Eye and head position control is an ideal paradigm for studying how central nervous system (CNS) mechanisms interact to stabilize a multidimensional motor system. Head movement signals detected by the semicircular canals are mediated through vestibulo-ocular and vestibulocollic pathways that link each of the three semicircular canals to a set of eye and head muscles. For tasks necessitating compensatory eye and head movements, the CNS will program muscles to respond in specific combinations rather than to generate an infinite variety of muscle contraction patterns. In order that these movements may be coordinated, the following basic requirements must be fulfilled by the CNS: (1) the appropriate muscles must be selected (spatial); (2) each participating muscle must be activated or inactivated in proper temporal relationship to the others (temporal); and (3) the appropriate amount of excitation or inhibition must be exerted on the pertinent motoneurons (quantitative). Single unit recording in animals engaging in motor behavior, introduced by H. H. Jasper and E. V. Evarts, made it possible to analyze temporal and quantitative aspects of neuronal activities in various parts of the CNS in relation to various movement parameters. However, the important issue of how the spatial or spatiotemporal output patterns of multiple muscle contraction are generated in the CNS still must be addressed.

Stimulation of individual semicircular canals produces canal-specific eye, neck, and body movements. The plane of eye and head movements produced by canal stimulation parallels that of the stimulated canal; that is, they are almost coplanar (Suzuki and Cohen, 1964) (Fig. 41–1). Therefore, a signal from each semicircular canal must be distributed to a proper set of eye and neck muscles to induce compensatory eye and neck movements in the same plane as the plane of the stimulated canal. Obviously a given motor control signal may be economically distributed by way of a single neuron with divergent branches to multiple target sites that participate in cocontraction of muscles to produce a purposeful movement. While the convergence of different inputs onto single neurons has been analyzed extensively, the divergent properties of single neurons have not been described because of technical difficulties. Since the details of the divergent properties of single vestibulospinal (VS) axons in the spinal cord are not yet understood well, this chapter reviews the neural circuitry of the vestibulocollic pathways and summarizes our recent studies on the morphology of single VS axons.

Labyrinthine Influences on Neck Motoneurons

Stimulation of the whole vestibular nerve evokes bilateral excitatory and inhibitory postsynaptic potentials (EPSPs and IPSPs) in neck motoneurons (Wilson and Yoshida, 1969a; Akaike et al., 1973c). The latencies of these evoked EPSPs and IPSPs range from 1.4 to 3.7 ms and most of them are disynaptic from the labyrinth. Stimulation of Deiters' nucleus (the lateral vestibular nucleus; LVN) evokes monosynaptic EPSPs in dorsal neck motoneurons at C2 and C3 (Wilson and Yoshida, 1969b) and stimulation of the medial vestibular nucleus (MVN) evokes monosynaptic IPSPs (Wilson and Yoshida, 1969c). Therefore, labyrinth-evoked disynaptic IPSPs in neck motoneurons were considered to be mediated by the MVN. Inhibitory medial vestibulospinal tract (MVST) neurons have a slow to medium conduction velocity, whereas excitatory MVST neurons have a fast conduction velocity (Akaike et al., 1973c). Felpel (1972) showed that these IPSPs were blocked by strychnine but not by bicuculline, and thus the putative neurotransmitter for these IPSPs was glycine. In these early studies, the origin of the labyrinthine input to neck motoneurons was not taken into account. According to Wilson and Maeda (1974), stimulation of the bilateral anterior and posterior semicircular canal nerves evokes EPSPs and IPSPs, respectively, and stimulation of the horizontal canal nerve evokes ipsilateral IPSPs and contralateral EPSPs. Generally this organized pattern of short-latency postsynaptic potentials (PSPs) is seen in all dorsal neck motoneurons examined, although there are some differences between motoneurons innervating biventer, complexus, and splenius muscles. To determine the contribution of the lateral vestibulospinal tract (LVST) and the MVST for these PSPs, lesions were made in either the LVST or the MVST in the brain stem (Wilson and Maeda, 1974). When the LVST was lesioned on the side ipsilateral to the recorded motoneurons, the disynaptic EPSPs evoked from the ipsilateral anterior canal nerve disappeared, but the disynaptic potentials evoked from the other five canal nerves remained unchanged. When the MVST was cut on the same side as the recorded motoneurons, only disynaptic IPSPs evoked from the ipsilateral horizontal and posterior canal nerves disappeared. Lesioning the MVST on the side ipsilateral to the recorded motoneurons abolished all disynaptic potentials evoked from the contralateral canal nerves (Fig. 41–2A). A similar analysis was performed for effects of semicircular canal nerves on flexor neck motoneurons (Fukushima et al., 1979b). Flexor neck muscle (sternocleidomastoid) motoneurons receive disynaptic excitatory inputs from the three contralateral semicircular canals via the ipsilateral MVST and disynaptic inhibitory inputs from the three ipsilateral canals via the ipsilateral MVST (Fig. 41–2B).

