J. Physiol. (1977), 272, pp. 755-767 With 3 text-firea Prntted in Great Britain
755
DIRECT EXCITATORY INTERACTIONS BETWEEN SPINAL MOTONEURONES OF THE CAT
BY P. GOGAN, J. P. GUERITAUD, GINETTE HORCHOLLE-BOSSAVIT AND SUZANNE TYC-DUMONT From the Department of Physiology, University of G(teborg, Sweden, the Laboratoire de Physiologie nerveuse, LA 204, Gif-sur- Yvette, 91190, France, and Unite' de Neurobiologie, INSERM- Marseille, France (Received 23 March 1977) SUMMARY
1. Ninety-seven spinal motoneurones were identified by their antidromic invasion following stimulation of the muscle nerve and submitted to a series of four tests to reveal a possible direct excitation between motoneurones. 2. Threshold differentiation, refractoriness, hyperpolarization and collision revealed antidromically induced depolarizations in fourteen of the ninety-seven tested motoneurones. 3. The parameters of the antidromically induced depolarizations indicate a short latency, a low amplitude and independence with regard to the membrane polarization. 4. It is concluded that the antidromically induced depolarizations reached the impaled motoneurone via a route other than its own axon. 5. The mechanism may involve either electrotonic interactions between neighbouring motoneurones or excitatory recurrent collaterals between synergist motoneurones. INTRODUCTION
A short latency facilitation induced in cat spinal motoneurones by synchronous antidromic excitation of adjacent motoneurones was described by Nelson in 1966. In the same year, Grinnell (1966) demonstrated the existence of a recurrent excitation of frog spinal motoneurones evoked by antidromic stimulation. More recently, Magherini, Precht & Schwindt (1976) have shown that this recurrent excitation of frog motoneurones is transmitted electrotonically between the soma-dendritic regions of the motoneurone membranes. In the present study,we reconsider the possibility of electrotonic interactions between spinal motoneurones of the cat. Evidence in favour of
P. GOGAN AND OTHERS 7566 electrotonic coupling has been described in some areas of the mammalian nervous system, for instance in the mesencephalic nucleus of the rat (Baker & Llnas, 1971), in the lateral vestibular nucleus of the rat (Korn, Sotelo & Crepel, 1973) and in the inferior olive of the cat (Llinas, Baker & Sotelo, 1974). Electrotonic interactions have also been studied in another population of motoneurones, those of the abducens oculomotor nucleus (Gogan, Gueritaud, Horcholle-Bossavit & Tyc-Dumont, 1974). We describe here membrane depolarizations which are induced by antidromic excitation of the motoneurone pool but which are not dependent on antidromic invasion ofthe impaled cell. The characteristics of the antidromically induced depolarization may fit the hypothesis of electrotonic interaction between spinal motoneurones but it is also necessary to consider the as yet incompletely published description by Cullheim & Kellerth (1976a, b), of proximal axon collaterals ending on synergist motoneurones, as a possible mechanism of origin. METHODS Preparation. The results are taken from experiments performed on four adult cats, anaesthetized with a-chloralose (40-60 mg/kg) after initial induction of anaesthesia with ether.A laminectomywasmade from L4to L7and the dorsal roots L6,L7 and SI were cut. The following nerves were dissected free in one hind limb and were mounted for stimulation: anterior biceps (AB), extensor digitorum longus (EDL), flexor digitorum longus (FDL), lateral gastrocnemius (LG), medial biceps (MB), medial gastrocnemius (MG), the whole peroneal nerve (PER), plantaris (PL), popliteus (Pop), semimembranosus (Sm), posterior biceps and semitendinosus (St), soleus (Sol), tibialis anterior (TA) and tibialis posterior (TP). Blood pressure was monitored throughout the experiment and during recording, the CO2 of the expired air was measured and maintained at 4 %. The animals were paralysed with gallamine triethidiodide and artificially respired. Recording and stimulation. Glass micropipettes filled with 2 M-K citrate, with tip diameters of 1-2-2-5 jum and resistances of 2-5 MD, were used both for intracellular recording from motoneurones in the segments L6-L7 of the spinal cord and for direct stimulation using intracellularly applied constant current pulses. Intracellularly recorded motoneurones were identified by antidromic invasion and the antidromic volley elicited by stimulation of the muscle nerve was recorded simultaneously from the surface of the spinal cord at the same level. The threshold stimulation intensity required to evoke this antidromic volley was defined as T and all subsequent stimuli were expressed as multiples of T. In each case the data were digitalized and the extracellular field potential subtracted from the intracellular record, thus eliminating by calculation the contribution of the former. This procedure, however, never revealed antidromically induced potentials which were not already visible to the eye, and thus the illustrations presented in the following Figures are taken from the original recordings.
DIRECT INTERACTIONS BETWEEN MOTONEURONES
757
RESULTS
Ninety-seven motoneurones were satisfactorily impaled randomly in the segments L6 and L7 of the spinal cord, and identified by antidromic invasion following stimulation of one of the dissected muscle nerves. The tests used to investigate the existence of direct interactions between motoneurones were based on antidromic activation of the motoneurone pool following stimulation of the muscle nerve. If such interactions exist, the excitation of one motoneurone should give rise to a depolarization in other related cells. However, since using this technique the motoneurones are fired nearly synchronously, the occurrence of a full antidromically initiated spike in the recorded neurone could mask any such event. Thus, in order to reveal antidromically evoked depolarizations, it was necessary to avoid the generation of a full spike in the recorded cell. This was achieved (1) by threshold differentiation of antidromic invasion, the antidromic stimulation strength being set just below threshold for the impaled motoneurone; (2) during refractoriness of the impaled cell by giving a second shock to the muscle nerve; (3) by blocking the cell with hyperpolarizing current and (4) by collision of the antidromic spike with an intrasomatically evoked orthodromic action potential. Each impaled motoneurone was systematically submitted to this series of tests. The threshold for antidromic invasion was determined and expressed as a multiple of the threshold for the antidromic volley (T) and the effect of increasing the strength of the peripheral stimulation was observed. Latencies of antidromic excitation were measured at different stimulus intensities. Little variation of antidromic latency was seen between motbneurones of the same pool, using supramaximal stimuli. The refractory period of the IS-SD component of the antidromic spike of the impaled cell was observed and the refractory period of the antidromic volley, representing compound action potentials of the antidromic volley in motor axons and of the antidromic motoneurone field potentials, was also determined. Hyperpolarizing currents (10-20 nA) were used to reveal the M spike and the latency, amplitude and refractory period were noted. Finally, the collision technique was applied to reveal any antidromically induced depolarization reaching the impaled motoneurone via a channel other than its own axon. Fig. 1 illustrates a typical recording where an antidromically induced depolarization was revealed by threshold differentiation and during the refractory period in a hyperpolarized LG motoneurone. The strength of stimulation of the muscle nerve was set straddling the threshold for antidromic invasion and a small depolarization was observed when stimulation failed to initiate a full antidromic spike. This depolarization appeared with
P. GOGAN AND OTHERS 758 a latency 300 ,ssec longer than that of the full antidromic spike (A). In B, the first antidromically induced depolarization is followed by a second small depolarization which sometimes appeared for the same strength of stimulation. The antidromically induced depolarization could also be seen in the hyperpolarized cell (Fig. 1 D, E) where the IS-SD components of the action potential were blocked and only the M spike remained. In this case, stimulation of the muscle nerve at 1'2 T evoked a simple M spike (D)
G
A
A
A
A
''-
J
D_%_ A A
1-2 T
1
K'
A 1-4 T
AA
Fig. 1. For legend see facing page.
