- Email: [email protected]
The central pattern generator for forelimb locomotion in the cat
Progress in Brain Research, Vol. 143 ISSN 0079-6123 Copyright ß 2004 Elsevier BV. All rights reserved
CHAPTER 11
The central pattern generator for forelimb locomotion in the cat Takashi Yamaguchi* Graduate School of Engineering, Yamagata University, Yamagata 992-8510, Japan
Abstract: This chapter addresses the neuronal organization of the central pattern generator (CPG) for forelimb locomotion in the cat. The focus is on fictive locomotion, as induced by repetitive stimulation of the cervical lateral funiculus (CLF) in the decerebrate preparation. The stimulus time-locked responses of cervical motoneurons (MNs) that follow each current pulse of such repetitive stimulation include disynaptic EPSPs, trisynaptic EPSPs and IPSPs, and polysynaptic PSPs. Their amplitudes are all phase-related to selected aspects of the locomotor rhythm, as determined by the pattern of discharge in nerves to selected muscles. Correlation analysis reveals rhythmic modulation between the responses of extensor and flexor MNs, thereby suggesting mutual interactions between pathways mediating the CLF effects on MNs. It is argued that the shortest CLF pathway to MNs via the CPG is disynaptic: i.e., with only a single intercalated interneuron (IN) between the descending command axons and the MNs. A second proposal is that such intercalated INs contribute to interacting, reverberating IN circuits (i.e., the CPG) through mutual excitation.
Introduction
forelimb CPG must have some properties that differ from its hindlimb counterpart because the two sets of limbs subserve different functions during locomotion. The hindlimbs provide the majority of the propulsive force, whereas the forelimbs are used largely for propping, steering, and supporting the head’s weighted mass (Vidal et al., 1986; Graf et al., 1997). Accordingly, my focus has been on the neuronal organization of the forelimb CPG in the cat, with the progress to date summarized in the next section.
The spatio-temporal pattern of muscle activity of the forelimbs during locomotion is primarily generated in the cervical enlargement of the spinal cord. This circuitry can be termed the central pattern generator (CPG) for forelimb locomotion. The CPGs for both the forelimbs and the hindlimbs (one for each limb) are controlled in parallel, i.e., ipsilateral or bilateral excitation of the midbrain locomotor region (MLR) elicits a coordinated locomotor rhythm in both sets of limbs (Shik et al., 1966; Jordan et al., 1979; Amemiya and Yamaguchi, 1984; Grillner and Walle´n, and Pearson, Chapters 1 and 12 of this volume). Furthermore, the spinal cord pathways activating the locomotor rhythm of both sets of limbs descend through the same region of the lateral funiculus at the upper cervical level of the spinal cord (Steeves and Jordan, 1980; Yamaguchi, 1986). Nevertheless, it is intuitively obvious that the
Overview of the forelimb CPG The forelimb CPG is activated by tonic locomotor command signals traveling through pathways descending in the lateral funiculus (Yamaguchi, 1986). These pathways presumably have projections into the rostral enlargement (C6–7) of the spinal cord. This was shown by demonstrating that transection of the lateral funiculus just rostral to C6 abolishes MLRevoked stepping in the decerebrate cat, whereas a
*Corresponding author: Tel.: þ 81-238-26-3396; Fax: þ 81-238-26-3177; E-mail: [email protected]
115 DOI: 10.1016/S0079-6123(03)43011-2
116
transection below C7 does not (Hishinuma and Yamaguchi, 1990). The localized projection of the command implies that command-sensitive ‘input’ (first-order) interneurons (INs) of the forelimb CPG located in the rostral segments of the cervical enlargement receive the command, and bring into play the CPG. In the immobilized, decerebrate cat, many INs in the cervical enlargement fire rhythmically during fictive forelimb locomotion. This activity produces rhythmic, alternating discharges of motoneurons (MNs) supplying elbow extensors and flexors (Yamaguchi, 1992). Each IN emits action potentials in a specific phase of the step cycle, i.e., the ‘active’ phase for that particular IN. To date, four major CPG groups of IN have been identified. (1) INs termed ‘f ’ are active within the flexor phase. Since the flexor phase is short, ‘f ’ neurons are short bursting. Flexor, phase-related neurons with longer active periods are termed ‘hf i’ (‘h i’ means expansion of active periods beyond the onset and offset of relevant phases). They are active throughout the entire flexor phase. They start firing in the late extensor phase or just prior to the flexor phase, and cease their discharge in the first one-third (‘mediumbursting hf i’ INs) or middle (‘long-bursting’) one-third of the immediately following extensor phase. (2) Those termed ‘f þ e’ start firing in the late flexor phase, and cease their discharge in the following extensor phase. Their ‘short’, ‘medium’, and ‘long-bursting’ subcategories stop firing in the first, second and last one-third of the extensor phase, respectively. (3) INs termed ‘e’ discharge from the early extensor phase to the middle one-third (‘short-bursting’), last one-third (‘medium-bursting’) of the extensor phase, or immediately before the transition to the flexor phase (‘long-bursting’). Neurons termed ‘hei’ are active throughout the entire extensor phase. (4) INs termed ‘e þ f ’ start firing in the extensor phase and stop in the next flexor phase. Their subcategories comprise ‘short’, ‘medium’, and ‘long-bursting’ types, which commence their discharge prior to the flexor phase, in the middle
one-third of the extensor phase, and in the first one-third of the extensor phase, respectively. The above-described nomenclature and activity patterns show that f, hf i and e þ f INs are ‘flexorlike’, while e, hei and f þ e INs are ‘extensor-like’. It is assumed that these INs can be first-order, in-betweenorder, or last-order, the latter exciting MNs to either flexor or extensor muscles. Figure 1 shows the approximate spinal cord location of these CPG IN types. In the dorsoventral axis, hf i and e þ f INs are located rather dorsally (mainly in laminae V–VI), whereas e and f þ e INs are located ventrally (mainly in the lamina VII). Along the rostrocaudal axis, e INs are located caudally (mainly C7–T1) and e þ f INs are more rostral (mainly in C5–6). The hf i and f þ e groups have a bimodal distribution of locations, one rostral (mainly C5–6) and the other caudal (C7–8). As mentioned earlier, the input INs of the CPG are presumably located in the C6–7 segments, but we have not been able to specify the first-order neurons solely on the basis of their active phase, because there are several types of rhythmically discharging INs at the C6–7 level. The forelimb CPG includes INs that send rhythmic excitation and inhibition to forelimb extensor and flexor MNs (Go¨dderz et al., 1990). These ‘output’ INs, the last-order ones of the CPG are, by definition, rhythmically active during fictive locomotion, and they are identifiable among the INs shown in the topographic maps of Fig. 1. Following the identification of such last-order INs by antidromic stimulation of MN pools (i.e., nuclei), their location has been compared to those of the total population of CPG INs (Terakado and Yamaguchi, 1990; Ichikawa et al., 1991). The last-order INs to the flexor MNs are localized in the more rostral C5–7 segments. Those of f and e þ f type are located in the dorsal intermediate zone, whereas those of f þ e and e type are located in the ventral intermediate zone. The former (f and e þ f ) INs are active in virtually the same phase as their target MNs, thereby suggesting that they are excitatory. In contrast, the f þ e and e INs are outof-phase with flexor MNs, thereby suggesting that they have an inhibitory function. The last-order CPG INs to extensor MNs are distributed in a column throughout the cervical
117
Fig. 1. Spinal cord location of cervical INs contributing to the forelimb CPG for locomotion. Four separate sets of spinal cord crosssection sketches are shown for the four major IN cell types (hf i, f þ e, e, e þ f ), which, together with their subtypes, are defined in the text. The asterisks (*; uppermost row, C7/8 and C8/T1) are for INs which exhibited a double burst of discharge; the C7/8 neuron showing one burst like hf i INs and the others like short-bursting e INs at the late extensor phase; and, the C8/T1 neuron showing short-bursting e þ f and medium-bursting e behavior. Notations with two segments separated by slash mean boundaries between the two. (Reproduced, in part, from Yamaguchi, 1992 with permission from Center for Academic Publications Japan.)
