- Email: [email protected]
Activity in the mesencephalic locomotor region during locomotion
EXPERIMENTAL
NEUROLOGY
82,609-622
(1983)
Activity in the Mesencephalic Locomotor Region during Locomotion E. GARCIA-RILL, R. D. SKINNER, AND J. A. FITZGERALD' Department of Anatomy, University of Arkansas for Mediral Sciences, Little Rock, Arkansas 7220.5 Received May 9, 1983: revision received July 13, I983 The activity of single neurons in the mesencephalic locomotor region (MLR) was recorded extracellularly in cats during spontaneous locomotion on a treadmill. Although stimulation of the MLR is required to induce locomotion on a treadmill after a precollicular-postmamillary brain stem transection in the cat, spontaneous locomotion may occur after a precollicular-premamillary transection. The activity of flexor and extensor muscles of each limb also was recorded by EMG. Nearly 50% of the MLR neurons exhibited rhythmic firing patterns during locomotion In about one-half of those cells, unit firing patterns could be correlated with the EMG activity in one or more muscles by using spike-tapers averaging. Single MLR neurons were found to be correlated to EMG activity in a single limb, and others were related to the EMG from muscles in two limbs or in all four limbs. Passive movement or stoppage of the limb(s) did not abolish rhythmicity in these neurons. In addition. somatosensory stimulation did not appear to affect the firing patterns of MLR neurons. Averaged EMGs of correlated forelimb muscles revealed a postspike mean latency of 7.1 ms. These measurements agreed well with reports of a I - to 1.5ms delay in MLR projections to reticulospinal neurons and a 5- to 6-ms delay (postspike) in reticulospinal activity correlated to EMGs during Iocomotion. These findings suggest that (a) MLR neurons are rh~hmically active during locomotion, (b) the activity of MLR neurons can be correlated with that of EMGs in one or more limbs. (c) rhythmicity in MLR neurons may be independent of phasic sensory input. and (d) the downstream influence of the MLR may be relayed, at least in part. via reticulospinal neurons.
Abbreviations: BC-bra&mm conjunctivum, CFN-cuneiform nucleus, MLR-mesencephalic locomotor region, NRG-nucleus reticularis ~~nt~ll~a~s, PLS-~ntobuI~r locomotor strip, PPN-~dunculo~ntine nucleus. ’ This work was supported by grants from the National Institutes of Health (NS- 16 143) and the National Science Foundation (ISP-80 I I447 ). 609
0014-4886/83 $3.00 Copyright ‘C’ 1982 hg Academic Press, inc. 411 nghrs of reproducmn in anr form reser\cd
610
GARCIA-RILL,
SKINNER,
AND
FITZGERALD
INTRODUCTION The mesencephalic locomotor region (MLR) is in the posterior midbrain of the cat and when stimulated, using low-amplitude current pulses (20 to 60 PA) at high frequencies (20 to 60 Hz), induces controlled locomotion on a treadmill (12, 27). This occurs when a precollicular-postmamillary transection is accomplished and a moving treadmill is placed beneath the limbs while the animal’s weight is supported (27). These findings imply that descending influences of the MLR can ultimately modulate locomotion oscillators present in the spinal cord (12, 18-20). We recently described the presence of small, direct projections to the MLR from the motor cortex and from the entopeduncular nucleus (6) and substantia nigra (8) the main outputs of the basal ganglia. That series of studies also determined that neuronal elements, and not fibers of passage, were responsible for the effects of electrical stimulation of the MLR (9-l 1). Those neurons were located in the region of the cuneiform nucleus (CFN) and the pedunculopontine nucleus (PPN), which is embedded in the brachium conjunctivum. The efferent fibers of the MLR were studied using anterograde transport of tritiated amino acids away from an injection site physiologically identified (by electrically inducing locomotion) as the MLR (9). One MLR efferent projection was found to travel along Probst’s tract in the dorsolateral medullary reticular formation and corresponded to the pontobulbar locomotor strip (PLS). The PLS is a tail-like extension of the MLR and stimulation of this region can also induce locomotion on a treadmill (28). A second MLR efferent pathway was found to project to the nucleus reticularis gigantocellularis (NRG), which is a source of reticulospinal projections (2). It had already been established that stimulation of the MLR would induce monosynaptic responses in some reticulospinal neurons (24). The reticulospinal projection system must remain intact in order for stimulation of the MLR to induce locomotion (33). The activity of reticulospinal neurons in relation to locomotion movements recently was described (30, 3 1). All those reports are suggestive of the presence of major locomotion modulatory mechanisms at the level of the MLR. However, no reports on the activity of MLR neurons during locomotion on a treadmill have been forthcoming. This gap in the literature makes it impossible to determine if neurons in the MLR have a tonic or phasic, rhythmic or nonrhythmic influence on downstream centers. One difficulty has been the necessity of electrical stimulation of the MLR in order to induce locomotion. Fortunately, however, if a precollicular-premamillary transection (or even a “decerebration” anterior to the thalamus) is carried out, the animal will locomote spontaneously (12, 23). We studied single-unit activity in the MLR in the spontaneous locomotion preparation to determine (a) if MLR neurons are active during locomotion,
UNIT RECORDING
OF MLR DURING
L~OMOTION
611
(b) if MLR neuronal activity is related to muscular activity during the step cycle, and (c) the exact location of MLR neurons related to locomotor events. Preliminary findings have been reported (7). METHODS Adult cats (3 to 4 kg) were anesthetized with a short-acting barbiturate (sodium methohexital, 11 mg/kg, i.v.) during surgical procedures. The carotid arteries were ligated, the bone overlying the occipital cortex was removed, and the cortex and tentorium overlying the corpora quadrigemina were extirpated. A precollicular-premamillary transection then was made using a spatula. This transection renders the preparation insentient to pain, nevertheless, all wound margins and pressure points were in~ltrated with local anesthetic agents (injectable procaine, topical xylocaine). A recording well was centered over the inferior colliculus and secured with dental cement to bolts implanted in the skull. Bipolar hooked-wire electrodes were implanted in muscles representative of flexors and extensors in forelimbs and hind limbs. The electromyograms (EMGs) were recorded on FM tape. The activity of the MLR neurons was recorded extracellularly using glass micropipets (1.6 M potassium citrate in a saturated solution of fast green dye, 20 to 40 Ma resistance) introduced through the recording well. The locus of selected neurons was marked by deposition of fast green dye at the recording site. After the recordings the animals were killed by an overdose of barbiturate, perfused via the previously ligated carotids with 10% phosphatebuffered Formalin, and the loci of dye spots and extent of transection verified in frozen 40-pm sections stained with cresyl violet. The taped activity of MLR neurons was correlated with concurrent EMG activity by using spike-triggered averaging. Each channel of EMG was “rectified” by digitizing, squaring all points, and normalizing. Each channel then was averaged using the MLR action potential to (spike) trigger each trial. Typically, 20 ms of EMG activity (in OS-ms bins) that followed each trigger was averaged for 100 to 2000 trials (spikes). These procedures were similar to those previously used in the analysis of neuronal activity in relation to quadripedal movements in cats (31). Each channel of EMG was averaged for the same trials (21%). The mean of each average was calculated as the average of activity occurring in the first 4 ms after each spike. The criteria for accepting a correlation consisted of a change in activity in excess of two standard deviations away from the mean and of at least three consecutive bins (1.5 ms) in duration. Because the basal EMG also was “rectified” and summed, negative (below the mean) correlations were accepted. The square of the EMG was used in the average as it is related to the power in the signal. In this manner, significant EMG events become evident over background
612
GARCIA-RILL,
activity with a smaller number unsquared signal.