DIRECT INTERACTIONS BETWEEN MOTONEURONES 759 while stimulation at 14 T evoked an additional small depolarization superimposed on the falling phase of the M spike. The latency of the depolarization in E is similar to that of the antidromically induced depolarization shown in A, though the exact onset is difficult to assess. That these depolarizations differ in origin from the antidromically initiated spike may be seen during the refractoriness of the M spike (Fig. 1 F-K). With a long interval between stimuli the induced depolarization appeared superimposed on each of the two M spikes (F, C) but as the stimulus interval was reduced, a point was reached where the second M spike was still observed but the depolarization on its falling phase was already refractory (H). At a critical smaller interval (J) where the second M spike was also completely refractory (at this point, the amplitude of the antidromic volley was reduced showing that part, though not all, of the axon population was refractory), an all-or-none potential was occasionally initiated, which was smaller than the M spike and had a latency 600 usec longer, being similar to that ofthe second antidromically initiated depolarization seen in response to a single shook in B. This all-or-none event was never observed in response to a single antidromic shock alone. Finally when the shock interval was still further reduced, the second shock was delivered during complete refractoriness of the motoneurone pool, as indicated by the absence of a second antidromic volley, and remained ineffective. A similar phenomenon is described Fig. 2. The records were obtained from an AB + St motoneurone and the refractory periods for the IS and the the SD components of the antidromic action potential are illustrated in A and B. Occasionally, a small all-or-none action potential was superimposed on the M spike; this can be seen in one trace in Fig. 2 B. By gradually hyperpolarizing the motoneurone (C), the IS and M spikes were distingguished and the respective latencies and amplitudes noted. Fig. 1. Antidromically induced depolarization in a LG motoneurone. Intracellular recording during antidromic activation by stimulation of the muscle nerve. Resting potential: -70 mV. A and B, antidromic stimulation at 1-3 T. When antidromic spike fails, one (A, arrow) and sometimes two (B, arrow) depolarizations are revealed. C, antidromic stimulation at 5 T. D, hyperpolarization to reveal M spike. E, at 1-4 T, a depolarization appears on the falling phase of the M spike (arrow). F-K, refractory period of M spike at T during hyperpolarization. F, with sufficient delay, two nearly identical M spikes. G, H, with shorter delays, the depolarization superimposed on the second M spike becomes refractory before the second M spike. J, when stimuli are given at a critical interval (0-8 msec), an all-or-none depolarization appears with an amplitude different from that of the M spike and with a latency of 2-3 msec; the second M spike is blocked by refractoriness. K, for a shorter interval, this depolarizaton does not occur. Calibration: A-C, 2 mV; D-K, 4 mV. Time: 2 msec.
760 P. GOGAN AND OTHERS In D, two shocks were applied to the muscle nerve at an interval of 1-2 msec and a two component additional depolarization was seen on the falling phase of each M spike thus antidromically evoked. Reduction of the interstimulus interval in E resulted in the disappearance of the second component of the antidromically induced depolarization evoked by the
D
A
A
B
A
A
A
A
C
A
Fig. 2. For legend see facing page.
DIRECT INTERACTIONS BETWEEN MOTONEURONES 761 second stimulus and this was completely refractory at the interval shown in F. As the inter-shock interval was reduced, the second component of the first depolarization became masked by the onset of the M spike evoked by the second stimulus, as can be seen in F. The antidromically induced depolarization thus became refractory before the M spike. At the interval shown in G, the second stimulus was delivered during the period of complete refractoriness and only the first M spike and following two component depolarization were seen. Occasionally, an all-or-none local response was initiated and though this could occur in a non-hyperpolarized cell (Fig. 2B), it was never observed in response to a single shock to the muscle nerve but only when two stimuli were spaced at a critical interval. To ascertain whether the antidromically induced depolarization reached the impaled motoneurone via its own axon or via other motoneurones, the technique of collision was applied to all the cells studied. In Fig. 3, during the refractory period for antidromic invasion (A-D), a critical interval between shocks to the muscle nerve (B) revealed two successive depolarizations after the first action potential. The earlier of these had the same latency as the full antidromic spike evoked by the second shock and was thus considered as the M spike. When the interval between stimuli was further reduced (C) the M spike failed, indicating that the axon of the impaled motoneurone was refractory but the second depolarization persisted with a latency identical to that in Fig. 3B. Still greater diminution of the inter-shock interval also abolished this later depolarization, together with the antidromic volley, thus indicating that all motoneurones were refractory. The effect of collision of orthodromically and antidromically initiated Fig. 2. Intracellular recordings from AB/St motoneurone during antidromic stimulation at 5 T. Membrane resting potential: -75 mV. A, upper trace: intracellular recordings; lower trace: surface recording of antidromic volley. Partial blockade of second antidromic spike revealing the IS component. B, further reduction of shock intervals to show M spike on which an all-or-none potential occasionally appears. C, gradual hyperpolarization of the cell. Upper trace: IS spike; middle trace: M spike; lower trace: afferent volley recorded from the surface of the spinal cord. D-G, refractory period of M spike in hyperpolarized cell. D, two identified M spikes with a two component depolarization (arrow) superimposed on the falling phase of each. F, the second M spike is becoming refractory, revealing the underlying depolarization when it fails. G, the second antidromic stimulus is given at a critical interval during refractoriness of the motoneurone and only the first M spike and the two following depolarizations are seen. Occasionally an all-or-none potential of the same amplitude as that in B is fired. Shock artifacts indicated by A. Five superimposed sweeps for each recording. Calibration: A-C, 10 mV; D-G, 1 mV. Time: 2 msec.
