The Site of Synaptic Activity and Impulse Initiation in an Identified Motoneurone in the Crab Brain

With some exceptions the cell soma of mammalian central neurones provides a stage between the dendrites and the axon where pre-synaptic activity occurs at the level of graded post-synaptic potentials which sum to affect a relatively low-threshold region of regenerative activity at the initial segment of the axon (Araki & Otani, 1955; Coombs, Curtis & Eccles, 1957; Fatt, 1957; Fuortes, Frank and Becker, 1957). In crustacean motoneurones, however, the cell body does not normally lie between the presynaptic inputs and the main axon; instead, all synaptic contacts are spread over the dendritic arborizations. The integration of the incoming excitatory and inhibitory potentials by interaction of various electrotonically spreading local potentials has been assumed to take place at the confluence of the dendrite branches (Bullock & Horridge, 1965; Maynard, 1966). The final result is thought to affect one or more trigger points on the main axon.

Recently there has been evidence of regenerative spike activity within the dendrites of mammalian hippocampal neurones (Purpura, McMurtry, Leonard & Malliani, 1966) and from the dendrites of crustacean sensory cells (Mellon & Kennedy, 1964; Mendelson, 1966; Pabst & Kennedy, 1967). The central neurones of crustaceans have not been shown to have spikes in their dendrites perhaps because of the difficulty of knowing whether the recording electrode is really within one of the branches of a nerve cell’s dendritic field and not in the main axon. However, spikes of different amplitudes from within one nerve cell in crayfish abdominal ganglia have been reported (motoneurones: Takeda & Kennedy, 1964; interneurones: Preston & Kennedy, 1960; Kennedy & Mellon, 1964; Takeda & Kennedy, 1965) and shown to be regenerative impulses invading the main axon from side branches.

Crustacean motoneurones have numerous excitatory and inhibitory units converging upon them (Maynard, 1966; Knights, 1966; Sandeman, 1967) and to explain their full integrative ability one can propose that the region of the main axon where the dendrites branch off is equivalent to the cell soma of the vertebrate motoneurone in that its threshold to spike initiation is either higher than that of the main axon or unable to conduct a spike at all. Regenerative spikes, even if present in the dendritic system of such a cell, would not propagate beyond their own dendritic branches, and the initial part of the axon would collect only electrotonically spreading potentials. The presence of a non-propagating region at the start of a crustacean sensory axon has been demonstrated in the crayfish stretch-receptor (Edwards & Ottoson, 1958) but there is no information with regard to the initial segment of crustacean motoneurones.

This paper reports an investigation of the sites of synaptic activation and impulse initiation of a particular motoneurone that causes the reflex withdrawal of the crab eye-cup. All references are to this one identifiable neurone. A single volley of impulses in known afferent nerves produces a characteristic burst of impulses in the motoneurone whose peripheral axon is 30–50 μ in diameter and easily penetrated with a microelectrode. The pathway of the axon through the brain from its probable synaptic sites is known (Sandeman, 1969). The eye-withdrawal reflex is one of the least complicated described for a crustacean: it is not under proprioceptive control and the link between afferent tracts and the reflex axon appears to be monosynaptic.

Crab brains were isolated and perfused with saline (Pantin, 1948) as described previously (Sandeman, 1967). Extracellular electrical recordings from the axon in the optic tract were made with a suction electrode which enclosed the whole tract. The amplitude of the spikes from this axon is ten to twenty times larger than from any other in the tract when recorded in this way. A portion of the optic tract was de-sheathed to allow penetration of the axon with a 20–40 MΩ micropipette filled with 3 M-KC1. Extracellular stimulation of, and recording from, the axon within the de-sheathed brain was carried out with a saline-filled glass micro-electrode broken back to have an internal tip diameter of 20–30 μ. Before filling, the electrode tip was smoothed by holding it close to a heated element.

A system of spatial co-ordinates relating to a fixed microscope eyepiece graticule allowed fairly accurate positioning of the extracellular electrode, but since landmarks on the brain such as the entry point of the cerebral artery were slightly different from one crab to another, this electrode could initially be placed only within about 30 μ of the neurone. Exact placing of the electrode over the axon was achieved by applying short (0·2 msec.) stimulus pulses to the de-sheathed brain surface and recording the evoked action potential with an intracellular electrode at the periphery. The test electrode for mapping sensitivity was then moved across the brain until a point was found at which the threshold for spike initiation was the lowest. It was found that this point corresponded to the position from which the largest extracellular potential could be recorded following antidromic or orthodromic activation of the axon.