SKINNER,
AND
FITZGERALD
of trials than does the standard rectified,
RESULTS Recordings were carried out only in preparations which showed s~n~neous locomotion (N = 15). Less than half of all transected animals walked spontaneously, and then for periods lasting 1 to 8 h. Although treadmill speed remained constant, the locomotion exhibited in these preparations varied considerably. Generally, locomotion started in one limb, was followed by the contralateral limb after a few cycles and then was joined by the other pair of limbs for periods lasting 1 to 30 min. Locomotion stopped spontaneously and was marked by moments of unilateral or bilateral rigidity, then relaxation, and ultimately continued cycling. These preparations allowed the correlation of single-cell activity with many aspects of muscular activity, rhythmic and nonrhythmic. A total of 96 neurons was recorded in the MLR for periods of 5 to 45 min during spontaneous locomotion. The presence of an A-B break on the rising phase of the action potentials of each cell was considered adequate evidence that the activity of fibers of passage was not being recorded. Each neuron studied was tested for (a) auditory input to determine ~rni~~le~~e sampling in the inferior colliculus was taking place, and (b) responses to jaw opening to determine if the recording electrode was positioned in the mesencephalic trigeminal nucleus. Several “tracks” through the central gray nucleus failed to sample any neurons in that nucleus showing rhythmic firing patterns. Recordings of all central gray neurons were, therefore, discarded (as were recordings of inferior colliculus and trigeminal neurons) following localization of dye spots in that region. The remainder of the neurons studied (N = 96) was localized within the CFN and PPN. The histo~o~~lly verified positions of these cells corresponded with locomotion-inducing stimulation sites in previous experiments (6,8-lo), which also pointed to a heterogeneous population of cells at the level of the MLR. Neurons recorded in the CFN and PPN which showed rhythmic patterns were found interspersed with other neurons showing no correlation with locomotor movements. This mixed population was assumed to be part of the MLR, although we can provide no direct evidence to suggest that nonrh~hmi~lly active neurons within the CFN and PPN were an integral part of the MLR. Attempts to further characterize these neurons in terms of their descending projections proved unsuccessful in this type of preparation. High-frequency (300 to 500 Hz) stimulation of the NRG or Probst’s tract at currents necessary to induce antidromic responses (two to three times the threshold for inducing locomotion) in MLR neurons produced postures and movements of sufficient violence to displace
UNIT
RECORDING
OF MLR DURING
613
L~OMOTION
the recording electrode. Collision testing proved unwieldy because antidromic stimulation blocked spontaneous locomotion as well. The MLR neurons (according to the above criteria) were tested unsuccessfully for responses to phasic somatosensory stimulation of the limbs and proximal areas. Passive movement of the limbs during quiescent episodes also failed to elicit responses in rh~hmically active neurons. Passive manipulation during spontaneous locomotion did not produce a significant change in MLR neuronal activity, but did induce active limb withdrawal or resistance, which also was occasionally sufficient to displace the microelectrode. Of the neurons localized in the CFN and PPN, more than 50% showed rhythmic activity. This activity changed with the frequency of the step cycle, and was generally composed of bursting firing patterns which increased in frequency toward the end of the burst. Bursts were 0.2 to 1.0 s in duration and were interrupted by quiescent periods of 0.2 to 0.8 s. However, when spike-triggered averaging was used to test for a correlation between unit activity and EMG activity, only 23% of the MLR neurons showed rhythmic activity in relation to at least one alternating EMG pattern. Those rhythmically active neurons not showing a positive correlation may have been contributing to overall rhythmic patterning by the MLR or were related to the activity of muscles which did not have implanted EMG leads. Of those cells which were found to be correlated with EMG activity (N = 22). 32% were related to EMG activity in muscles of one limb (either ipsi- or contralateral). 45% were related to EMG activity in muscles of two limbs and 23% were related to EMG activity in muscles of all four limbs. Table 1 shows a breakdown of the types of activity, rhythmic and nonrhythmic, found in this population of MLR neurons. Cells denoted as “related” were those found to be correlated by spike-triggered averaging to at least one TABLE
1
Mesencephaiic Locomotor Region Units Localized by Fast Green Deposition in Spontaneously Locomoting Brain Stem-Transected Cats
Total number studied Not related to locomotion Rh~hmica~iy active Rh~hmicafly active in relation to alternating EMGs Of the last group: Related to EMGs of one limb Related to EMGs of two limbs Related to EMGs of all limbs
No. cells
Percentage
96 40 34
100 42 35
22
23
7 10 5
32 45 23
614
GARCIA-RILL,
SKINNER,
AND
FITZGERALD
EMG. Spike-triggered averaging showed positive correlations with one particular muscle (usually a flexor) in some cells. In other neurons a negative correlation with the antagonist muscle also was observed, and, in those related to more than one limb, a positive correlation with concurrently contracting muscles in other limbs also was evident. Figure 1, lower left, shows the location of a fast green dye spot deposited at the site (boundary of the CFN and PPN or brachium conjunctivum) where the activity of an MLR neuron was recorded. This figure also shows the rhythmic firing pattern (long burst increasing in frequency) of this neuron together with concurrent EMG records of ipsilateral and contralateral forelimb flexors during spontaneous locomotion. The activity in this neuron appears to precede the activity of the contralateral forelimb flexor. This correlation is better observed in a subsequent segment of activity of the same neuron (Fig. 2). During this episode, stepping frequency increased and the activity of the neuron was diminished in concert with that of the EMG of the con-
MLR UNIT ACTIVITY
I PSI ‘FORELIMB FLEXOR
1
EMG CONTRA
1 s
rRECORDlNG SITE
FIG. I. Activity of a single mesencephahc locomotor region (MLR) neuron (top record) during spontaneous locomotion in a precollicular-premamillary transected cat. The location of this recording is shown in the drawing at bottom left. The activity of EMGs of ipsilateral (IPSI) and contralateral (CONTRA) forelimb flexors is shown in the middle and bottom records, respectively. BC-brachium conjunctivum, CFN-cuneiform nucleus, CG-central gray, K-inferior colliculus, P2-stereotactic plane.
UNIT RECORDING
OF MLR DURING
615
L~OMOTION
tralateral flexor. The EMG activity of the ipsilateral flexor appeared undiminished and was comparable in size to that of the previous segment (Fig. I). The possible correlations of the activity of the neuron shown in Figs. 1 and 2 were tested using spike-triggered averaging. The results of this analysis are shown in Fig. 3. A si~ni~cant positive correlation with the EMG activity in the contralateraf forelimb flexor was clearly evident at a 7-ms latency. In addition, a negative correlation with the EMG of the ipsilateral forelimb flexor was present at a 12.5ms latency. This neuron was typical of those showing activity correlated with EMGs in more than one limb. Figure 4 shows the spike-triggered averages of a different MLR neuron (located in the PPN). This neuron showed a positive correlation with the ipsilateral forelimb flexor at a 7.5ms latency, together with a negative correlation with the antagonist, the ipsilateral forelimb extensor at a 55ms latency. Although the contralateral extensor is somewhat concurrently active when the ipsilateral flexor is contracting, no statistical correlation with that muscle was found. This MLR neuron was representative of those showing a correlation with the activity of muscles in only one limb. The mean latency from the trigger (spike) to the beginning of a significant correlation with forelimb EMG activity for the entire population of MLR neurons was 7.1 + 3.2 ms (standard error of the mean, range 5.5 to 16 ms). The mean latency for correlated hind limb EMG activity was 6 to 7 ms
MLR UNIT ACTIVITY
IPSI FORELIMB FLEXOR
EMG CONTRA
I
s
FIG. 2. Activity of the same MLR neuron (top record) as in Fig. 1. during a different episode of spontaneous locomotion. The EMG activity of ipsilateral and contralateral forelimb flexors is shown in the middle and bottom records, respectively. The activity of this cell was used to calculate the spike-triggered averages shown in Fig. 3.
616
GARCIA-RILL,
SKINNER,
AND FITZGERALD
t’i,;“h-:; 1
MLR
;:-r
o.
UNIT
CF b
0.00 sroo--
10.00
15.00
.
MS
myj
CE b
0. 00
5. 00
20.00 MS
.
