VpOt) as a function of presynaptic depolarizing current (Ipre) from experiment shown in A. Current is more effective in releasing transmitter after hyperpolarizing pulses (A) than after conditioning depolarization (]).
According to the delayed rectification hypothesis, the results shown in Fig. 7 should be attributable solely to changes in presynaptic membrane resistance. This being the case, the relation between presynaptic depolarization and e.p.s.p. amplitude should be unchanged by the conditioning pulses, provided the presynaptic recording electrode is, in fact, recording
420 A. R. MARTIN AND G. L. RINGHAM the depolarization at the region of synaptic contact. This relation is shown in the synaptic transfer curves of Fig. 8, taken from the same experiment as Fig. 7. Electrode positions are shown in the inset. Acontrolserieswithout conditioning was obtained first (open circles). Responses were then obtained with hyperpolarizing (triangles) and depolarizing pulses (squares). Finally, a second control series was obtained (filled circles). There is little detectable shift in the transfer curves, except for an apparent slight increase in synaptic efficacy in the final control series. If anything, the effect of the conditioning 7-
0
6
>-
@0
3-
VanM Vi
100 PM00
0
.yu.Vprer~mv
10
20
30
40
so
60
70
so
90
100
Fig. 8. Synaptic transfer curves from experiment of Fig. 7. O. *: initial and final control points; /\: after hyperpolarizing conditioning pulses; E]: after depolarizing conditioning pulses. Inset shows electrode positions with dashed line indicating approximate position of synaptic contact. Conditioning pulses have little or no effect on synaptic transfer.
hyperpolarization was to cause an apparent slight decrease in release, as compared with conditioning depolarization. As will be discussed below, such a shift could be accounted for by a slight misplacement of the presynaptic recording electrode. These results 'indicate, then, that there is no observable effect of conditioning pulses on the release mechanism when the presynaptic recording electrode is properly placed in the synaptic region. As a further test of the hypothesis, experiments were done in which the presynaptic recording electrode was deliberately misplaced. One such experiment is shown in Fig. 9. The electrode positions were such that the presynaptic record electrode was between the synapse and the current
421 SYNAPTIC TRANSFER electrode (inset). With this arrangement the actual depolarization at the synapse is only a fraction of that recorded. The effect of an increase in membrane resistance, and hence in the space constant, should be to increase the fraction of the recorded depolarization reaching the synapse and thus cause an apparent' shift in the synaptic transfer curve to the left. Such a shift was observed after hyperpolarizing conditioning pulses (triangles), compared to the control responses (open and filed circles). Conversely, depolarizing conditioning pulses shifted the transfer curve to 7 6
A
5
>4
0
Vpre .z
3
-__
1100pjm
v
10
20
30
40
50
60
70
V1,re (mY)
80
90
100 110
120
Fig. 9. Effects of conditioning pulses on synaptic transfer curves with electrode positions as shown in inset. 0, *: initial and final control responses; A: responses after 40 msec hyperpolarizing conditioning pulse 42 mV in amplitude; El: responses after 22 mV, 40 msec depolarizing pulse. Test pulse 4 msec in duration. Apparent facilitation by hyperpolarization and depression by depolarization due to changes in presynaptic space constant (see text).
the right (squares), as would be expected if the space constant were decreased and a smaller fraction of the recorded presynaptic depolarization reached the synapse. Finally, one experiment was done in which the recording electrode was placed farther than appropriate from the current electrode; i.e. the synapse was closer than the recording electrode to the current source. The results are shown in Fig. 10. With this arrangement the electrode should record only a fraction of the actual presynaptic depolarization. An increase in space constant should increase this fraction so that the amount of depolarization required to produce a given post-synaptic response appears to increase. Thus the synaptic transfer curve should be shifted to
A. R. MARTIN AND G. L. RINGHAM the right. Such a shift was obtained with hyperpolarizing conditioning pulses (triangles), compared to the initial controls (open circles). In other words, hyperpolarizing conditioning pulses appeared to depress, rather than facilitate, transmitter release. The effect of depolarizing conditioning pulses was less clear. The transfer curve (squares) was to the left of that obtained with the hyperpolarizing pulses, as expected, but not to the left of the initial control series. However, the final control series (filled circles), obtained immediately after the trials with depolarizing conditioning pulses, was shifted to the right, indicating a loss in synaptic efficacy throughout the experiment, and most of the responses with depolarizing conditioning pulses were, in fact, to the left of the final controls. 422
7 0
6-
E
4 3
-
2
-
~~~~~~~0
-
Vpro
r
,
100jpm
vPOj~)
~~~~~~0
1
10
20
30
40
so
60
70
80
90
100
Vpre(mV) Fig. 10. Effects of conditioning pulses on synaptic transfer with electrode positions as shown in inset. Current electrode is closer to synapse than to presynaptic recording electrode. 0: initial control without conditioning pulse; A: conditioning hyperpolarization (32 mV, 40 msec) produces apparent depression of responses; F]: conditioning depolarization (20 mV, 40msec) produced slight facilitation ofresponses with respect to final control (-). Reversal of effects from previous Figure is due to difference in electrode arrangement (see text).
The facilitation and depression observed in the experiments described above were clearly dependent on electrode position and were consistent with the idea that changes in presynaptic membrane resistance were the
SYNAPTIC TRANSFER 423 underlying cause. This conclusion was reinforced by experiments in which the effects of conditioning pulses on transmitter release by action potentials were examined. In this case, if conditioning pulses are applied in the region of the synapse any resulting facilitation or depression of release should be independent of the position of the presynaptic recording electrode. In eight out of nine such experiments no effect of either hyperpolarizing or depolarizing conditioning pulses was observed. Records from one such experiment are shown in Fig. 11. The presynaptic action potential (middle trace), initiated by a depolarizing current pulse (upper trace), produced an e.p.s.p. of about 4 mV (lower trace). When the action potential was preceded by a 20 mV hyperpolarization 80 msec in duration (part B) the amplitude of the e.p.s.p. was unchanged. B
A r
t
--
1-
--- r
---
200 nA
1 mV 5-in50
-
mmm!~,
___b ]
mV
2
Li 20 msec
Fig. 11. Absence of effect of hyperpolarizing conditioning pulse on transmitter release by propagated action potential. A, depolarizing current pulse (upper trace) produces action potential in presynaptic fibre (middle trace), followed by e.p.s.p. in cell (lower trace). B, same as A but with stimulus preceded by hyperpolarizing conditioning pulse. E.p.s.p. amplitude unchanged.
