Sequential Stimulation of Guinea Pig Cerebellar ...

Naturwissenschaften 82, 201-203 (1995)

© Springer-Verlag 1995

Sequential Stimulation of Guinea Pig Cerebellar Cortex in Vitro Strongly Affects Purkinje Cells via Parallel Fibers D. Heck Max-Planck-Institut fur biologische Kybernetik, D-72076 Tubingen The cortex of the cerebellum is distin

guished from other cortices by its large

population of thin unmyelinated fibers

(the parallel fibers) which reach from

each point in the cerebellar cortex a few

millimeters to the left and to the right [4,

8, 14). Parallel fibers originate from a T

shaped bifurcation of ascending axons of

granular cells. These relay signals from

one of the main systems of afferent fibers,

the mossy fibers, to the Purkinje cells.

The latter are the only output cells of the

cerebellar cortex. The parallel fibers con

tact the flattened dendritic trees of the

Purkinje cells by synapses which have

been shown to be excitatory [6). Purkinje

cells receive input from nearly 200000

parallel fibers [8). These synapses are

the large majority of synapses in the

cerebellar cortex. Parallel fibers also con

tact the dendrites of stellate and basket in

terneurons, the inhibitory axons of which

are oriented at right angles to the parallel

fibers.

The dispersion of signals from each

parallel fiber onto rows of Purkinje cells situated at various distances from the origin has been interpreted in various ways. One suggestion was that the whole arrangement transforms time intervals in to space intervals and vice versa, and therefore acts as a set of delay lines [3). Naturwissenschaften 82 (1995)

However, the delays produced by in dividual parallel fibers are too short to be of any use in motor control, a function which is commonly attributed to the cerebellum. A different interpretation proposed for the system of parallel fibers is that of a tissue globally conducting signals at a fixed speed [2). In fact, the anatomy suggests that an ordered sequential input in the mossy fiber system "moving" across the cerebel lar cortex in the direction of the parallel fibers at their speed of conduction should produce something like a tidal wave of spikes in the parallel fiber system, i.e., a very strong excitation of the target neurons (Fig. 1). This may explain why other authors, working with static input from the mossy fiber system [1, 16, 17), found so little in fluence of parallel fibers on Purkinje cells that the effectiveness of parallel fiber synapses was all together called in ques tion [1, 12, 13]. I tackled this issue experimentally using an in vitro slice preparation of guinea pig cerebellar cortex (Figs. 2, 3). Acute slices were prepared from adult guinea pigs (500-600 g). Animals were decapitated under ether anesthesia and the cerebellum was quickly excised. Slices were cut by means of a vibratome (Campbden, UK)


in a transversal direction, so as to contain long stretches of parallel fibers. They were stored at room temperature in a holding chamber that contained artificial cerebro spinal fluid (ACSF) containing (ruM): NaCI (124), KCl (5), NaH 2 P0 4 (1.2), MgS0 4 ' 7 H 20 (1.3), NaHCO] (26), CaCl 2 ' 2H 2 0 (2.5) and d-glucose (10) continuously gassed with carbogen (95070 02' 5% CO 2), Recordings were carried out in a separate chamber where the slices lay fully submerged at 36°C. Microelec trodes for intracellular recordings were filled with 3 M KCl and 2% neurobiotin (Vector) and had resistances of 70-120 MO. Neurobiotin was injected into the cell with depolarizing current pulses for the later visualization of the cell [lll. After insertion of the stimulat ing electrodes into the granular layer, the white substance between the stimulating and the recording site was cut in order to prevent excitation of Purkinje cells via subcortical fibers (Fig. 3). The effects of the stimuli on Purkinje cells could no lon ger be observed when the glutamatergic transmission was blocked by addition of 20 11M of the selective non-NMDA an tagonist CNQX (6-cyano-7-nitroquinoxa line-2,3-dione) and 200 11M of the selec tive NMDA antagonist AP-5 (2- amino-5 phosphonopentanoic acid) to the bathing medium. Sequential stimuli were delivered to thc granular layer of the slices by means of an array of electrodes, as illustrated in Fig. 2. A more detailed description of the generation of "moving" stimuli is given in [10). When recording the activity of parallel fibers extracellularly in the molecular lay er, strong presynaptic activity could be observed when the sequence of stimula 201

tion imitated movement of the input towards the recording site [9]. The record ed amplitudes of the signals were greatest for "velocities" between 0.3 and 0.5 m/s. The optimal velocity is in the range of the

A

velocities of conduction in parallel fibers as measured directly in various animals

