Acta physiol. scand. 1969. 77. 68-84 From the Institutes of Physiology and Neurophysiology, University of Oslo, Norway

The Firing Pattern of Dorsal Spinocerebellar Tract Neurones during Inhibition

BY

J. K. s. JANSEN, K. NICOLAYSEN and L. WALLOE

Received 6 December 1968

Abstract

JANSEN, J. K. S., K. NICOLAYSEN and L. WALLOE. The firing pattern of dorsal spzno- cerebellar tract neurones during inhibition. Acta physiol. scand. 1969. 77. 68- 84.

Dorsal spinocerebellar tract neurones adequately activated by Ia afferents from the tibialias anterior ( T A ) , extensor digitorum longus (EDL) or flexor digitorum longus (FDL) muscles are regularly inhibited by group I afferents from the gastrocnemius-soleus (GS) nerve. The inhibition is maintained throughout the duration of GS nerve stimulation. During inhibition the irregularity of firing may be unchanged or moderately reduced. Correspondingly the serial dependency between neighbouring intervals may be unchanged or reduced. The inhibitory input usually caused a constant reduction in firing frequency at various levels of excitatory drive or occasionally a greater reduction at higher levels of excitatory drive. The latter effect was associated with increased regularity of firing and reduced serial dependency. Comparable inhibitory effects were obtained by repetitive stimulation of the contralateral somato sensory- cortex. I t is suggested that the inhibitory effects can be accounted for by post-synaptic in- hibitory mechanisms.

Certain features of the transfer signals through the Clarke’s column relay of the dorsal spinocerebellar tract (DSCT) can be explained in terms of a simple model of the second order neurone (Wallere 1968, Jansen and Wallere 1968). This applies to the discharge pattern of the second order neurones adequately activated through their primary afferent fibres. Some of the predictions of the model, such as the size of the unitary EPSPs and the degree of convergence of primary afferent fibres on to each second order neurone, have been supported by direct record of the synaptic ac- tivity of Clarke column cells (Eide et at. 196933).

The DSCT neurones also receive inhibitory inputs from primary afferent fibres (Holmqvist, Lundberg and Oscarsson 1956, Eccles, Oscarsson and Willis 1961. Hongo and Okada 1967). The excitatory coupling of the DSCT relay is known to be highly specific in the sense that a given second order neurone is activated only from one particular muscle. Some of the inhibitory input is supplied by group I afferent fibres from muscles not directly connected to the second order neurone (Hongo and Okada 1967, Jansen, Nicolaysen and Wal l~e 1967).

68

69 Two types of inhibitory mechanisms have been reported to act at the Clarke

column synapses. The one is the traditional postsynaptic inhibition presumably mediated via one interneurone (Eccles et al. 1961, Hongo and Okada 1967). The second is a presynaptic inhibition on which the evidence is less direct. It consists of reduction in the size of DSCT volleys and excitability changes of the primary af- ferent terminals following suitable conditioning ( Eccles, Schmidt and Willis 1963, Jankowska, Jukes and Lund 1965) and of the pharmacological properties of the in- hibitory effect (Jankowska, Jukes and Lund 1964).

The purpose of the present experiments was to study the interaction of excitation and inhibition on the firing pattern of DSCT neurones. Controlled excitation of these was conveniently obtained by muscle stretch. Preliminary studies (Jansen et al. 1967) suggested that inhibitory effects could regularly be obtained from GS group I afferents acting on signals from stretch receptors of FDL or pretibial flexors. Our main experiment has therefore been to study the effect of a constant amount of in- hibition (obtained by repetitive, electrical stimulation of group I GS afferents) on TA, EDL or FDL second order neurones under varying amount of excitatory drive. The results can be interpreted on the basis of the existing model of synaptic excita- tion of the DSCT second order neurone.

INHIBITION A N D NEURONAL FIRING PATTERN

Methods The experiments were performed on nembutal anesthetized cats. One hind leg was denervated except for the nerves to the anterior tibia1 ( T A ) , the extensor digitorum longus (EDL) and the flexor digitorum longus (FDL) muscIes. The tendons of these muscles could be attached to a muscle stretcher or loaded with weights. The gastrocnemius-soleus (GS) nerve was cut peripherally and mounted for electrical stimulation. The shock intensity was determined in relation to the afferent volley recorded from the L VII dorsal root entry zone. The ipsilateral ventral roots L VI, L \'I1 and S I were always cut. The activity of single DSCT axons was picked up in the ipsilateral dorsolateral funicle of the cord a t T h X-XI1 level and recorded on film and magnetic tape. The criteria for unit identification and details of the methods have been described in full previously (Jansen et a2. 1966).

To obtain periods of reasonably steady state firing of the neurones, the initial one sec of the transients at the beginning and after an inhibitory input was excluded from the analysis. The duration of the inhibitory stimulus was limited to approximately 10 sec. Consequently the number of firing intervals in these periods might be too small to determine the serial corre- lation coefficient reliably. To get a measure of the first order serial dependency we have in- stead used the linear regression line describing the relationship between the duration of one interval and the average duration of the following interval (see Jansen et aE. 1966).

