Brain Research, 120 (1977) 1-15

1

© Elsevier/North-Holland Biomedical Press, Amsterdam - Printed in The Netherlands

Research Reports

D Y N A M I C R E L A T I O N S B E T W E E N N A T U R A L VESTIBULAR I N P U T S A N D ACTIVITY O F F O R E L I M B E X T E N S O R MUSCLES IN T H E D E C E R E B R A T E CAT. I. M O T O R O U T P U T D U R I N G S I N U S O I D A L L I N E A R ACCELERATIONS

JOHN H. ANDERSON, JOHN F. SOECHTING and CARLO A. TERZUOLO Laboratory of Neurophysiology, University of Minnesota Medical School, Minneapolis, Minn. 55455 (u.s.A.)

(Accepted May 6th, 1976)

SUMMARY Decerebrate cats were subjected to sinusoidal linear accelerations along the animal's horizontal and vertical axes, while recording the E M G activity of both triceps brachii muscles. This activity was found to be sinusoidally modulated in response to the accelerations and thus phase and gain relations between motor output and input acceleration could be obtained. They were found to be the same for accelerations along each of the three axes. In particular the gain dropped by 14-20 dB over a frequency range from 0.2 to 1.0 Hz and the phase of the motor output showed a lag of 40-60 ° at 1.0 Hz. Thus, it was concluded that (1) the dynamic behavior ofutricular and saccular receptors is the same, (2) the changes in motor activity observed during accelerations along the vertical axis are mostly due to the activation of saccular afferents, and (3) the motor output cannot simply result from vestibular afferent activities being relayed directly to the spinal motoneurons via the vestibulo-spinal tracts.

INTRODUCTION Vestibular inputs play a large role in the control of posture; yet the dynamic characteristics of the relations between natural vestibular inputs and spinal motor outputs, i.e. the postural vestibular reflexes, are not entirely known. Investigations with human subjects zl-z3 and trained dogs 3s have demonstrated that under normal, unrestrained conditions it is difficult to isolate the vestibular component in the overall reflex motor behavior elicited by imposing accelerations, since other sensory inputs participate. Thus it is desirable, as a first step, to utilize a simpler preparation. To this end the decerebrate preparation can be used, where powerful vestibular reflexes are

known to be present in the forelimb extensor muscles, particularly after section of the spinal cord at low thoracic leveP 7. Moreover, it is already known that under these experimental conditions a frequency analysis can be used to quantitatively relate the motor output to a natural vestibular input 2. Experiments were therefore designed with the aim of defining the individual contribution to the motor output by canal and macular afferents in order to determine, and eventually to model, the behavior of the central integrative mechanisms whereby vestibular inputs are transformed into postural reflexes. In this paper we shall present the experimental data obtained by imposing linear accelerations along the cat's longitudinal, transverse and vertical axes, thus activating macular afferents 15. We shall then proceed in the subsequent two papers1, 34 to consider the motor output to angular accelerations as well as to combinations of linear and angular accelerations. In all cases deductions concerning the dynamic properties of the central integrative mechanisms are predicated on the assumption that the dynamic characteristics of the relations between natural vestibular inputs and the activity in the primary afferents from macular and canal receptors are essentially the same in the cat as in the squirrel monkey, where they have been extensively studied~,6,12. Thus it becomes possible, if certain properties are satisfied by the system, to isolate the dynamic characteristics of the central mechanism responsible for the postural vestibular reflexes in the decerebrate cat. METHODS Decerebrate cats were subjected to sinusoidal, linear accelerations while recording the electromyograms (EMG) of the extensor muscles, triceps brachii, of both forelimbs. Adult cats, weighing 2.5-3.5 kg, were first anesthetized with ketamine (Ketalar, initial dose was 15 mg/kg). After transecting the spinal cord at the Thlz level, to avoid afferent activities from the hindlimbs and to enhance the vestibular reflexes, a decerebration was performed by electrocoagulation in the anterior 3 plane of the Horsley-Clarke coordinates. Frequently, the decerebration was performed in the evening preceding the experiment in order to ensure a more stable preparation. Forelimb deafferentation, labyrinthectomy and bilateral destruction of the cerebellar nuclei were performed under sterile conditions. The labyrinth was destroyed (two days prior to decerebration) by means of a blunt probe introduced through the round window of the exposed bulla. Bilateral destruction of all cerebellar nuclei and surrounding white matter (16-24 days prior to decerebration) was accomplished by electrocoagulation using two electrodes placed stereotaxically in the appropriate horizontal plane by a posterior approach. Currents (10-15 mA) were applied (30 sec) at each of the several electrode positions needed to encompass the entire region to be destroyed. The destruction of the nuclei was indicated, in the course of the procedure, by the appearance of an intense extensor hypertonus in each forelimb and was later verified histologically. Forelimb deafferentation was performed by sectioning the dorsal roots of spinal segments C5-T1. The head, neck, and trunk of the animal were encased in a plaster cast to eliminate unwanted neck and body movements and to allow fixation of the animal to an