‘F b
FIG. 3. Left side-a dorsal view of the midbrain, medulla (MED) and cervical enlargement (CE). Right side-spike-triggered averages of EMGs of muscles (MUS) in the contralateral (CFcontralateral flexor, CE-contralateral extensor) and ipsilateral (IF-ipsilateral flexor, IE-ipsilateral extensor) forelimbs during spontaneous locomotion. Each EMG was averaged for the 20-ms period following the occurrence of each of 1000 spikes (occurring at point 0.00) of an MLR neuron (sfar). The mean of each average is the mean of the activity during the first 4 ms and is indicated by the solid line. The dotted line indicates the activity at two standard deviations above the mean. Filled bins, each 0.5 ms in width, indicate activity in excess of two standard deviations above or below the mean. Periods of correlation having durations of three or more bins (1.5 ms) were present in the averages of CF (increase at a 7-ms latency) and IF (decrease at a 12.5-ms latency). SC-superior colliculus, TRANS-precollicular-premamillary brain stem transection.
longer. Figure 5 shows the histologically verified loci of all MLR neurons which were rhythmically active and which exhibited a significant correlation with at least one EMG. The distribution was limited to the CFN and PPN. The MLR neurons which did not show a correlation with EMG activity (not shown) were located among those shown in Fig. 5, often sampled in the same microelectrode “track.” DISCUSSION Our results demonstrate that (i) a significant proportion of MLR neurons are rhythmically active during locomotion, (ii) the activity of some of those
UNIT RECORDING
OF MLR DURING
L~OMOTION
617
neurons can be shown to be correlated with cyclic EMG activity in particular limb muscles by using spike-triggered averaging, and (iii) MLR neurons correlated with locomotor movements are situated in the CFN and PPN. Until the advent of a series of studies outlining the afferent and efferent connections of the MLR (9), it was not known if stimulation-induced locomotion at that site was due to activation of fibers of passage or of neuronal somata (12). Those reports determined that the motor cortex, entopeduncular nucleus, and substantia nigra projections were directed at the same site which, when stimulated, elicited locomotion on a treadmill (6, 8). Intracellular recordings in proximity to a locomotion-inducing site revealed the presence of functionally varied synaptic effects after activation of pallidal afferent fibers to the MLR ( 10). Anterograde and retrograde transport away from the physiologically identified MLR also revealed its input-output organization (9). More recently, we reported that controlled locomotion can be induced by localized application of pharmacologic agents in the MLR (11). The presence of dendritic or somatic receptors is necessary for the expression of those effects, again indicative of neuronal, not axonal, substrates being responsible for MLR-induced locomotion. Still lacking, however, was any evidence demonstrating that MLR neuronal activity was in any way related to locomotion. The findings described herein provide convincing evidence that a significant propo~ion of MLR neurons are active rh~hmically during s~ntaneous locomotion. In some of these cells, a significant correlation between the rhythmic activity and alternating EMG patterns can be established. These results indicate that the MLR may be providing not only a rhythmic background of activity for downstream centers, but may also be “driving” some of those centers. The MLR is necessary for eliciting stimulation-induced locomotion in a cat with a precollicular-postmamillary transection. Because spontaneous locomotion can occur after a precollicular-premamilla~ transection, the implication is that the MLR or parallel systems are “released” by the more anterior transection (12, 18,25). The fact that MLR unit activity is still rhythmic and related to EMG activity in this preparation speaks to the importance of this region in the rhythmic modulation of locomotion. How does the MLR influence downstream centers? Our previous results indicated the presence of two main descending projections of the MLR (9). First, efferent fibers in the region of Probst’s tract in the dorsolateral medullaxy reticular fo~ation were noted. This projection coincides with the location of the PLS, which other investigators stimulated to induce locomotion (28). This tract appears to be in or near Probst’s tract, which is part of the trigeminospinal system (17). That this pathway may be partially involved in modulating locomotion is supported by studies demonstrating locomotion on a treadmill in the precollicular-postmamillary cat after stimulation of the pinna f 1).
618
GARCIA-RILL,
SKINNER,
AND FITZGERALD
I 0. 00
LJ 5.00 ‘A
20.00 MS'
0: 00
5: 00
'
lb. 00
Ii.00
zb. 00
15.00
20.00
MS
0.00
. -0.00 5’00 MS
'
m*j 0. 00
5.00
10.00
MS
FIG. 4. Spike-triggered averages of EMGs of forelimb flexors and extensors that follow the activity of another MLR neuron. Significant correlations were present in the IF (increase at a 7.5-ms latency) and its antagonist IE (decrease at a 5.5-ms latency). Abbreviations as in Fig. 3.
There is a second downstream projection of the MLR, that to the NRG (9), a source of reticulospinal projections (2). Stimulation of the MLR was reported to induce monosynaptic responses at a l- to 1S-ms latency in reticulospinal neurons (24). We subsequently demonstrated the presence of this pathway anatomically (9). This projection may well be the primary element through which the MLR modulates downstream centers. Lesion of the ventrolateral quadrant of the white matter of the spinal cord is known to abolish MLR-induced locomotion (33). This finding implies that the reticulospinal (and perhaps vestibulospinal) projections are essential for the manifestation of MLR effects. Recent expe~ments in which the activity of reticulospinal neurons was recorded during spontaneous locomotion in the “thalamic” cat, showed these neurons to be rhythmically active and correlated with EMG activity in limb muscles (30, 31). The use of spike-triggered averaging in those experiments revealed a 5- to 6-ms latency for averaged EMGs in forelimb muscles following reticulospinal neuron activity (3 1). Allowing for the reported l- to IS-ms delay from the MLR to the NRG (24), the mean latency reported herein of 7.1 ms for MLR activity preceding EMG
UNIT RECORDING
PlO
OF MLR DURING
Pi5
619
L~OMOTION
P20
FIG. 5. Locations of MLR neurons which showed rhythmic activity in relation to locomotor movements at various stereotactic planes. The focalized fast green dye deposits were limited to the CFN and BC, in which the PPK is ern~dd~.