The slope of this regression line depends on the first order dependency between intervals. There is, however, no unique relation between this slope and the first order serial correlation coefficient, but in view of the uniform behaviour of this group of DSCT neurones (Jansen e t al. 1966) it is considered sufficiently accurate for the present purposes.

Results

During the experiments to be reported the dorsolateral funicle of the spinal cord was explored by a micropipette to find axons of DSCT neurones monosynaptically ac- tivated by group I afferents from TA, EDL or FDL, and also activated by stretch of the appropriate muscle. Once a unit was identified according to criteria previously described (Jansen and Rudjord 1965) the effect of repetitive stimulation of the GS nerve was examined. In some preliminary experiments the inhibitory effect of dif-

70 J. K. S , J A N S E N , K. NICOLAYSEN A N D L. W A L L o E

ferent frequencies of nerve stimulation was determined. No certain reduction in firing frequency was found for stimulus rates of less than 10 sec-l. At higher rates the inhibitory effect increased approximately linearly up to frequencies of about 120 sec-l. Beyond this rate the inhibitory effect was only slightly increased. Since the purpose of the present experiments was to study the effect of inhibition on the firing pattern of the neurones, the frequency of stimulation was routinely kept at 100 sec-l, and the shock intensity was maximal for group I afferent fibres in the following ex- periments. If there was noticeable reduction in firing frequency during GS nerve stimulation, the firing frequency of the unit was changed by varying the degree of static stretch of the appropriate muscle and the effect of the same inhibitory stimulus was determined at the various levels of excitatory drive.

GS inhibition of TA-EDL I@ in units. Most of our material consists of DSCT units activated from primary endings of the TA and EDL muscles. Such units are usually inhibited by group I afferent volleys in the GS nerve (Jansen et al. 1967). In the present series of experiments the effect of group I GS stimulation was examined on 18 TA-EDL Ia units and the firing frequency was perceptibly reduced in 16 of them. Seven of these were studied at a reasonable number of different states and for a sufficient period of time to permit a systematic description. The data from these seven units constitute most of the material on which the following account is based. Supporting data were obtained from the other units.

A characteristic example of the effect is presented in Fig. 1. Record A shows the unit firing without any load on the tendon. This background firing was often quite regular and at a rate of about 10 imp. sec-l. The cell was silenced immediately by GS nerve stimulation at 100 sec-', and the inhibition persisted throughout the period of stimulation. Just after the inhibition there was usually a transient rebound in- crease in firing frequency. With a 50 g load on the TA tendon (Fig. 1B) the firing frequency of the unit increased to about 40 imp. sec-l. The inhibitory stimulus again reduced the firing frequency. The reduction was great initially and recovered partly in the following period. But the inhibition was clearly maintained throughout the 14 sec of inhibitory stimulation. To illustrate the time course of the inhibition the firing frequency of the same unit with a 200 g load on the TA tendon has been plotted in Fig. 1C. During the inhibitory stimulus the firing frequency dropped immediately and recovered partly during the first few sec. Thereafter the inhibitory effect re- mained approximately constant, and the firing frequency of the cell stayed between 50 and 55 imp. sec-l till the end of the inhibition.

The inhibition illustrated in Fig. 1 was due to activation of group I afferent fibres of the GS nerve. This was controlled by dorsal root records of the afferent volley. The inhibition might be present at shock intensities as low as 120 per cent of group I threshold. The inhibitory effect increased with increasing shock strength till all group I fibres were activated. Group I maximal shocks were routinely employed in the fol- lowing experiments.

The important observation of Fig. 1 from the point of view of firing pattern of the

INHIBITION AND NEURONAL FIRING PATTERN 71

6 l i l o s e c

C

I I I I I I I I t 1 I I

0 1 2 3 4 5 6 7 8 9 10 1 1 1 2 S e c

Fig. 1 . Effect of repetitive (100/sec) GS nerve stimulation on Ia DSCT neuron activated by stretch of TA. Intensity of nerve stimulation just below group I maximum. A. No load on TA. B. 50 g load on T A tendon. 7 sec cut out in the middle of record. C. Plot 0;f firing frequency against time of same unit during GS inhibition. 200 g load on TA tendon.

cells is that the inhibition is maintained and gives rize to an approximately steady state firing after the initial overshoot. The following statistical description of the firing pattern requires steady state periods to be meaningful. Therefore, the first sec of the transient after the onset of inhibition was excluded and interval distribution and serial dependency of the rest of the spike trains was determined. The small drift which commonly was present also in the late period of inhibition was ignored, since it will not have any significant influence on the following rather crude statistical description.