In the second type of experiment antidromic impulses were initiated by stimulating the axon intracellularly at the periphery and were picked up from different parts of the brain with the extracellular mapping electrode, now used for recording. Orthodromic impulses were initiated by stimulating the pre-synaptic afferent nerves in the tegumentary or oculomotor nerve bundles (Fig. 1). Antidromic responses recorded in the brain could be averaged on a Biomac special-purpose computer, but as latencies for pre-synaptic stimulation vary, orthodromic responses could not be averaged. Where a comparison between the amplitudes of the antidromic and orthodromic responses is drawn, only single sweep records are used. In all cases except where stated, the figures show tracings of the original photographed records.

The points within the brain where the axon is most sensitive to focally applied stimuli coincide with its path as shown from preparations where dye has been injected into it. However, different parts of the axon differ in their sensitivity to anodal and cathodal stimulation. A 0·2 msec, cathodal pulse from the focal mapping electrode applied anywhere along the peripheral portion of the neurone (points 1–9 in Fig. 2) produces a single spike after a latency of about 1 msec, or less, but an anodal stimulus of equal intensity produces no spike. These points are marked negative in Fig. 2. However, just past the point within the brain where it swings sharply away from the midline (point 8 in Fig. 2) the axon becomes more and more sensitive to anodal pulses, and in one preparation a point was found where the axon was sensitive to cathodal and anodal pulses of equal amplitude. Usually the neurone responded preferentially to one or other stimulus polarity. As the electrode was moved even more proximally and laterally along the axon, only anodal pulses produced axon spikes and cathodal pulses of equal intensity had no effect. These points are marked positive on Fig. 2. The placing of the electrode was equally critical over regions of positive or negative sensitivity. Threshold intensities of stimulation were approximately the same for both with the exception of points 6 and 12 on Fig. 2 which will be considered later. Moving the indifferent electrode to different parts of the preparation had no effect on the sensitivities of the various portions of the axon. The actual stimulus intensities needed to initiate a spike vary at all points with vertical electrode movements but are most affected by lateral movements; a lateral displacement of 6–10 μ off-target being sometimes enough to halve the effectiveness of the stimulus. Repetitive stimulation of the axon up to 100/sec. with either anodal or cathodal pulses, depending on which was appropriate to the area stimulated, resulted in axon spikes which followed the stimulus one-to-one.

A relatively high-intensity cathodal pulse applied to some of the positive-pulsesensitive areas (points 12–14 in Fig. 2) often produced a train of axon spikes similar to that obtained by pre-synaptic stimulation via the tegumentary nerve. These impulses would not follow a repetitive stimulus at much more than 10/sec., a value close to the 7–10 impulse/sec. limit for pre-synaptic stimulation. During this cathodal stimulation an antidromic compound potential in the tegumentary nerve followed the high-frequency stimulus one-to-one, showing that the relevant sensory input lies near this point. Cathodal pulses at this critical place stimulate the afferent fibres and lead to trans-synaptic activation of the axon: anodal pulses at the same point stimulate the axon directly.

A distinction between the propagating and non-propagating portions of the axon may be drawn on the basis of conduction velocities which are measurable for actively conducted spikes. In a non-propagating region the impulse travels so fast that it appears to arrive simultaneously at all points along the length, although with the usual decrease in size at points remote from the potential source. A second criterion which is sometimes applicable for establishing the propagating or non-propagating nature of the conducting pathway is to record with a focal electrode the shapes of the extracellular wave-forms of impulses passing along it. A dior triphasic potential with a predominant negative phase can indicate active conduction, and a monophasic positive potential can indicate passive electrotonic spread (Fatt, 1957; Freygang & Frank, 1959; Terzuolo and Araki, 1961; Murakami, Watanabe & Tomita, 1961; Takeuchi & Takeuchi, 1962; Werman, 1963; Dudel, 1963. For exceptions see: Brooks & Eccles, 1947; Katz & Miledi, 1964; Eccles, 1964).

In the crab brain an exact measurement of the conduction velocity of impulses between two chosen points on the axon was obtained by comparing the arrival time of impulses at the central focal recording electrode with those appearing at the peripheral extracellular recording electrode. The focal electrode was moved along the axon and the conduction velocities for the different parts were measured for both antidromic and orthodromically travelling impulses (Fig. 3). Potentials conducted in either direction between points 1 and 8 (Fig. 4) have a conduction velocity of between 0·94 and 1·2 m./sec. but appeared virtually simultaneously along the whole proximal portion of the nerve from points 8 to 12 in Fig. 4.