P. GOGAN AND OTHERS 762 action potentials in the impaled motoneurones is illustrated in Fig.3E-H. It was assessed that the axon of the impaled motoneurone was blocked when the shock to the muscle nerve was delivered within less than once the antidromic conduction time after the initiation of an orthodromic action
E
A
_0061[m
0
_-
F
B
I! G
k
C-I-
-10 nA -
44i _o * J_ a-lwA V
H
*"n
D
Fig. 3. For legend see facing page.
DIRECT INTERACTIONS BETWEEN MOTONEURONES 763 potential, by a depolarizing current pulse via the micro-electrode. Under these conditions the antidromically induced depolarization was revealed as the two action potentials in the impaled motoneurone collided (Fig. 3E). Its latency was delayed with respect to the initiation of the full antidromic action potential. Occasionally, it was observed that when a depolarizing current was applied via the micro-electrode during the collision period in TABTi 1. Parameters of motoneurones tested
Motoneurones Peroneus Pop + Tpost
Sm+MB EDL MG St Long. AB FDL LG St Long. MG
AB+Sm AB+Sm AB+Sm
Threshold differentiation + 0 + + + Not tested Not tested 0 + 0 + + + +
Latency of the depolarization measured from Amplitude of antidromic spike the depolarization
(#Aec)
(T&V)
-200 +200 +200 -300 + 600 0 +200 +200 + 300 + 750 + 700 +300 +700 + 600
200 800 2000 700 1200 300 ? 1000 600 250 500 500 900 200
Fig. 3. Intracellular recordings from a FDL motoneurone during antidromic stimulation at 5 T. Membrane resting potential: -32 mV. Upper trace: intracellular recording; lower trace: antidromic volley recorded on the surface of the spinal cord. Five superimposed sweeps in each case. A, refractory period; the second antidromic shock shows the IS component of the spike. B, further reduction of the interval between the shocks reveals the M spike followed by a depolarization (arrow). C, shorter interval suppressed the M spike, whereas the depolarization disappears only when the antidromic volley is also refractory (D). E, direct activation of the motoneurone is followed by an antidromic stimulation applied during the collision time. This stimulation gives rise to a depolarization with a latency 200 #asec longer than the latency of the full antidromic spike obtained without intracellular stimulation. F, same procedure as E, but duration of intracellular stimulation was widened to 8 msec; an all-or-none potential is added on the depolarization. G, with membrane hyperpolarization, the depolarization is maintained and the all-or-none nature of the potential appears. H, stronger hyperpolarization blocks this potential whereas the depolarization i4 still observed. Calibration: 10 mV, Time: square pulses 1 Msec.
P. GOGAN AND OTHERS 764 order to increase the motoneurone membrane excitability, a small all-ornone response was superimposed on the antidromically induced depolarization (F). This local response was often observed in the depolarized cell. When the excitability was reduced by passing hyperpolarizing current (Fig. 3 G), it was observed that the local response was an all-or-none event which was suppressed by further hyperpolarization (Fig. 3H), although the antidromically induced depolarization remained. Antidromically induced depolarizations, as thus described, were found in fourteen of the ninety-seven motoneurones which were tested. Their parameters are summarized in Table 1. It can be seen that latencies of the antidromically induced depolarization with respect to the initiation of the antidromic action potential in the impaled motoneurones are in the range - 200 to + 750 ssec. In eleven motoneurones the antidromically induced depolarization was delayed, in one the latencies were identical and in the remaining two cases the depolarization preceded antidromic spike invasion by 200 and 300 Itsec respectively. The amplitude of the antidromically induced depolarization was small, generally less than 1 mV. Antidromic threshold differentiation was possible in nine cases. These depolarizations were observed in eleven extensor motoneurones and in three flexor motoneurones but this apparent difference may simply have been due to the random tracking in the spinal cord. DISCUSSION
These results show that antidromic activation of a motoneurone pool, in a preparation in which the dorsal roots were cut, may evoke small transitory depolarizations in a small percentage of the tested motoneurones which are of an origin different from that of the antidromic spike invasion of these particular cells. Such antidromically induced depolarizations can only be observed in an impaled cell in the absence of the full antidromic action potential since the latter masks all small amplitude phenomena of almost identical latency. Antidromic spike invasion could be differentiated from the antidromically induced depolarizations by their different thresholds to stimulation of the muscle nerve and by their different refractory period characteristics. Antidromically induced depolarizations may also be evidenced during collision of the antidromic action potential of the impaled cell with an orthodromically initiated spike. Together, these facts indicate that the antidromically induced depolarizations thus characterized originate via a route other than the axon of the cell in question, but are the result of the antidromic excitation of neighbouring cells. The mechanism by which such an excitation is transmitted remains a subject for hypothesis. However, it is shown that it implies a rapid trans-
DIRECT INTERACTIONS BETWEEN MOTONEURONES 765 mission between motoneurones. The latency difference between antidromic spike invasion and the onset of an antidromically induced depolarization varied from -200 psec to +750psec. Based on current knowledge, two alternative hypotheses may be proposed concerning the mechanism involved. The first is that of an electrotonic interaction between certain motoneurones, though the absence of a description of gap junctions between mammalian spinal motoneurones would seem to diminish the postulate of an electrotonic coupling. However, by analogy with that which has recently been suggested in the spinal cord of the frog, it may be possible to envisage electrotonic interactions occurring between dendritic regions of the motoneurones. In experiments on frog spinal motoneurones, in which dendritic spikes have been demonstrated (Czeh, 1972), Magherini et al. (1976) have described a recurrent e.p.s.p. supposed to be transmitted between the soma-dendritic regions of neighbouring motoneurones by electrotonic transmission. Since junctional specialization (gap junctions) are said to be