The distribution of intervals. On adequate excitation the DSCT units usually generate unimodal and fairly symmetric interval distributions (Jansen et al. 1966). During steady state inhibition most units still generated the same type of histograms as long as their mean firing-frequency remained above 10-20 imp. sec-'. This is illustrated in Fig. 2. With a 50 g load on the TA tendon the mean interval was 15.6 msec and the distribution symmetrical (Fig. 2A). During GS stimulation with the same load the mean interval increased to 26.2 msec, but the distribution retained its general shape (Fig. 2B). For comparison the interval distribution obtained with a 20 g load on the tendon is shown in C . The mean interval was now 29.9 msec; rather

72 J. K. S. JANSEN, K. NICOLAYSEN AND L. WALLBE

0 32 4 8 0 41 1u 192 16 96 ms.c miec

Fig. 2. Distribution of intervals during maintained inhibition. A. Control period. 50 g load on TA tendon. B, as A during 100/sec, group I max GS nerve stimulation. C. Control period, same unit, 20 g load on TA tendon. D, as C during GS nerve stimulation. N (number of intervals) X (mean interval), SD (standard deviation of distribution) given for each histogram.

close to that obtained during inhibition with a 50 g load, and the two histograms (Fig. 2B and C ) are very similar. Accordingly the general shape of the interval distribution may remain essentially unchanged after the addition of an inhibitory synaptic process. Of the seven TA-EDL units studied for a prolonged period at a number of different firing levels, four did not show obvious changes in their interval distribution during inhibition. The same type of behaviour has been reported for second order neurones of the cochlear nucleus by Goldberg and Greenwood ( 1966). I t should, however, be realized, and it will appear from the following that mere inspection of interval histograms is not a sensitive method for detecting small changes.

Other units showed characteristic changes in the shape of the interval distribu- tions during inhibition. One type of change occurred when the firing frequency was reduced to less than approximately 20 imp. sec-l by the inhibition. As illustrated in Fig. 3B the interval histogram then became positively skewed with a long tail. The histograms of Fig. 3C and D are from the same unit now firing at a higher rate, and they illustrate a different alteration of the interval distribution. During excitation only (Fig. 3C) the distribution was broad, almost rectangular. With inhibition added the distribution was much narrower with definite tails on both ends rather like a Gaussian distribution (Fig. 3D). A comparison of Fig. 3A and D is useful. In these

INHIBITION AND NEURONAL FIRING PATTERN

P

0.20 N=576 0.20

0.10 0 1 0

73 N-218 r=62.2

S D.25.70 B r

-

A

0.20

P

0.10

N=356

50=4.97 P D x-23 8

0.20 - N=8Ol C x= 14.3

50=507

0.10 -

0 16 0 32 4 8 m n c l6 mses

Fig. 3. Distribution of intervals during maintained inhibition. A. Control period. 50 g load on TA tendon. B, as A during 100/sec, group I max GS nerve stimulation. C. Control period, same unit, 150 g load on T A tendon. D, as C during GS nerve stimulation.

two situations the unit is firing at approximately the same mean rate, in the one subjected to excitation only (Fig. 3A) in the second to a mixed excitation and in- hibition (Fig. 3D). This type of change was seen consistently for two of the seven TA-EDL units, but as will appear from the following section, it is frequently found to a smaller degree in the present material.

The irregularity of firing. The degree of irregularity of firing is most conveniently measured by the coefficient of variation (CV) which is the ratio between the standard deviation (SD) and the mean interval of the sample. Exposed only to its specific excitatory input the coefficent of variation of DSCT cells is usually approxi- mately constant over the entire frequency range (Wallcle 1968). This is simply dem- onstrated by a plot of the SD against the mean interval. This plot is well described by a straight line. During inhibition the observations would usually obey the same relationship. This is illustrated by the data from the unit presented in Fig. 4A. The data from some of the other units, however, suggested that the firing pattern during inhibition might be slightly less irregular than that of purely excitatory firing. For one unit this was definitely so, as illustrated in Fig. 4B. In this graph the points ob- tained during inhibition are clearly located below the control values. The reduction in variability was accompanied by a change of the interval distribution of the type illustrated in Fig. 3C and D.

I n this type of presentation small systematic changes in CV might be obscured by the random changes in variability from trial to trial. The CV of all the inhibitory

J. K. S. JANSEN, K. NICOLAYSEN AND L. WALL0E

30-

20

x f 8

10

-

-

A

6 0

: ro "

o'6 0.6

periods observed in the seven TA-EDL neurones has therefore been plotted against the CV of the immediately proceeding control period of purely excitatory firing (Fig. 5). I t appears that in all but 5 of 34 inhibitory periods there was a slight to moderate reduction in the CV. The data from the unit of Fig. 4B is presented with a separate symbol in Fig. 5.

Accordingly, the conclusion is reached that the irregularity of firing of these DSCT units is usually slightly to moderately reduced by a steady state group I inhibitory input.

-

Serial dependency of intervals. When subjected to a purely excitatory input there is a strong negative serial dependency between neighbouring intervals (Jansen et al.

/ 0.L

0.3

0 2 0.2 0 3 0.L 0.5 0.6 C V E

Fig. 5. Coefficient of variation

(CV = ") during GS inhibition (ordinate) plotted against the CV of the immediately preceding control period (abscissa). Data from seven DSCT Ia units activated by TA stretch. Periods with mean frequencies of firing of less than 20 implsec excluded. +, data from same unit as Fig. 4B.