{5,6, 15]. However, extracelluar recording is not sufficient to eliminate doubts [1, 12, 13] about the effectiveness of parallel fibers on Purkinje cells, I therefore impaled a number of Purkinje cell bodies with microelectrodes in order to observe the ef fects of the "moving" stimuli on their membrane potential. Purkinje cells in

i¥FfflI

slices cut in the transverse plane are less viable than in slices cut in the sagittal plane. Still, in a first series of experiments seven cells could be recorded for a long enough time to test their responses to stimulation of the granular layer with the electrode array. Five out of these seven cells did not respond or responded only weakly to stimulation of the granular lay er, indicating that only a few or none of the activated parallel fibers reached the

////// o Imsl

-5

-4

-3

-2

-1

B

~ o

/ / / -,/ / / lmsl

-2.5

-2

-1.5

-0.5

Fig. I. The granule cells and their axons that bifurcate in the molecular layer to form the so called parallel fibers. The box on the right rep resents a Purkinje cell dendritic tree receiving excitatory input from the parallel fibers. Black and white arrows indicate the traveling direc tion and the position of parallel fiber action potentials as they would appear at time zero when a sequential stimulation of the granular cells was started some ms before. The figures below each granular cell indicate the times in the past (expressed in ms) when an incoming stimulus excited that cell. For simplicity, we assume in the following that the delay in troduced by the rising part of the granular cell axon is negligible. A) Sequential mossy fiber input "moves" along the granular layer at the speed of conduction in parallel fibers. In this case, each new input excites a granular cell in a certain position at the moment when the previ ously elicited spike traveling in the molecular layer (in the direction of the "moving" input, black arrows) reaches that position. Thus, every new input adds to the wave of spikes al ready traveling in the molecular layer. The total number of jointly traveling spikes increases over a distance corresponding to the length of one branch of a parallel fiber and saturates for longer distances. In contrast, spikes traveling on the parallel fiber branches running in the opposite direction (while arrows) are widely dispersed. B) If the velocity of the "moving" mossy fiber input is greater than that of con duction in parallel fibers (e.g_, twice as fast), no wave of synchronous spikes will build up. The same is true for velocities slower than those of parallel fibers

202

~

Gl

PCl

000

'-------------------------"""Ml

Fig. 2. The layer of granular cells was stimulated electrically through a comb of II electrodes (S/-SI/) arranged in a row parallel to the direction of the parallel fibers, with an interelectrode spacing of 130-140 flm. By sequentially switching the stimulus current from one electrode to the next, a "movement" of the stimulus could be imitated. Inversion of the sequence produced a "movement" in the opposite direction. Different "velocities" could be generated by altering the time interval between consecutive stimuli. During both the intra- and extracellular experiments, the responses to different "movement" velocities and to the two opposite directions were recorded. (S/-SI1 stimulating electrodes, R recording electrode, ML molecular layer, PCL Purkinje cell layer, GL granular layer)

I+- .1.1 mm -.-., PC

.~'

Fig. 3. Nissl-stained slice of the cerebellar cortex as used in the experiments. Two separate pieces of the granular layer are visible. They appear as two dark bands crossing the picture from left to right. The upper strip of granular layer was stimulated with an array of stimulating electrodes. At the sites where the electrodes had penetrated the slice, white cell-free islands can be seen in the Nissl picture. The intracellularly recorded Purkinje cell, the responses of which to "moving" input are described in the text and shown in Fig. 4, was filled with neurobiotin for later staining. In the section of the slice shown, the stained Purkinje cell is only weakly visible. Its position in the molec ular layer is marked by the arrow (PC). The distance between the closest stimulating electrode and the Purkinje cell is indicated in the picture and was 1.1 mm in this experiment. The white substance was cut along the scattered line in order to prevent signal conduction through subcortical fibers Naturwissenschaften 82 (1995)


TOWARDS

AWAY

A

0,35 m/s

B

0.35 m/s

E

10 ms

observed cell. This may have geometrical reasons (e.g., cuts were not exactly parallel to the direction of parallel fibers) or phys iological reasons, i.e., that a large fraction of granule cells did not survive the slice preparation. In two cases, Purkinje cells strongly responded to stimulation of the granular layer with the electrode array and their responses to "moving" inputs could be tested (Fig. 4). As was to be ex pected from the extracellular data, this difference was greatest for "velocities" at or near the parallel fiber conductance velocity, which was around 0.35 m/s in the example shown in Fig. 4. When the stimulus-induced excitation depolarized the cells above threshold, spikes were elicited. Together with earlier findings [9], the data described in this paper argue strong ly in favor of an interpretation of the cerebellar cortex as a sequence-address able memory. The increased simple spike activity of selected Purkinje cells represents the con tribution of the cerebellum to the control of the movement, the planning and/or ex ecution of which caused the occurrence of the sequential mossy fiber activity. The interpretation of the cerebellar cortex as a detector for "moving" activity in the mossy fiber system [2] offers an elegant explanation of the geometry of the parallel fiber system. In addition, the Naturwissenschaften 82 (1995)