The differences in shape between antidromically and orthodromically conducted impulses at successive points along the nerve are also shown in Fig. 4. The antidromic wave-forms from each point are the averaged result of 256 sweeps whereas the orthodromic responses are single sweeps. Potentials conducted in either direction show the same general change from an initially diphasic or triphasic impulse with a predominant negative phase at the periphery (point 1), to a monophasic positive-going one at the central end of the axon (point 12). The inversion of the wave-form takes place around points 8–9 and the potential becomes monophasic positive in the region of the axon which is preferentially sensitive to anodal focal stimuli (points 10–14, Fig. 2). Together, the shape, latency and stimulus sensitivity changes indicate that the initial part of the axon from points 12 to 6 (Fig. 4) conducts impulses passively whereas distal to point 6 spikes are actively propagated.

Two focal points were revealed by consistent amplitude changes corresponding to electrode movements. As the electrode was moved proximally along the axon, so the negative phase of both the orthodromic and antidromic potentials increased in size until a maximum was reached in the vicinity of point 6 (Fig. 4). Tracking along the most proximal portion of the fibre revealed a similar focal point for the maximum positive potential (point 11, Fig. 4). In addition these two points of maxima were the places of lowest threshold to directly applied stimuli (see Discussion).

As well as the general changes in potential shape and amplitude mentioned above, specific differences exist between the antidromic and orthodromic passage of an impulse past any single point on the axon. These are shown in Fig. 5. At the periphery the wave-forms differ mainly in shape as follows: antidromic spikes produce a sharp initial positive phase followed by a slower negative phase of about the same amplitude. A final slow positive phase is apparent in some recordings. Orthodromic responses recorded from the same point are more obviously triphasic with the negative phase predominant.

At the place of maximum negative spike (called the negative spike focus) both the initial positive and following negative phases of the antidromic response are larger than in the orthodromic response. However, the most striking change in amplitude is obtained at the positive spike focus (point 11 in Fig. 4). Here the large-amplitude positive antidromic potentials are three times the size of their orthodromic equivalents. Because none of the results were averaged, a number of additional potentials are always present in the recordings, especially from the positive spike focus. These stem from neurones nearby which are also activated by the pre-synaptic stimulus.

The above differences in shape and amplitude following orthodromic and antidromic stimuli (Fig. 5) may in part be due to the geometry of the recording situation and the activation in the orthodromic case of many units lying near the electrode, but neither of these can fully explain the considerable difference between the amplitudes of the antidromic and orthodromic responses recorded from the positive spike focus. This amplitude change could be a consequence of resistive changes brought about in this region by the activation of the synapse which would short out the amplitude of a passively spreading potential.

To test this, antidromic impulses were timed to arrive at the proximal end of the fibre at the same time as an excitatory post-synaptic potential was generated by a presynaptic stimulus just subthreshold for an orthodromic spike. Previous intracellular recordings from the axon have shown that an EPSP is generated within it by this means (Sandeman, 1969). Recordings from the negative and positive spike foci are compared in Fig. 6; the decrease in amplitude following simultaneous pre-synaptic stimulation is about 45% for the positive spike focus but only about 23% for the negative spike focus. Some difference of this sort is to be expected if the positive spike focus fies closer to the synapse than the negative spike focus. Repeated stimulation of the pre-synaptic input resulted in a gradual decrease in the attenuating effect on the antidromic response, caused presumably by the fatigue of the synapse. A similar graded reduction of the size of the antidromic response was achieved by gradually increasing or decreasing the pre-synaptic stimulus, but in all cases the percentage decrease of the antidromic potential at the positive spike focus was greater than that at the negative spike focus for a given stimulus.

In addition to the monophasic positive response of the reflex motoneurone, records from the positive spike focus often showed compound action potentials closely following the pre-synaptic stimulus (Fig. 7). These compound potentials follow one-to-one a repetitive pre-synaptic stimulus up to 100/sec. and clearly originate from two size-classes of fibre. The initial short latency component has a lower threshold than the subsequent longer-lasting component. This graded series of records, in which increasing presynaptic stimuli were combined with antidromic stimulation of the axon, shows that the antidromic response was diminished only when the second component of the compound potential was present. Also during purely orthodromic stimulation a postsynaptic spike never arose in the motoneurone unless the second component in the compound spike was present. These results indicate but do not unequivocably establish a causal relation between the compound spike and the orthodromic response. The second part of the compound potential presumably represents the activity of the afferent fibres near their termination and synaptic junction with the motoneurone.

