3 14-325 (1984)
of Cerebellar Tremor on Proprioceptive but Not Visual Feedback D. FLAMENT,
AND J. HORE
Department of Physiology, University of Western Ontario, London, Ontario N6A 5C1, Canada Received August 26, 1983; revision received December 14, 1983 We studied the influence of proprioceptive and visual feedback on cerebellar tremor which occurred after arm perturbations and aher voluntary elbow flexions. Cerebellar tremor was produced in monkeys by reversibly cooling through two probes implanted lateral and medial to the dentate nucleus. Cerebellar tremor was synchronized in different trials to torque pulse onset and to the end, but not the start, of voluntary movements. Addition of loads to the handle held by the monkey (increases in spring stiffness, viscosity, constant torque, and inertial load) changed the amplitude and frequency of tremor that follows arm perturbations or voluntary movements in the same way. In both situations EMG activity in each cycle of tremor followed stretch of its own muscle and attained a peak near peak velocity irrespective of the mechanical load. Removal of visual feedback did not alter the characteristics of the tremor or the associated EMG activity. We concluded that cerehellar intention tremor, which occurs when attempting to hold the arm in an intended position, is driven by stretch-evoked peripheral feedback and not by voluntary corrections based on vision.
INTRODUCTION Holmes ( 10-12) described two situations in which cerebellar tremor was prominent. This tremor occurred during attempts to maintain a limb at a specified position against gravity and mechanical forces and during movements, especially toward their cessation. In the first situation he suggested that tremor was due to voluntary corrections of defective postural fixation, and in the second, to a failure of uniform deceleration complicated by secondary or correcting jerks. Some studies of cerebellar tremor in monkeys with lesions of the deep cerebellar nuclei concluded that the tremor following movements results from a voluntary attempt to correct for an initial error in goal-directed movements (7, 8). One possibility is that the voluntary corrections are mediated by visual feedback (14). 314 0014-4886184 $3.00 Copyright Q 1984 by Academic Press, Inc. All [email protected]
of reproduction in any form reserved.
An alternative view is that cerebellar tremor results from an abnormality in proprioceptive feedback from the limbs to the cerebral cortex (7, 16, 17). Evidence for this view is that mechanical changes in the limb changed the amplitude and frequency of the tremor (5,9,21). Although in theory tremor could result from an increase in gain in long-loop reflexes (20), no increase was found in the magnitude of the long-latency EMG responses to limb perturbations in monkeys that showed a marked cerebellar tremor (2 1). Instead there was loss of a predictive signal [intended component of Evarts and Tanji (6)] in perturbation-evoked discharge of motor cortex neurons during cerebellar dysfunction (22). Tremor was thought to result because motor cortex discharge followed muscle stretch instead of leading it. In view of the opinion of Holmes (11) that all tremors occuning in cerebellar disease are not of the same nature, the first aim of our study was to determine whether or not cerebellar tremor evoked by limb perturbations and that following limb movements had the same characteristics. Roth these situations required that the arm be held in a tixed position. The second aim was to determine to what extent cerebellar tremor was dependent on proprioceptive and visual feedback. The results showed that the characteristics of tremor that follow perturbations and that follow movements are the same. In both cases tremor was altered equally by changes in the load on the handle held by the monkey but was not affected by the exclusion of visual feedback of limb position.