In one of the nine experiments a marked facilitating effect of hyperpolarization was observed, together with a depression of the e.p.s.p. following conditioning depolarization. This one observation remains unexplained. It may be that the Na conductance mechanism associated with the action potential was partially inactivated and that Na inactivation was reduced by the conditioning pulse. A resulting increase in peak action potential amplitude would then produce an increase in the e.p.s.p. size. Unfortunately, the action potential itself was obscured by the stimulus artifact so that accurate measurement of its amplitude with and without the conditioning pulses was not possible and such an explanation must remain tentative.
424
A. R. MARTIN AND G. L. RINGHAM DISCUSSION
The results presented here indicate that the characteristics of synaptic transfer at a vertebrate central synapse are very similar to those at the squid giant synapse and the frog neuromuscular junction. There are, however, a number of points of quantitative difference, the chief one being the amplitude of the post-synaptic response. Saturation of transmitter release was apparently reached with presynaptic depolarizations of about 100 mV when the e.p.s.p. itself was only a few mV in amplitude. The maximum e.p.s.p. was sometimes as small as 2 mV in amplitude and rarely greater than 10 mV. Although the reversal potential for the e.p.s.p. is not known, it seems unlikely that such small responses were anywhere near post-synaptic saturation (cf. Katz & Miledi, 1970), so that the limited amplitude of the responses must be attributed to a limited ability to release transmitter. One possibility is that a single action potential, or a corresponding electrotonic depolarization of 100 mV or more, releases most of the available transmitter from the terminal, as has been suggested for a synapse in the hatchet-fish medulla (Auerbach & Bennett, 1969). However, this conclusion would not be consistent with the observation that maximal release can be increased by increasing duration of the presynaptic depolarization from 2 to 4 msec (e.g. Fig. 3). A more likely explanation lies in the nature of the presynaptic Ca conductance increase underlying the transmitter release process (Katz & Miledi, 1969, 1971). If one supposes the increase in Ca conductance is graded, at least between about 40 mV and 100 mV depolarization (Katz & Miledi, 1970, fig. 9), then increases in presynaptic depolarization will result in increased Ca entry and subsequent transmitter release. At some point, however, further conductance increases associated with still larger depolarizing pulses will be offset by a decreased driving force for Ca entry as the membrane potential approaches the Ca equilibrium potential. Consequently, increases in presynaptic depolarization beyond such a point can be envisioned to produce no further increase in Ca current into the terminal and thus no further increase in transmitter release. The extreme case of these considerations is illustrated dramatically at the squid giant synapse, where prolonged depolarization of large amplitude completely suppresses Ca entry and transmitter release until the depolarization is removed (Katz &; Miledi, 1967c, 1969, 1971). One consequence of the position of the synaptic transfer curve relation to the abscissa is that the action potential produces maximal transmitter release, its amplitude being more than 100 mV (e.g. Figs. 1 and 2). Thus, small changes in action potential amplitude can be expected to have no effect on transmitter release, although increases in duration could produce an increase in release. The general low level of transmitter release is, of
425 SYNAPTIC TRANSFER course, to be expected at a central synapse on an interneurone which is presumably required to integrate activity from a number of presynaptic inputs. The shifts in the synaptic transfer curves with electrode position (Fig. 6) are consistent with the assumption that the region of synaptic contact on the pre-synaptic axon is relatively discrete, compared with the space constant of the fibre, and agrees with the morphological observations reported in the previous paper (Ringham, 1975) thatthe spread of candidate dendrites along the length of the medial portion of the cord was of the order of 150 /sm. It seems likely that the synapse under consideration involves only a single dendritic contact, since the e.p.s.p. is relatively small and a multiplicity of contacts is probably not required to achieve such a small amount of synaptic input to the cell. It seems reasonable to conclude, then, that in the present experiments the presynaptic membrane was uniformly depolarized and that the measurements of presynaptic depolarizations were within + 10 % of the actual values. The results presented in Figs. 7-11 indicate that the facilitation and depression of transmitter release by conditioning pulses are artifacts of electrode position and due to changes in the space constant of the presynaptic fibre as postulated by Katz & Miledi (1967c, 1971). It is possible that a real effect on the release mechanism could be cancelled out by a slight misplacement of the presynaptic recording electrode. For example, if the recording electrode were too far from the current electrode an increase in space constant following hyperpolarization would produce an apparent depression in transmitter release; i.e. a shift of the transfer curve to the right. This might mask a true facilitating effect of the hyperpolarizing conditioning pulse. However, such a possibility must be rejected on the grounds that conditioning pulses had no effect on release by propagated action potentials. This work was supported by Research Grant no. NS-09660 and Fellowship no. NS-50,849, from the U.S.P.H.S.
REFERENCES
AuERBACH, A. A. & BENNETr, M. V. L. (1969). Chemically mediated transmission at a giant fiber synapse in the central nervous system of a vertebrate. J. gen. Physiol. 53, 183-210. BLOEDEL, J. R., GAGE, P. W., LinHAs, R. & QUASTEL, D. M. J. (1966). Transmitter release at the squid giant synapse in the presence of tetrodotoxin. Nature, Lond. 212, 49-50. BLOEDEL, J. R., GAGE, P. W., LLmNAs, R. & QUASTEL, D. M. J. (1967). Transmission across the squid giant synapse in the presence of tetrodotoxin. J. Physiol. 188, 52-53P.