enlargement. The INs of f and e þ f type, which are presumably inhibitory (i.e., for the reasons provided earlier) are mainly located more rostrally (C6–7), and those of f þ e and e type (presumably excitatory) are more caudal (C7-T1). Note further that potentially excitatory last-order INs, especially those to flexor MNs, are usually located in the same spinal cord segment as their corresponding firstorder INs. In summary, the forelimb CPG receives descending tonic impulses via its first-order input INs at the C6–7 level. Then, the full array of CPG INs develops its locomotor rhythm, and distributes rhythmic
excitatory and inhibitory synaptic volleys to MNs by way of various last-order output INs.
The stimulus time-locked-response approach to studying the forelimb CPG What kinds of signal processing take place between input and output CPG INs? There could be multiple interneuronal paths connecting the two groups of cells. Signal transmission through these paths and mutual interactions among them could conceivably generate a locomotor rhythm. Is it possible to reveal the
118
organization of these interneuronal paths, and delineate the characteristics of their signal transmission? At first glance, one approach would be to analyze MN responses to a single stimulus pulse delivered within a descending pathway involved in the locomotor command. When the CPG is at rest, however, so too are most of its constituent INs. Therefore, a CPG’s IN paths can be investigated only when the CPG is active. We use fictive locomotion for this purpose, as evoked by repetitive stimulation of the upper cervical lateral funiculus. Under these conditions, Fig. 2 shows that locomotor discharges in nerves supplying selected flexor and extensor muscles include spikes resembling the action potentials of MNs. These are time-locked to each stimulus, and they are modulated rhythmically (Kinoshita and Yamaguchi, 2001). Such stimulus time-locked discharges presumably result from activity in multiple IN paths within the forelimbs’ locomotor CPG (Shefchyk and Jordan, 1985; Degtyarenko et al.,
Fig. 2. Stimulus time-locked muscle neurogram discharges during fictive locomotion (A). Forelimb fictive locomotion evoked by 33 Hz stimulation of the spinal cord’s lateral funiculus at the C2 level. Such stimulation (Stim) produces rhythmic alternating discharges in nerves supplying flexor (Fl) and extensor (Ext) muscles (B). Expanded timescale records of a transition phase from an extensor to a flexor burst of neurographic discharge. Note that MN action potential-like discharges (at arrows) were induced by each stimulus pulse (at the dotted vertical lines). (Reproduced, in part, from Kinoshita and Yamaguchi, 2001 with permission from Elsevier Science.)
1998). Thus, the CPG network can be investigated indirectly by studying the stimulus time-locked responses of MNs during fictive locomotion.
Analysis and interpretation of stimulus-time-locked responses of MNs We have made intracellular (IC) recordings in MNs supplying elbow flexor and extensor muscles during fictive locomotion, which was produced as described in the previous section (Kinoshita and Yamaguchi, 2001). Figure 3 shows that under these conditions, both flexor and extensor MNs exhibit stimulus time-locked disynaptic excitatory postsynaptic potentials (EPSPs), trisynaptic inhibitory postsynaptic potentials (IPSPs) and polysynaptic EPSPs. All such postsynaptic potentials (PSPs) are rhythmically modulated with a stereotypic pattern. The disynaptic EPSPs of flexor MNs are facilitated in the flexor phase of locomotion, whereas those of extensor MNs are facilitated mainly at the transition from the flexor to the extensor phase. The depth of such modulation is greater in flexor versus extensor MNs. The amplitude of trisynaptic IPSPs change in parallel with that of the disynaptic EPSPs of antagonistic MNs. This is a strong indication that the trisynaptic IPSPs of flexor MNs are facilitated in the flexor-to-extensor transition phase, whereas those of extensor MNs are facilitated in the flexor phase. After transection of the lateral funiculus to abolish disynaptic EPSPs, another type of stimulus time-locked EPSP is found in extensor MNs. It is trisynaptic, and it presumably results from excitation transmitted through the ventral part of the spinal cord. It is clearly facilitated rhythmically in the extensor phase. The earlier results are summarized in Fig. 4. This schematic diagram proposes that rhythmic modulation of MNs during locomotion is due to the operation of three subsystems of the CPG. These control flexor bursting (F phase of the step), the initiation of extensor bursts (the E1 phase, prior-to-foot-placement phase) and the maintenance of extensor bursts [the E2–3 stance (thrust) phase].
119
Fig. 3. Exemplary IC recordings made in MNs supplying flexor and extensor muscles during fictive locomotion (A–B). The upper two traces are as in Fig. 2A. The third trace shows IC recordings from MNs supplying the flexor, brachialis (Br) and extensor, triceps brachii, longus (TLo) muscle. The fourth trace shows a portion of the IC-recorded responses on an expanded timescale. The fifth (lowest) trace shows the timing of the stimulus (Stim) pulses. In the two latter traces, note the arrows, which show that each stimulus pulse was followed sequentially by early depolarization, hyperpolarization and late depolarization. (Reproduced, in part, from Kinoshita and Yamaguchi, 2001 with permission from Elsevier Science.)