Five Cebus monkeys were trained to hold a handle, pivoted at the elbow, within a target zone (width 12 deg). The monkeys were rewarded with grape drink either for returning the handle to target within 200 ms after a torque pulse perturbation, or for making flexions and extensions by moving the handle between two target positions separated by 40 deg of arc. In the latter step-tracking task, the monkeys were rewarded if the handle arrived in the new target within 0.7 s of the target jump, and remained within target for 0.4 s. Thus the monkeys had to make prompt and accurate movements to gain reward. Target and handle position were displayed on an oscilloscope in front of the monkey. An opaque plate prevented the monkeys from seeing their arms. In some experiments the handle position display (cursor) was randomly removed at the time of the torque pulse onset or target jump thereby eliminating visual feedback of limb position. The handle held by the monkey constituted primarily an inertial load. This load was altered by adding constant torque through the torque motor attached to the handle, spring stiffness by feedback of the position signal to the motor, viscosity by
feedback of the velocity signal, and increased inertia by addition of mass to the handle. The constant torques, which were applied in either the flexion or extension directions, varied from 10 to 20 g in magnitude as measured at the end of the manipulandum, 11 cm from its center of rotation. All perturbations were produced by torque pulses of 40 ms duration that initially stretched the biceps. Reversible Lesions. Two cryoprobe sheaths (1.3 mm in diameter) were implanted stereotaxically under pentobarbital anesthesia (35 mg/kg, i.p.) one iateral to the dentate (coordinates P 6.5, L 7.0, V -5.0) and the other through the region of the interpositus (P 8.0, L 3.0, V -4.0) ipsilateral to the forearm tested. Lateral and medial sheaths were inserted at caudorostral angles of 40 and 20 deg, respectively, to the vertical and were pushed past the targets by 2 mm. Temperature was measured by a thermocouple attached to the outside of the sheaths 4 mm from their tips. Histologic confirmation of sheath positions was obtained after killing the animals by an overdose of pentobarbital and perfusion with saline and 10% Formalin. Figure 1 shows the loci of the sheaths for three monkeys whose results are illustrated in the figures. The dashed lines are estimates of the 20°C isotherms (3), the temperature at which synaptic transmission is blocked (2), when sheath temperature was 10°C. The isotherms indicated that when the lateral sheath was cooled to lO”C, the lateral portion of the dentate nucleus was affected in all three monkeys. Cooling the medial sheath to 10°C affected the medial portion of the dentate and a major part of the interpositus. In addition, cooling through the medial sheath may have affected transmission in the fastigial nucleus. In all experiments cooling was through both the lateral and medial sheaths. To confirm that the effects of cooling described under Results were due to a reversible lesion of the nuclei and not of the overlying cerebellar tissue, tests were conducted in which the cooling probe was withdrawn approximately 5 mm within the implanted sheath while a constant cooling rate was maintained. During this procedure the temperature of the probe sheath at the approximate locus of the nuclei returned to 33°C and the EMG responses and movement parameters were indistinguishable from those observed under control conditions. Data Acquisition and Analysis. Two parameters were recorded: (i) handle position by means of a thin-film potentiometer coupled to the handle and (ii) intramuscular EMG activity of the biceps and triceps by a pair of stainlesssteel, Teflon-coated wires inserted into the muscle during each experiment. The EMG was amplified, filtered (bandwidth 50 to 1000 Hz), and full wave rectified. The data were recorded on-line at a sampling rate of 1000 Hz and block-averaged into 5-ms bins by a PDP-I l/40 computer. Velocity was obtained by digital differentiation of position. The data were stored in digital form on magnetic tape for later retrieval and off-line analysis.
FIG. 1. Positions of medial and lateral cooling probe sheaths in BE, JO, and BZ. Frontal sections are shown for each monkey at 20,40, 60, and 80% of the rostral-caudal extent of the dentate nucleus. Dotted lines are estimated isotherms for tissue temperature of 20°C when sheath reference temperature was 1O’C. Midline is at 0 mm. D-dentate nucleus (cross hatching), IP-interpositus nucleus (vertical hatching), F-fast&&l nucleus (horizontal hatching): LVlateral vestibular nucleus (unshaded), P-sheath position (black). Histology from monkey BE was shown elsewhere as M28L (22).