426
A. R. MARTIN AND G. L. RINGHAM
FJRUKAWA, T., SASAOKA, T. & HOSOYA, Y. (1959). Effects of tetrodotoxin on the neuromuscular junction. Jap. J. Physiol. 9, 143-152. HAMA, K. (1962). Some observations on the fine structure of the giant synapse in the stellate ganglion of the squid, Dorythenphi8 bleckeri. Z. Zellforsch. mikrosk. Anat. 56, 437-444. KATZ, B. (1969). The Releae of Neural Transmitter Substances, pp. 33-49. Liverpool: Liverpool University Press. KATZ, B. & MILEDI, R. (1967a). Tetrodotoxin and neuromuscular transmission. Proc. R. Soc. B. 167, 8-22. KATZ, B. & MILEDI, R. (1967 b). The release of acetylcholine from nerve endings by graded electric pulses. Proc. Roy. Soc. B. 167, 23-38. KATZ, B. & MILEDI, R. (1967c). A study of synaptic transmission in the absence of nerve impulses. J. Physiol. 192, 407-436. KATZ, B. & MIILEDI, R. (1969). Tetrodotoxin-resistant electric activity in presynaptic terminals. J. Physiol. 203, 459-487. KATZ, B. & MILEDI, R. (1970). A further study of the role of calcium in synaptic transmission. J. Physiol. 207, 789-801. KATZ, B. & MILEDI, R. (1971). The effect of prolonged depolarization on synaptic transfer in the stellate ganglion of the squid. J. Physiol. 216, 503-512. KUSANO, K. ( 1968). Further study of the relationship between pre- and post-synaptic potentials in the squid giant synapse. J. gen. Physiol. 52, 326-345. MARTIN, A. R. & RINGHAM, G. (1974). Synaptic transfer curves at a vertebrate central synapse. J. Physiol. 242, 84-85P. RINGHAM, G. L. (1975). Localization and electrical characteristics of a giant synapse in the spinal cord of the lamprey. J. Physiol. 251, 395-407.
409
SYNAPTIC TRANSFER AT A VERTEBRATE CENTRAL NERVOUS SYSTEM SYNAPSE
BY A. R. MARTIN AND G. L. RINGHAM* From the Department of Physiology, University of Colorado Medical Center, Denver, Colorado 80220 U.S.A.
(Received 17 January 1975) SUMMARY
1. The relation between presynaptic depolarization and transmitter release was examined at a synapse between a Mifller axon and a lateral interneurone in the spinal cord of the lamprey. Two micro-electrodes, one for passing current and the other for recording the resulting voltage change, were placed in the presynaptic axon; a single electrode for recording the post-synaptic potential produced by release of transmitter was placed in the post-synaptic cell. 2. When action potentials were blocked with tetrodotoxin, brief depolarizing pulses in the presynaptic fibre were as effective as the action potential had been in producing transmitter release. 3. The release process had an apparent threshold depolarization of 40-50 mV and saturated- at presynaptic depolarizations of the order of 100 mV. Increasing the duration of the presynaptic pulse increased the maximum level of release. 4. Displacing the presynaptic voltage recording electrode from the position of synaptic contact toward the current passing electrode increased the apparent depolarization required to produce a given level of transmitter release. This shift in the input-output relation was consistent in magnitude with the voltage attenuation between the presynaptic recording electrode and the synapse expected from the space constant of the fibre. 5. The effect of conditioning hyperpolarization and depolarization of the presynaptic fibre on subsequent transmitter release by brief depolarizing pulses was examined. No effect was observed when the presynaptic recording electrode was in the region of synaptic contact. When the presynaptic electrode was not so positioned, conditioning effects were observed which depended on electrode position and could be attributed to changes * Present address: Department of Physiology, University of Utah College of Medicine, Salt Lake City, Utah 84132, U.S.A.
410 A. R. MARTIN AND G. L. RINGHAM in the space constant of the presynaptic fibre. No conditioning effects were observed on transmitter release by the action potential. INTRODUCTION
During the normal process of chemical synaptic transmission an action potential in the presynaptic nerve causes release of transmitter on to the post-synaptic cell, producing a conductance change in the post-synaptic membrane. The release of transmitter is mediated by an influx of Ca ions into the presynaptic nerve as a result of the depolarization associated with the arrival of the action potential (see Katz, 1969). Studies of the relation between presynaptic depolarization and transmitter release are normally limited by the presence of an all-or-nothing action potential in the presynaptic nerve. However, the action potential may be blocked with tetrodotoxin without interfering with the transmitter release process (Furukawa, Sasaoka & Hosoya, 1959), thus enabling graded depolarizations of the presynaptic nerve. Studies of this kind have been carried out on the neuromuscular junction of the frog (Katz & Miledi, 1967 a, b) and on the giant synapse of the squid (Bloedel, Gage, Llinas & Quastel, 1966; Katz & Miledi, 1967 c; Kusano, 1968), but only at the squid synapse was it possible to record the magnitude of the presynaptic potential changes. In the experiments to be reported here, studies of the relation between presynaptic polarization and transmitter release have been carried out at the synapse described in the preceding paper (Ringham, 1975), occurring in the lamprey spinal cord between a Mifller axon and a lateral interneurone. Apart from extending previous observations on the squid to a vertebrate central nervous system synapse, the preparation has the advantage that the presynaptic element is a continuous cylindrical axon which forms a discrete synaptic contact less than 100 I=m in extent, probably with a single short dendrite extending from the post-synaptic cell. This is in contrast with the somewhat more complicated geometry of the squid synapse (Hama, 1962) in which the presynaptic terminal is of the order of 1 mm in extent, and consequently, may be subject to non-uniform depolarization because of the electrotonic decrement along its length. A preliminary report of some of the present observations has been published previously (Martin & Ringham, 1974). METHODS The method of preparing and mounting the lamprey spinal cord was as described in the previous paper (Ringham, 1975). Two micro-electrodes were placed in the presynaptic fibre, one for passing depolarizing currents and the other for recording the presynaptic potential changes. A third electrode was placed in the lateral interneurone to record the excitatory post-synaptic potential (e.p.s.p.). In the initial experiments the presynaptic recording electrode was placed as close as possible to