Fig. 4. A proposal on the neuronal mechanisms and pathways underlying the stimulus-time-locked responses of flexor and extensor MNs during fictive forelimb locomotion. F-MN, flexor motoneurons; E-MN, extensor motoneurons; open circles, excitatory INs; closed circles, inhibitory INs. The fictive step phases shown are flexion (F), prior-to-foot-placement (E1), and stance (E2). Other abbreviations: LF, lateral funiculus; VF, ventral funiculus. See the text for further details.
Further thoughts on the stimulus time-locked PSPs of MNs The above-described repetitive electrical stimulation of the lateral funiculus could excite not only
command pathways for locomotion, but other descending pathways, as well. For example, disynaptic excitation of MNs can be evoked by several other descending effects, and these responses could be modulated by the locomotor rhythm if the forelimb CPG is coactivated by the same stimulus. This type of possibility is shown in Fig. 5A. One such ‘nonlocomotor’ descending pathway is the corticospinal tract, most of whose axons descend in the dorsolateral funiculus. In fact, stimulation of the medullary pyramid evokes disynaptic EPSPs in forelimb MNs (mainly to flexors), and these are clearly rhythmically modulated during forelimb fictive locomotion (Seki et al., 1997). This suggests that the corticospinal tract and the forelimb CPG share common last-order INs. Similarly, the presence of such interneuronal sharing has been explored for the cervical CPG and the vestibulospinal tract (Kitama et al., 1996); trisynaptic EPSPs evoked from vestibular afferents in forelimb extensor MNs are rhythmically modulated during fictive locomotion. There has been no demonstration to date, however, of these two descending pathways (and the CPG) having their own private last-order INs for connection to MNs. It is well
120
Fig. 5. Schema for the descending control of cervical MNs. The diagrams indicate possible ways that the stimulus-time-locked responses of cervical MNs are subjected to the dual influences of descending effects mediated by stimulation (Stim) of the lateral funiculus (LF) of the spinal cord and the forelimb locomotor CPG, with the latter circuitry also activated by the same descending effect. See the text for further details.
known, however, that at the lumbosacral level of the cat spinal cord, the locomotor hindlimb CPG and its descending and sensory inputs share common lastorder INs, and that these come into play for the integration of locomotor and nonlocomotor movements (Edgley and Jankowska, 1987; Edgley et al., 1988; Shefchyk et al., 1990). A similar organization is known at the cervical level for interactions between the locomotor forelimb CPG and cutaneous reflex pathways (Seki and Yamaguchi, 1997). A second, as yet untested explanation of the basis for stimulus time-locked EPSPs of MNs is that the locomotor command, itself, sends direct parallel excitation to the CPG and its last-order INs (Fig. 5B). Such a feed-forward connection would lower the threshold of the last-order INs to their subsequent rhythmic excitation by the CPG. If the input INs of the CPG also function as its output INs, then their locomotor discharge must occur in bursts. Such bursts of action potentials can be generated either by network properties, including reverberation (Fig. 5C), intrinsic membrane properties, such as plateau-potential-based discharge (Hounsgaard et al., 1988), or their varying combinations. It is not yet known how these mechanisms relate to the stimulus time-locked EPSPs described earlier. If we assume that a reverberating effect is a feature of the locomotor forelimb CPG, however, we
can then propose that the CPG includes three kinds of networks. One is for flexor bursting, with INs mediating stimulus time-locked disynaptic EPSPs of flexor INs. The other two must control the disynaptic and trisynaptic EPSPs of extensor INs, respectively. The former of these two EPSPs is facilitated mainly at the transition from the flexor to extensor phases in the step cycle, i.e., when a reverberating network initiates extensor bursting. In contrast, the trisynaptic EPSPs of extensor INs are rhythmically facilitated throughout the full extensor phase, this being controlled by a third reverberating network. The locomotor rhythm and its spatio-temporal pattern of muscle activity could presumably be produced by synaptic interactions among the above three reverberating networks. Their respective inhibitory INs, which produce the trisynaptic IPSPs of flexor and extensor INs, appear to be excited by the axon collaterals of the INs of disynaptic pathways (Fig. 4). Thus, it will be valuable to test for the respective inhibitory INs also projecting axon collaterals to the antagonistic reverberating networks. It is also necessary to consider how late stimulus time-locked EPSPs might relate to reverberating networks. For example, the late, polysynaptic EPSPs of flexor and extensor MNs are facilitated (increase in amplitude) during the flexor and extensor phases of the step cycle, respectively. If the INs producing the early stimulus time-locked MN EPSPs are components of reverberating networks, they may be reexcited by these networks, such that the late, polysynaptic MN EPSPs are produced by the same INs that produce the early MN EPSPs.
Concluding remarks Locomotion can result from continuous, repetitive stimulation of various supraspinal structures (Shik et al., 1966, 1968; Orlovsky, 1969; Mori et al., 1977, 1998). This shows that the brain stem sends ‘tonic’ impulses to the locomotor CPGs of the forelimbs and hindlimbs. These convert the tonic input into a locomotor rhythm, which builds up within the CPG’s circuitry. The output INs of each CPG distribute the locomotor rhythm to relevant MN pools. This distribution role may also be possessed by the CPG’s input
121
INs, which receive the tonic command, but such a role has never been shown. For the moment then, it is parsimonious to propose that multiple interneuronal paths should be traceable from input to output INs within the CPG. If so, the identification of such INs and their pathways is critical for advancing understanding of locomotor CPGs. Recall, however, that such interneuronal paths become active (i.e., open) when the CPG, itself, is active. When such pathways are open, their input–output relations and transfer functions can be analyzed indirectly by studying MN responses to stimulus pulses delivered to supraspinal command centers and their descending pathways. Generation of any central nervous system (CNS) rhythm driven by tonic excitation is a highly nonlinear process. Linear summation of the linear transfer functions of the interneuronal paths alone cannot model such a rhythm. Rather, nonlinear transfer functions and integration are also required. Nonlinear transfer functions and integration are ubiquitous within CPGs, because the single IN, itself, has many nonlinear properties, including its threshold for eliciting action potentials, persistent inward currents (plateau potentials) and so on. As a result, it is unlikely that any single nonlinear process within a CPG will ever be shown to be the predominant mechanism underlying the CPG’s capacity to generate a rhythm. In CPG circuitry consisting of many (indeed thousands of ) INs, any tiny nonlinearity can result in large, macroscopic change of state of the CPG. This process is termed self-organization, which, in turn affects elementary processes. Rhythmic modulation of stimulus time-locked MN responses is such an elementary process. Therefore, instead of searching for nonlinear processes within CPG circuitry, it seems much more fruitful to advance our understanding of how the CNS modulates stimulus time-locked MN responses during fictive locomotion. It must be recognized that many questions and points have yet to be investigated, including the study of individual INs within the pathways that comprise the forelimb’s locomotor CPG.