Stimulation of macular nerves is technically more difficult (Suzuki et al., 1969), and little is known about the neural connections between otolith organs and neck motoneurons. Stimulation of the utricular nerve evokes IPSPs ipsilaterally and EPSPs contralaterally in neck extensor motoneurons at latencies as short as 2.0 ms (Wilson et al., 1977). Since disynaptic EPSPs evoked by contralateral whole vestibular nerve stimulation were abolished by transection of the MVST (Akaike et al., 1973c), the utricular-evoked EPSPs must be conveyed by the MVST. The LVST is excitatory (Lund and Pompeiano, 1968; Wilson and Yoshida, 1969b; Grillner et al., 1970). Since it is known that utricular and saccular afferents to Deiters' nucleus are abundant (Stein and Carpenter, 1967; Peterson, 1970), it is likely that otolith effects are also relayed polysynaptically via the LVST to neck motoneurons. Stimulation of the saccular nerve, on the other hand, evokes ipsilateral EPSPs and contralateral IPSPs in neck motoneurons via di- or trisynaptic pathways (Wilson et al., 1977), whose details remain to be examined.

As reviewed above, vestibular effects on neck muscles have been extensively analyzed electrophysiologically, but our knowledge about the vestibular connections to neck motoneurons is (p. 260 )

still limited. Most of these studies were confined to the analysis of inputs to motoneurons of the biventer-complexus muscles. The vestibular input to other neck muscle motoneurons has not been examined, although the large suboccipital muscles are important for head rotation at the atlantoaxial joint. Vidal et al. (1986) clearly demonstrated that, depending on the direction of an intended head movement, different articulations of the neck come into play, indicating a functional compartmentalization of the cervical column. This observation suggests that neck muscles at different portions of the cervical column may be innervated differentially regarding vestibular inputs. Therefore, further systematic analysis of vestibular inputs onto motoneurons of different neck muscles and motoneurons of the same muscles at different cervical levels is necessary. (p. 261 )

Location of Vestibular Nucleus Neurons Projecting to Neck Motoneurons

The location of vestibular neurons projecting to neck motoneurons was examined by recording antidromic spikes evoked from neck motoneuron pools. Many neurons in the LVN are activated from the ventral horn at C6 to T2 (Abzug et al., 1974). Neurons projecting to the ventral horn at C3 are located in the LVN, MVN, and descending vestibular nucleus (DVN), but among them, the neurons sending axons to the ipsilateral LVST are only found in the LVN and DVN, whereas the neurons to the MVST are found in all three nuclei (Rapoport et al., 1977a).

Stimulation of the MVN and the MVST.evokes monosynaptic IPSPs in neck motoneurons, suggesting that inhibitory neurons are located in the MVN (Wilson and Yoshida, 1969c). Since stimulation in the MVN may activate passing fibers from other nuclei, a more direct experimental method was used to identify the location of excitatory and inhibitory neurons acting on neck motoneurons. The activity of single vestibular neurons, recorded extracellularly, was used to trigger an averager while recording intracellular potentials from neck motoneurons (Rapoport et al., 1977b; Uchino and Hirai, 1984). Inhibitory neurons were found in the MVN, DVN, and LVN, and excitatory neurons in the LVN (Rapoport et al., 1977b). Uchino et al. (1984) further characterized these inhibitory and excitatory neurons in terms of their inputs from different semicircular canals (see also Isu et al., 1988). Excitatory vestibular neurons receiving anterior canal input are located in the lateral half and the ventral two thirds of the MVN and in the rostral and ventral two thirds of the DVN. Only a few are found in the ventral LVN (Uchino and Hirai, 1984). Among excitatory vestibular nucleus neurons receiving posterior canal input, the neurons projecting to the contralateral upper cervical spinal cord are located in the rostral one third of the DVN and partially in the MVN, and the group projecting ipsilaterally is mainly located in the ventral part of the LVN (Isu et al., 1988). A more detailed description of the location of these neurons can be found in Chapter 42 in this volume.

Anatomy of Vestibulospinal Projection

In spite of a wealth of electrophysiologic reports on vestibular inputs to neck motoneurons, there are far fewer studies of anatomic connections between vestibular nuclei and neck motoneurons. Most of these anatomic studies used the classical degeneration method. Only a few data are available that have been obtained from autoradiography (Holstege and Kuypers, 1982) and horseradish peroxidase (HRP) application (Peterson and Coulter, 1977; Kneisley et al., 1978).

Originally, the LVST was believed to originate entirely in Deiters' nucleus and to descend ipsilaterally, terminating in all levels of the spinal cord (Brodal et al., 1962). However, recent electrophysiologic studies indicate that neurons in the DVN also send axons into the LVST (Rapoport et al., 1977a). Pompeiano and Brodal (1957), using a retrograde degeneration method, reported that the LVN was organized somatotopically; neurons projecting to the lumbosacral spinal cord were found in the dorsocaudal part of the nucleus, neurons projecting to the thoracic spinal cord in the ventral region of the caudal third of the nucleus, and neurons projecting to the cervical spinal cord in the rostral third and the ventral part of the middle third of the nucleus. This somatotopic organization was reexamined by electrophysiologic methods in which neurons were activated antidromically by stimulating axons at different spinal levels (Wilson et al., 1967a; Peterson, 1970; Akaike, 1973, 1983). The results showed that the neurons projecting as far as the lumbar spinal cord predominate dorsally, whereas the neurons projecting as far as the cervical spinal cord predominate ventrally, although there is considerable overlap.