activity in forelimb muscles agrees well with those reports. Our data suggest that the correlation between MLR unit activity and EMGs during locomotion is mediated via the reticulospinal system. Not only is the latency appropriate, but the duration of the correlation is similar for MLR spike-triggered averages (this report) compared with reticulospinal spike-triggered averages (3 1). Previous studies of reticulospinal neurons during s~ntaneous locomotion reported the presence of clear responses to phasic somatosenso~ stimulation in these neurons (30, 31). In addition, passive manipulation of the limbs leads to a reduction in rhythmicity in reticulospinal neurons (30, 3 1). These responses are thought to be mediated by the spinobulbospinal pathway (29). Our experiments reveal no such responses in MLR neurons. The lack of afferent modulation in MLR activity indicates this region may be a central rhythmic modulator the output of which comes under the influence of phasic aRerent input only at lower levels (perhaps reticulospinal and/or spinal). This afferent input is not absolutely essential for the MLR to induce locomotion because it is still possible to induce rhythmic activity in motoneurons after stimulation of the MLR in the “fictive locomotion” (paralyzed) preparation (13). Our findings also demonstrated that neurons which are active rhythmically in relation to locomotor movements are situated in the CFN and PPN. That the CFN is part of the MLR was suggested in the original description of brain stem-induced locomotion (27). On the other hand, only recently has the PPN been associated with the MLR (6-10). The PPN is embedded in the brachium conjunctivum and Long has been known to receive basal ganglia (14, 21) and cortical (15) afferent fibers. However, only the most posterior part of the PPN appears to be within the MLR, because stimulation of the anterior PPN does not induce locomotion after an appropriate brain stem transection (6-10). The PPN as a whole was reported to have reciprocal
620
GARCIA-RILL,
SKINNER,
AND FITZGERALD
connections with the pallidum (5, 16, 2 1, 32), the substantia nigra (5, 6, 8, 16), and the subthalamic nucleus (4, 5, 22). As such, it has been included as an integral part of “extrapyramidal circuitry” (26). It is then logical to suggest that the link between the basal ganglia and the MLR, which includes at least a portion of the PPN, participates in processes related to the functions of both regions. Because the basal ganglia are involved in the preparation for movement (3), and the MLR is involved in the modulation of locomotion, one suggestion is that this link may participate in the preparation for movements which include locomotion. Under this model, if the link between the basal ganglia and the MLR were absent, initiation of locomotion would be impaired. This type of deficit is present in Parkinson’s disease. Moreover, the observed inability of Parkinson patients to stop walking when started may be due to a “locking” of the locomotion oscillators to the sensory feedback generated by leg swing. A similar observation was reported in animals with caudate nucleus ablation (34). The peripheral input to spinal locomotion oscillators is, under normal circumstances, modulated by descending central projections (12, 18, 23). Deficits in starting and stopping locomotor movements could become manifest if these central connections were compromised. Much additional evidence is necessary to substantiate this rudimentary functional model, and the foregoing discussion is intended to provide a general basis for further investigation. REFERENCES I. AOKI, M., AND S. MORI. 1981. Locomotion elicited by pinna stimulation in the acute precollicular-postmamillary decerebrate cat. Bruin Rex 214: 424-428. 2. BRODAL, A. 1956. Anatomical aspects of the reticular formation of the pons and medulla oblongata. Prog. Neurobiol. 2: 240-255. 3. BUCHWALD, N. A., C. D. HULL, M. S. LEVINE, AND J. VILLABLANCA. 1975. The basal ganglia and the regulation of response and cognitive sets. Pages 171-189 in M. A. B. BRAZIER, Ed.. Growth and Development of the Brain. Raven Press, N.Y. 4. GROSSMAN, A. R., AND A. JACKSON. 1981. The efferent projections of the subthalamic nucleus (of Luys) in the rat, with special reference to a previously undescribed projection. J. Physiol. (London) 319: 108P. 5. EDLEY, S. M., AND A. M. GRAYBIEL. 1980. Connections of the nucleus tegmenti pedunculopontinus, pars compacta (TPc) in cat. Amt. Rec. 196: 129A. 6. GARCIA-RILL, R. D. SKINNER, AND S. GILMORE. 1981. Pallidal projections to the mesencephalic locomotor region (MLR) in the cat. Am. J. Amt. 161: 31 l-321. 7. GARCIA-RILL, W., R. D. SKINNER. AND J. A. FITZGERALD. 1982. Mesencephalic locomotor region (MLR) unit activity during locomotion. Sot. Neurosci. Abstr. 8: 167. 8. GARCIA-RILL, E., R. D. SKINNER, M. B. JACKSON, AND M. M. SMITH. 1983. Connections of the mesencephalic locomotor region (MLR). I. Substantia nigra afferents. Bruin Res. Bull. 10: 57-62. 9. GARCIA-RILL, E., R. D. SKINNER, S. A. GILMORE, AND R. OWINGS. 1983. Connections of
UNIT RECORDING
OF MLR DURING
LOCOMOTION
621
the mesencephalic locomotor region (MLR). II. Afferents and efferents. Bruin Rex Bull. 10: 63-71.
10. GARCIA-RILL, E. 1983. Connections of the mesencephalic locomotor region (MLR). III. Intracellular recordings. Brain Res. Bull. 10: 73-8 I. 11. GARCIA-RILL, E., R. D. SKINNER. AND J. A. FITZGERALD. 1983. Locomotion induced by chemical activation of the mesencephalic locomotor region (MLR). Sot. Neurosci. .4bstr. 9: 357.
12. GRILLNER, S.. AND M. L. SHIK. 1973. On the descending control of the lumbosacral spinal cord from the “mesencephalic locomotor region.” Acta Physiol. Scund. 87: 320-333. 13. JORDAN, L. M., C. A PRATT, AND J. E. MENZIES. 1979. Locomotion evoked by brain stem stimulation: occurrence without phasic segmental afferent input. Bruin Rex 177: 204207.
14. Kuo, J. S.. AND M. B. CARPENTER. 1973. Organization of pallido-thalamic projections in the rhesus monkey. J. Comp. Neural. 151: 201-236, 1973. 15. KUYPERS, H. G. J. M., AND D. G. LAWRENCE. 1967. Cortical projections to the red nucleus and the brain stem in the rhesus monkey. Brain Rex 4: 15 l-188. 16. LARSEN, K. D.. AND J. SUTIN. 1978. Output organization of the feline entopeduncular and subthalamic nuclei. Bruin Res. 157: 21-3 1. 17. MATSUSHITA, M., N. OKADO, M. IKEDA, AND Y. HOSOYA. 1981. Descending projections from the spinal and mesencephalic nuclei of the trigeminal nerve to the spinal cord in the cat. A study with the horseradish peroxidase technique. J. Comp. Neural. 196: l73187. 18. MORI, S., M. L. SHIK, AND A. S. YACODNI~-SYN. 1977. Role of pontine tegmentum for locomotor control in mesencephalic cat. J. Neurophysiol. 40: 284-295. 19. MORI, S., H. NISHIMLJRA, C. KURAKAMI, T. YAMAMURA, AND M. AOKI. 1978. Controlled locomotion in the mescncephalic cat: distribution of facilitory and inhibitory regions within pontine tegmentum. J. Neurophysiol. 41: I580- 159 I. 20. MORI, S., K. KAWAHARA, T. SAKAMOTO, M. AOKI, AND T. TOMIYAMA. 1982. Setting and resetting of level of postural muscle tone in decebrate cat by stimulation of brain stem. J. Neurophysiol.
48: 737-748.
21. NAUTA, W. J. H.. AND W. R. MEHLER. 1966. Projections of the lentiform nucleus in the monkey. Bruin Res. 1: 3-42. 22. NAUTA, H. J. W.. AND M. COLE. 1978. Efferent projections of the subthalamic nucleus: an autoradiographic study in monkey and cat. J. Comp. Neural. 180: I-16. 23. ORLOVSKY,G. N. 1969. Spontaneous and induced locomotion of the thalamic cat. Biophysics 14: 1154-l 162. 24. ORLOVSKY, G. N. 1970. Connections between reticulospinal neurons and locomotor regions of the brainstem. Biophysics 15: 58-64. 25. ORLOVSKY, G. N. 1972. The effect of different descending systems on flexor and extensor activity during locomotion. Brain Res. 40: 359-37 1. 26. SAPER, C. B., AND A. D. LOEWU. 1982. Projections of the pedunculopontine tegmental nucleus in the rat: evidence for additional extrapyramidal circuitry. Brain Rex 252: 367372.
SHIK, M. L., F. V. SEVERIN, AND G. N. ORLOVSKY. 1966. Control of walking and running by means of electrical stimulation of the mid-brain. Biophysics 11: 756-765. 28. SHIK, M. L., AND A. S. YAGODNITSYN. 1977. The pontobulbar locomotor strip. Neuro27.
physiology
9: 72-74.
29. SHIMAMURA, M., I. KOGURE, AND S. WADA. 1980. Three types of reticular neurons involved in the spino-bulbo-spinal reflex of cats. Bruin Res. 186: 99-l 13.
622
GARCIA-RILL,
SKINNER,
AND
FITZGERALD
30. SHIMAMURA, M., I. KOGURE, AND S. WADA. 1982. Reticular neuron activities associated with locomotion in thalamic cats. Brain Rex 231: 51-62. 3 I. SHIMAMURA, M., AND I. KOCURE. 1983. Discharge patterns of reticulospinal neurons corresponding with quadrupedal leg movement in thalamic cats. Bruin Res. 230: 27-34. 32. SKINNER, R. D., E. GARCIA-RILL, AND S. GILMORE. 1979. Entopeduncular nucleus projections to the mesencephalic locomotor region. Sot. Neurosci. Abstr. 5: 79. 33. STEEVES, J. D., AND L. M. JORDAN. 1980. Localization of a descending pathway in the spinal cord which is necessary for controlled treadmill locomotion. Neurosci. Left. 20: 283-288. 34. VILLABLANCA, J., R. J. MARCUS, AND C. E. OLMSTEAD. 1976. Effects of caudate nuclei or frontal cortical ablation in cats. 1. Neurology and gross behavior. Exp. Neural. 52: 389420.