INHIBITION A X D NEURONAL FIRING PATTERN 75

1

2 5 50 2 5 5 0

I ' i - t (rnsec) 'i-1 ( m s e c )

Fig. 6. Mean duration of conditional interval distributions (ordinate) plotted against the duration of the preceding interval (abscissa). Group sizes of preceding intervals were 3.2 msec in A, 1.6 msec in B (lower curve) and 3.2 msec (upper curve). A, control period, mean frequency of 64 imp/sec with a 50 g load on TA tendon. Total number of intervals 991. SD of means less than 0.8 msec for all data groups containing more than 20 observations. Slope constant of least square regression line - 0.53. Linear correlation coefficient (xi-xi - 1) = -0.98. 0 Same excitatory drive. Data obtained during GS inhibitory stimulation reducing the mean frequency of firing to 38 imp/sec. Total no of intervals 371. SD of means less than 1.5 msec for all groups of more than 20 intervals. Slope constant of least square regression line -0.60. Linear correlation coefficient (xi-xi -1) = -0.96. B. another cell, control period, mean frequency 70 implsec with a 150 g load on TA tendon. Total number of intervals 801. SD of means less than 1.1 msec for all groups of more than 20 intervals. Slope constant of regression line -0.67. Linear correlation coefficient (xi---xi - I) =0.97. 0 same excitatory drive. Data obtained during GS stimulation reducing the mean frequency of firing to 42 imp/sec. Total number of intervals 356. SD of means less than 1.5 msec for all groups of more than 20 obs. Slope constant of regression line -0.29. Linear correlation coefficient (xi-xi-1) = -0.84. The data in A are taken from the unit of Fig. 2 and those of B from that of Fig. 3.

1966). Short and long intervals tend to occur alternately. This can be demonstrated by the relationship between the duration of neighbouring intervals, i.e. joint interval histograms. In the present work we have determined the average duration of inter- vals following intervals of a given duration, that is the average value of the con- ditional distribution of intervals (see Methods). Jansen et al. (1966) showed that this mean value of the conditional distributions was an approximately linearly de- creasing function of the duration of the conditioning interval, and that the slope of the regression line was constant at all levels of excitation (their Fig. 10). This is due to the fact that the degree of serial dependency is relatively independent of the firing frequency or the excitatory drive of the cell.

When inhibition was added to the input the serial dependency was still present. For some units the slope of the regression line was unchanged or only slightly reduced indicating that the degree of serial dependency was similar (Fig. 6A) as long as the cell was firing at a reasonable frequency (20 to 60 imp. sec-l was the range actually observed). For other units the nummerical value of the slope of the regression was definitely reduced (Fig. 6B) indicating a decreased serial dependency. All our ob- servations on TA-EDL units are presented in Fig. 7, which is a scatter diagram of the slope constant of the regression line during inhibition (ordinate) against that of the immediately preceding control period. There appears to be different degrees of

76 J. K. S. JANSEN, K. NICOLAYSEN AND L. WALL0E

-0.8,

0 0 .

o o ot

x

4A x

Fig. 7. Slope constants obtained as illustrated in Fig. 6 during inhibition (ordinate) plotted against the corresponding slope constant of the immediately preceding control period (abscissa). -o.2L- -0 .2 - 0.4 -0.6 L1 -0.8 Data 30 imp/sec with mean has been frequencies excluded. of firing Data from less than dif-

ferent units have been given separate symbols. 'E

reduction of the slope constant. To show that each unit tended to behave systemati- cally they have been given separate symbols. One group of three units showed no or slight reduction in slope constant (circles). The other units regularly exhibited varying degrees of reduced slope constant during inhibition. Spike trains with mean frequencies of less than 30 imp. sec.' have been excluded from the material.

The main finding of this section is that DSCT units could fire with approximately the same degree of serial dependency during inhibition, but that the serial dependency was frequently reduced. Furthermore, the units with decreased dependency were the ones exhibiting the most pronounced changes in the shape of the interval distribu- tions in the direction of increased regularity of firing. For instance, the data from the unit illustrated in Fig. 3 are represented by x, and those of Fig. 4B by filled triangles in Fig. 7. The unit of Fig. 2 and 4.4, on the other hand, is represented by the open circles of Fig. 7.