Acknowledgments This work was supported by a Grant-in-Aid for Scientific Research (#10480223 to T.Y.), the Japanese Ministry of Education, Science, Sports and Culture.
Abbreviations CLF Cn CNS CPG EPSP IC IN IPSP MLR MN PSP Tn
cervical lateral funiculus nth cervical segment central nervous system central pattern generator excitatory postsynaptic potential intracellular interneuron inhibitory postsynaptic potential midbrain locomotor region motoneuron postsynaptic potential nth thoracic segment
References Amemiya, M. and Yamaguchi, T. (1984). Fictive locomotion of the forelimb evoked by stimulation of the mesencephalic locomotor region in the decerebrate cat. Neurosci. Lett., 50: 91–96. Degtyarenko, A.M., Simon, E.S. and Burke, R.E. (1998). Locomotor modulation of disynaptic EPSPs from the mesencephalic locomotor region in cat motoneurons. J. Neurophysiol., 80: 3284–3296. Edgley, S.A. and Jankowska, E. (1987). An interneuronal relay for group I and II muscle afferents in the midlumbar segments of the cat spinal cord. J. Physiol. (Lond.), 389: 647–674. Edgley, S.A, Jankowska, E. and Shefchyk, S.J. (1988). Evidence that midlumbar neurones in reflex pathways from group II afferents are involved in locomotion in the cat. J. Physiol. (Lond.), 403: 57–71. Go¨dderz, W., Illert, M. and Yamaguchi, T. (1990). Efferent pattern of fictive locomotion in the cat forelimb: with special reference to radial motor nuclei. Eur. J. Neurosci., 2: 663–671. Graf, W., Keshner, E., Richmond, F.J., Shinoda, Y., Statler, K. and Uchino, Y. (1997). How to construct and move a cat’s neck. J. Vest. Res., 7: 219–237. Hishinuma, M. and Yamaguchi, T. (1990). Axonal projection of descending pathways responsible for eliciting forelimb stepping into the cat cervical spinal cord. Exp. Brain Res., 82: 597–605. Hounsgaard, J., Hultborn, H., Jespersen, B. and Kiehn, O. (1988). Bistability of alpha-motoneurones in the decerebrate cat and in the acute spinal cat after intravenous 5-hydroxytryptophan. J. Physiol. (Lond.), 405: 345–367. Ichikawa, Y., Terakado, Y. and Yamaguchi, T. (1991). Lastorder interneurones controlling activity of elbow extensor motoneurones during forelimb fictive locomotion in the cat. Neurosci. Lett., 121: 37–39.
122 Jordan, L.M., Pratt, C.A. and Menzies, J.E. (1979). Locomotion evoked by brain stem stimulation: occurrence without phasic segmental afferent input. Brain Res., 177: 204–207. Kinoshita, M. and Yamaguchi, T. (2001). Stimulus time-locked responses of motoneurons during forelimb fictive locomotion evoked by repetitive stimulation of the lateral funiculus. Brain Res., 904: 31–42. Kitama, T., Seki, K. and Yamaguchi, T. (1996). Labyrinthine reflex activity during forelimb fictive locomotion in the decerebrate cat. Jpn. J. Physiol., Suppl. 46: S142. Mori, S., Shik, M.L. and Yagodnitsyn, A.S. (1977). Role of pontine tegmentum for locomotor control in mesencephalic cat. J. Neurophysiol., 40: 284–295. Mori, S., Matsui, T., Kuze, B., Asanome, M., Nakajima, K. and Matsuyama, K. (1998). Cerebellar-induced locomotion: reticulospinal control of spinal rhythm generating mechanism in cats. Ann. N.Y. Acad. Sci., 860: 94–105. Orlovsky, G.N. (1969). Spontaneous and induced locomotion of the thalamic cat. Biofizika, 14: 1095–1102. Seki, K. and Yamaguchi, T. (1997). Cutaneous reflex activity of the cat forelimb during fictive locomotion. Brain Res., 753: 56–62. Seki, K., Kudo, N., Kolb, F. and Yamaguchi, T. (1997). Effects of pyramidal tract stimulation on forelimb flexor motoneurons during fictive locomotion in cats. Neurosci. Lett., 230: 195–198. Shefchyk, S.J. and Jordan, L.M. (1985). Excitatory and inhibitory postsynaptic potentials in alpha-motoneurons produced during fictive locomotion by stimulation of the
mesencephalic locomotor region. J. Neurophysiol., 53: 1345–1355. Shefchyk, S.J., McCrea, D., Kriellaars, D., Fortier, P. and Jordan, L.M. (1990). Activity of interneurons within the L4 spinal segment of the cat during brainstem-evoked fictive locomotion. Exp. Brain Res., 80: 290–295. Shik, M.L., Severin, F.V. and Orlovsky, G.N. (1966). Control of walking and running by means of electrical stimulation of the midbrain. Biophysics, 11: 659–666. Shik, M.L., Orlovsky, G.N. and Severin, F.V. (1968). Locomotion of the mesencephalic cat evoked by pyramidal stimulation. Biofizika, 13: 127–135. Steeves, J.D. and Jordan, L.M. (1980). Localization of a descending pathway in the SC which is necessary for controlled treadmill locomotion. Neurosci. Lett., 20: 283–288. Terakado, Y. and Yamaguchi, T. (1990). Last-order interneurones controlling activity of elbow flexor motoneurones during forelimb fictive locomotion in the cat. Neurosci. Lett., 111: 292–296. Vidal, P.P., Graf, W. and Berthoz, A. (1986). The orientation of the cervical vertebral column in unrestrained awake animals. I. Resting position. Exp. Brain Res., 61: 549–559. Yamaguchi, T. (1986). Descending pathways eliciting forelimb stepping in the lateral funiculus: experimental studies with stimulation and lesion of the cervical cord in decerebrate cats. Brain Res., 379: 125–136. Yamaguchi, T. (1992). Activity of cervical neurons during forelimb fictive locomotion in decerebrate cats. Jpn. J. Physiol., 42: 501–514.