NEUROLOGY
82,609-622
(1983)
Activity in the Mesencephalic Locomotor Region during Locomotion E. GARCIA-RILL, R. D. SKINNER, AND J. A. FITZGERALD' Department of Anatomy, University of Arkansas for Mediral Sciences, Little Rock, Arkansas 7220.5 Received May 9, 1983: revision received July 13, I983 The activity of single neurons in the mesencephalic locomotor region (MLR) was recorded extracellularly in cats during spontaneous locomotion on a treadmill. Although stimulation of the MLR is required to induce locomotion on a treadmill after a precollicular-postmamillary brain stem transection in the cat, spontaneous locomotion may occur after a precollicular-premamillary transection. The activity of flexor and extensor muscles of each limb also was recorded by EMG. Nearly 50% of the MLR neurons exhibited rhythmic firing patterns during locomotion In about one-half of those cells, unit firing patterns could be correlated with the EMG activity in one or more muscles by using spike-tapers averaging. Single MLR neurons were found to be correlated to EMG activity in a single limb, and others were related to the EMG from muscles in two limbs or in all four limbs. Passive movement or stoppage of the limb(s) did not abolish rhythmicity in these neurons. In addition. somatosensory stimulation did not appear to affect the firing patterns of MLR neurons. Averaged EMGs of correlated forelimb muscles revealed a postspike mean latency of 7.1 ms. These measurements agreed well with reports of a I - to 1.5ms delay in MLR projections to reticulospinal neurons and a 5- to 6-ms delay (postspike) in reticulospinal activity correlated to EMGs during Iocomotion. These findings suggest that (a) MLR neurons are rh~hmically active during locomotion, (b) the activity of MLR neurons can be correlated with that of EMGs in one or more limbs. (c) rhythmicity in MLR neurons may be independent of phasic sensory input. and (d) the downstream influence of the MLR may be relayed, at least in part. via reticulospinal neurons.
Abbreviations: BC-bra&mm conjunctivum, CFN-cuneiform nucleus, MLR-mesencephalic locomotor region, NRG-nucleus reticularis ~~nt~ll~a~s, PLS-~ntobuI~r locomotor strip, PPN-~dunculo~ntine nucleus. ’ This work was supported by grants from the National Institutes of Health (NS- 16 143) and the National Science Foundation (ISP-80 I I447 ). 609
0014-4886/83 $3.00 Copyright ‘C’ 1982 hg Academic Press, inc. 411 nghrs of reproducmn in anr form reser\cd
610
GARCIA-RILL,
SKINNER,
AND
FITZGERALD
INTRODUCTION The mesencephalic locomotor region (MLR) is in the posterior midbrain of the cat and when stimulated, using low-amplitude current pulses (20 to 60 PA) at high frequencies (20 to 60 Hz), induces controlled locomotion on a treadmill (12, 27). This occurs when a precollicular-postmamillary transection is accomplished and a moving treadmill is placed beneath the limbs while the animal’s weight is supported (27). These findings imply that descending influences of the MLR can ultimately modulate locomotion oscillators present in the spinal cord (12, 18-20). We recently described the presence of small, direct projections to the MLR from the motor cortex and from the entopeduncular nucleus (6) and substantia nigra (8) the main outputs of the basal ganglia. That series of studies also determined that neuronal elements, and not fibers of passage, were responsible for the effects of electrical stimulation of the MLR (9-l 1). Those neurons were located in the region of the cuneiform nucleus (CFN) and the pedunculopontine nucleus (PPN), which is embedded in the brachium conjunctivum. The efferent fibers of the MLR were studied using anterograde transport of tritiated amino acids away from an injection site physiologically identified (by electrically inducing locomotion) as the MLR (9). One MLR efferent projection was found to travel along Probst’s tract in the dorsolateral medullary reticular formation and corresponded to the pontobulbar locomotor strip (PLS). The PLS is a tail-like extension of the MLR and stimulation of this region can also induce locomotion on a treadmill (28). A second MLR efferent pathway was found to project to the nucleus reticularis gigantocellularis (NRG), which is a source of reticulospinal projections (2). It had already been established that stimulation of the MLR would induce monosynaptic responses in some reticulospinal neurons (24). The reticulospinal projection system must remain intact in order for stimulation of the MLR to induce locomotion (33). The activity of reticulospinal neurons in relation to locomotion movements recently was described (30, 3 1). All those reports are suggestive of the presence of major locomotion modulatory mechanisms at the level of the MLR. However, no reports on the activity of MLR neurons during locomotion on a treadmill have been forthcoming. This gap in the literature makes it impossible to determine if neurons in the MLR have a tonic or phasic, rhythmic or nonrhythmic influence on downstream centers. One difficulty has been the necessity of electrical stimulation of the MLR in order to induce locomotion. Fortunately, however, if a precollicular-premamillary transection (or even a “decerebration” anterior to the thalamus) is carried out, the animal will locomote spontaneously (12, 23). We studied single-unit activity in the MLR in the spontaneous locomotion preparation to determine (a) if MLR neurons are active during locomotion,
UNIT RECORDING
OF MLR DURING
L~OMOTION
611
(b) if MLR neuronal activity is related to muscular activity during the step cycle, and (c) the exact location of MLR neurons related to locomotor events. Preliminary findings have been reported (7). METHODS Adult cats (3 to 4 kg) were anesthetized with a short-acting barbiturate (sodium methohexital, 11 mg/kg, i.v.) during surgical procedures. The carotid arteries were ligated, the bone overlying the occipital cortex was removed, and the cortex and tentorium overlying the corpora quadrigemina were extirpated. A precollicular-premamillary transection then was made using a spatula. This transection renders the preparation insentient to pain, nevertheless, all wound margins and pressure points were in~ltrated with local anesthetic agents (injectable procaine, topical xylocaine). A recording well was centered over the inferior colliculus and secured with dental cement to bolts implanted in the skull. Bipolar hooked-wire electrodes were implanted in muscles representative of flexors and extensors in forelimbs and hind limbs. The electromyograms (EMGs) were recorded on FM tape. The activity of the MLR neurons was recorded extracellularly using glass micropipets (1.6 M potassium citrate in a saturated solution of fast green dye, 20 to 40 Ma resistance) introduced through the recording well. The locus of selected neurons was marked by deposition of fast green dye at the recording site. After the recordings the animals were killed by an overdose of barbiturate, perfused via the previously ligated carotids with 10% phosphatebuffered Formalin, and the loci of dye spots and extent of transection verified in frozen 40-pm sections stained with cresyl violet. The taped activity of MLR neurons was correlated with concurrent EMG activity by using spike-triggered averaging. Each channel of EMG was “rectified” by digitizing, squaring all points, and normalizing. Each channel then was averaged using the MLR action potential to (spike) trigger each trial. Typically, 20 ms of EMG activity (in OS-ms bins) that followed each trigger was averaged for 100 to 2000 trials (spikes). These procedures were similar to those previously used in the analysis of neuronal activity in relation to quadripedal movements in cats (31). Each channel of EMG was averaged for the same trials (21%). The mean of each average was calculated as the average of activity occurring in the first 4 ms after each spike. The criteria for accepting a correlation consisted of a change in activity in excess of two standard deviations away from the mean and of at least three consecutive bins (1.5 ms) in duration. Because the basal EMG also was “rectified” and summed, negative (below the mean) correlations were accepted. The square of the EMG was used in the average as it is related to the power in the signal. In this manner, significant EMG events become evident over background
612
GARCIA-RILL,
activity with a smaller number unsquared signal.