Inhibition as a function of the rxcitatory drive. The efficiency of the inhibition was measured as the reduction in average frequency in the steady state period. The results are lnost conveniently presented as diagrams of the frequency of firing during inhibition against frequency of firing in the control period with the same excitatory drive on the cell (Fig. 8 ) . Since there was always a slow reduction in firing frequency during prolonged excitation the excitatory frequency of firing is given as the mean of the firing frequencies in the period before and after the inhibitory period. This manipulation did not introduce any qualitative changes in the results. If the inhibi- tory input had no effect, the data in diagrams like those of Fig. 8 would lie on the line through the origin with a slope of unity. This line has been drawn on the plots to illustrate the inhibitory effects. I t appears that the observations are reasonably well described by linear regression lines. This applies to all the 7 DSCT units examined. The correlation coefficient was as high as 0.99 for all units. This means

INHIBITION AND NEURONAL FIRING PATTERN 7 7

F. I i m p l s * c l F* l imp/seci

Fig. 8. Effect of GS nerve stimulation at different levels of excitatory drive. Mean frequency of firing during inhibition (ordinate) plotted against the mean frequency of the preceding and succeeding control periods. A and B from two different units. Slopes of least square regression lines, 1.06 and 0.84 respectively.

that the observations can be adequately described by a simple equation of the fol- lowing type:

Fi = a * F, + b

The value of the constant b gives the intercept with Fi axis, and it had values be- tween 4 . 8 and -27.7 imp. sec" for the seven units. Of greater interest is perhaps the value of a which gives the slope of the regression line. The two units of Fig. 8 were selected to illustrate the extreme values of a, which ranged from 1.1 (Fig. 8A) to 0.84 (Fig. 8B). Accordingly, a constant amount of inhibition can cause constant reduction in firing frequency of the DSCT cells over a reasonable frequency range, or it can have an increasing effect with increasing frequency of firing. I t should be pointed out that there was no simple relationship between the values of the two constants a and b in the present limited material. There was for instance no sugges- tion that a more powerful inhibition giving a large value of b was associated with a small value of the slope constant a.

However, comparing the values of the slope constant with the effect of the in- hibition on the variability, histogram shape and serial dependency a significant cor- relation appears to be present, even in the present small material. The units with small values of the slope constant were the ones with a reduced variability and a decreased serial dependency during inhibition. The unit of Fig. 8B is the unit il- lustrating increased regularity in Fig. 3 and decreased serial dependency in Fig. 7 (x) . Similarly the unit illustrating increased regularity in Fig. 4B had a slope constant of 0.88. The unit of Fig. 8A, on the other hand, retained its firing pattern during in- hibition (Figs. 2, 4A and 7 0). Accordingly the reduced variability, the reduced serial dependency and slope constants less than one appear to be a consequence of the same inhibitory mechanism.

78 J. K. S. JANSEN. K. NICOLAYSEN AND L. W A L I . 0 E

A -' I I 1 1 I

BL

l l lOsec

C Cortcr s t i m 10OJsec

I m plss c I I

. . 0 6 e 9 lb 1; 1; 13 14Sec

Fig. 9. Effect of repetetive stimulation of contralateral postcruciate gyrus at 100/sec, pulse duration 1 msec, intensity 1 mA. Unit activated by FDL, stretch. '4, background firing. B, FDL stretched about 3.5 mm. About 9 sec cut out from middle of record B. Period of cortical stimulation indicated by continuous lines under records A and B. C, Plot of firing frequency against time during cortical stimulation. Same unit as A and B. FDL stretched about 6 mm.

The value of the present slope constant of inhibition is of obvious functional sig- nificance. Since the relationship between muscle length and firing frequency of Ia DSCT units is linear (Jansen et al. 1966) inhibitory slope constants smaller than one represent a reduction of sensitivity of the system. Assuming that the signals from the various DSCT units ultimately converge on Purkinje cells of the cerebellum, it is probably meaningful to compute the average slope constant of the present material. For the present seven units it was 0.94.

Inhibition of FDL l a units. The inhibitory effect of GS nerve stimulation on the group Ia signal from the FDL muscle in the DSCT was reported briefly in a pre- ceding paper (Jansen et al. 1967). I n the present series of experiments 14 units of this sort have been examined and all of them were found to be inhibited by repetitive stimulation of the GS nerve of group I strength.

The general effects of inhibition were the same as those described above for the inhibition TA-EDL units. This applies to the shape of the distribution of intervals which might be essentially unchanged during inhibition or might show a reduced variability. The reduction in variability was again associated with slope constants appreciably smaller than unity in the Fi-F,, plots and a reduction in serial depend- ency. I n these respects the FDL units were apparently so similar to the TA-EDL units that no further description is needed. The following account will therefore

INHIBITION AND NEURONAL FIRING PATTERN 79

Fig. 10. Comparison of effects of cortical inhibition o and GS in- hibition +. Data from same unit as Fig. 9. Same type of plot as Fig. 8. Slopes of least square linear regression lines, 0.96 for cortical inhibition, 0.93 for GS inhibition.

s o -

LO -

a 10 2 0 30 LO 50 6 0 F e l m W s c r l

concentrate on certain additional observations on inhibition observed in the DSCT Ia units.

Stimulat ion of post cruciate gyrus. Hongo and Okada (1967) reported IPSPs in group Ia activated Clarke’s column neurones following stimulation of the post- cruciate gyrus of the cat. The cortical inhibition was mediated by pyramidal tract fibres, and apparently caused only by post-synaptic inhibitory mechanisms. To peri- pheral nerve stimulation, on the other hand, they obtained evidence for presynaptic as well as postsynaptic inhibitory effects. We, therefore, found it of interest to com- pare the firing pattern during cortical inhibition with that elicited from group I afferents of peripheral nerves.