extremely rare in frog motoneurones (Sotelo & Taxi, 1970; Stensaas & Stensaas, 1971), it has recently been proposed that appositional contacts of dendritic electrogenic membranes may be an adequate structure for electrotonic transmission (Szekely & Kosaras, 1976). Close appositions of dendritic membranes have also been described in cat spinal motoneurones (Matthews, Willis & Williams, 1971; Scheibel & Scheibel, 1975) and recent electrophysiological results suggest that the rising phase of the I a afferent fibre monosynaptic e.p.s.p. in cat spinal motoneurones may result from an electrotonic coupling (Werman & Carlen, 1976). This corresponds to the theoretical model proposed by Rall in 1967. The second hypothesis is based on the recent demonstration by Culheim & Kellerth (1976a, b) of motoneurone axon collaterals which originate close to the motoneurone soma and form synaptic terminals on synergistic motoneurones. In this case, the antidromically induced depolarizations might represent small amplitude e.p.s.p.s from synergist motoneurones which are also antidromically activated by stimulation of the muscle nerve. Since the antidromic spike latencies for motoneurones from a given motoneurone pool were observed to be relatively uniform, it would perhaps then be necessary to postulate a rapid transmission in the cases where the difference in latency was only 200 pusec. It is difficult to reconcile the short latency of the depolarizations with the conventional values given for synaptic delays and impulse propagation in their axon collateral. In these experiments it was interesting to note that local responses were sometimes initiated from an antidromically induced depolarization, as illustrated in the three figures. These were all or none phenomena of larger amplitude than the antidromically induced depolarization. They
766
P. GOGAN AND OTHERS could be facilitated or blocked by modification of the membrane polarization showing them to be transmembrane depolarizations generated by the membrane of the impaled cell. In addition, the observation that such local responses could be initiated even during blockade of the axon of the recorded cell probably indicates that they are generated in the dendritic region of the recorded neurone and conducted in a decremental manner to the recording site in the soma. If it is assumed that the local responses observed were action potentials of dendritic origin, it is also of interest to note that they were initiated only under conditions of critical timing. In each case, the dendritic spike was associated with an antidromically induced depolarization and appeared as an all or none potential change superimposed on the latter. One might thus hypothesize that the dendritic spike occurs only when the motoneurone is sufficiently depolarized or when the timing of two successive antidromic shocks is such that temporal and/or spatial summation of several antidromically induced depolarizations may result. In conclusion, it is impossible to distinguish between these two hypotheses on the basis of the results presented here, and indeed, short of direct demonstration by simultaneous recording from two motoneurones with such interaction, it is difficult to see how the problem might be resolved electrophysiologically. The characteristics of the depolarizations described in this paper might fit either situation and it is perhaps unwise to assume that the two hypotheses should be mutually exclusive. Our gratitude is due to Dr E. Jankowska for her invaluable help in the preparation of the experiments, for all her advice in the course of the experiments and for her detailed discussion of the manuscript. We would also like to thank Mrs R. Larsson for her excellent technical assistance. This work was supported by a twinning grant from the European Training Program in Brain and Behaviour Research. REFERENCES BAERT, R. & LLINAs, R. (1971). Electrotonic coupling between neurones in the rat mesencephalic nucleus. J. Phygiol. 212, 45-63. CuLrLHEIM, S. & KETLERTH, J.-O. (1976a). Combined light and electron microscopical tracing of lumbar motor axon collaterals and their synaptic terminals in the cat, after intracellular injection of horseradish peroxidase. Acta phy8iol. Wcand. 98, suppl. 440. COULHaEm, S. & KELLERTH, J.-O. (1976b). Combined light and electron microscopic tracing of neurons, including axons and synaptic terminals, after intracellular injection of horseradish peroxidase. Neuromci. Lett, 2, 307-313. CzEH, G. (1972). The role of dendritic events in the initiation of monosynaptic spikes in frog motoneurones. Brain Re8. 39, 505-509. GoG.AN, P., GuERITAUD, J. P., HORCHOT.T-BossAvrI, G. & Tyc-DuMoNr, S. (1974). Electrotonic coupling between motoneurones in the abducens nucleus of the cat. Expl Brain Re8. 21, 139-154.
DIRECT INTERACTIONS BETWEEN MOTONEURONES 767 GRINNELL, A. D. (1966). A study of the interaction between motoneurones in the frog spinal cord. J. Physiol. 182, 612-648. KORN, H., SOTELO, C. & CREPEL, F. (1973). Electrotonic coupling between neurons in rat lateral vestibular nucleus. Expl Brain Re8. 16, 255-275. LLINAs, R., BAKER, R. & SOTELO, C. (1974). Electrotonic coupling between neurons in cat inferior olive. J. Neurophysiol. 37, 560-571. MAGHERINI, P. C., PRECHT, W. & SCHWINDT, P. C. (1976). Evidence for electrotonic coupling between frog motoneurons in the in situ spinal cord. J. Neurophysiol. 39, 474-483. MATTHEWS, M.A., WAis, W. D. & WiLLAkMs, V. (1971). Dendrite bundles in lamina IX of cat spinal cord: a possible source for electrical interaction between motoneurons. Anat. Rec. 171, 313-328. NELSON, P. G. (1966). Interaction between spinal motoneurons of the cat. J. Neurophysiol. 29, 275-287. RALL, W. (1967). Distinguishing theoretical potentials computed for different somadentritic distribution of synaptic input. J. Neurophysiol. 30, 1138-1168. SCHEIBEL, M. E. & SCHEIBEL, A. B. (1975). Dendrites as neuronal couplers: the dentritic bundle. In Golgi Centennial Symposium, ed SANTN, M., pp. 347-354. New York: Raven. SOTELO, C. & TAXI, J. (1970). Ultrastructural aspects of electrotonic junctions in the spinal cord of the frog. Brain Re8. 17, 137-141. STENSAAS, L. J. & STENSAAS, S. S. (1971). Light and electron microscopy of motoneurons and neuropile in the amphibian spinal cord. Brain Re8. 31, 67-84. SZEKELY, G. & KosAlAs, B. (1976). Dendro-dendritic contacts between frog motoneurons shown with the cobalt labelling technique. Brain Ree. 108, 194-198. WERMAN, R. & CARLEN, P. L. (1976). Unusual behaviour of the Ia EPSP in cat spinal motoneurons. Brain Re8. 112, 395-401.