CHAPTER 11
The central pattern generator for forelimb locomotion in the cat Takashi Yamaguchi* Graduate School of Engineering, Yamagata University, Yamagata 992-8510, Japan
Abstract: This chapter addresses the neuronal organization of the central pattern generator (CPG) for forelimb locomotion in the cat. The focus is on fictive locomotion, as induced by repetitive stimulation of the cervical lateral funiculus (CLF) in the decerebrate preparation. The stimulus time-locked responses of cervical motoneurons (MNs) that follow each current pulse of such repetitive stimulation include disynaptic EPSPs, trisynaptic EPSPs and IPSPs, and polysynaptic PSPs. Their amplitudes are all phase-related to selected aspects of the locomotor rhythm, as determined by the pattern of discharge in nerves to selected muscles. Correlation analysis reveals rhythmic modulation between the responses of extensor and flexor MNs, thereby suggesting mutual interactions between pathways mediating the CLF effects on MNs. It is argued that the shortest CLF pathway to MNs via the CPG is disynaptic: i.e., with only a single intercalated interneuron (IN) between the descending command axons and the MNs. A second proposal is that such intercalated INs contribute to interacting, reverberating IN circuits (i.e., the CPG) through mutual excitation.
Introduction
forelimb CPG must have some properties that differ from its hindlimb counterpart because the two sets of limbs subserve different functions during locomotion. The hindlimbs provide the majority of the propulsive force, whereas the forelimbs are used largely for propping, steering, and supporting the head’s weighted mass (Vidal et al., 1986; Graf et al., 1997). Accordingly, my focus has been on the neuronal organization of the forelimb CPG in the cat, with the progress to date summarized in the next section.
The spatio-temporal pattern of muscle activity of the forelimbs during locomotion is primarily generated in the cervical enlargement of the spinal cord. This circuitry can be termed the central pattern generator (CPG) for forelimb locomotion. The CPGs for both the forelimbs and the hindlimbs (one for each limb) are controlled in parallel, i.e., ipsilateral or bilateral excitation of the midbrain locomotor region (MLR) elicits a coordinated locomotor rhythm in both sets of limbs (Shik et al., 1966; Jordan et al., 1979; Amemiya and Yamaguchi, 1984; Grillner and Walle´n, and Pearson, Chapters 1 and 12 of this volume). Furthermore, the spinal cord pathways activating the locomotor rhythm of both sets of limbs descend through the same region of the lateral funiculus at the upper cervical level of the spinal cord (Steeves and Jordan, 1980; Yamaguchi, 1986). Nevertheless, it is intuitively obvious that the
Overview of the forelimb CPG The forelimb CPG is activated by tonic locomotor command signals traveling through pathways descending in the lateral funiculus (Yamaguchi, 1986). These pathways presumably have projections into the rostral enlargement (C6–7) of the spinal cord. This was shown by demonstrating that transection of the lateral funiculus just rostral to C6 abolishes MLRevoked stepping in the decerebrate cat, whereas a
*Corresponding author: Tel.: þ 81-238-26-3396; Fax: þ 81-238-26-3177; E-mail: [email protected]
115 DOI: 10.1016/S0079-6123(03)43011-2
116
transection below C7 does not (Hishinuma and Yamaguchi, 1990). The localized projection of the command implies that command-sensitive ‘input’ (first-order) interneurons (INs) of the forelimb CPG located in the rostral segments of the cervical enlargement receive the command, and bring into play the CPG. In the immobilized, decerebrate cat, many INs in the cervical enlargement fire rhythmically during fictive forelimb locomotion. This activity produces rhythmic, alternating discharges of motoneurons (MNs) supplying elbow extensors and flexors (Yamaguchi, 1992). Each IN emits action potentials in a specific phase of the step cycle, i.e., the ‘active’ phase for that particular IN. To date, four major CPG groups of IN have been identified. (1) INs termed ‘f ’ are active within the flexor phase. Since the flexor phase is short, ‘f ’ neurons are short bursting. Flexor, phase-related neurons with longer active periods are termed ‘hf i’ (‘h i’ means expansion of active periods beyond the onset and offset of relevant phases). They are active throughout the entire flexor phase. They start firing in the late extensor phase or just prior to the flexor phase, and cease their discharge in the first one-third (‘mediumbursting hf i’ INs) or middle (‘long-bursting’) one-third of the immediately following extensor phase. (2) Those termed ‘f þ e’ start firing in the late flexor phase, and cease their discharge in the following extensor phase. Their ‘short’, ‘medium’, and ‘long-bursting’ subcategories stop firing in the first, second and last one-third of the extensor phase, respectively. (3) INs termed ‘e’ discharge from the early extensor phase to the middle one-third (‘short-bursting’), last one-third (‘medium-bursting’) of the extensor phase, or immediately before the transition to the flexor phase (‘long-bursting’). Neurons termed ‘hei’ are active throughout the entire extensor phase. (4) INs termed ‘e þ f ’ start firing in the extensor phase and stop in the next flexor phase. Their subcategories comprise ‘short’, ‘medium’, and ‘long-bursting’ types, which commence their discharge prior to the flexor phase, in the middle
one-third of the extensor phase, and in the first one-third of the extensor phase, respectively. The above-described nomenclature and activity patterns show that f, hf i and e þ f INs are ‘flexorlike’, while e, hei and f þ e INs are ‘extensor-like’. It is assumed that these INs can be first-order, in-betweenorder, or last-order, the latter exciting MNs to either flexor or extensor muscles. Figure 1 shows the approximate spinal cord location of these CPG IN types. In the dorsoventral axis, hf i and e þ f INs are located rather dorsally (mainly in laminae V–VI), whereas e and f þ e INs are located ventrally (mainly in the lamina VII). Along the rostrocaudal axis, e INs are located caudally (mainly C7–T1) and e þ f INs are more rostral (mainly in C5–6). The hf i and f þ e groups have a bimodal distribution of locations, one rostral (mainly C5–6) and the other caudal (C7–8). As mentioned earlier, the input INs of the CPG are presumably located in the C6–7 segments, but we have not been able to specify the first-order neurons solely on the basis of their active phase, because there are several types of rhythmically discharging INs at the C6–7 level. The forelimb CPG includes INs that send rhythmic excitation and inhibition to forelimb extensor and flexor MNs (Go¨dderz et al., 1990). These ‘output’ INs, the last-order ones of the CPG are, by definition, rhythmically active during fictive locomotion, and they are identifiable among the INs shown in the topographic maps of Fig. 1. Following the identification of such last-order INs by antidromic stimulation of MN pools (i.e., nuclei), their location has been compared to those of the total population of CPG INs (Terakado and Yamaguchi, 1990; Ichikawa et al., 1991). The last-order INs to the flexor MNs are localized in the more rostral C5–7 segments. Those of f and e þ f type are located in the dorsal intermediate zone, whereas those of f þ e and e type are located in the ventral intermediate zone. The former (f and e þ f ) INs are active in virtually the same phase as their target MNs, thereby suggesting that they are excitatory. In contrast, the f þ e and e INs are outof-phase with flexor MNs, thereby suggesting that they have an inhibitory function. The last-order CPG INs to extensor MNs are distributed in a column throughout the cervical
117
Fig. 1. Spinal cord location of cervical INs contributing to the forelimb CPG for locomotion. Four separate sets of spinal cord crosssection sketches are shown for the four major IN cell types (hf i, f þ e, e, e þ f ), which, together with their subtypes, are defined in the text. The asterisks (*; uppermost row, C7/8 and C8/T1) are for INs which exhibited a double burst of discharge; the C7/8 neuron showing one burst like hf i INs and the others like short-bursting e INs at the late extensor phase; and, the C8/T1 neuron showing short-bursting e þ f and medium-bursting e behavior. Notations with two segments separated by slash mean boundaries between the two. (Reproduced, in part, from Yamaguchi, 1992 with permission from Center for Academic Publications Japan.)