In agreement with Hongo and Okada (1967) the cortical inhibition appeared to be most easily elicited from the medial region of the post-cruciate gyrus, and this was the stimulation site in the following experiments. The cortex was stimulated repetively at a rate of 100 sec-l. The pulse duration was usually 1 msec and the threshold intensity was about 1 mA. I n search for inhibitory effects the shock in- tensity was increased up to 5 mA. These are stimulus parameters of the same order of magnitude as those employed for instance by Hearn et al. (1962) in order to activate pyramidal tract neurones. A definite reduction in firing frequency was seen for all of the 13 Ia DSCT units examined in this way. Examples of the effect are presented in Fig. 9. The cortical inhibition appears to be similar to that elicited from the GS nerve, as illustrated in Fig. 2. Particularly important is the finding that the cortical inhibition is maintained, after an initial transient overshoot, throughout the period of stimulation (Fig. 9). Accordingly the firing pattern during cortical inhibition can be statistically described in the steady state period.

The effect of a constant cortical inhibition at various levels of excitatory drive is

80 J . K. S. JANSEN, K. NICOLAYSEN AND L. WALL0E

C V I

a.4 -

0.3-

0.2

Fig. 11. Interval distributions during cortical inhibition ( B ) and-GS inhibition (C) . A. In- terval distribution of control period immediately preceding B. N, X and SD as in Fig. 2.

. + -f

+ f + . . +

b + + +

Fig. 12. Coefficient of variation of interval distribu- tions during cortical inhibition and GS inhibi- tion + (ordinate) plotted against the coefficient of

I

illustrated in Fig. lo? in a diagram like that of Fig. 8. I t appears that the reduction in firing frequency was slightly greater at higher frequencies of firing. The slope constant of the best fitting regression line was 0.96. The effect of group I GS nerve stimulation is shown in the same diagram from comparison. In this cell the GS in- hibition was slightly less effective than that elicited from the cerebral cortex. The slope constant of the GS inhibition was 0.93, very nearly the same as that of the cortical inhibition.

The distribution of intervals might also be similar during cortical and peripheral inhibition. This is illustrated in Fig. 11. The histogram obtained during the control period was as usual unimodal and fairly symmetric. During cortical stimulation the firing frequency was reduced from 56.1 to 37.6 imp, sec but the distribution of intervals was still of the same shape. For comparison the interval histogram obtained during GS nerve stimulation is shown in Fig. 11C. It is indistinguishable from that obtained during cortical inhibition.

The degree of irregularity of firing might also be similar during cortical and pe- ripheral Ia inhibition. In Fig. 12 the data from one unit is presented in a scatter diagram like that of Fig. 5. The variation in CV was somewhat greater than usual

INHIBITION AND NEURONAL FIRING PATTERN 81

0 0 * .

25 5 0 Xi-, I m s e c l

+ + o +

t C4

0

0

+ f O

0

Fig. 13. A. Mean duration of conditional interval distributions (ordinate) plotted against the duration of preceding interval (abscissa) as in Fig. 6. Group sizes of preceding intervals 3.2 msec. All data from same unit. Control period, mean frequency of firing 56 imp/sec. Total no of intervals 353. SD of mean less than 1 m e c for all data groups of more than 20 observa- tions. Slope constant of least square regression line -0.44. Linear correlation coefficient xi- xi-1) = -0.93. 0 Same during inhibition by cortical stimulation (100/sec). Mean frequency of firing 37.5 imp/sec. Total no of intervals 380. SD of mean less than 1.6 msec for all data groups of more than 20 intervals. Slope constant of least square regression line -0.57. Linear correlation coefficient (xi-xc- 1) =0.97.+Same unit during inhibition from GS nerve. Mean frequency of firing 43 implsec which was a reduction of 13 imp/sec from the preceding control period. Total number of observations 314. SD of mean less than 1.3 msec for all data groups of more than 20 intervals. Slope constant of least square regression line -0.47. Linear cor- relation coefficient (xi-xi-1) = -0.97. B. Slope constants obtained as illustrated in A during cortical inhibition 0, and during GS inhibition + (ordinate) plotted against the corresponding slope constant of the immediately preceding control period (abscissa). All data from same unit as A. Periods with firing frequencies of less than 25 imp/sec have been excluded.

for this unit, but it appears that there was rather little difference between the CV of the control period and that during inhibition, and secondly that the data from cortical and peripheral inhibition are located in the same area of the diagram.