SYNAPTIC TRANSFER
411
the region of synaptic contact on the fibre, which was medial and usually within 150 ,um rostral to the lateral cell. The presynaptic stimulating electrode was then placed a short distance (usually within 300 ,um) away. However, the voltage electrode sometimes produced damage in the presynaptic region, resulting in partial or complete loss of the e.p.s.p. In later experiments the electrodes were arranged so that the synapse was on one side of the presynaptic current electrode and the voltage recording electrode an equal distance on the other side. As the decrement in presynaptic depolarization along the cylindrical fibre should have been the same in either direction, the voltage electrode was assumed to record the same potential changes as were occurring at the synapse. For the experiments in which the effects of conditioning pulses were examined, the conditioning pulses and test depolarizations were applied to the same presynaptic current electrode. Tetrodotoxin (Sankyo) was added to all bathing solutions at a concentration of 0-2 fig/ml. RESULTS
After a lateral interneurone had been impaled and current and voltage electrodes placed in the appropriate Miller fibre, tetrodotoxin was added to the perfusing fluid to block initiation of propagated action potentials. Action potentials in the fibre, evoked by brief depolarizing pulses, were not blocked immediately and their amplitudes, together with those of the corresponding e.p.s.p.s in the cell, could be monitored as block progressed. After block was complete, e.p.s.p.s could then be produced by presynaptic depolarization without the intervention of action potentials. One such sequence of records is shown in Fig. 1. In each frame the upper record is from the presynaptic fibre, the lower from the interneurone. The left column shows responses obtained during the development of tetrodotoxin block. In the first frame a presynaptic action potential of about 85 mV amplitude produced an e.p.s.p. in the cell of about 2 mV. As the action potential failed (second and third frames) the e.p.s.p. also diminished in size. In the right column, obtained after the block was complete, depolarizing pulses 2 msec in duration and similar in amplitude to the failing action potentials on the left produced similar e.p.s.p.s. Thus the artificial pulses were similar in efficacy to the action potentials in producing transmitter release at the synapse. The relation between e.p.s.p. amplitude and presynaptic depolarization, obtained from the experiment described above, is shown in Fig. 2. With both the action potentials (curve A) and the electrotonic pulses (P) there was an apparent 'threshold' for production of a detectable e.p.s.p. of about 50 mV, and after about 90 mV presynaptic depolarization no further increase in e.p.s.p. amplitude occurred. The early part of each curve is linear on a semilogarithmic plot (not shown) with a tenfold increase in e.p.s.p. amplitude being produced by a 24 mV increment in presynaptic depolarization. The fact that the electrotonic pulses were slightly less effective in producing transmitter release than were the failing action
412 A. R. MARTIN AND G. L. RINGHAM potentials could suggest that the presynaptic recording electrodes were positioned a short distance from the synapse, toward the current passing electrode. The recorded presynaptic depolarization would then be somewhat larger than that actually occurring at the synapse. However, if this were the case, the slope of the curve (P) should have been less than that obtained with the action potentials and the plateau level the same. A more likely explanation can be obtained by examining Fig. 1 more closely and noting that the depolarizing pulses were slightly shorter in duration than the action potentials. Thus they would be expected to release less
transmitter. I
i
U'
VpreI
4
2
S 'I~~
__
__
a
3
6 %ftam _wrq~_
V"~
, r
]2mV
~10msec Fig. 1. Intracellular records from presynaptic axon (upper trace) and postsynaptic cell (lower trace) in lamprey spinal cord. First column: failure of presynaptic action potential after addition of tetrodotoxin 02 ,ug/ml. to the bathing solution. First action potential is normal. Failure during second and third frames is accompanied by reduction in amplitude of post-synaptic potential (e.p.s.p.). Second column: e.p.s.p.s produced by artificial presynaptic depolarizing pulses 2 msec in duration after tetrodotoxin block. Responses are similar to those produced by failing action potentials. Tetrodotoxin was present in experiments illustrated by all subsequent Figures except Fig. 11.
The effect of pulse duration on transmitter release is illustrated in Fig. 3. E.p.s.p.s produced by pulses of 4 msec duration are shown in Fig. 3A. In Fig. 3B, synaptic transfer curves are shown for a different experiment in which pulse durations of 1, 2 and 4 msec were used. There was little difference between the responses to 1 msec and 2 msec pulses except at
SYNAPTIC TRANSFER 413 large depolarizations; with 4 msec pulses the transfer curve was shifted to the left by about 15 mV and the maximum response increased by 2-3 mV. However, the maximum rate of change of e.p.s.p. amplitude with depolarization was unaltered, again being of the order of tenfold for a 25 mV increment in peak presynaptic depolarization. The dependence of release on duration is somewhat different than that reported for the neuromuscular junction of the frog (Katz & Miledi, 1967b), where the maximum effect of increasing pulse duration occurred between durations of 1 and 2 msec.
3 A
P
0
Aoe I
0
I
20
I
I
40
I
I
60
I
I
80
I
I
100
Vpre(mV) Fig. 2. Synaptic transfer curves from experiment shown in Fig. 1. Abscissa: amplitude of presynaptic depolarization; ordinate: e.p.s.p. amplitude. Curve A (@) was obtained with failing action potentials during development of tetrodotoxin block, curve P (A) with 2 msec depolarizing pulses after block was complete.