enlargement. The INs of f and e þ f type, which are presumably inhibitory (i.e., for the reasons provided earlier) are mainly located more rostrally (C6–7), and those of f þ e and e type (presumably excitatory) are more caudal (C7-T1). Note further that potentially excitatory last-order INs, especially those to flexor MNs, are usually located in the same spinal cord segment as their corresponding firstorder INs. In summary, the forelimb CPG receives descending tonic impulses via its first-order input INs at the C6–7 level. Then, the full array of CPG INs develops its locomotor rhythm, and distributes rhythmic
excitatory and inhibitory synaptic volleys to MNs by way of various last-order output INs.
The stimulus time-locked-response approach to studying the forelimb CPG What kinds of signal processing take place between input and output CPG INs? There could be multiple interneuronal paths connecting the two groups of cells. Signal transmission through these paths and mutual interactions among them could conceivably generate a locomotor rhythm. Is it possible to reveal the
118
organization of these interneuronal paths, and delineate the characteristics of their signal transmission? At first glance, one approach would be to analyze MN responses to a single stimulus pulse delivered within a descending pathway involved in the locomotor command. When the CPG is at rest, however, so too are most of its constituent INs. Therefore, a CPG’s IN paths can be investigated only when the CPG is active. We use fictive locomotion for this purpose, as evoked by repetitive stimulation of the upper cervical lateral funiculus. Under these conditions, Fig. 2 shows that locomotor discharges in nerves supplying selected flexor and extensor muscles include spikes resembling the action potentials of MNs. These are time-locked to each stimulus, and they are modulated rhythmically (Kinoshita and Yamaguchi, 2001). Such stimulus time-locked discharges presumably result from activity in multiple IN paths within the forelimbs’ locomotor CPG (Shefchyk and Jordan, 1985; Degtyarenko et al.,
Fig. 2. Stimulus time-locked muscle neurogram discharges during fictive locomotion (A). Forelimb fictive locomotion evoked by 33 Hz stimulation of the spinal cord’s lateral funiculus at the C2 level. Such stimulation (Stim) produces rhythmic alternating discharges in nerves supplying flexor (Fl) and extensor (Ext) muscles (B). Expanded timescale records of a transition phase from an extensor to a flexor burst of neurographic discharge. Note that MN action potential-like discharges (at arrows) were induced by each stimulus pulse (at the dotted vertical lines). (Reproduced, in part, from Kinoshita and Yamaguchi, 2001 with permission from Elsevier Science.)
1998). Thus, the CPG network can be investigated indirectly by studying the stimulus time-locked responses of MNs during fictive locomotion.
Analysis and interpretation of stimulus-time-locked responses of MNs We have made intracellular (IC) recordings in MNs supplying elbow flexor and extensor muscles during fictive locomotion, which was produced as described in the previous section (Kinoshita and Yamaguchi, 2001). Figure 3 shows that under these conditions, both flexor and extensor MNs exhibit stimulus time-locked disynaptic excitatory postsynaptic potentials (EPSPs), trisynaptic inhibitory postsynaptic potentials (IPSPs) and polysynaptic EPSPs. All such postsynaptic potentials (PSPs) are rhythmically modulated with a stereotypic pattern. The disynaptic EPSPs of flexor MNs are facilitated in the flexor phase of locomotion, whereas those of extensor MNs are facilitated mainly at the transition from the flexor to the extensor phase. The depth of such modulation is greater in flexor versus extensor MNs. The amplitude of trisynaptic IPSPs change in parallel with that of the disynaptic EPSPs of antagonistic MNs. This is a strong indication that the trisynaptic IPSPs of flexor MNs are facilitated in the flexor-to-extensor transition phase, whereas those of extensor MNs are facilitated in the flexor phase. After transection of the lateral funiculus to abolish disynaptic EPSPs, another type of stimulus time-locked EPSP is found in extensor MNs. It is trisynaptic, and it presumably results from excitation transmitted through the ventral part of the spinal cord. It is clearly facilitated rhythmically in the extensor phase. The earlier results are summarized in Fig. 4. This schematic diagram proposes that rhythmic modulation of MNs during locomotion is due to the operation of three subsystems of the CPG. These control flexor bursting (F phase of the step), the initiation of extensor bursts (the E1 phase, prior-to-foot-placement phase) and the maintenance of extensor bursts [the E2–3 stance (thrust) phase].
119
Fig. 3. Exemplary IC recordings made in MNs supplying flexor and extensor muscles during fictive locomotion (A–B). The upper two traces are as in Fig. 2A. The third trace shows IC recordings from MNs supplying the flexor, brachialis (Br) and extensor, triceps brachii, longus (TLo) muscle. The fourth trace shows a portion of the IC-recorded responses on an expanded timescale. The fifth (lowest) trace shows the timing of the stimulus (Stim) pulses. In the two latter traces, note the arrows, which show that each stimulus pulse was followed sequentially by early depolarization, hyperpolarization and late depolarization. (Reproduced, in part, from Kinoshita and Yamaguchi, 2001 with permission from Elsevier Science.)