Turning finally to the effect of cortical inhibition on the serial dependency be- tween neighbouring intervals the data from the unit of Fig. 10 and 12 are presented in Fig. 13. Part A of this Fig. is a plot of the type already employed in Fig. 6. I t gives the average duration of conditional intervals (ordinate) as a function of the pre- ceding interval. The frequency of firing in the control period was 56.1 imp.sec l. The regression line then had a slope constant of --0.44 (filled circles). During cortical inhibition the mean frequency was reduced to 37.5 imp. sec-l and the slope constant of the regression line was -0.58 (open circles). The third group of data in Fig. 14A (crosses) was obtained during GS inhibition of the same cell. During the GS inhibi- tion the frequency of firing was 43.3 imp.sec.l and this was a reduction of 13 imp.sec from the preceding control period. The slope constant of the conditional mean in- tervals was 0.47 during the inhibition. These three values of the slope constant were

G 9 3 0 0 3 . Actn physiol. scand. Vol. 77: 1-2

82 J. K. S. J A N S E N , K. NICOLAYSEN A N D L. W A L L B E

all within the range of variation of this particular cell. All the data from this cell are presented in the diagram of Fig. 13B. This is a plot of the slope constant of the conditional mean intervals during inhibition (ordinate) against the same dope constant during the immediately preceding control period. The values might be somewhat greater or smaller than the control values, but this is probably not signifi- cant. The important finding is that the values obtained during cortical and peripheral inhibition are distributed over the same area of the graph. Accordingly, the serial dependency may remain essentially unchanged during cortical as well as during GS Ia inhibition.

Discussion

As mentioned in the Introduction presynaptic as well as postsynaptic inhibitory mechanisms have been described operating on the Clarke’s column synaptic transfer (Eccles et al. 1963, Hongo and Okada 1967). I t we accept our earlier analysis and simplified model of the excitatory synaptic coupling of the Clarke’s column relay (Jansen and Wallere 1968, Wallere 1968) predictions can be made about the probable effects of the various types of inhibitory mechanisms on the firing pattern of these neurones. Let us consider three different situations with a constant amount of in- hibition acting on neurones at different levels of excitatory drive. The first and the simplest is that of presynaptic inhibition. This will cause only a reduced size of EPSPs generated by each of the primary afferent fibres. With no input to the second order neurone, the inhibition will have no effect, and the reduction in excitation will increase directly with the intensity of the afferent input. Accordingly one expects a reduced slope in the F,-F, diagrams (cf. W a l l ~ e 1968). I n addition the presynaptic mechanism should reduce the irregularity of firing, since this is primarily determined by the size of the unitary EPSPs. The third predicted effect of presynaptic inhibition is a reduced serial dependency between the duration of neighbouring intervals. This is because the serial dependency appears to be due mainly to the particular pattern of activity in the input fibres of the DSCT cells (Wallere 1968), and with reduced amplitudes of the EPSPs this effect would be less pronounced. This particular group of effects was indeed found for some of the cells of the present material. But it should be pointed out that some of the inhibitory effects that were found can not easily be explained by a presynaptic mechanism. This applies particularly to the inhibition of the backgrund firing which was a regular finding in the present material.

Consider next a postsynaptic inhibition which acts by hyperpolarizing the second order neurones without any concomitant shunting of the excitatory synaptic currents. Eide et aZ. (1969a) found a linear relationship between the firing frequency and the transmembrane current for Clarke column neurone over a wide range of frequencies. They also found an algebraical summation of excitatory synaptic and injected mem- brane currents. Accordingly a steady hyperpolarizing inhibition which is caused by a steady inward membrane current in the soma will cause a constant reduction in the firing frequency over the entire range and thus give Fi-F, diagrams with a slope

INHIBITION AND NEURONAL FIRING PATTERN 83 of one. Furthermore since the soma conductance is unchanged by the inhibition, the size of the EPSPs will be unchanged. Therefore, one would not expect appreciable changes in the firing pattern of the neurones.

Consider finally a postsynaptic inhibition which acts predominantly by shunting the excitatory synaptic currents. The quantity of excitatory current “lost” by the shunting would increase directly with the excitatory drive and give Fi-F, diagrams with reduced slopes. At the same time the size of the EPSPs will be reduced with consequences on the firing pattern as outlined for the presynaptic inhibition. In ad- dition to these effects which presumably might be very similar to those of the pre- synaptic inhibition, postsynaptic inhibition with shunting will also inhibit the back- ground firing of the cells.

Returning now to the observations presented in the Result section, it appears that the significant finding is that the DSCT neurones examined exhibited a continuous spectrum of behaviour. At the one extreme there are the units without appreciable changes in the interval-distribution and the serial dependency between intervals. These units had slopes close to one in the F,-F, diagrams. The other extreme is represented by units with marked reduction in the irregularity of firing and a reduced serial dependency. This was associated with slopes significantly smaller than one in the F,-F, diagrams. Intermediate types of behaviour were also observed.