Attempts to depolarize the presynaptic axon for more than 4 msec were frustrated by the onset of delayed rectification, after which the micropipette could not pass enough current to depolarize the fibre by more than about 40 mV. However, as noted in the previous paper (Ringham, 1975), the synapse was electrically as well as chemically coupled, and prolonging the depolarization resulted in an increase in the post-synaptic response because of summation of the electrical coupling potential with the e.p.s.p. One such experiment is shown in Fig. 4, in which pulses of 50 msec duration were used. Electrical coupling to the cell can be seen with both hyperpolarizing and depolarizing pulses, and whea the initial depolarization was sufficient, e.p.s.p.s. appeared on top of the coupling potentials. The synaptic transfer curve for this experiment is shown in Fig. 5. The
A. R. MARTIN AND G. L. RINGHAM post-synaptic responses in the hyperpolarizing region and the responses in the early part of the depolarizing region were due to electrical coupling. At about 35-40 mV depolarization the e.p.s.p-. appeared and reached its maximum amplitude at about 80 mV peak depolarization. Beyond this, the response continued to increase at a rate consistent with that expected for the electrical coupling potential alone (dashed line). The contribution of the 414
A
VPre
f0
vPost
49 8>~~~~~~~
4-
50 mV
2-
-g I2 mV 5 msec
0
100
50
150
V~re(MV) Fig. 3. Effect of pulse duration on transmitter release. A, e.p.s.p.s (V,,,.) produced by presynaptic depolarizing pulses (Ipr) 4 msec in duration. B, synaptic transfer curves obtained from one preparation with presynaptic pulses, 1, (0), 2 (A) and 4 (N) msec duration. Increasing duration increases synaptic efficacy.
coupling potential to the over-all response was estimated by measuring the break in the post-synaptic potential at the end of the depolarizing pulse (Fig. 4). This was then plotted against peak depolarization in Fig. 5 (open circles) to obtain the dashed line. It should be pointed out that the coupling potential is produced by the steady-state depolarization, not the peak. Because the points are plotted against peak depolarization, the dashed line does not represent the transfer curve for the electrical coupling, but rather the contribution of the coupling to the total response.For example, the point at 94 mV peak depolarization corresponded to a steady-state presynaptic depolarization of only 37 mV, and the coupling potential contributed 2-3 mV to the total post-synaptic response of 7-7 mV. When
SYNAPTIC TRANSFER 415 the coupling potential amplitudes are plotted against steady-state depolarization (points represented by x 's, Fig. 5), it can be seen that the electrical coupling is highly rectifying, as reported previously (Ringham, 1975). 1 Vpre
4
vpost
2
5
36
- j'
5omV
_mewt---_ i_ ]2 mV 10 msec Fig. 4. Responses to presynaptic depolarizing pulses 50 msec in duration, showing both e.p.s.p. and electrical coupling between presynaptic axon and post-synaptic cell. In last three frames e.p.s.p.s are superimposed on coupling potentials. Breaks at beginning and end of pulses are due to capacitative coupling between current-passing and recording electrodes. Note delayed rectification in presynaptic fibre.
The over-all synaptic transfer curve shown in Fig. 5 represents, of course, a highly artificial situation. Normally the presynaptic depolarization is brief and the chemical and electrical responses do not sum because of synaptic delay. In Fig. 4, and in the experimental records shown in the previous figures, the rising and falling phases of the electrotonic potentials in both the cell and the fibre and the onset of the e.p.s.p.s were obscured by large capacitative artifacts due to coupling between the current passing electrode and the two recording electrodes. Because of such artifacts it was not possible to measure synaptic delay accurately. In other experiments in which e.p.s.p.s were produced by action potentials the delay was in the range of 1-3 msec at about 100 C.
Effects of electrode position The shape of the synaptic transfer curves and their position in relation to the abscissa (Vpre) is obviously dependent on the location of the
416 A. R. MARTIN AND G. L. RINGHAM presynaptic voltage recording electrode with respect to the synapse. Initially regions of synaptic contact were located by exploring with the presynaptic current electrode to find the position which gave a maximum post-synaptic response with moderate depolarization. The region of synaptic contact was almost invariably 0-150,um rostral to the position of the interneurone, and in later experiments the synapse was simply assumed to be located in the middle of this region. Thus, localization of the synapse was probably in error by less than + 1001tm. During several experiments a check was made to determine the probable error in the synaptic transfer curve due to such errors in electrode positioning. One such
+10 B E
6
0
X
2
4-
20
2
40
60
80
100
120 140 160+
~~~~Vpre(miV)
Fig. 5. Synaptic transfer curve for experiment of Fig. 4 with 50 msec presynaptic hyperpolarizing and depolarizing pulses. For points indicated by 0 and * abscissa represents peak presynaptic depolarization. *: total amplitude of post-synaptic response; 0, contribution of electrical coupling potential. Difference between these two curves is e.p.s.p. amplitude. x, coupling potential amplitude replotted with abscissa representing steadystate presynaptic depolarization. These points plus those in hyperpolarizing quadrant represent synaptic transfer curve for electrical coupling potential.
experiment is shown in Fig. 6, with the usual recording arrangement in the inset. Current was passed through electrode 2 and presynaptic potential changes recorded with electrode 1. With this arrangement, disruption of the synapse by the recording micropipette was avoided. The depolarization at the synapse, VA, is related to the recorded depolarization, V1, by VA = V1 exp - (x - d)/A, where A is the space constant of the fibre, x the distance from the current passing electrode to the synapse and d the inter-electrode distance. Thus when x = d, the recorded depolarization, V1, should equal
SYNAPTIC TRANSFER 417 the actual presynaptic depolarization, VA. Curve A of Fig. 6 was obtained in this way, using 30 msec presynaptic pulses. After curve A was obtained, the functions of the two electrodes were reversed, electrode 1 becoming the current electrode and electrode 2 the voltage electrode. This arrangement resulted in curve B. In this case the presynaptic depolarization, VB, should be related to the recorded depolarization, V2, by VB = V2 exp -x/A. If one now assumes that equal post-synaptic responses indicate equal presynaptic depolarizations, then when the post-synaptic responses are the same for the two curves, VA = VB and the displacement of the 12+ 10 8
~~~~~~Fibre
//
-80
60
40
20
~
_20
cow ~~~~~2
40
60
80
100 120 140 160+
Vpre(mV
4-
Fig. 6. Effect of electrode position. Location of presynaptic electrodes with respect to synapse shown in inset. Distance d was 257 jum; distance x assumed to be similar. Curve A obtained with electrode 2 as currentpassing electrode, electrode 1 as presynaptic recording electrode. Curve B obtained with electrode functions reversed. Shift in transfer curve for small responses (V2 - V1) is consistent with that expected with a presynaptic space constant of 1 mm (see text).