In addition to the LVST, a second tract from the vestibular nucleus to the spinal cord was found and termed the medial vestibulospinal tract. It was thought to originate only from the MVN and to descend bilaterally in the medial longitudinal fasciculus (MLF) and in the ventral funiculus close to the midline of the spinal cord (Nyberg-Hansen, 1964a). The majority of degenerating fibers of this tract terminate in the cervical spinal cord and disappear below the middle thoracic spinal cord. The degenerating fibers on the ipsilateral side outnumber those on the contralateral side. According to Petras (1967), the fibers are abundant in cervical segments above the cervical enlargement, but rapidly diminish in number in segments through the enlargement, making it difficult to identify fibers below cervical levels. These anatomic findings are in good agreement with the electrophysiologic result that only 11% of MVST axons project below T1 (Akaike, 1973). More recent electrophysiologic studies added some information about the origin of neurons to the MVST. MVST neurons are found not only in the MVN, but also in the DVN (Kawai et al., 1969; Wilson and Yoshida, 1969c) and in the LVN (Akaike, 1973; Akaike et al., 1973a).

The termination sites of the VS tract have been studied with the Marchi method, but reliable details could be determined only with the introduction of the Nauta method. Nyberg-Hansen and Mascitti (1964) showed that degenerating fibers terminate in the entirety of lamina VII and in the neighboring central and also medial parts of lamina VIII at all spinal levels following LVN lesions. No terminations were found on motoneurons, except on cells of the ventromedial group in the thoracic spinal cord. Terminal distribution of LVN neurons in the upper cervical spinal cord was shown by Petras (1967). Terminations were massive in segments throughout the cervical (C5 through Tl) and lumbosacral (L4 through S5) enlargements, but fewer degenerated fibers appeared to be distributed in upper cervical segments (CI through C4) (Fig. 41–3). After a lesion in the MVN, degenerating fibers terminate in the dorsal half of lamina VIII and the neighboring portion of lamina VIII bilaterally, but no terminal degenerations were found on motoneurons (Nyberg-Hansen, 1964a). This author did not explicitly describe terminal degeneration in the upper cervical spinal cord, but in one figure in that paper degenerating terminal fibers are shown in the medial portion of laminae VIII and VII at C3 on both sides (see Fig. 41–3).

Unfortunately, only little information is available on the vestibular projection to the upper cervical spinal cord. In spite of the electrophysiologic demonstration of monosynaptic connections between VS neurons and neck motoneurons, no terminations of the LVST and MVST axons have been found on motoneurons in the degeneration studies (Nyberg-Hansen, 1964a; Nyberg-Hansen and Mascitti, 1964). Therefore, VS axons are thought to synapse on distal dendrites of motoneurons. In fact, neck motoneurons have long dendrites radiating in lamina VIII outside lamina IX (Rose, 1981; Keirstead and Rose, 1983).

Study of Vestibulospinal Axon Morphology in the Cat

To fill the gap between electrophysiologic and anatomic results, we examined the morphology of single VS axons in the cervical spinal cord in the cat. For this purpose, single MVST or LVST axons were penetrated in the ventral funiculus of the cervical spinal cord. They were electrophysiologically identified as such by their monosynaptic responses to stimulation of the vestibular nerves and also by their direct responses to stimulation of the MLF or the LVST, respectively. After this identification, HRP was iontophoretically injected into the axons (Shinoda et al., 1986a, 1988a, 1988b, 1990). The trajectory of single MVST and LVST axons was reconstructed on serial transverse sections of the upper cervical spinal cord. (p. 262 )

Conclusion

Our present data show a morphologic correlate of monosynaptic connections between vestibular nucleus neurons and axial motoneurons, including neck motoneurons, and provide evidence for the existence of multiple axon collaterals from single LVST and MVST axons at different segments of the cervical spinal cord.

(p. 265 ) Single LVST axons may control the excitability of different dorsal axial muscles simultaneously at multisegmental levels by these collaterals and may function to coordinate activation of the muscles of the neck, trunk, and limbs. Single MVST axons have extensive terminals at multiple segments of the upper cervical spinal cord and seem to innervate more than one motor nucleus of neck muscles by diverging axon collaterals, even at the same segment. Accordingly, this result suggests the possibility that a single VS axon may specify a spatial pattern of cocontraction of multiple neck muscles. More specifically, a single MVST axon receiving input from a particular semicircular canal may diverge onto a particular set of multiple motor nuclei of neck muscles, which compose a functional synergy for a proper compensatory head movement. Further studies on functional synergies of neck muscles that are mediated by a single VS axon are necessary to arrive at an understanding of the neural mechanisms used to maintain posture that are controlled by the vestibular system.