RESULTS Implantation of the cooling sheaths did not produce any observable disorder in limb performance as the monkeys moved in their cages or reached for food. In contrast, during cooling of both probes to lO”C, the monkeys showed the classic signs of cerebellar ataxia, including terminal tremor, when reaching for food and when attempting to place the food in their mouths. To study tremor quantitatively the monkeys were trained to hold a handle with the elbow resting on a pivot. This restricted the tremor to movements at the elbow. Initially study was made of EMG responses during tremor atier limb
VILIS, AND HORE
perturbations. In this situation tremor in different trials was synchronized to the onset of the perturbation. Characteristics of Tremor after Perturbations. Under normal conditions the monkeys responded to torque pulse perturbations that initially stretched the biceps (the agonist) by returning the limb to target without overshoot or oscillation when a constant force loaded either the triceps or biceps (Fig. 2A, IL-thin line). This return movement was associated with a burst of EMG in the biceps which peaked about 50 ms after perturbation onset, and was followed by phasic activity in the triceps (the antagonist). When a constant force was applied to the handle that loaded the triceps (Fig. 2A), the phasic activity in the triceps occurred at a latency of 70 to 100 ms which was before the start of stretch of that muscle (dashed line) on the return movement. This early antagonist activity was presumably generated to help stop the limb from overshooting the target when the constant force was in the direction that assisted overshoot. As reported elsewhere (2 1,22), cooling the cerebellar nuclei (Fig. 2-thick line) had only a small effect on the biceps response but jtbolished the early triceps response. During cooling the EMG response in the triceps now followed stretch of the triceps and started a series of oscillations (tremor). Comparison of Fig. 2A and B reveals that a clear difference in timing of the antagonist response between the normal and cooled situations occurred only when a constant force loaded the antagonist. To investigate whether or not during cerebellar dysfunction EMG responses of the antagonist continued to follow stretch of the antagonist for different A
FIG. 2. Effect of constant force on antagonist (triceps) response to a perturbation that initially stretched the biceps during normal conditions (thin line) and during cerebellar nuclear cooling (thick line). Each trace is the average of 15 responses. A-small constant force opposed the triceps, B-small constant force opposed the biceps. Arrows indicate onset and offset of 40-ms perturbation. Dashed line is drawn when velocity first attained zero (at start of triceps stretch). Pos-position, Vel-velocity, Bi-biceps, Tri-triceps. Calibrations: target, 12 deg; velocity, 200 deg/s. Monkey JO.
return velocities, torque pulses of different magnitudes were applied with a constant force opposing the triceps (Fii. 3). In this situation stretch of the triceps (when velocity hrst returned to zero) occurred at different times after torque pulse onset. Under normal conditions the antagonist (triceps) response occurred at a latency of 70 to 100 ms which was before stretch of the triceps for all three magnitudes (Fig. 3A). During cooling, triceps onset occurred at progressively longer latencies as the torque pulse increased in size. This was consistent with its onset being determined by the onset of stretch of the triceps. Whereas under normal conditions no relation existed between velocity and triceps EMG, during cooling the peak of the triceps EMG was correlated with peak velocity (triangles, Fig. 3A). The decrease in magnitude of the biceps response during cooling was not a consistent finding in this task in all monkeys. It has been ascribed to loss of set during cerebellar dysfunction (13). Eficts of Difirent Mechanical Conditions. A second situation in which time of onset of stretch of the antagonist muscle can be varied is by applying different mechanical loads to the limb. Figure 4 compares responses during cerebellar cooling for the normal manipulandum (Fig. 4A and B-thin line) and when the manipulandum acted as a linear spring (Fig. 4A-thick line) and with an inertial load added to the manipulandum (Fig. 4B-thick line). With the handle acting as a spring there was a more rapid decrease in velocity and therefore earlier onset of triceps stretch, whereas with increased inertia
FIG. 3. Effect of cooling cerebellar nuclei on antagonist (triceps) EMG response to perturbations of three different magnitudes of force. A-normal conditions, B-cooling the cerebellar nuclei. Each trace is the average of 15 responses to perhubations that initially stretched the biceps. Arrows indicate onset and offset of the 40-ms perturbation. Dashed line is when velocity 6rst attained zero (at start of triceps stretch). A constant force was applied that loaded the triceps. Triangles indicate peak of the return velocity. Pos-position, Vel-velocity. Calibrations: target, 12 deg velocity, 200 de&s. Monkey BZ.