755
DIRECT EXCITATORY INTERACTIONS BETWEEN SPINAL MOTONEURONES OF THE CAT
BY P. GOGAN, J. P. GUERITAUD, GINETTE HORCHOLLE-BOSSAVIT AND SUZANNE TYC-DUMONT From the Department of Physiology, University of G(teborg, Sweden, the Laboratoire de Physiologie nerveuse, LA 204, Gif-sur- Yvette, 91190, France, and Unite' de Neurobiologie, INSERM- Marseille, France (Received 23 March 1977) SUMMARY
1. Ninety-seven spinal motoneurones were identified by their antidromic invasion following stimulation of the muscle nerve and submitted to a series of four tests to reveal a possible direct excitation between motoneurones. 2. Threshold differentiation, refractoriness, hyperpolarization and collision revealed antidromically induced depolarizations in fourteen of the ninety-seven tested motoneurones. 3. The parameters of the antidromically induced depolarizations indicate a short latency, a low amplitude and independence with regard to the membrane polarization. 4. It is concluded that the antidromically induced depolarizations reached the impaled motoneurone via a route other than its own axon. 5. The mechanism may involve either electrotonic interactions between neighbouring motoneurones or excitatory recurrent collaterals between synergist motoneurones. INTRODUCTION
A short latency facilitation induced in cat spinal motoneurones by synchronous antidromic excitation of adjacent motoneurones was described by Nelson in 1966. In the same year, Grinnell (1966) demonstrated the existence of a recurrent excitation of frog spinal motoneurones evoked by antidromic stimulation. More recently, Magherini, Precht & Schwindt (1976) have shown that this recurrent excitation of frog motoneurones is transmitted electrotonically between the soma-dendritic regions of the motoneurone membranes. In the present study,we reconsider the possibility of electrotonic interactions between spinal motoneurones of the cat. Evidence in favour of
P. GOGAN AND OTHERS 7566 electrotonic coupling has been described in some areas of the mammalian nervous system, for instance in the mesencephalic nucleus of the rat (Baker & Llnas, 1971), in the lateral vestibular nucleus of the rat (Korn, Sotelo & Crepel, 1973) and in the inferior olive of the cat (Llinas, Baker & Sotelo, 1974). Electrotonic interactions have also been studied in another population of motoneurones, those of the abducens oculomotor nucleus (Gogan, Gueritaud, Horcholle-Bossavit & Tyc-Dumont, 1974). We describe here membrane depolarizations which are induced by antidromic excitation of the motoneurone pool but which are not dependent on antidromic invasion ofthe impaled cell. The characteristics of the antidromically induced depolarization may fit the hypothesis of electrotonic interaction between spinal motoneurones but it is also necessary to consider the as yet incompletely published description by Cullheim & Kellerth (1976a, b), of proximal axon collaterals ending on synergist motoneurones, as a possible mechanism of origin. METHODS Preparation. The results are taken from experiments performed on four adult cats, anaesthetized with a-chloralose (40-60 mg/kg) after initial induction of anaesthesia with ether.A laminectomywasmade from L4to L7and the dorsal roots L6,L7 and SI were cut. The following nerves were dissected free in one hind limb and were mounted for stimulation: anterior biceps (AB), extensor digitorum longus (EDL), flexor digitorum longus (FDL), lateral gastrocnemius (LG), medial biceps (MB), medial gastrocnemius (MG), the whole peroneal nerve (PER), plantaris (PL), popliteus (Pop), semimembranosus (Sm), posterior biceps and semitendinosus (St), soleus (Sol), tibialis anterior (TA) and tibialis posterior (TP). Blood pressure was monitored throughout the experiment and during recording, the CO2 of the expired air was measured and maintained at 4 %. The animals were paralysed with gallamine triethidiodide and artificially respired. Recording and stimulation. Glass micropipettes filled with 2 M-K citrate, with tip diameters of 1-2-2-5 jum and resistances of 2-5 MD, were used both for intracellular recording from motoneurones in the segments L6-L7 of the spinal cord and for direct stimulation using intracellularly applied constant current pulses. Intracellularly recorded motoneurones were identified by antidromic invasion and the antidromic volley elicited by stimulation of the muscle nerve was recorded simultaneously from the surface of the spinal cord at the same level. The threshold stimulation intensity required to evoke this antidromic volley was defined as T and all subsequent stimuli were expressed as multiples of T. In each case the data were digitalized and the extracellular field potential subtracted from the intracellular record, thus eliminating by calculation the contribution of the former. This procedure, however, never revealed antidromically induced potentials which were not already visible to the eye, and thus the illustrations presented in the following Figures are taken from the original recordings.
DIRECT INTERACTIONS BETWEEN MOTONEURONES
757
RESULTS
Ninety-seven motoneurones were satisfactorily impaled randomly in the segments L6 and L7 of the spinal cord, and identified by antidromic invasion following stimulation of one of the dissected muscle nerves. The tests used to investigate the existence of direct interactions between motoneurones were based on antidromic activation of the motoneurone pool following stimulation of the muscle nerve. If such interactions exist, the excitation of one motoneurone should give rise to a depolarization in other related cells. However, since using this technique the motoneurones are fired nearly synchronously, the occurrence of a full antidromically initiated spike in the recorded neurone could mask any such event. Thus, in order to reveal antidromically evoked depolarizations, it was necessary to avoid the generation of a full spike in the recorded cell. This was achieved (1) by threshold differentiation of antidromic invasion, the antidromic stimulation strength being set just below threshold for the impaled motoneurone; (2) during refractoriness of the impaled cell by giving a second shock to the muscle nerve; (3) by blocking the cell with hyperpolarizing current and (4) by collision of the antidromic spike with an intrasomatically evoked orthodromic action potential. Each impaled motoneurone was systematically submitted to this series of tests. The threshold for antidromic invasion was determined and expressed as a multiple of the threshold for the antidromic volley (T) and the effect of increasing the strength of the peripheral stimulation was observed. Latencies of antidromic excitation were measured at different stimulus intensities. Little variation of antidromic latency was seen between motbneurones of the same pool, using supramaximal stimuli. The refractory period of the IS-SD component of the antidromic spike of the impaled cell was observed and the refractory period of the antidromic volley, representing compound action potentials of the antidromic volley in motor axons and of the antidromic motoneurone field potentials, was also determined. Hyperpolarizing currents (10-20 nA) were used to reveal the M spike and the latency, amplitude and refractory period were noted. Finally, the collision technique was applied to reveal any antidromically induced depolarization reaching the impaled motoneurone via a channel other than its own axon. Fig. 1 illustrates a typical recording where an antidromically induced depolarization was revealed by threshold differentiation and during the refractory period in a hyperpolarized LG motoneurone. The strength of stimulation of the muscle nerve was set straddling the threshold for antidromic invasion and a small depolarization was observed when stimulation failed to initiate a full antidromic spike. This depolarization appeared with
P. GOGAN AND OTHERS 758 a latency 300 ,ssec longer than that of the full antidromic spike (A). In B, the first antidromically induced depolarization is followed by a second small depolarization which sometimes appeared for the same strength of stimulation. The antidromically induced depolarization could also be seen in the hyperpolarized cell (Fig. 1 D, E) where the IS-SD components of the action potential were blocked and only the M spike remained. In this case, stimulation of the muscle nerve at 1'2 T evoked a simple M spike (D)
G
A
A
A
A
''-
J
D_%_ A A
1-2 T
1
K'
A 1-4 T
AA
Fig. 1. For legend see facing page.