transfer curve should be such that V1/V2 = exp - dIA. In the experiment shown, the inter-electrode distance, d, was about 260 /um, and for small post-synaptic responses the shift in the curves (indicated by dashed lines) was consistent with a space constant of about 1 mm. The shiftwas relatively larger for larger responses (e.g. VpO8t = 6 mV), the estimated value for A being about 0 7 mm. These values are consistent with direct measures of A (Ringham, 1975), the smaller value for the larger depolarizations being due to delayed rectification. If we assume that with the original arrangement the voltage electrode (1) was within + 70 um from the appropriate position, then the recorded presynaptic depolarizations should be within
418 A. R. MARTIN AND G. L. RINGHAM + 10 % of those actually occurring at the synapse, assuming a space constant of 0-7 mm. It seems likely that the synapse between the axon and the interneurone is formed on a single dendrite, although candidate dendrites occur morphologically along a region of axon about 150 ,um in extent (Ringham, 1975). If several synaptic contacts were made over this region their depolarizations would not be uniform. The information obtained above indicates that if such were the case the variation in depolarization from one end of the synaptic region to the other would be about + 10 % of the depolarization in the middle of the region.
Effects of presynaptic conditioning pulses In experiments on the squid giant synapse, the e.p.s.p. resulting from a given presynaptic depolarizing pulse was increased when the depolarization was preceded by a hyperpolarizing conditioning pulse and decreased when the conditioning pulse was in the depolarizing direction (Bloedel et al. 1967; Katz & Miledi, 1967c, 1970, 1971; Kusano, 1968). Thus, the synaptic transfer curves were shifted along the abscissa to the left by conditioning hyperpolarization and to the right by conditioning depolarization. These observations were initially attributed to direct effects on the transmitter release mechanism, possibly related to inactivation of the Ca conductance mechanism by depolarization and removal of resting inactivation by hyperpolarization. However, Katz & Miledi (1971) demonstrated that the effects could be abolished by intracellular injection of tetraethylammonium. Since such injection also abolishes delayed rectification, they suggested that the effects of the conditioning pulses were due simply to changes in presynaptic membrane resistance and the consequent alteration in the space constant of the presynaptic terminal. On hyperpolarization, for example, the space constant would increase and a subsequent depolarizing pulse would spread more completely into the distal portion of the nerve terminal, thus increasing transmitter release. At the lamprey synapse, where the region of synaptic contact is relatively restricted and the presynaptic axon fully accessible for penetration by micropipettes, this hypothesis could be tested directly. Fig. 7A shows the effect of 30 msec conditioning pulses on e.p.s.p.s produced by a subsequent 4 msec depolarization. In frame 1 responses to two successive stimuli are superimposed; in the first only the test depolarization was applied while in the second this was preceded by a hyperpolarizing conditioning pulse. After the conditioning pulse both the subsequent fibre depolarization and the corresponding e.p.s.p. were increased in size. In frame 2 the same procedure was followed with a depolarizi conditioning pulse. In this case both the subsequent fibre depolarization and the e.p.s.p. were decreased. The results obtained with the hyperpolarizing and depolarizing conditioning pulses are summarized in Fig. 7B, in
SYNAPTIC TRANSFER 419 which e.p.s.p. amplitude is plotted against presynaptic current. After hyperpolarization (triangles) a given current was more effective in producing transmitter release than after depolarization (squares). A I
2
VP,. V
-] 50 mV
e-
aPM ","I
Iw ~4 ] 2 mV 10 msec
7-
0
B 0
65-
E
I 321I
20
60
I
I
I
I
I
100 140 180 220 260
I
300 340
I
I
380 420
Ipr(nA) Fig. 7. Effects of hyperpolarizing and depolarizing conditioning pulses. A, presynaptic potential changes (Vpre) and e.p.s.p.s (VP) produced by 4 msec presynaptic current pulses (Ipre) with and without preceding conditioning pulses 30 msec in duration. 1: hyperpolarizing conditioning pulse increases both presynaptic depolarization and e.p.s.p. produced by subsequent test pulse. 2: depolarizing conditioning pulse produces reduction in both presynaptic depolarization and e.p.s.p. B, e.p.s.p. amplitude (VpOt) as a function of presynaptic depolarizing current (Ipre) from experiment shown in A. Current is more effective in releasing transmitter after hyperpolarizing pulses (A) than after conditioning depolarization (]).
According to the delayed rectification hypothesis, the results shown in Fig. 7 should be attributable solely to changes in presynaptic membrane resistance. This being the case, the relation between presynaptic depolarization and e.p.s.p. amplitude should be unchanged by the conditioning pulses, provided the presynaptic recording electrode is, in fact, recording
420 A. R. MARTIN AND G. L. RINGHAM the depolarization at the region of synaptic contact. This relation is shown in the synaptic transfer curves of Fig. 8, taken from the same experiment as Fig. 7. Electrode positions are shown in the inset. Acontrolserieswithout conditioning was obtained first (open circles). Responses were then obtained with hyperpolarizing (triangles) and depolarizing pulses (squares). Finally, a second control series was obtained (filled circles). There is little detectable shift in the transfer curves, except for an apparent slight increase in synaptic efficacy in the final control series. If anything, the effect of the conditioning 7-
0
6
>-
@0
3-
VanM Vi
100 PM00
0
.yu.Vprer~mv
10
20
30
40
so
60
70
so
90
100
Fig. 8. Synaptic transfer curves from experiment of Fig. 7. O. *: initial and final control points; /\: after hyperpolarizing conditioning pulses; E]: after depolarizing conditioning pulses. Inset shows electrode positions with dashed line indicating approximate position of synaptic contact. Conditioning pulses have little or no effect on synaptic transfer.