VILIS, AND HOW
the triceps stretch was delayed. This figure illustrates that for all three mechanical conditions onset of the first phasic response in the triceps followed stretch of the triceps and its peak occurred near the peak of the return velocity during cerebellar dysfunction. Similarly the second burst in the biceps followed stretch of the biceps and its peak occurred near the next peak of velocity. These same relationships then occurred for each subsequent cycle of tremor. Characteristics of Tremor aJ2er Movements. A question that remained was whether cerebellar tremor, which occurred in the new intended position after movements, had the same characteristics as that following perturbations. To investigate this the characteristics of tremor in the two situations were studied under various mechanical conditions. As reported elsewhere (21), the frequency of tremor that followed perturbations was increased by constant forces (loads) and decreased by addition of mass (inertia), whereas the amplitude of the tremor was increased by constant forces and decreased by an increase in viscous resistance (Fig. 5-Pert). Exactly the same changes occurred in the tremor that followed movements (Fig. 5-Move). This indicates that the tremor in both situations was likely to be generated by the same mechanism. To compare the EMG characteristics of tremor after movements with those of tremor after perturbations, it was necessary to synchronize the tremor in the former situation on some parameter of movement to enable averaging of EMG activity. When position and velocity records were synchronized to the start of movement no clear synchronization of the tremor after different movements occurred. This is illustrated for two monkeys in Fig. 6A. However, when records were synchronized to the end of movement, defined as when
FIG. 4. Effect of a spring force and increased inertial load on EMG responsesto limb perturbations during ceretzellar nuclear cooling. Thin line, normal manipulandum (same in A and B). Thick line A-manipulandum acting as a spring; B-adding mass to manipulandum and increasing force of perturbation so that the same initial peak velocity was achieved. Arrows indicate onset and offset of 40-ms perturbation. Pos-position, Vel-velocity, Bi-biceps, Tri-triceps. Calibrations: target, 12 deg; velocity, 100 de&s. Monkey BZ.
hertio Large Load
FIG. 5. Cerehellar tremor that followed voluntary movements (Move) and perturbations (Pert) for four different mechanical load conditions. Single trials of manipulandum position. Monkey BE.
velocity f&t returned to zero, tremor after the different movements was in phase (Fig. 6B). To confirm that EMG activity in tremor after movements had the same characteristics as that after perturbations, tremor was studied afkr movements A
FIG. 6. Superimposed position and velocity records of individual flexion movements during cexebellar nuclear cooling synchronized to A-start of the movement, B-end of the movement. Arrow at 0 indicates synchronization point. Calibrations: target, 12 deg, velocity, 100 de&. Upper four series of traces, monkey BZ; lower four series of traces, monkey JO.
VILIS, AND HORE
made with the normal manipulandum and with added mass. Figure 7 illustrates that when traces were synchronized to the end of movement, EMG activity of each muscle in each cycle of tremor followed stretch of that muscle for both mechanical conditions. Lack of Znzruence of Vision. The influence of vision on cerebellar tremor was investigated by comparing the tremor when visual feedback of the handle position was present and when it was absent. In these experiments visual feedback was excluded in 25 to 50% of the trials by unexpectedly removing the handle cursor from the screen at the time of the perturbation or target jump. In both tremor after perturbations (Fig. 8A) and tremor after voluntary movements (Fig. 8B), the removal of visual feedback (lower record of each pair) had no effect on the amplitude or frequency of tremor or on the associated EMG activity. DISCUSSION Humans with lesions of the cerebellum have an intention tremor when holding the limb steady or when performing goal-directed movements (lo12). The same disorder is seen in monkeys with lesions of the deep cerebellar nuclei [e.g., (1, 4, 7, S)]. The finding in our study that the characteristics of cerebellar tremor that follows limb perturbations or voluntary movements changed in the same way when different loads were applied to the limb indicates that these tremors are likely to be the result of the same mechanism. What is the mechanism that generates cerebellar tremor? Earlier studies emphasized the concept that tremor was a result of voluntary corrections
FIG. 7. Effect of inertial load on EMG activity during cerebellar tremor following movements, Thin line, normal manipulandum; thick line, 100-g mass added to manipulandum. Averages of 15 movements. Movements synchronized to end of movement (when velocity equalled zero). Arrow at 0 indicates synchronization point. Pos-position, Vel-velocity, Bi-biceps, Tritriceps. Calibrations: target, 12 deg; velocity, 100 [email protected]
FIG. 8. Effect of removal of vision on cembellar intention tremor. A-averages of tremors after perturbations; B-averages of tremors a&r movements. All traces during cooling of the cerebellar nuclei. Upper trace of each pair is with handle cursor displayed (vision), lower trace of each pair is with handle cursor removed (no vision). Movement traces were synchronized to end of movement (when velocity tirst equalled zero). Averages of 22 trials (perturbation with vision), 18 trials &rturbation witb no vision), 29 trials (movement with vision), 17 trials (movement with no vision). Arrows indicate onset and o&et of 40-ms perturbation (A) and synchronization point (B). Pos-position, Vel-velocity, Bi-biceps, Tri-triceps. Calibrations: target, 12 deg; velocity, 100 deg/s. A-monkey MO, B-monkey JO.