SKINNER,
AND
FITZGERALD
of trials than does the standard rectified,
RESULTS Recordings were carried out only in preparations which showed s~n~neous locomotion (N = 15). Less than half of all transected animals walked spontaneously, and then for periods lasting 1 to 8 h. Although treadmill speed remained constant, the locomotion exhibited in these preparations varied considerably. Generally, locomotion started in one limb, was followed by the contralateral limb after a few cycles and then was joined by the other pair of limbs for periods lasting 1 to 30 min. Locomotion stopped spontaneously and was marked by moments of unilateral or bilateral rigidity, then relaxation, and ultimately continued cycling. These preparations allowed the correlation of single-cell activity with many aspects of muscular activity, rhythmic and nonrhythmic. A total of 96 neurons was recorded in the MLR for periods of 5 to 45 min during spontaneous locomotion. The presence of an A-B break on the rising phase of the action potentials of each cell was considered adequate evidence that the activity of fibers of passage was not being recorded. Each neuron studied was tested for (a) auditory input to determine ~rni~~le~~e sampling in the inferior colliculus was taking place, and (b) responses to jaw opening to determine if the recording electrode was positioned in the mesencephalic trigeminal nucleus. Several “tracks” through the central gray nucleus failed to sample any neurons in that nucleus showing rhythmic firing patterns. Recordings of all central gray neurons were, therefore, discarded (as were recordings of inferior colliculus and trigeminal neurons) following localization of dye spots in that region. The remainder of the neurons studied (N = 96) was localized within the CFN and PPN. The histo~o~~lly verified positions of these cells corresponded with locomotion-inducing stimulation sites in previous experiments (6,8-lo), which also pointed to a heterogeneous population of cells at the level of the MLR. Neurons recorded in the CFN and PPN which showed rhythmic patterns were found interspersed with other neurons showing no correlation with locomotor movements. This mixed population was assumed to be part of the MLR, although we can provide no direct evidence to suggest that nonrh~hmi~lly active neurons within the CFN and PPN were an integral part of the MLR. Attempts to further characterize these neurons in terms of their descending projections proved unsuccessful in this type of preparation. High-frequency (300 to 500 Hz) stimulation of the NRG or Probst’s tract at currents necessary to induce antidromic responses (two to three times the threshold for inducing locomotion) in MLR neurons produced postures and movements of sufficient violence to displace
UNIT
RECORDING
OF MLR DURING
613
L~OMOTION
the recording electrode. Collision testing proved unwieldy because antidromic stimulation blocked spontaneous locomotion as well. The MLR neurons (according to the above criteria) were tested unsuccessfully for responses to phasic somatosensory stimulation of the limbs and proximal areas. Passive movement of the limbs during quiescent episodes also failed to elicit responses in rh~hmically active neurons. Passive manipulation during spontaneous locomotion did not produce a significant change in MLR neuronal activity, but did induce active limb withdrawal or resistance, which also was occasionally sufficient to displace the microelectrode. Of the neurons localized in the CFN and PPN, more than 50% showed rhythmic activity. This activity changed with the frequency of the step cycle, and was generally composed of bursting firing patterns which increased in frequency toward the end of the burst. Bursts were 0.2 to 1.0 s in duration and were interrupted by quiescent periods of 0.2 to 0.8 s. However, when spike-triggered averaging was used to test for a correlation between unit activity and EMG activity, only 23% of the MLR neurons showed rhythmic activity in relation to at least one alternating EMG pattern. Those rhythmically active neurons not showing a positive correlation may have been contributing to overall rhythmic patterning by the MLR or were related to the activity of muscles which did not have implanted EMG leads. Of those cells which were found to be correlated with EMG activity (N = 22). 32% were related to EMG activity in muscles of one limb (either ipsi- or contralateral). 45% were related to EMG activity in muscles of two limbs and 23% were related to EMG activity in muscles of all four limbs. Table 1 shows a breakdown of the types of activity, rhythmic and nonrhythmic, found in this population of MLR neurons. Cells denoted as “related” were those found to be correlated by spike-triggered averaging to at least one TABLE
1
Mesencephaiic Locomotor Region Units Localized by Fast Green Deposition in Spontaneously Locomoting Brain Stem-Transected Cats
Total number studied Not related to locomotion Rh~hmica~iy active Rh~hmicafly active in relation to alternating EMGs Of the last group: Related to EMGs of one limb Related to EMGs of two limbs Related to EMGs of all limbs
No. cells
Percentage
96 40 34
100 42 35
22
23
7 10 5
32 45 23
614
GARCIA-RILL,
SKINNER,
AND
FITZGERALD
EMG. Spike-triggered averaging showed positive correlations with one particular muscle (usually a flexor) in some cells. In other neurons a negative correlation with the antagonist muscle also was observed, and, in those related to more than one limb, a positive correlation with concurrently contracting muscles in other limbs also was evident. Figure 1, lower left, shows the location of a fast green dye spot deposited at the site (boundary of the CFN and PPN or brachium conjunctivum) where the activity of an MLR neuron was recorded. This figure also shows the rhythmic firing pattern (long burst increasing in frequency) of this neuron together with concurrent EMG records of ipsilateral and contralateral forelimb flexors during spontaneous locomotion. The activity in this neuron appears to precede the activity of the contralateral forelimb flexor. This correlation is better observed in a subsequent segment of activity of the same neuron (Fig. 2). During this episode, stepping frequency increased and the activity of the neuron was diminished in concert with that of the EMG of the con-
MLR UNIT ACTIVITY
I PSI ‘FORELIMB FLEXOR
1
EMG CONTRA
1 s
rRECORDlNG SITE
FIG. I. Activity of a single mesencephahc locomotor region (MLR) neuron (top record) during spontaneous locomotion in a precollicular-premamillary transected cat. The location of this recording is shown in the drawing at bottom left. The activity of EMGs of ipsilateral (IPSI) and contralateral (CONTRA) forelimb flexors is shown in the middle and bottom records, respectively. BC-brachium conjunctivum, CFN-cuneiform nucleus, CG-central gray, K-inferior colliculus, P2-stereotactic plane.
UNIT RECORDING
OF MLR DURING
615
L~OMOTION
tralateral flexor. The EMG activity of the ipsilateral flexor appeared undiminished and was comparable in size to that of the previous segment (Fig. I). The possible correlations of the activity of the neuron shown in Figs. 1 and 2 were tested using spike-triggered averaging. The results of this analysis are shown in Fig. 3. A si~ni~cant positive correlation with the EMG activity in the contralateraf forelimb flexor was clearly evident at a 7-ms latency. In addition, a negative correlation with the EMG of the ipsilateral forelimb flexor was present at a 12.5ms latency. This neuron was typical of those showing activity correlated with EMGs in more than one limb. Figure 4 shows the spike-triggered averages of a different MLR neuron (located in the PPN). This neuron showed a positive correlation with the ipsilateral forelimb flexor at a 7.5ms latency, together with a negative correlation with the antagonist, the ipsilateral forelimb extensor at a 55ms latency. Although the contralateral extensor is somewhat concurrently active when the ipsilateral flexor is contracting, no statistical correlation with that muscle was found. This MLR neuron was representative of those showing a correlation with the activity of muscles in only one limb. The mean latency from the trigger (spike) to the beginning of a significant correlation with forelimb EMG activity for the entire population of MLR neurons was 7.1 + 3.2 ms (standard error of the mean, range 5.5 to 16 ms). The mean latency for correlated hind limb EMG activity was 6 to 7 ms
MLR UNIT ACTIVITY
IPSI FORELIMB FLEXOR
EMG CONTRA
I
s
FIG. 2. Activity of the same MLR neuron (top record) as in Fig. 1. during a different episode of spontaneous locomotion. The EMG activity of ipsilateral and contralateral forelimb flexors is shown in the middle and bottom records, respectively. The activity of this cell was used to calculate the spike-triggered averages shown in Fig. 3.
616
GARCIA-RILL,
SKINNER,
AND FITZGERALD
t’i,;“h-:; 1
MLR
;:-r
o.
UNIT
CF b
0.00 sroo--
10.00
15.00
.
MS
myj
CE b
0. 00
5. 00
20.00 MS
.