Fig. 4. A proposal on the neuronal mechanisms and pathways underlying the stimulus-time-locked responses of flexor and extensor MNs during fictive forelimb locomotion. F-MN, flexor motoneurons; E-MN, extensor motoneurons; open circles, excitatory INs; closed circles, inhibitory INs. The fictive step phases shown are flexion (F), prior-to-foot-placement (E1), and stance (E2). Other abbreviations: LF, lateral funiculus; VF, ventral funiculus. See the text for further details.
Further thoughts on the stimulus time-locked PSPs of MNs The above-described repetitive electrical stimulation of the lateral funiculus could excite not only
command pathways for locomotion, but other descending pathways, as well. For example, disynaptic excitation of MNs can be evoked by several other descending effects, and these responses could be modulated by the locomotor rhythm if the forelimb CPG is coactivated by the same stimulus. This type of possibility is shown in Fig. 5A. One such ‘nonlocomotor’ descending pathway is the corticospinal tract, most of whose axons descend in the dorsolateral funiculus. In fact, stimulation of the medullary pyramid evokes disynaptic EPSPs in forelimb MNs (mainly to flexors), and these are clearly rhythmically modulated during forelimb fictive locomotion (Seki et al., 1997). This suggests that the corticospinal tract and the forelimb CPG share common last-order INs. Similarly, the presence of such interneuronal sharing has been explored for the cervical CPG and the vestibulospinal tract (Kitama et al., 1996); trisynaptic EPSPs evoked from vestibular afferents in forelimb extensor MNs are rhythmically modulated during fictive locomotion. There has been no demonstration to date, however, of these two descending pathways (and the CPG) having their own private last-order INs for connection to MNs. It is well
120
Fig. 5. Schema for the descending control of cervical MNs. The diagrams indicate possible ways that the stimulus-time-locked responses of cervical MNs are subjected to the dual influences of descending effects mediated by stimulation (Stim) of the lateral funiculus (LF) of the spinal cord and the forelimb locomotor CPG, with the latter circuitry also activated by the same descending effect. See the text for further details.
known, however, that at the lumbosacral level of the cat spinal cord, the locomotor hindlimb CPG and its descending and sensory inputs share common lastorder INs, and that these come into play for the integration of locomotor and nonlocomotor movements (Edgley and Jankowska, 1987; Edgley et al., 1988; Shefchyk et al., 1990). A similar organization is known at the cervical level for interactions between the locomotor forelimb CPG and cutaneous reflex pathways (Seki and Yamaguchi, 1997). A second, as yet untested explanation of the basis for stimulus time-locked EPSPs of MNs is that the locomotor command, itself, sends direct parallel excitation to the CPG and its last-order INs (Fig. 5B). Such a feed-forward connection would lower the threshold of the last-order INs to their subsequent rhythmic excitation by the CPG. If the input INs of the CPG also function as its output INs, then their locomotor discharge must occur in bursts. Such bursts of action potentials can be generated either by network properties, including reverberation (Fig. 5C), intrinsic membrane properties, such as plateau-potential-based discharge (Hounsgaard et al., 1988), or their varying combinations. It is not yet known how these mechanisms relate to the stimulus time-locked EPSPs described earlier. If we assume that a reverberating effect is a feature of the locomotor forelimb CPG, however, we
can then propose that the CPG includes three kinds of networks. One is for flexor bursting, with INs mediating stimulus time-locked disynaptic EPSPs of flexor INs. The other two must control the disynaptic and trisynaptic EPSPs of extensor INs, respectively. The former of these two EPSPs is facilitated mainly at the transition from the flexor to extensor phases in the step cycle, i.e., when a reverberating network initiates extensor bursting. In contrast, the trisynaptic EPSPs of extensor INs are rhythmically facilitated throughout the full extensor phase, this being controlled by a third reverberating network. The locomotor rhythm and its spatio-temporal pattern of muscle activity could presumably be produced by synaptic interactions among the above three reverberating networks. Their respective inhibitory INs, which produce the trisynaptic IPSPs of flexor and extensor INs, appear to be excited by the axon collaterals of the INs of disynaptic pathways (Fig. 4). Thus, it will be valuable to test for the respective inhibitory INs also projecting axon collaterals to the antagonistic reverberating networks. It is also necessary to consider how late stimulus time-locked EPSPs might relate to reverberating networks. For example, the late, polysynaptic EPSPs of flexor and extensor MNs are facilitated (increase in amplitude) during the flexor and extensor phases of the step cycle, respectively. If the INs producing the early stimulus time-locked MN EPSPs are components of reverberating networks, they may be reexcited by these networks, such that the late, polysynaptic MN EPSPs are produced by the same INs that produce the early MN EPSPs.
Concluding remarks Locomotion can result from continuous, repetitive stimulation of various supraspinal structures (Shik et al., 1966, 1968; Orlovsky, 1969; Mori et al., 1977, 1998). This shows that the brain stem sends ‘tonic’ impulses to the locomotor CPGs of the forelimbs and hindlimbs. These convert the tonic input into a locomotor rhythm, which builds up within the CPG’s circuitry. The output INs of each CPG distribute the locomotor rhythm to relevant MN pools. This distribution role may also be possessed by the CPG’s input
121
INs, which receive the tonic command, but such a role has never been shown. For the moment then, it is parsimonious to propose that multiple interneuronal paths should be traceable from input to output INs within the CPG. If so, the identification of such INs and their pathways is critical for advancing understanding of locomotor CPGs. Recall, however, that such interneuronal paths become active (i.e., open) when the CPG, itself, is active. When such pathways are open, their input–output relations and transfer functions can be analyzed indirectly by studying MN responses to stimulus pulses delivered to supraspinal command centers and their descending pathways. Generation of any central nervous system (CNS) rhythm driven by tonic excitation is a highly nonlinear process. Linear summation of the linear transfer functions of the interneuronal paths alone cannot model such a rhythm. Rather, nonlinear transfer functions and integration are also required. Nonlinear transfer functions and integration are ubiquitous within CPGs, because the single IN, itself, has many nonlinear properties, including its threshold for eliciting action potentials, persistent inward currents (plateau potentials) and so on. As a result, it is unlikely that any single nonlinear process within a CPG will ever be shown to be the predominant mechanism underlying the CPG’s capacity to generate a rhythm. In CPG circuitry consisting of many (indeed thousands of ) INs, any tiny nonlinearity can result in large, macroscopic change of state of the CPG. This process is termed self-organization, which, in turn affects elementary processes. Rhythmic modulation of stimulus time-locked MN responses is such an elementary process. Therefore, instead of searching for nonlinear processes within CPG circuitry, it seems much more fruitful to advance our understanding of how the CNS modulates stimulus time-locked MN responses during fictive locomotion. It must be recognized that many questions and points have yet to be investigated, including the study of individual INs within the pathways that comprise the forelimb’s locomotor CPG.