for initial errors (8, 12). Recognition of error could be based on proprioceptive or visual feedback. The finding in our study that tremor was unaltered by exclusion of visual feedback appears to rule out this latter possibility. This result is consistent with the observation of Holmes (10) that movements were equally “ataxic” whether a patients eyes were open or closed. An alternative view is that cerebellar tremor results from an abnormality in proprioceptive feedback loops to the motor cortex (7, 16, 17). Evidence for tremor being driven by proprioceptive feedback from stretched muscles comes from the finding that tremor is affected by altering the mechanical load on the limb. For example, addition of inertia to the limb decreased the velocity of stretch of muscle in each cycle of tremor and decreased the frequency of the tremor (Figs. 4, 5, 7). Addition of elastic stiffness increased the velocity of muscle stretch and increased the tremor frequency (Fig. 4). If cerebellar tremor was driven entirely by a central oscillator, addition of these loads to the handle would not have at&ted the frequency of the tremor. In theory tremor could result from an increase in gain in supraspinal servo loops (20). This has been suggested to be one of the factors that contributes to postural tremor, when standing, in patients with late cerebellar atrophy (15). Similarly in the present study an increase in strength of stretch reflexes during cerebellar cooling could have contributed to their destabilizing action.
Although EMG responses at 30 and 50 ms after the initial perturbation were not increased in magnitude during cerebellar cooling, there was an increase in the duration of EMG activity in the stretched muscle (Fig. 2B). Another possibility is an increase in loop time in supraspinal loops (increase in phase lag) ( 19,2 1). Evidence for this has come from experiments involving limb perturbations in monkeys. Analysis of motor cortex discharge response to limb perturbations led Evarts and Tanji (6) to conclude that there were two components: a reflex component (latency 20 ms) that depended on the direction of the perturbation, and an intended component (latency about 50 ms) that depended on the direction in which the monkey wanted to move. Vilis and Hore (22) found that during cerebellar nuclear cooling the first component was largely unchanged but the second component was abolished. They suggested that in their paradigm, which involved the monkey resisting the perturbation without overshooting the target on the return movement, the second excitatory cortical component in antagonist related neurons provided a signal to the antagonist muscle so that it was activated prior to antagonist muscle stretch. This second (50 ms) cortical component could have been generated via a short-latency central pathway involving the cerebellum, perhaps via an efference copy mechanism [Fig. 7 in (22)]. Such a pathway would provide a mechanism for phase advance of antagonist EMG over antagonist muscle stretch. Loss of this phase advanced or predictive signal during cerebellar dysfunction would cause EMG responses to be delayed as they would now by driven only by stretch reflexes in part through motor cortex. The same argument then applies for each cycle of the subsequent tremor. Those previous results indicated that the fundamental cause of cerebellar tremor after limb perturbations was loss of a predictive mechanism that normally provided phase advance of EMG signals over muscle stretch [cf. (18)]. Loss of this mechanism during cerebellar dysfunction forces the motor system to rely on stretch reflexes that follow muscle stretch. This produces an increase in phase lag. Our results showed that cerebellar tremor after goaldirected movements had the same characteristics as that after limb perturbations. We therefore conclude that cerebellar tremor of the limb (i) occurs as a result of loss of a cerebellardependent predictive mechanism and (ii) is a series of alternating stretch reflexes in agonists and antagonists in part through motor cortex. REFERENCES 1. ARING, C. D., AND J. F. FULTON. 1936. Relation of the cerebrum to the cerebellum. Part 2. Cerebellar tremor in the monkey and its absence after removal of the principal excitable areas of the cerebral cortex (area 4 and 6a, upper part). Arch. Neural. Psychiatry 35: 439466.