‘F b
FIG. 3. Left side-a dorsal view of the midbrain, medulla (MED) and cervical enlargement (CE). Right side-spike-triggered averages of EMGs of muscles (MUS) in the contralateral (CFcontralateral flexor, CE-contralateral extensor) and ipsilateral (IF-ipsilateral flexor, IE-ipsilateral extensor) forelimbs during spontaneous locomotion. Each EMG was averaged for the 20-ms period following the occurrence of each of 1000 spikes (occurring at point 0.00) of an MLR neuron (sfar). The mean of each average is the mean of the activity during the first 4 ms and is indicated by the solid line. The dotted line indicates the activity at two standard deviations above the mean. Filled bins, each 0.5 ms in width, indicate activity in excess of two standard deviations above or below the mean. Periods of correlation having durations of three or more bins (1.5 ms) were present in the averages of CF (increase at a 7-ms latency) and IF (decrease at a 12.5-ms latency). SC-superior colliculus, TRANS-precollicular-premamillary brain stem transection.
longer. Figure 5 shows the histologically verified loci of all MLR neurons which were rhythmically active and which exhibited a significant correlation with at least one EMG. The distribution was limited to the CFN and PPN. The MLR neurons which did not show a correlation with EMG activity (not shown) were located among those shown in Fig. 5, often sampled in the same microelectrode “track.” DISCUSSION Our results demonstrate that (i) a significant proportion of MLR neurons are rhythmically active during locomotion, (ii) the activity of some of those
UNIT RECORDING
OF MLR DURING
L~OMOTION
617
neurons can be shown to be correlated with cyclic EMG activity in particular limb muscles by using spike-triggered averaging, and (iii) MLR neurons correlated with locomotor movements are situated in the CFN and PPN. Until the advent of a series of studies outlining the afferent and efferent connections of the MLR (9), it was not known if stimulation-induced locomotion at that site was due to activation of fibers of passage or of neuronal somata (12). Those reports determined that the motor cortex, entopeduncular nucleus, and substantia nigra projections were directed at the same site which, when stimulated, elicited locomotion on a treadmill (6, 8). Intracellular recordings in proximity to a locomotion-inducing site revealed the presence of functionally varied synaptic effects after activation of pallidal afferent fibers to the MLR ( 10). Anterograde and retrograde transport away from the physiologically identified MLR also revealed its input-output organization (9). More recently, we reported that controlled locomotion can be induced by localized application of pharmacologic agents in the MLR (11). The presence of dendritic or somatic receptors is necessary for the expression of those effects, again indicative of neuronal, not axonal, substrates being responsible for MLR-induced locomotion. Still lacking, however, was any evidence demonstrating that MLR neuronal activity was in any way related to locomotion. The findings described herein provide convincing evidence that a significant propo~ion of MLR neurons are active rh~hmically during s~ntaneous locomotion. In some of these cells, a significant correlation between the rhythmic activity and alternating EMG patterns can be established. These results indicate that the MLR may be providing not only a rhythmic background of activity for downstream centers, but may also be “driving” some of those centers. The MLR is necessary for eliciting stimulation-induced locomotion in a cat with a precollicular-postmamillary transection. Because spontaneous locomotion can occur after a precollicular-premamilla~ transection, the implication is that the MLR or parallel systems are “released” by the more anterior transection (12, 18,25). The fact that MLR unit activity is still rhythmic and related to EMG activity in this preparation speaks to the importance of this region in the rhythmic modulation of locomotion. How does the MLR influence downstream centers? Our previous results indicated the presence of two main descending projections of the MLR (9). First, efferent fibers in the region of Probst’s tract in the dorsolateral medullaxy reticular fo~ation were noted. This projection coincides with the location of the PLS, which other investigators stimulated to induce locomotion (28). This tract appears to be in or near Probst’s tract, which is part of the trigeminospinal system (17). That this pathway may be partially involved in modulating locomotion is supported by studies demonstrating locomotion on a treadmill in the precollicular-postmamillary cat after stimulation of the pinna f 1).
618
GARCIA-RILL,
SKINNER,
AND FITZGERALD
I 0. 00
LJ 5.00 ‘A
20.00 MS'
0: 00
5: 00
'
lb. 00
Ii.00
zb. 00
15.00
20.00
MS
0.00
. -0.00 5’00 MS
'
m*j 0. 00
5.00
10.00
MS
FIG. 4. Spike-triggered averages of EMGs of forelimb flexors and extensors that follow the activity of another MLR neuron. Significant correlations were present in the IF (increase at a 7.5-ms latency) and its antagonist IE (decrease at a 5.5-ms latency). Abbreviations as in Fig. 3.
There is a second downstream projection of the MLR, that to the NRG (9), a source of reticulospinal projections (2). Stimulation of the MLR was reported to induce monosynaptic responses at a l- to 1S-ms latency in reticulospinal neurons (24). We subsequently demonstrated the presence of this pathway anatomically (9). This projection may well be the primary element through which the MLR modulates downstream centers. Lesion of the ventrolateral quadrant of the white matter of the spinal cord is known to abolish MLR-induced locomotion (33). This finding implies that the reticulospinal (and perhaps vestibulospinal) projections are essential for the manifestation of MLR effects. Recent expe~ments in which the activity of reticulospinal neurons was recorded during spontaneous locomotion in the “thalamic” cat, showed these neurons to be rhythmically active and correlated with EMG activity in limb muscles (30, 31). The use of spike-triggered averaging in those experiments revealed a 5- to 6-ms latency for averaged EMGs in forelimb muscles following reticulospinal neuron activity (3 1). Allowing for the reported l- to IS-ms delay from the MLR to the NRG (24), the mean latency reported herein of 7.1 ms for MLR activity preceding EMG
UNIT RECORDING
PlO
OF MLR DURING
Pi5
619
L~OMOTION
P20
FIG. 5. Locations of MLR neurons which showed rhythmic activity in relation to locomotor movements at various stereotactic planes. The focalized fast green dye deposits were limited to the CFN and BC, in which the PPK is ern~dd~.