Fig. 4. The responses of a Purkinje cell to input "moving" at two different velocities (0.35 m/s in A- D and 0.5 m/s in E, F) towards the recorded Purkinje cell and away from it (A, B, C, D, E, F). In each example, a superposition of ten consecutive trials is shown. The initial rectangular downward deflection is due to the overlap of the 11 consecutive stimulus artifacts, the negative peaks of which are clipped. In this experiment, the distance between the Purkinje cell and the nearest stimulating electrode of the comb was about 1 mm. A, B) The different effects of the two "movement" directions were sometimes as clear as shown here. The speed of the "movement" was 3.5 mis, and thus close to the speed of con ductance in parallel fibers. No spikes could be elicited with "movements" away from the recording site (A), whereas "movement" towards it always produced an action potential (B). C, D) The experimental paradigm is the same as in (A, B) but the cell was more excitable during this period. Under these conditions stimuli "moving" away from the cell sometimes also triggered action potentials (C). As before, the "movement" towards the cell was clearly more effective in exciting it (D). There is, however, an additional interesting effect to be observed here: during this period of higher excitability the spikes produced by "movements" towards the cell are time-locked to the stimulus (six out of ten spikes lie within a period of less than 2 ms) (D). E, F) As was to be expected (see Fig. 1), at a vclocity (0.50 m/s) slightly higher than that of parallel fiber conductance, the effects of input "moving" towards the cell become weaker. Also the dependence on direc tion is less pronounced, even if the stimulus "moving" away from the cell (E) was still less effective than the stimulus "moving" towards it (F)

puzzling weak effect of single parallel fibers on Purkinje cells now finds an ex

planation. It may be an important prere

quisite for filtering out noise or "nonop timal" mossy fiber sequences. These experimellls, together with the one of Garwicz and Andersson [7], rehabil itate the parallel fibers as the main intra cortical relay between input and output in the cerebellar cortex. Ascending axons of granular cells may have the effects which other authors ascribe to them [1, 12, 13], but under dynamical conditions (which may be a specialty of the cerebellum) their effect is certainly overshadowed by that of their long horizontal branches, the parallel fibers.

I am very grateful to V. Braitenberg, A. Borst, and F. Sultan for constructive com ments, ideas, and help with the manu script. Thanks are also due to M. Dorte mann, 1. Hemasi, V. Staiger, and U. Seipt for excellent technical assistance and to S. Wurth for proof-reading.

Experiments comply with the Principles of Animal Care, publication No. 85-23, revised 1985 of the National Institutes of Health and also with the laws of the respective country in which the experiments were performed, the German Law on the Protection of Animals.


Recei ved August 17 and December 12, 1994

1. Bower, 1. M., Woolston, D. C: J. Neurophysiol. 49, 745 (1983) 2. Braitenberg, V.: J. Theor. Neurobiol. 2, 237 (1983) 3. Braitenberg, v., Atwood, R. P.: 1. Compo Neurol. 109, I (1958) 4. Brand, S., Dahl, A. L., Mugnaini, E.: Exp. Brain Res. 26, 39 (1976) 5. Crepel, F., Dhanjal, S. S., Garthwaite, 1.: 1. Physiol. (Lond.) 316, 127 (1981) 6. Eccles, J. C, L1inas, R., Sasaki, K.: Exp. Brain Res. 1, 17 (1966) 7. Garwicz, M., Andersson, G.: ibid. 88, 615 (l992) 8. Harvey, R.1., Napper, R. M. A.: J. Compo Neurol. 274, 151 (l988) 9. Heck, D.: Neurosci. Lett. 157, 95 (1993) 10. Heck, D.: 1. Neurosci. Meth. (in press) 11. Horikawa, K., Armstrong, W. E.: 1. Neurosci. 25, 1 (l988) 12. L1inas, R., in: The Cerebellum, New Vistas, p. 189 (S. L. Palay, V. Chan-Palay, eds.). New York: Springer 1982 13. L1inas, R., Sugimori, M., in: The Cerebellum Revisited, p. 167 (R. L1imis, C. Sotelo, eds.). New York: Springer 1992 14. Mugnaini, E.: 1. Compo Neurol. 220, 7 ( 1983) 15. Nicholson, C, Llinas, R.: 1. Neurophysiol. 34,509(1971) 16. Shambes, G. M., Beermann, D. H., Welker, w.: Brain Res. 157, 123 (l978) 17. Sharp, F. R., Gonzales, M. F., Sharp, 1. w., Sagar, S. M.: 1. Compo Neurol. 284, 621 (1989)

203