The sequence of events in the axon can be reconstructed as follows : local (or spike) potentials in the dendrites generate a post-synaptic potential which is conveyed electrotonically along a non-propagating part of the axon to an impulse-initiating point. From here the spike is propagated actively along the axon but its effect also spreads passively back towards the synaptic site. The evidence for this view comes from the measured changes in shape and latency of the conducted potential and the different sensitivities of the various portions of the axon to positive and negative stimuli.

The change of shape of the extracellular wave-form is not always sufficient by itself to establish the conducting nature of the membrane (Eccles, 1964), but in the crab brain the change in shape is correlated with latency measurements which show that the monophasic potential appears virtually simultaneously along the whole proximal length of the neurone. Distally, where the extracellular potential is triphasic and predominantly negative, a conduction velocity of 0·9–1·2 m./sec. can be measured. The sensitivity of the system to positive and negative pulses at different points along its axon can also be explained if the initial portion of the axon is non-excitable. In such a system the possible directions of current flow during direct electrical and presynaptic stimulation are shown in Fig. 8.

A negative pulse applied through a focal electrode creates an outward current across the membrane beneath the electrode, and a more diffuse current flow inwards in the surrounding regions. This has the effect of depolarizing the interior of the membrane beneath the electrode. If the current pulse is strong enough and the electrode is positioned over an excitable portion of the axon (as in Fig. 8,a) a regenerative spike will be initiated and will rapidly create an inward current. The flow of current in the surrounding membrane is therefore reversed so that it depolarizes and the spike is propagated. If, on the other hand, a negative pulse is applied to a non-excitable part of the neurone (as in Fig. 8,b) thus depolarizing the region beneath the electrode, a spike will not be initiated because the surrounding membrane is subjected to a continual hyperpolarization by the inward peripheral current. A positive pulse over the same non-excitable region would, if large enough, produce an outward current along the neurone and depolarize the active region, and if reaching threshold a spike would follow (Fig. 8 c).

During synaptic activation of this conceptual system the release of the transmitter lowers the impedance of the post-synaptic membrane (Hagiwara & Tasaki, 1958) resulting in an inward current flow beneath the synaptic terminals and consequent outward depolarizing current flow along the rest of the system (Fig. 8 d). When this outward current is large enough to spread to the region of lowest threshold, the axon fires.

There is some evidence, however, that in the axon under examination there are added complications. Extracellular recording reveals two areas where the extracellular spikes are significantly larger than at other points on the neurone. Also these areas are more sensitive to focal electrical stimulation. These foci may be little more than points where the electrode can make particularly good contact with the axon, and it is also true that relative stimulus intensities are judged only from the size of the applied voltages. Nevertheless, the consistent position of the focal points in different preparations, and the absence of any obvious ‘sealing in’ of the electrode tip argues that the current flow during antidromic spike invasion is concentrated at these meaningful points. The presence of just such a ‘sink’ and ‘source’ of current flow has been described for goldfish Mauthner cells (Furshpan & Furukawa, 1962).

The mechanism of excitation which best fits the experimental results is set out in Fig. 9. During an antidromic response the spike actively invades the neurone as far as the negative spike focus, at which point a large inward current is concentrated. This current is drawn from the positive spike focus situated about 300 μm away and returned along the inside of the axon in the non-excitable portion separating the two focal points. During orthodromic stimulation the outward current generated by synaptic activation is similarly channelled first out through the negative spike focus then reversed by the active phase of the regenerative spike that is generated at the negative spike focus. Whether the channelling is caused by low-resistance areas of the membrane at the positive spike foci is yet to be determined. Such a sink/source system may be a general feature for arthropod motoneurones, and if present in the antennular motoneurones of the crayfish it may provide an explanation for the inability of axon spikes to invade the dendrites there (Maynard, 1966). Also, in the insect Rhodnius prolixus some motoneurones are particularly heavily sheathed for about 120 μm. as they leave the neuropile area of the ganglion and enter the nerve bundle (Wigglesworth, 1959). The sheath does not extend into the dendrite region nor for any distance along the nerve, and any excitatory currents in this region would almost certainly be concentrated at the two ends of the short-sheathed portion.

The Biomac 500 special-purpose computer was obtained with a grant provided to Dr G. A. Horridge by the Medical Research Council.