2. BROOKS, V. B. 1983. Study of brain function by local, reversible cooling. Rev. Physiof. B&hem. Phurmucol. 951 l-109. 3. BROOKS, V. B., I. B. KO~~I~~KAYA, A. ATKIN, F. E. HORVATH, AND M. UNO. 1973. Effects of cooling dentate nucleus on tmcking-task performance in monkeys. J. Neurophysiol. 36: 974-995. 4. CARREA, R. M. E., ANDF. A. METTLER. 1947. Physiologic consequences following extensive movals of the cerebellar cortex and deep cembellar nuclei and &ect of secondary cerebral ablations in the primate. J. Comp. Neural. 87: 169-288. 5. CHASE, R. A., J. K. CULLEN, S. A. SULLIVAN, AND A. K. OMMAYA. 1965. Modification of intention tremor in man. Nature (London) 206: 485-487. 6. EVARTS, E. V., AND J. TANJI. 1976. Reflex and intended responses in motor cortex pyramidal tract neurons of monkey. J. Newophysiol. 39: 1069-1080. 7. GOLDBERGER, M. E., AND J. H. GROWDON. 1973. Pattern of recovery following cerebellar deep nuclear lesions in monkeys. Exp. Neurof. 39: 307-322. 8. GROWDON, J. H., W. W. CHAMBERS, AND C. N. Lru. 1967. An experimental study of cerebellar dyskinesia in the rhesus monkey. Bruin 90: 603-632. 9. HEWER, R. L., R. COOPER,AND M. H. MORGAN. 1972. An investigation into the value of treating intention tremor by weighting the affected arm. Brain 95: 579-590. 10. HOLMES, G. 19 17. The symptoms of acute cerebellar injuries due to gunshot injuries. Bruin 40: 461-535. 11. HOLMES, G. 1922. The Croonian Lectures on the clinical symptoms of cercbellar disease and their interpretation. Lancef 1Olk 1177-l 182. 12. HOLMES, G. 1939. The cerebellum of man. Brain 62: l-30. 13. HORE, J., AND T. VILIS. 1982. Contribution of the dentate nucleus to set.Abstr. Sot. Neurosci. 8: 445. 14. ITO, M. 1979. Is the cerebellum reaify a computer? Trends Neurosci. 2~ 122-126. 15. MAURITZ, K. H., C. SCHMITT, AND J. DICHGANS. 198 1. Delayed and enhanced long latency retlexes as the possible cause of postural tremor in late cerebellar atrophy. Bruin 104: 97116. 16. MEYER-LOHMANN, J., B. CONRAD, K. MATSUNAMI, AND V. B. BROOKS. 1975. Effects of dentate cooling on precentral unit activity following torque pulse injections into elbow movements. Bruin Res. 94: 237-251. 17. MURPHY, J. T., H. C. KWAN, W. A. MACKAY, ANDY. C. WONG. 1975. Physiological basis of cerebellar dysmetria. Can. J. Neural. Sci. 2: 279-284. 18. NEILSON, P. D., AND J. W. LANCE. 1978. Reflex transmission characteristica during voluntary activity in normal man and patients with movement disorders. Pages 263-299 in J. E. DESMEDT, Ed., Cerebral Motor Control in Man: Long Loop Mechanisms, Progress in Clinical Neurophysiology, Vol. 4. Karger, Basel. 19. STEIN, R. B., AND M. N. O&U~T~RELI. 1976. Tremor and other oscillations in neuromuscular systems.Biof. Cybern. 22: 147-157. 20. STEIN, R. B., AND M. N. OCUZTORELI. 1978. Reflex involvement in the generation and control of tremor and clonus. Pages 28-50 in J. E. DESMEDT, Ed., Physiological Tremor, Pathological Tremors and Clonus, Progress in Clinical Neurophysiology, Vol. 5. Larger, Basel. 21. VILIS, T., AND J. HORE. 1977. Effects of changes in mechanical state of limb on cerebellar intention tremor. J. Neurophysiol. 40: 12 14-1224. 22. VIUS, T., AND J. HORE. 1980. Central neural mechanisms contributing to cerebellar tremor produced by limb perturbations. J. Neurophysiol. 43: 279-291.