Acknowledgments This work was supported by a Grant-in-Aid for Scientific Research (#10480223 to T.Y.), the Japanese Ministry of Education, Science, Sports and Culture.
Abbreviations CLF Cn CNS CPG EPSP IC IN IPSP MLR MN PSP Tn
cervical lateral funiculus nth cervical segment central nervous system central pattern generator excitatory postsynaptic potential intracellular interneuron inhibitory postsynaptic potential midbrain locomotor region motoneuron postsynaptic potential nth thoracic segment
References Amemiya, M. and Yamaguchi, T. (1984). Fictive locomotion of the forelimb evoked by stimulation of the mesencephalic locomotor region in the decerebrate cat. Neurosci. Lett., 50: 91–96. Degtyarenko, A.M., Simon, E.S. and Burke, R.E. (1998). Locomotor modulation of disynaptic EPSPs from the mesencephalic locomotor region in cat motoneurons. J. Neurophysiol., 80: 3284–3296. Edgley, S.A. and Jankowska, E. (1987). An interneuronal relay for group I and II muscle afferents in the midlumbar segments of the cat spinal cord. J. Physiol. (Lond.), 389: 647–674. Edgley, S.A, Jankowska, E. and Shefchyk, S.J. (1988). Evidence that midlumbar neurones in reflex pathways from group II afferents are involved in locomotion in the cat. J. Physiol. (Lond.), 403: 57–71. Go¨dderz, W., Illert, M. and Yamaguchi, T. (1990). Efferent pattern of fictive locomotion in the cat forelimb: with special reference to radial motor nuclei. Eur. J. Neurosci., 2: 663–671. Graf, W., Keshner, E., Richmond, F.J., Shinoda, Y., Statler, K. and Uchino, Y. (1997). How to construct and move a cat’s neck. J. Vest. Res., 7: 219–237. Hishinuma, M. and Yamaguchi, T. (1990). Axonal projection of descending pathways responsible for eliciting forelimb stepping into the cat cervical spinal cord. Exp. Brain Res., 82: 597–605. Hounsgaard, J., Hultborn, H., Jespersen, B. and Kiehn, O. (1988). Bistability of alpha-motoneurones in the decerebrate cat and in the acute spinal cat after intravenous 5-hydroxytryptophan. J. Physiol. (Lond.), 405: 345–367. Ichikawa, Y., Terakado, Y. and Yamaguchi, T. (1991). Lastorder interneurones controlling activity of elbow extensor motoneurones during forelimb fictive locomotion in the cat. Neurosci. Lett., 121: 37–39.
122 Jordan, L.M., Pratt, C.A. and Menzies, J.E. (1979). Locomotion evoked by brain stem stimulation: occurrence without phasic segmental afferent input. Brain Res., 177: 204–207. Kinoshita, M. and Yamaguchi, T. (2001). Stimulus time-locked responses of motoneurons during forelimb fictive locomotion evoked by repetitive stimulation of the lateral funiculus. Brain Res., 904: 31–42. Kitama, T., Seki, K. and Yamaguchi, T. (1996). Labyrinthine reflex activity during forelimb fictive locomotion in the decerebrate cat. Jpn. J. Physiol., Suppl. 46: S142. Mori, S., Shik, M.L. and Yagodnitsyn, A.S. (1977). Role of pontine tegmentum for locomotor control in mesencephalic cat. J. Neurophysiol., 40: 284–295. Mori, S., Matsui, T., Kuze, B., Asanome, M., Nakajima, K. and Matsuyama, K. (1998). Cerebellar-induced locomotion: reticulospinal control of spinal rhythm generating mechanism in cats. Ann. N.Y. Acad. Sci., 860: 94–105. Orlovsky, G.N. (1969). Spontaneous and induced locomotion of the thalamic cat. Biofizika, 14: 1095–1102. Seki, K. and Yamaguchi, T. (1997). Cutaneous reflex activity of the cat forelimb during fictive locomotion. Brain Res., 753: 56–62. Seki, K., Kudo, N., Kolb, F. and Yamaguchi, T. (1997). Effects of pyramidal tract stimulation on forelimb flexor motoneurons during fictive locomotion in cats. Neurosci. Lett., 230: 195–198. Shefchyk, S.J. and Jordan, L.M. (1985). Excitatory and inhibitory postsynaptic potentials in alpha-motoneurons produced during fictive locomotion by stimulation of the
mesencephalic locomotor region. J. Neurophysiol., 53: 1345–1355. Shefchyk, S.J., McCrea, D., Kriellaars, D., Fortier, P. and Jordan, L.M. (1990). Activity of interneurons within the L4 spinal segment of the cat during brainstem-evoked fictive locomotion. Exp. Brain Res., 80: 290–295. Shik, M.L., Severin, F.V. and Orlovsky, G.N. (1966). Control of walking and running by means of electrical stimulation of the midbrain. Biophysics, 11: 659–666. Shik, M.L., Orlovsky, G.N. and Severin, F.V. (1968). Locomotion of the mesencephalic cat evoked by pyramidal stimulation. Biofizika, 13: 127–135. Steeves, J.D. and Jordan, L.M. (1980). Localization of a descending pathway in the SC which is necessary for controlled treadmill locomotion. Neurosci. Lett., 20: 283–288. Terakado, Y. and Yamaguchi, T. (1990). Last-order interneurones controlling activity of elbow flexor motoneurones during forelimb fictive locomotion in the cat. Neurosci. Lett., 111: 292–296. Vidal, P.P., Graf, W. and Berthoz, A. (1986). The orientation of the cervical vertebral column in unrestrained awake animals. I. Resting position. Exp. Brain Res., 61: 549–559. Yamaguchi, T. (1986). Descending pathways eliciting forelimb stepping in the lateral funiculus: experimental studies with stimulation and lesion of the cervical cord in decerebrate cats. Brain Res., 379: 125–136. Yamaguchi, T. (1992). Activity of cervical neurons during forelimb fictive locomotion in decerebrate cats. Jpn. J. Physiol., 42: 501–514.