hyperpolarization was to cause an apparent slight decrease in release, as compared with conditioning depolarization. As will be discussed below, such a shift could be accounted for by a slight misplacement of the presynaptic recording electrode. These results 'indicate, then, that there is no observable effect of conditioning pulses on the release mechanism when the presynaptic recording electrode is properly placed in the synaptic region. As a further test of the hypothesis, experiments were done in which the presynaptic recording electrode was deliberately misplaced. One such experiment is shown in Fig. 9. The electrode positions were such that the presynaptic record electrode was between the synapse and the current
421 SYNAPTIC TRANSFER electrode (inset). With this arrangement the actual depolarization at the synapse is only a fraction of that recorded. The effect of an increase in membrane resistance, and hence in the space constant, should be to increase the fraction of the recorded depolarization reaching the synapse and thus cause an apparent' shift in the synaptic transfer curve to the left. Such a shift was observed after hyperpolarizing conditioning pulses (triangles), compared to the control responses (open and filed circles). Conversely, depolarizing conditioning pulses shifted the transfer curve to 7 6
A
5
>4
0
Vpre .z
3
-__
1100pjm
v
10
20
30
40
50
60
70
V1,re (mY)
80
90
100 110
120
Fig. 9. Effects of conditioning pulses on synaptic transfer curves with electrode positions as shown in inset. 0, *: initial and final control responses; A: responses after 40 msec hyperpolarizing conditioning pulse 42 mV in amplitude; El: responses after 22 mV, 40 msec depolarizing pulse. Test pulse 4 msec in duration. Apparent facilitation by hyperpolarization and depression by depolarization due to changes in presynaptic space constant (see text).
the right (squares), as would be expected if the space constant were decreased and a smaller fraction of the recorded presynaptic depolarization reached the synapse. Finally, one experiment was done in which the recording electrode was placed farther than appropriate from the current electrode; i.e. the synapse was closer than the recording electrode to the current source. The results are shown in Fig. 10. With this arrangement the electrode should record only a fraction of the actual presynaptic depolarization. An increase in space constant should increase this fraction so that the amount of depolarization required to produce a given post-synaptic response appears to increase. Thus the synaptic transfer curve should be shifted to
A. R. MARTIN AND G. L. RINGHAM the right. Such a shift was obtained with hyperpolarizing conditioning pulses (triangles), compared to the initial controls (open circles). In other words, hyperpolarizing conditioning pulses appeared to depress, rather than facilitate, transmitter release. The effect of depolarizing conditioning pulses was less clear. The transfer curve (squares) was to the left of that obtained with the hyperpolarizing pulses, as expected, but not to the left of the initial control series. However, the final control series (filled circles), obtained immediately after the trials with depolarizing conditioning pulses, was shifted to the right, indicating a loss in synaptic efficacy throughout the experiment, and most of the responses with depolarizing conditioning pulses were, in fact, to the left of the final controls. 422
7 0
6-
E
4 3
-
2
-
~~~~~~~0
-
Vpro
r
,
100jpm
vPOj~)
~~~~~~0
1
10
20
30
40
so
60
70
80
90
100
Vpre(mV) Fig. 10. Effects of conditioning pulses on synaptic transfer with electrode positions as shown in inset. Current electrode is closer to synapse than to presynaptic recording electrode. 0: initial control without conditioning pulse; A: conditioning hyperpolarization (32 mV, 40 msec) produces apparent depression of responses; F]: conditioning depolarization (20 mV, 40msec) produced slight facilitation ofresponses with respect to final control (-). Reversal of effects from previous Figure is due to difference in electrode arrangement (see text).
The facilitation and depression observed in the experiments described above were clearly dependent on electrode position and were consistent with the idea that changes in presynaptic membrane resistance were the
SYNAPTIC TRANSFER 423 underlying cause. This conclusion was reinforced by experiments in which the effects of conditioning pulses on transmitter release by action potentials were examined. In this case, if conditioning pulses are applied in the region of the synapse any resulting facilitation or depression of release should be independent of the position of the presynaptic recording electrode. In eight out of nine such experiments no effect of either hyperpolarizing or depolarizing conditioning pulses was observed. Records from one such experiment are shown in Fig. 11. The presynaptic action potential (middle trace), initiated by a depolarizing current pulse (upper trace), produced an e.p.s.p. of about 4 mV (lower trace). When the action potential was preceded by a 20 mV hyperpolarization 80 msec in duration (part B) the amplitude of the e.p.s.p. was unchanged. B
A r
t
--
1-
--- r
---
200 nA
1 mV 5-in50
-
mmm!~,
___b ]
mV
2
Li 20 msec
Fig. 11. Absence of effect of hyperpolarizing conditioning pulse on transmitter release by propagated action potential. A, depolarizing current pulse (upper trace) produces action potential in presynaptic fibre (middle trace), followed by e.p.s.p. in cell (lower trace). B, same as A but with stimulus preceded by hyperpolarizing conditioning pulse. E.p.s.p. amplitude unchanged.
In one of the nine experiments a marked facilitating effect of hyperpolarization was observed, together with a depression of the e.p.s.p. following conditioning depolarization. This one observation remains unexplained. It may be that the Na conductance mechanism associated with the action potential was partially inactivated and that Na inactivation was reduced by the conditioning pulse. A resulting increase in peak action potential amplitude would then produce an increase in the e.p.s.p. size. Unfortunately, the action potential itself was obscured by the stimulus artifact so that accurate measurement of its amplitude with and without the conditioning pulses was not possible and such an explanation must remain tentative.