activity in forelimb muscles agrees well with those reports. Our data suggest that the correlation between MLR unit activity and EMGs during locomotion is mediated via the reticulospinal system. Not only is the latency appropriate, but the duration of the correlation is similar for MLR spike-triggered averages (this report) compared with reticulospinal spike-triggered averages (3 1). Previous studies of reticulospinal neurons during s~ntaneous locomotion reported the presence of clear responses to phasic somatosenso~ stimulation in these neurons (30, 31). In addition, passive manipulation of the limbs leads to a reduction in rhythmicity in reticulospinal neurons (30, 3 1). These responses are thought to be mediated by the spinobulbospinal pathway (29). Our experiments reveal no such responses in MLR neurons. The lack of afferent modulation in MLR activity indicates this region may be a central rhythmic modulator the output of which comes under the influence of phasic aRerent input only at lower levels (perhaps reticulospinal and/or spinal). This afferent input is not absolutely essential for the MLR to induce locomotion because it is still possible to induce rhythmic activity in motoneurons after stimulation of the MLR in the “fictive locomotion” (paralyzed) preparation (13). Our findings also demonstrated that neurons which are active rhythmically in relation to locomotor movements are situated in the CFN and PPN. That the CFN is part of the MLR was suggested in the original description of brain stem-induced locomotion (27). On the other hand, only recently has the PPN been associated with the MLR (6-10). The PPN is embedded in the brachium conjunctivum and Long has been known to receive basal ganglia (14, 21) and cortical (15) afferent fibers. However, only the most posterior part of the PPN appears to be within the MLR, because stimulation of the anterior PPN does not induce locomotion after an appropriate brain stem transection (6-10). The PPN as a whole was reported to have reciprocal
620
GARCIA-RILL,
SKINNER,
AND FITZGERALD
connections with the pallidum (5, 16, 2 1, 32), the substantia nigra (5, 6, 8, 16), and the subthalamic nucleus (4, 5, 22). As such, it has been included as an integral part of “extrapyramidal circuitry” (26). It is then logical to suggest that the link between the basal ganglia and the MLR, which includes at least a portion of the PPN, participates in processes related to the functions of both regions. Because the basal ganglia are involved in the preparation for movement (3), and the MLR is involved in the modulation of locomotion, one suggestion is that this link may participate in the preparation for movements which include locomotion. Under this model, if the link between the basal ganglia and the MLR were absent, initiation of locomotion would be impaired. This type of deficit is present in Parkinson’s disease. Moreover, the observed inability of Parkinson patients to stop walking when started may be due to a “locking” of the locomotion oscillators to the sensory feedback generated by leg swing. A similar observation was reported in animals with caudate nucleus ablation (34). The peripheral input to spinal locomotion oscillators is, under normal circumstances, modulated by descending central projections (12, 18, 23). Deficits in starting and stopping locomotor movements could become manifest if these central connections were compromised. Much additional evidence is necessary to substantiate this rudimentary functional model, and the foregoing discussion is intended to provide a general basis for further investigation. REFERENCES I. AOKI, M., AND S. MORI. 1981. Locomotion elicited by pinna stimulation in the acute precollicular-postmamillary decerebrate cat. Bruin Rex 214: 424-428. 2. BRODAL, A. 1956. Anatomical aspects of the reticular formation of the pons and medulla oblongata. Prog. Neurobiol. 2: 240-255. 3. BUCHWALD, N. A., C. D. HULL, M. S. LEVINE, AND J. VILLABLANCA. 1975. The basal ganglia and the regulation of response and cognitive sets. Pages 171-189 in M. A. B. BRAZIER, Ed.. Growth and Development of the Brain. Raven Press, N.Y. 4. GROSSMAN, A. R., AND A. JACKSON. 1981. The efferent projections of the subthalamic nucleus (of Luys) in the rat, with special reference to a previously undescribed projection. J. Physiol. (London) 319: 108P. 5. EDLEY, S. M., AND A. M. GRAYBIEL. 1980. Connections of the nucleus tegmenti pedunculopontinus, pars compacta (TPc) in cat. Amt. Rec. 196: 129A. 6. GARCIA-RILL, R. D. SKINNER, AND S. GILMORE. 1981. Pallidal projections to the mesencephalic locomotor region (MLR) in the cat. Am. J. Amt. 161: 31 l-321. 7. GARCIA-RILL, W., R. D. SKINNER. AND J. A. FITZGERALD. 1982. Mesencephalic locomotor region (MLR) unit activity during locomotion. Sot. Neurosci. Abstr. 8: 167. 8. GARCIA-RILL, E., R. D. SKINNER, M. B. JACKSON, AND M. M. SMITH. 1983. Connections of the mesencephalic locomotor region (MLR). I. Substantia nigra afferents. Bruin Res. Bull. 10: 57-62. 9. GARCIA-RILL, E., R. D. SKINNER, S. A. GILMORE, AND R. OWINGS. 1983. Connections of
UNIT RECORDING
OF MLR DURING
LOCOMOTION
621
the mesencephalic locomotor region (MLR). II. Afferents and efferents. Bruin Rex Bull. 10: 63-71.
10. GARCIA-RILL, E. 1983. Connections of the mesencephalic locomotor region (MLR). III. Intracellular recordings. Brain Res. Bull. 10: 73-8 I. 11. GARCIA-RILL, E., R. D. SKINNER. AND J. A. FITZGERALD. 1983. Locomotion induced by chemical activation of the mesencephalic locomotor region (MLR). Sot. Neurosci. .4bstr. 9: 357.
12. GRILLNER, S.. AND M. L. SHIK. 1973. On the descending control of the lumbosacral spinal cord from the “mesencephalic locomotor region.” Acta Physiol. Scund. 87: 320-333. 13. JORDAN, L. M., C. A PRATT, AND J. E. MENZIES. 1979. Locomotion evoked by brain stem stimulation: occurrence without phasic segmental afferent input. Bruin Rex 177: 204207.
14. Kuo, J. S.. AND M. B. CARPENTER. 1973. Organization of pallido-thalamic projections in the rhesus monkey. J. Comp. Neural. 151: 201-236, 1973. 15. KUYPERS, H. G. J. M., AND D. G. LAWRENCE. 1967. Cortical projections to the red nucleus and the brain stem in the rhesus monkey. Brain Rex 4: 15 l-188. 16. LARSEN, K. D.. AND J. SUTIN. 1978. Output organization of the feline entopeduncular and subthalamic nuclei. Bruin Res. 157: 21-3 1. 17. MATSUSHITA, M., N. OKADO, M. IKEDA, AND Y. HOSOYA. 1981. Descending projections from the spinal and mesencephalic nuclei of the trigeminal nerve to the spinal cord in the cat. A study with the horseradish peroxidase technique. J. Comp. Neural. 196: l73187. 18. MORI, S., M. L. SHIK, AND A. S. YACODNI~-SYN. 1977. Role of pontine tegmentum for locomotor control in mesencephalic cat. J. Neurophysiol. 40: 284-295. 19. MORI, S., H. NISHIMLJRA, C. KURAKAMI, T. YAMAMURA, AND M. AOKI. 1978. Controlled locomotion in the mescncephalic cat: distribution of facilitory and inhibitory regions within pontine tegmentum. J. Neurophysiol. 41: I580- 159 I. 20. MORI, S., K. KAWAHARA, T. SAKAMOTO, M. AOKI, AND T. TOMIYAMA. 1982. Setting and resetting of level of postural muscle tone in decebrate cat by stimulation of brain stem. J. Neurophysiol.
48: 737-748.
21. NAUTA, W. J. H.. AND W. R. MEHLER. 1966. Projections of the lentiform nucleus in the monkey. Bruin Res. 1: 3-42. 22. NAUTA, H. J. W.. AND M. COLE. 1978. Efferent projections of the subthalamic nucleus: an autoradiographic study in monkey and cat. J. Comp. Neural. 180: I-16. 23. ORLOVSKY,G. N. 1969. Spontaneous and induced locomotion of the thalamic cat. Biophysics 14: 1154-l 162. 24. ORLOVSKY, G. N. 1970. Connections between reticulospinal neurons and locomotor regions of the brainstem. Biophysics 15: 58-64. 25. ORLOVSKY, G. N. 1972. The effect of different descending systems on flexor and extensor activity during locomotion. Brain Res. 40: 359-37 1. 26. SAPER, C. B., AND A. D. LOEWU. 1982. Projections of the pedunculopontine tegmental nucleus in the rat: evidence for additional extrapyramidal circuitry. Brain Rex 252: 367372.
SHIK, M. L., F. V. SEVERIN, AND G. N. ORLOVSKY. 1966. Control of walking and running by means of electrical stimulation of the mid-brain. Biophysics 11: 756-765. 28. SHIK, M. L., AND A. S. YAGODNITSYN. 1977. The pontobulbar locomotor strip. Neuro27.
physiology
9: 72-74.
29. SHIMAMURA, M., I. KOGURE, AND S. WADA. 1980. Three types of reticular neurons involved in the spino-bulbo-spinal reflex of cats. Bruin Res. 186: 99-l 13.
622
GARCIA-RILL,
SKINNER,
AND
FITZGERALD
30. SHIMAMURA, M., I. KOGURE, AND S. WADA. 1982. Reticular neuron activities associated with locomotion in thalamic cats. Brain Rex 231: 51-62. 3 I. SHIMAMURA, M., AND I. KOCURE. 1983. Discharge patterns of reticulospinal neurons corresponding with quadrupedal leg movement in thalamic cats. Bruin Res. 230: 27-34. 32. SKINNER, R. D., E. GARCIA-RILL, AND S. GILMORE. 1979. Entopeduncular nucleus projections to the mesencephalic locomotor region. Sot. Neurosci. Abstr. 5: 79. 33. STEEVES, J. D., AND L. M. JORDAN. 1980. Localization of a descending pathway in the spinal cord which is necessary for controlled treadmill locomotion. Neurosci. Left. 20: 283-288. 34. VILLABLANCA, J., R. J. MARCUS, AND C. E. OLMSTEAD. 1976. Effects of caudate nuclei or frontal cortical ablation in cats. 1. Neurology and gross behavior. Exp. Neural. 52: 389420.