DIRECT INTERACTIONS BETWEEN MOTONEURONES 759 while stimulation at 14 T evoked an additional small depolarization superimposed on the falling phase of the M spike. The latency of the depolarization in E is similar to that of the antidromically induced depolarization shown in A, though the exact onset is difficult to assess. That these depolarizations differ in origin from the antidromically initiated spike may be seen during the refractoriness of the M spike (Fig. 1 F-K). With a long interval between stimuli the induced depolarization appeared superimposed on each of the two M spikes (F, C) but as the stimulus interval was reduced, a point was reached where the second M spike was still observed but the depolarization on its falling phase was already refractory (H). At a critical smaller interval (J) where the second M spike was also completely refractory (at this point, the amplitude of the antidromic volley was reduced showing that part, though not all, of the axon population was refractory), an all-or-none potential was occasionally initiated, which was smaller than the M spike and had a latency 600 usec longer, being similar to that ofthe second antidromically initiated depolarization seen in response to a single shook in B. This all-or-none event was never observed in response to a single antidromic shock alone. Finally when the shock interval was still further reduced, the second shock was delivered during complete refractoriness of the motoneurone pool, as indicated by the absence of a second antidromic volley, and remained ineffective. A similar phenomenon is described Fig. 2. The records were obtained from an AB + St motoneurone and the refractory periods for the IS and the the SD components of the antidromic action potential are illustrated in A and B. Occasionally, a small all-or-none action potential was superimposed on the M spike; this can be seen in one trace in Fig. 2 B. By gradually hyperpolarizing the motoneurone (C), the IS and M spikes were distingguished and the respective latencies and amplitudes noted. Fig. 1. Antidromically induced depolarization in a LG motoneurone. Intracellular recording during antidromic activation by stimulation of the muscle nerve. Resting potential: -70 mV. A and B, antidromic stimulation at 1-3 T. When antidromic spike fails, one (A, arrow) and sometimes two (B, arrow) depolarizations are revealed. C, antidromic stimulation at 5 T. D, hyperpolarization to reveal M spike. E, at 1-4 T, a depolarization appears on the falling phase of the M spike (arrow). F-K, refractory period of M spike at T during hyperpolarization. F, with sufficient delay, two nearly identical M spikes. G, H, with shorter delays, the depolarization superimposed on the second M spike becomes refractory before the second M spike. J, when stimuli are given at a critical interval (0-8 msec), an all-or-none depolarization appears with an amplitude different from that of the M spike and with a latency of 2-3 msec; the second M spike is blocked by refractoriness. K, for a shorter interval, this depolarizaton does not occur. Calibration: A-C, 2 mV; D-K, 4 mV. Time: 2 msec.
760 P. GOGAN AND OTHERS In D, two shocks were applied to the muscle nerve at an interval of 1-2 msec and a two component additional depolarization was seen on the falling phase of each M spike thus antidromically evoked. Reduction of the interstimulus interval in E resulted in the disappearance of the second component of the antidromically induced depolarization evoked by the
D
A
A
B
A
A
A
A
C
A
Fig. 2. For legend see facing page.
DIRECT INTERACTIONS BETWEEN MOTONEURONES 761 second stimulus and this was completely refractory at the interval shown in F. As the inter-shock interval was reduced, the second component of the first depolarization became masked by the onset of the M spike evoked by the second stimulus, as can be seen in F. The antidromically induced depolarization thus became refractory before the M spike. At the interval shown in G, the second stimulus was delivered during the period of complete refractoriness and only the first M spike and following two component depolarization were seen. Occasionally, an all-or-none local response was initiated and though this could occur in a non-hyperpolarized cell (Fig. 2B), it was never observed in response to a single shock to the muscle nerve but only when two stimuli were spaced at a critical interval. To ascertain whether the antidromically induced depolarization reached the impaled motoneurone via its own axon or via other motoneurones, the technique of collision was applied to all the cells studied. In Fig. 3, during the refractory period for antidromic invasion (A-D), a critical interval between shocks to the muscle nerve (B) revealed two successive depolarizations after the first action potential. The earlier of these had the same latency as the full antidromic spike evoked by the second shock and was thus considered as the M spike. When the interval between stimuli was further reduced (C) the M spike failed, indicating that the axon of the impaled motoneurone was refractory but the second depolarization persisted with a latency identical to that in Fig. 3B. Still greater diminution of the inter-shock interval also abolished this later depolarization, together with the antidromic volley, thus indicating that all motoneurones were refractory. The effect of collision of orthodromically and antidromically initiated Fig. 2. Intracellular recordings from AB/St motoneurone during antidromic stimulation at 5 T. Membrane resting potential: -75 mV. A, upper trace: intracellular recordings; lower trace: surface recording of antidromic volley. Partial blockade of second antidromic spike revealing the IS component. B, further reduction of shock intervals to show M spike on which an all-or-none potential occasionally appears. C, gradual hyperpolarization of the cell. Upper trace: IS spike; middle trace: M spike; lower trace: afferent volley recorded from the surface of the spinal cord. D-G, refractory period of M spike in hyperpolarized cell. D, two identified M spikes with a two component depolarization (arrow) superimposed on the falling phase of each. F, the second M spike is becoming refractory, revealing the underlying depolarization when it fails. G, the second antidromic stimulus is given at a critical interval during refractoriness of the motoneurone and only the first M spike and the two following depolarizations are seen. Occasionally an all-or-none potential of the same amplitude as that in B is fired. Shock artifacts indicated by A. Five superimposed sweeps for each recording. Calibration: A-C, 10 mV; D-G, 1 mV. Time: 2 msec.