424
A. R. MARTIN AND G. L. RINGHAM DISCUSSION
The results presented here indicate that the characteristics of synaptic transfer at a vertebrate central synapse are very similar to those at the squid giant synapse and the frog neuromuscular junction. There are, however, a number of points of quantitative difference, the chief one being the amplitude of the post-synaptic response. Saturation of transmitter release was apparently reached with presynaptic depolarizations of about 100 mV when the e.p.s.p. itself was only a few mV in amplitude. The maximum e.p.s.p. was sometimes as small as 2 mV in amplitude and rarely greater than 10 mV. Although the reversal potential for the e.p.s.p. is not known, it seems unlikely that such small responses were anywhere near post-synaptic saturation (cf. Katz & Miledi, 1970), so that the limited amplitude of the responses must be attributed to a limited ability to release transmitter. One possibility is that a single action potential, or a corresponding electrotonic depolarization of 100 mV or more, releases most of the available transmitter from the terminal, as has been suggested for a synapse in the hatchet-fish medulla (Auerbach & Bennett, 1969). However, this conclusion would not be consistent with the observation that maximal release can be increased by increasing duration of the presynaptic depolarization from 2 to 4 msec (e.g. Fig. 3). A more likely explanation lies in the nature of the presynaptic Ca conductance increase underlying the transmitter release process (Katz & Miledi, 1969, 1971). If one supposes the increase in Ca conductance is graded, at least between about 40 mV and 100 mV depolarization (Katz & Miledi, 1970, fig. 9), then increases in presynaptic depolarization will result in increased Ca entry and subsequent transmitter release. At some point, however, further conductance increases associated with still larger depolarizing pulses will be offset by a decreased driving force for Ca entry as the membrane potential approaches the Ca equilibrium potential. Consequently, increases in presynaptic depolarization beyond such a point can be envisioned to produce no further increase in Ca current into the terminal and thus no further increase in transmitter release. The extreme case of these considerations is illustrated dramatically at the squid giant synapse, where prolonged depolarization of large amplitude completely suppresses Ca entry and transmitter release until the depolarization is removed (Katz &; Miledi, 1967c, 1969, 1971). One consequence of the position of the synaptic transfer curve relation to the abscissa is that the action potential produces maximal transmitter release, its amplitude being more than 100 mV (e.g. Figs. 1 and 2). Thus, small changes in action potential amplitude can be expected to have no effect on transmitter release, although increases in duration could produce an increase in release. The general low level of transmitter release is, of
425 SYNAPTIC TRANSFER course, to be expected at a central synapse on an interneurone which is presumably required to integrate activity from a number of presynaptic inputs. The shifts in the synaptic transfer curves with electrode position (Fig. 6) are consistent with the assumption that the region of synaptic contact on the pre-synaptic axon is relatively discrete, compared with the space constant of the fibre, and agrees with the morphological observations reported in the previous paper (Ringham, 1975) thatthe spread of candidate dendrites along the length of the medial portion of the cord was of the order of 150 /sm. It seems likely that the synapse under consideration involves only a single dendritic contact, since the e.p.s.p. is relatively small and a multiplicity of contacts is probably not required to achieve such a small amount of synaptic input to the cell. It seems reasonable to conclude, then, that in the present experiments the presynaptic membrane was uniformly depolarized and that the measurements of presynaptic depolarizations were within + 10 % of the actual values. The results presented in Figs. 7-11 indicate that the facilitation and depression of transmitter release by conditioning pulses are artifacts of electrode position and due to changes in the space constant of the presynaptic fibre as postulated by Katz & Miledi (1967c, 1971). It is possible that a real effect on the release mechanism could be cancelled out by a slight misplacement of the presynaptic recording electrode. For example, if the recording electrode were too far from the current electrode an increase in space constant following hyperpolarization would produce an apparent depression in transmitter release; i.e. a shift of the transfer curve to the right. This might mask a true facilitating effect of the hyperpolarizing conditioning pulse. However, such a possibility must be rejected on the grounds that conditioning pulses had no effect on release by propagated action potentials. This work was supported by Research Grant no. NS-09660 and Fellowship no. NS-50,849, from the U.S.P.H.S.
REFERENCES
AuERBACH, A. A. & BENNETr, M. V. L. (1969). Chemically mediated transmission at a giant fiber synapse in the central nervous system of a vertebrate. J. gen. Physiol. 53, 183-210. BLOEDEL, J. R., GAGE, P. W., LinHAs, R. & QUASTEL, D. M. J. (1966). Transmitter release at the squid giant synapse in the presence of tetrodotoxin. Nature, Lond. 212, 49-50. BLOEDEL, J. R., GAGE, P. W., LLmNAs, R. & QUASTEL, D. M. J. (1967). Transmission across the squid giant synapse in the presence of tetrodotoxin. J. Physiol. 188, 52-53P.
426
A. R. MARTIN AND G. L. RINGHAM
FJRUKAWA, T., SASAOKA, T. & HOSOYA, Y. (1959). Effects of tetrodotoxin on the neuromuscular junction. Jap. J. Physiol. 9, 143-152. HAMA, K. (1962). Some observations on the fine structure of the giant synapse in the stellate ganglion of the squid, Dorythenphi8 bleckeri. Z. Zellforsch. mikrosk. Anat. 56, 437-444. KATZ, B. (1969). The Releae of Neural Transmitter Substances, pp. 33-49. Liverpool: Liverpool University Press. KATZ, B. & MILEDI, R. (1967a). Tetrodotoxin and neuromuscular transmission. Proc. R. Soc. B. 167, 8-22. KATZ, B. & MILEDI, R. (1967 b). The release of acetylcholine from nerve endings by graded electric pulses. Proc. Roy. Soc. B. 167, 23-38. KATZ, B. & MILEDI, R. (1967c). A study of synaptic transmission in the absence of nerve impulses. J. Physiol. 192, 407-436. KATZ, B. & MIILEDI, R. (1969). Tetrodotoxin-resistant electric activity in presynaptic terminals. J. Physiol. 203, 459-487. KATZ, B. & MILEDI, R. (1970). A further study of the role of calcium in synaptic transmission. J. Physiol. 207, 789-801. KATZ, B. & MILEDI, R. (1971). The effect of prolonged depolarization on synaptic transfer in the stellate ganglion of the squid. J. Physiol. 216, 503-512. KUSANO, K. ( 1968). Further study of the relationship between pre- and post-synaptic potentials in the squid giant synapse. J. gen. Physiol. 52, 326-345. MARTIN, A. R. & RINGHAM, G. (1974). Synaptic transfer curves at a vertebrate central synapse. J. Physiol. 242, 84-85P. RINGHAM, G. L. (1975). Localization and electrical characteristics of a giant synapse in the spinal cord of the lamprey. J. Physiol. 251, 395-407.