Z Fig. 1. Coordinate axes. The animal's axes along which linear accelerations were applied are labeled X (longitudinal), Y (transverse), and Z (vertical). Positive acceleration along the X-axis is taken to be in the cat's forward direction. For the Y-axis, the positive acceleration is in the direction to the side being considered (i.e. to the right for the right triceps and to the left for the left triceps). Positive acceleration along the Z-axis is taken to be in the downward direction. The two diagrams at the left also show the displacement of the cilia in the utricle and sacculus due to the inertia of the otoliths, for acceleration along the Y-axis (ay) and Z-axis (az), respectively. The corresponding shearing forces are labeled Sy and Sz). aluminum frame. Also, the distal portion of the forelimbs was cast and fixed rigidly to the frame. The head was immobilized with the roof of the mouth inclined about 35 ° to the horizontal plane 8. The frame was mounted to a platform which was connected by a rod and crank mechanism to the shaft o f a DC motor. This arrangement translated a constant angular velocity of the Shaft into a sinusoidally modulated rectilinear motion of the platform. However, due to the mechanical arrangement of the crank and connecting rod, there was a displacement distortion (mostly second harmonic), of about 0.05. The amplitude of the displacement could be varied from 4- 10 to 4- 27 cm, while the modulation frequency could be varied, by changing the angular velocity of the crank, from 0.1 to 1.0 Hz. For displacement of 4- 27 cm, which was used for most experiments, the acceleration amplitude was 4- 0.025 × g at 0.15 Hz and 4- 1.1 × g at 1.0 Hz, where g is the acceleration magnitude due to gravity. By varying the orientation of the frame with respect to the platform, the motion could be directed along the cat's X (longitudinal), Y (transverse), or Z (vertical) axis, as illustrated in Fig. 1. The Y-axis is colinear with a linear segment joining the two labyrinths, while the X- and Z-axes intersect the Y-axis at the midpoint between the two labyrinths. An accelerometer attached to the frame was used to obtain a description of the input. To measure the output flexible, teflon coated wires (diameter of 0.003 in.) inserted into the muscles were used to sample, simultaneously, the activity of motor units of the triceps brachii muscles in both forelimbs (Fig. 2). This activity was amplified and used to trigger a discriminator, the output of which was a frequency modulated pulse train. To demonstrate the adequacy of this technique, the following was done. The activity was sampled by more than one electrode. The results obtained were

4

•

~

I'

•

I

i • ,,i .,!.

•

I

i

I

I

I

i

~, I ~,

I

I

I

I

I

I

I

~1

I

I

I

Fig. 2. Example of EMG data. From top to bottom: EMG from the right triceps, input acceleration, EMG from the left triceps. The sinusoidai acceleration was applied along the Y-axis. Note that the activity is 180° out of phase in the two muscles. Time scale is 100 msec/div. the same, in terms of phase and gain slope, as those obtained when only one electrode was used. Moreover, in a few experiments the E M G was rectified and integrated. Again the results were not different. An IBM 1800 computer was used for data acquisition and processing. Briefly, the input cycle was divided into a preselected number (from 11 to 71) of equal time segments (bins). The number of pulses occurring within each bin was counted and divided by the total number of pulses, thus providing a probability density for pulse occurrence la. An arbitrary number of cycles could be sampled and averaged. A Fourier analysis was then performed on the averaged binned data (see Fig. 3) to obtain the phase, gain and harmonic distortion. The phase is the difference, in degrees, between the maxima of the output and input fundamentals. The gain in decibels is defined as: 20

log10

( output amplitude

where the amplitudes are those of the fundamentals in imp./sec (output) and m/sec 2 (input). Harmonic distortion is defined as:

t/'

N i=2

(amplitudei) 2 amplitude1

where the subscript denotes the harmonic, 1 being the fundamental.

No normalization procedure was used in averaging the gain data obtained either in a given preparation or from different preparations. This was done to include in the data all uncertainties due to different levels of rigidity and reflex excitability as well as the limitations in the sampling of EMG data. Moreover, in order to compare data obtained from the same preparation for accelerations applied along different axes, normalization cannot be used. RESULTS

It was found that the motor output was sinusoidally modulated in response to sinusoidal accelerations applied along the three body axes of the cat (see Fig. 2). The total harmonic distortion in the output (see Methods), for the data which were retained, was usually between 0.1 and 0.3. This amount of distortion, mostly confined to the second harmonic, is compatible with that present in the input (see Methods). The phase and gain of higher order harmonics were randomly distributed and consequently uncorrelated to the input. Also, the phase and gain relations between the

A

B

200

200 RIGHT E M G

RIGHT EMG

o

W U] \ 0_

0

I

I

I

I

I

I--

0

H

12/

"--.2

EOO LEFT EMG

LEFT EMG

LI tn n

0

I 0

I 90

I

I

180 BEG

I

I

270

I

I

360

I

0

~

9O

I

I

180

I

I

I

I

270

nEG

Fig. 3. Examples of averaged binned data. In A the acceleration was applied along the X-axis, while in B it was applied along the Y-axis. The solid curves interpolating the data points ( × ) are the fundamentals computed from the Fourier series of the output. The input frequency was 0.4 H z and the displacement amplitude 4- 27 cm. Note that in A a forward acceleration (Ant) is associated with an increase in the probability density for the E M G activity in both triceps (labeled right and left EMG). In B, instead, the activity of the two triceps is 180 ° out of phase, increasing when the acceleration, applied along the Y-axis, is directed to the ipsilateral side. Note the appreciable phase lag o f the motor output with respect to the input acceleration.