P. GOGAN AND OTHERS 762 action potentials in the impaled motoneurones is illustrated in Fig.3E-H. It was assessed that the axon of the impaled motoneurone was blocked when the shock to the muscle nerve was delivered within less than once the antidromic conduction time after the initiation of an orthodromic action
E
A
_0061[m
0
_-
F
B
I! G
k
C-I-
-10 nA -
44i _o * J_ a-lwA V
H
*"n
D
Fig. 3. For legend see facing page.
DIRECT INTERACTIONS BETWEEN MOTONEURONES 763 potential, by a depolarizing current pulse via the micro-electrode. Under these conditions the antidromically induced depolarization was revealed as the two action potentials in the impaled motoneurone collided (Fig. 3E). Its latency was delayed with respect to the initiation of the full antidromic action potential. Occasionally, it was observed that when a depolarizing current was applied via the micro-electrode during the collision period in TABTi 1. Parameters of motoneurones tested
Motoneurones Peroneus Pop + Tpost
Sm+MB EDL MG St Long. AB FDL LG St Long. MG
AB+Sm AB+Sm AB+Sm
Threshold differentiation + 0 + + + Not tested Not tested 0 + 0 + + + +
Latency of the depolarization measured from Amplitude of antidromic spike the depolarization
(#Aec)
(T&V)
-200 +200 +200 -300 + 600 0 +200 +200 + 300 + 750 + 700 +300 +700 + 600
200 800 2000 700 1200 300 ? 1000 600 250 500 500 900 200
Fig. 3. Intracellular recordings from a FDL motoneurone during antidromic stimulation at 5 T. Membrane resting potential: -32 mV. Upper trace: intracellular recording; lower trace: antidromic volley recorded on the surface of the spinal cord. Five superimposed sweeps in each case. A, refractory period; the second antidromic shock shows the IS component of the spike. B, further reduction of the interval between the shocks reveals the M spike followed by a depolarization (arrow). C, shorter interval suppressed the M spike, whereas the depolarization disappears only when the antidromic volley is also refractory (D). E, direct activation of the motoneurone is followed by an antidromic stimulation applied during the collision time. This stimulation gives rise to a depolarization with a latency 200 #asec longer than the latency of the full antidromic spike obtained without intracellular stimulation. F, same procedure as E, but duration of intracellular stimulation was widened to 8 msec; an all-or-none potential is added on the depolarization. G, with membrane hyperpolarization, the depolarization is maintained and the all-or-none nature of the potential appears. H, stronger hyperpolarization blocks this potential whereas the depolarization i4 still observed. Calibration: 10 mV, Time: square pulses 1 Msec.
P. GOGAN AND OTHERS 764 order to increase the motoneurone membrane excitability, a small all-ornone response was superimposed on the antidromically induced depolarization (F). This local response was often observed in the depolarized cell. When the excitability was reduced by passing hyperpolarizing current (Fig. 3 G), it was observed that the local response was an all-or-none event which was suppressed by further hyperpolarization (Fig. 3H), although the antidromically induced depolarization remained. Antidromically induced depolarizations, as thus described, were found in fourteen of the ninety-seven motoneurones which were tested. Their parameters are summarized in Table 1. It can be seen that latencies of the antidromically induced depolarization with respect to the initiation of the antidromic action potential in the impaled motoneurones are in the range - 200 to + 750 ssec. In eleven motoneurones the antidromically induced depolarization was delayed, in one the latencies were identical and in the remaining two cases the depolarization preceded antidromic spike invasion by 200 and 300 Itsec respectively. The amplitude of the antidromically induced depolarization was small, generally less than 1 mV. Antidromic threshold differentiation was possible in nine cases. These depolarizations were observed in eleven extensor motoneurones and in three flexor motoneurones but this apparent difference may simply have been due to the random tracking in the spinal cord. DISCUSSION
These results show that antidromic activation of a motoneurone pool, in a preparation in which the dorsal roots were cut, may evoke small transitory depolarizations in a small percentage of the tested motoneurones which are of an origin different from that of the antidromic spike invasion of these particular cells. Such antidromically induced depolarizations can only be observed in an impaled cell in the absence of the full antidromic action potential since the latter masks all small amplitude phenomena of almost identical latency. Antidromic spike invasion could be differentiated from the antidromically induced depolarizations by their different thresholds to stimulation of the muscle nerve and by their different refractory period characteristics. Antidromically induced depolarizations may also be evidenced during collision of the antidromic action potential of the impaled cell with an orthodromically initiated spike. Together, these facts indicate that the antidromically induced depolarizations thus characterized originate via a route other than the axon of the cell in question, but are the result of the antidromic excitation of neighbouring cells. The mechanism by which such an excitation is transmitted remains a subject for hypothesis. However, it is shown that it implies a rapid trans-
DIRECT INTERACTIONS BETWEEN MOTONEURONES 765 mission between motoneurones. The latency difference between antidromic spike invasion and the onset of an antidromically induced depolarization varied from -200 psec to +750psec. Based on current knowledge, two alternative hypotheses may be proposed concerning the mechanism involved. The first is that of an electrotonic interaction between certain motoneurones, though the absence of a description of gap junctions between mammalian spinal motoneurones would seem to diminish the postulate of an electrotonic coupling. However, by analogy with that which has recently been suggested in the spinal cord of the frog, it may be possible to envisage electrotonic interactions occurring between dendritic regions of the motoneurones. In experiments on frog spinal motoneurones, in which dendritic spikes have been demonstrated (Czeh, 1972), Magherini et al. (1976) have described a recurrent e.p.s.p. supposed to be transmitted between the soma-dendritic regions of neighbouring motoneurones by electrotonic transmission. Since junctional specialization (gap junctions) are said to be