From our discussion above two types of interpretation appear to be possible. The one is that the inhibition was caused by a mixture of pre- and postsynaptic inhibition in variable proportions on the various units. Alternatively the inhibition may be purely postsynaptic with a variable degree of shunting of the excitatory currents. With the available material these two possibilities can probably not be distinguished. But on the principle of accepting the simplest model, at least as a working hypothesis, we tend to favour the second explanation. This is supported by the regular inhibi- tion also of the background firing. Secondly, the constant occurrence of a rebound increase in firing rate after the inhibition is a point in favour of a postsynaptic mechanism. The similarity between the inhibitory effect of GS group I input and the cortical stimulation points in the same direction. As mentioned. Hongo and Okada (1967) found no evidence of a presynaptic inhibitory component efter cortical stim- ulation. A postsynaptic inhibition causing different degrees of shunting of excitatory currents is also easily visualized morphologically. An inhibitory synaptic input distri- buted over variable distances along the dendrites might well exhibit the desired properties. It has been demonstrated by SzentAgothai and Albert (1955) that the primary afferent excitatory input of these neurones is located proximally on the dendrites.

Finally. some of the observations may be worth considering in a more extensive context. The high degree of linearity of the F,-F, diagrams indicates that the linearity of the synaptic transfer in the Clarke column relay (Jansen et al. 1966) is maintained also during a constant inhibitory input. The origin of this inhibition in stretch receptors of a related muscle and its tonic nature may also be significant. The content of the signal transmitted through this subdivision of the DSCT can be re-

84 J. K. S . JANSEN, K. NICOLAYSEN AND L. WALLOE

garded as information on muscle length, and the observed inhibition would in the simplest case represent a constant bias on this signal so that the apparent length that the cerebellum sees is reduced. Similar considerations apply to the cortical inhibition. I t is of interest that this is mediated by pyramidal tract fibres (Hongo and Okada 1967) which accordingly acquires a direct control over this input to the cerebellum.

The assistance of Mr. J. Rausandaksel is gratefully acknowledged. The work has been sup- ported by grants from Nansmfondet.

References ECCLES, J. C., 0. OSCARSSON and W. D. WILLIS, Synaptic action of group I and I1 afferent

fibres of muscle on the cells of the dorsal spinocerebellar tract. J . Physiol. (Lond.) 1961. 158. 517-543.

ECCLES, J. C., R. F. SCHMIDT and W. D. WILLIS, Inhibition of discharges into the dorsal and ventral spinocerebellar tracts. J . Neurophysiol. 1963. 26. 635-645.

EIDE, E., L. FEDINA, J. JANSEN, A. LUNDBERC and L. VYKLICK~, Properties of Clarke’s column neurones. Acta physiol. scand. 1969a. 77. 125-144.

EIDE, E., L. FEDINA, J. K. S. JANSEN, A. LUNDBERG and L. VYKLICK~, Unitary components in the activation of Clarke’s column cells. Acta physiol. scand. 1969b. 77. 145-158.

HERN, J. E. C., S. LANDGREN, G. G. PHILLIPS and R. PORTER, Selective excitation of cortico- fugal neurones by surface-anodal stimulation of the baboon’s motor cortex. J . Physiol. (Lond.) 1962. 161. 73-90.

HOLMQVIST, B., A. LUNDBERC and 0. OSCARSSON, Functional organization of the dorsal spino- cerebellar tract in the cat. V. Further experiments on convergence of excitatory and inhibi- tory actions. Acta physiol. scand. 19.56. 38. 76-90.

HONGO, T. and M. OKADA, Cortically evoked pre- and postsynaptic inhibition of impulse transmission to the dorsal spinocerebellar tract. Ex@. Brain. Res. 1967. 3. 163-177.

JANKOWSKA, E., M. G. M. JUKES and S. LUND, On the presynaptic inhibition to the dorsal spinocerebellar tract. J . Physiol. (Lond.) 1964. 177. 19-21P.

JANKOWSKA, E., M. G. M. JUKES and S. LUND, The pattern of pmsynaptic inhibition on transmission to the dorsal spinocerebellar tract of the cat. J . Physiol. (Lond.) 1965. 178.

JANSEN, J. K. S., K. NICOLAYSEN and T. RUDJORD, The discharge patterns of neurons of the dorsal spinocerebellar tract activated by static extension of the primary endings of muscle spindles. J. Neurophysiol. 1966. 29. 1061-1086.

JANSEN, J. K. S., K. NICOLAYSEN and L. WALLBE, On the inhibition of transmission to the dorsal spinocerebellar tract by stretch of various ankle muscles of the cat. Acta physiol. scand. 1967. 70. 362-368.

JANSEN, J. K. S. and T. RUDJORD, The dorsal spino-cerebellar tract: Response pattern of fibres to muscle stretch. Science 1965. 149. 1109-1 11 1.

JANSEN, J. K. S. and L. WALLBE, Transmission of signals from muscle stretch receptors to the dorsal spinocerebellar tract. In T h e Cerebellum in Health and Disease. Ed. W. S . Fields, W. H. Green, Inc. St. Louis, 1968. In press.

SZENTAGOTHAI. T. and A. ALBERT. The svnaDtoloev of Clarke’s column. Acta morbh. Acad. Sci.

17-18P.

, . Y l

Hung. 1955: 3. 43-5 1. WALLBE. L., Transfer of signals through a second order sensory neuron. Thesis. Oslo Uni- . -

versitv. 1968. 106 pp.