3f5 30

I z

10' 5' 0I

I

I

I

I

I

I

I

i

.1.20. (.3 60-

x

0 , - -

7 -180 0.1

012

71 J1 '7 '7 ~

0'.4

i

O'.S

I

I

,

018 '110

FREQ-HZ

Fig. 4. Phase and gain relations between input acceleration and motor output for linear accelerations along the X ( E>)- and Y ( × )-axes, as a function of input frequency. The two symbols represent the mean of the averages of the data from 8 animals, each animal being equally weighted. Data from both triceps brachii of each animal are included, the input reference for each of them being the appropriate positive acceleration. The vertical bars include plus and minus one standard deviation. The ordinate scale on the right of the gain plot gives the magnitude of the changes in the probability density of the EMG activity in terms of impulse/sec per m/secL This applies also to the subsequent Bode plots (Figs. 5 and 6).

output and the input, to be considered shortly, were not found to be significantly different for a nearly threefold change o f the input's amplitude. Finally, the system was found to be stationary in the sense that for a given input, applied at different times during the course o f an experiment, the output was the same. The subject o f lineatity will be further considered in the following two papers 1,~6 in the light o f additional data.

Horizontal accelerations Fig. 3 is a representative example o f data obtained with accelerations applied along the X (Fig. 3A) and Y (Fig. 3B) axis at one input frequency, 0.4 Hz. F r o m top to b o t t o m , in each case, are the averaged binned data from the right triceps, the input acceleration and the averaged binned data f r o m the left triceps. It can be seen that for accelerations along the X-axis the E M G activities in the t w o forelimb extensors are in

7 phase. On the contrary, when an acceleration is applied along the Y-axis, the two forelimb extensors are 180° out of phase (see also Fig. 2), as should be expected. Indeed, in the first case the activation of the macular receptors of the right and left labyrinths is in phase, while in the latter case their activation is 180° out of phase. A sign convention was chosen so that a positive acceleration is associated with forelimb extension. Therefore, a positive acceleration along the X-axis is in the cat's forward direction, while along the Y-axis it is directed to the ipsilateral side, i.e. to the right for the right forelimb and to the left for the left forelimb. Note that for accelerations along each of the axes the restoring shearing forces, resulting from the inertia of the otoliths and actually responsible for exciting the macular receptors, are in the same direction as the applied acceleration. This is shown in Fig. 1 for the acceleration ay, along the Y-axis, where the shearing force is labeled sy. Fig. 3 shows that the motor output lags the input acceleration for accelerations along both the X- and Y-axis. This behavior is more fully described in the Bode plots of Fig. 4 where the data points for the X (~7) and Y (x) axis are the mean values of the data obtained in 8 cats, each cat being equally weighted. Data from the triceps of both forelimbs were used. Note that for input frequencies below 0.15 Hz no reliable data could be obtained. This is not surprising since the acceleration amplitude is less than 0.025 g at these frequencies (see Methods). Several points need to be stressed. First of all, although the mean values for the phase and gain obtained during accelerations along the two axes are not too different, the gain may be less for accelerations along the X-axis than along the Y-axis. However, a larger sample size would be necessary to establish the significance of this observation. Nevertheless, there can be no doubt that the motor output lags the input acceleration at frequencies above 0.4 Hz, this lag reaching 40-60 ° at 1 Hz. If one accepts as being valid for the cat the data obtained by Goldberg and Fernandez '1,12 on the dynamic characteristics of the macular afferents in the squirrel monkey, the phase data just reported exclude that a direct vestibulo-spinal pathway is mainly responsible for the reflex postural adjustments (see Discussion). Note that the gain drops by about 14-20 db within the range of input frequencies used. Numerous factors can be mentioned which are likely to contribute to the scattering of the experimental data for both the gain and phase. In this context one has to take into account not only the factors mentioned under Methods but also the fact that the vestibular inputs are not relayed directly to the alpha motoneurons (see above), but instead are subject to extensive central processing. Under these conditions, and particularly if both excitatory and inhibitory actions are involved, the measured output can be greatly affected by the background excitability. Moreover, if both excitation and inhibition result from the application of a given input and converge upon a common element, then both the gain and the phase at the output of this element can be greatly altered by changing the ratio between excitation and inhibition. What we are able to exclude is that the scattering of data is due to changes in the central integrative mechanisms resulting from fluctuations in sensory inputs from the two limbs. Indeed, bilateral forelimb deafferentation (performed 5-25 days prior to the decerebration) did not significantly change the values of the gain or phase or reduce the scattering.

30-

il

;=,~.

20-

,!

I

Z 10

0 1BO

I

I

I

I

I

I

I

I

120

I

W

o

__I

I