extremely rare in frog motoneurones (Sotelo & Taxi, 1970; Stensaas & Stensaas, 1971), it has recently been proposed that appositional contacts of dendritic electrogenic membranes may be an adequate structure for electrotonic transmission (Szekely & Kosaras, 1976). Close appositions of dendritic membranes have also been described in cat spinal motoneurones (Matthews, Willis & Williams, 1971; Scheibel & Scheibel, 1975) and recent electrophysiological results suggest that the rising phase of the I a afferent fibre monosynaptic e.p.s.p. in cat spinal motoneurones may result from an electrotonic coupling (Werman & Carlen, 1976). This corresponds to the theoretical model proposed by Rall in 1967. The second hypothesis is based on the recent demonstration by Culheim & Kellerth (1976a, b) of motoneurone axon collaterals which originate close to the motoneurone soma and form synaptic terminals on synergistic motoneurones. In this case, the antidromically induced depolarizations might represent small amplitude e.p.s.p.s from synergist motoneurones which are also antidromically activated by stimulation of the muscle nerve. Since the antidromic spike latencies for motoneurones from a given motoneurone pool were observed to be relatively uniform, it would perhaps then be necessary to postulate a rapid transmission in the cases where the difference in latency was only 200 pusec. It is difficult to reconcile the short latency of the depolarizations with the conventional values given for synaptic delays and impulse propagation in their axon collateral. In these experiments it was interesting to note that local responses were sometimes initiated from an antidromically induced depolarization, as illustrated in the three figures. These were all or none phenomena of larger amplitude than the antidromically induced depolarization. They
766
P. GOGAN AND OTHERS could be facilitated or blocked by modification of the membrane polarization showing them to be transmembrane depolarizations generated by the membrane of the impaled cell. In addition, the observation that such local responses could be initiated even during blockade of the axon of the recorded cell probably indicates that they are generated in the dendritic region of the recorded neurone and conducted in a decremental manner to the recording site in the soma. If it is assumed that the local responses observed were action potentials of dendritic origin, it is also of interest to note that they were initiated only under conditions of critical timing. In each case, the dendritic spike was associated with an antidromically induced depolarization and appeared as an all or none potential change superimposed on the latter. One might thus hypothesize that the dendritic spike occurs only when the motoneurone is sufficiently depolarized or when the timing of two successive antidromic shocks is such that temporal and/or spatial summation of several antidromically induced depolarizations may result. In conclusion, it is impossible to distinguish between these two hypotheses on the basis of the results presented here, and indeed, short of direct demonstration by simultaneous recording from two motoneurones with such interaction, it is difficult to see how the problem might be resolved electrophysiologically. The characteristics of the depolarizations described in this paper might fit either situation and it is perhaps unwise to assume that the two hypotheses should be mutually exclusive. Our gratitude is due to Dr E. Jankowska for her invaluable help in the preparation of the experiments, for all her advice in the course of the experiments and for her detailed discussion of the manuscript. We would also like to thank Mrs R. Larsson for her excellent technical assistance. This work was supported by a twinning grant from the European Training Program in Brain and Behaviour Research. REFERENCES BAERT, R. & LLINAs, R. (1971). Electrotonic coupling between neurones in the rat mesencephalic nucleus. J. Phygiol. 212, 45-63. CuLrLHEIM, S. & KETLERTH, J.-O. (1976a). Combined light and electron microscopical tracing of lumbar motor axon collaterals and their synaptic terminals in the cat, after intracellular injection of horseradish peroxidase. Acta phy8iol. Wcand. 98, suppl. 440. COULHaEm, S. & KELLERTH, J.-O. (1976b). Combined light and electron microscopic tracing of neurons, including axons and synaptic terminals, after intracellular injection of horseradish peroxidase. Neuromci. Lett, 2, 307-313. CzEH, G. (1972). The role of dendritic events in the initiation of monosynaptic spikes in frog motoneurones. Brain Re8. 39, 505-509. GoG.AN, P., GuERITAUD, J. P., HORCHOT.T-BossAvrI, G. & Tyc-DuMoNr, S. (1974). Electrotonic coupling between motoneurones in the abducens nucleus of the cat. Expl Brain Re8. 21, 139-154.
DIRECT INTERACTIONS BETWEEN MOTONEURONES 767 GRINNELL, A. D. (1966). A study of the interaction between motoneurones in the frog spinal cord. J. Physiol. 182, 612-648. KORN, H., SOTELO, C. & CREPEL, F. (1973). Electrotonic coupling between neurons in rat lateral vestibular nucleus. Expl Brain Re8. 16, 255-275. LLINAs, R., BAKER, R. & SOTELO, C. (1974). Electrotonic coupling between neurons in cat inferior olive. J. Neurophysiol. 37, 560-571. MAGHERINI, P. C., PRECHT, W. & SCHWINDT, P. C. (1976). Evidence for electrotonic coupling between frog motoneurons in the in situ spinal cord. J. Neurophysiol. 39, 474-483. MATTHEWS, M.A., WAis, W. D. & WiLLAkMs, V. (1971). Dendrite bundles in lamina IX of cat spinal cord: a possible source for electrical interaction between motoneurons. Anat. Rec. 171, 313-328. NELSON, P. G. (1966). Interaction between spinal motoneurons of the cat. J. Neurophysiol. 29, 275-287. RALL, W. (1967). Distinguishing theoretical potentials computed for different somadentritic distribution of synaptic input. J. Neurophysiol. 30, 1138-1168. SCHEIBEL, M. E. & SCHEIBEL, A. B. (1975). Dendrites as neuronal couplers: the dentritic bundle. In Golgi Centennial Symposium, ed SANTN, M., pp. 347-354. New York: Raven. SOTELO, C. & TAXI, J. (1970). Ultrastructural aspects of electrotonic junctions in the spinal cord of the frog. Brain Re8. 17, 137-141. STENSAAS, L. J. & STENSAAS, S. S. (1971). Light and electron microscopy of motoneurons and neuropile in the amphibian spinal cord. Brain Re8. 31, 67-84. SZEKELY, G. & KosAlAs, B. (1976). Dendro-dendritic contacts between frog motoneurons shown with the cobalt labelling technique. Brain Ree. 108, 194-198. WERMAN, R. & CARLEN, P. L. (1976). Unusual behaviour of the Ia EPSP in cat spinal motoneurons. Brain Re8. 112, 395-401.
