PNJgersArchiv

Pfltigers Arch. 380, 79- 84 (1979)

EuropeanJournal of Physiology 9 by Springer-Verlag1979

Timing of Bilateral Cerebellar Output Evoked by Unilateral Vestibular Stimulation in the Frog N. Dieringer and W. Precht* Neurobiologische Abteilung, Max-Planck-Institut ffir Hirnforschung, Deutschordenstrasse 46, D-6000 Frankfurt/M., Federal Republic of Germany

Abstract. Electrical stimulation of one V I I I t h nerve evoked simple spike activity in Purkinje cells located on either side of the cerebellum. This cerebellar output was delayed by ca. 10 ms with respect to its mossy fiberparallel fiber input. The onset of the cerebellar output occurs on the average simultaneously on either side of the corpus cerebelli. The delay is explained by slowly rising EPSPs in PC induced by primary afferent and by second and higher order vestibular fibers. The latter inputs are stronger and terminate ipsi- and contralaterally in the granular layer. Key words: Cerebellum - Vestibular input - Onset of output - Delay line.

across the cerebellum between the bilateral auricles [19]. In this region many PC respond to electrical and natural vestibular stimulation of either side [t, 16]. Since the contralateral cerebellar cortex receives in addition to the transversing parallel fiber input also excitation from secondary vestibular neurons crossing the midline in the cerebellar commissure [12] the time of firing of contralateral PC will depend on the relative strength of each of these inputs and on the time of their arrival at a given PC. It was the aim of this study to find out whether the proposed delay line with a linearly increasing latency of the onset of PC is restricted to neighboring PC or whether it extends also into the contralateral half of the cerebellar dorsal rim.

Methods Introduction Time is a crucial parameter in the coordination of muscular activity. Since the cerebellum has long been assumed to be involved in movement coordination the timing of the output of Purkinje cells (PC) to a given input is of considerable interest. Braitenberg's hypothesis [2] that the cerebellar cortex functions as a kind of clock, implies that afferent cerebellar impulses propagate along beams of parallel fibers and thereby activate sequentially PC located on that beam. For neighboring PC (150 ~tm apart) it was shown that their time of firing differs by an interval determined by the conduction time of the parallel fibers connecting them [9]. Since a substantial number of parallel fibers seem to transverse the entire width of the cerebellar cortex [8] a mossy fiber input restricted to one side would activate PC on the contralateral side significantly later than those on the ipsilatera[ side due to the slow conduction velocity of parallet fibers (0.2 m/s; [9]). In the cerebellum o f the frog, the vestibular input is largely restricted to the posterior rim which extends * Supported by Deutsche Forschungsgemeinschaft(Pr. 158/1)

Experiments were carried out with 15 grassfrogs (Rana temporaria) immobilized with d-tubocurarine (0.5-0.9 rag/animal). Dissection, stimulation and recording procedures were similarto those described earlier [5,6,20]. Intensity of stimulation of the anterior branch of the VIIIth nerve was expressed in multiples of the threshold of the N 1 field potential (TN1)recorded in the ipsilateralvestibularnucleus and was restricted to 5 x TN1. Currents necessaryfor TN 1were similarly small (about 15 gA; 0.1 ms) in all experiments. The cerebellumwas positioned with the molecular layer upward and the surface was covered with a small drop of paraffin oil to prevent drying out of the surface and to increase isolation for recording of evoked field potentials. Responseswere averagedwith a specialpurpose computer (Nicolet 1074). PC were identified by the presence of complex spikes [15]. Intracellular records were obtained with aged K-citrate filled glass pipettes (10-20 MO) and extracellular recordings of field potentials were performed with 4 M NaC1pipettes having resistances of ca. 5 MO.

Results A. Field Potentials Electrical stimfilation of the VIIIth nerve evoked at the surface of the corpus cerebelli (molecular layer, ML) a long lasting ( > 25 ms) negative potential (N3) that partially overlaps with a positive field potential (P-

0031-6768/79/0380/0079/$01.20

80

Pflfigers Arch. 380 (1979)

N3

r

Fig. 1 A and B

_V[ '/,mac' AMPLITUDE(jJV)

DEPTH

?

....

,pm 200 ~

'

lOB a

u

11}1}

,,

B ~'~J~'~

-lOB a

-200

/

o

-300

i

J

f

31}0

~00 ~v['~,,,~ec~

i"-- %

~ 5 G 1 }

wave) with a longer latency and a shorter duration (Fig. 1). The amplitude of the P-wave was largest on either side of the middle of the corpus cerebelli. In the granular layer (GL) a large negative field potential is evoked. Its peak has the same latency as the peak of the superficially recorded P-wave (Fig. 1, A, B). The latency of onset of the negativity in the G L was 2.5 to 3.0 ms shorter than that in the M L ( N 2 - N 3 in Fig. 1 A). The depth profile of these evoked potentials shows that the P-wave reverses at about 200 gm below the surface (Fig. 1 B). Between 250 and 400 gm usually evoked spikes (PC-discharge; see below) were superimposed. The characteristics of the evoked field potentials are in good agreement with earlier description and interpretations of mossy fiber field potentials elicited by electrical stimulation of the cerebellar white matter [15]. Following the interpretation of these authors the Pwave at the surface can be explained by the generation of passive sources to active sinks occurring in the region of the s o m a of PC which generate action currents, These sinks can be seen in the PC layer and in the G L as a negative potential (N4) following the N 2 potential. Thus P-wave and N 4 potential are simultaneous events and can be used as indications of PC activity. The latencies of the field potentials depended on the stimulus intensity. To allow a comparison between animals, we therefore used the same relative stimulus

b

Field potentials recorded in the contralateral cerebellar corpus following stimulation of vestibular nerve. A Field potentials in the granular layer (500 [am deep) are only negative in polarity (N2, N4). Field potentials in the molecular layer (50 ~tm deep) consist of a negative potential (N3) and a positive potential (P). Note the difference in latency of N2 and N3 and the similarity in latency of the peaks of N4 and Pwaves. B Laminar field potentials recorded at A in steps of 100 gm. Note the reversal of the superficiaI P-wave around 200 •m and the small ripples superimposed on the reversed Pwave at a depths of 300- 400 gin. Lines mark the time at which the amplitudes were measured for the plot on the right. Dashed line indicates the field generated by activation of the granule cell-paralM fiber pathway. The second line indicates the field potentials generated by Purkinje cell activation. Each record is an average of 16 responses

intensity (5 x the threshold of the N 1 potential recorded in the ipsilateral vestibular nucleus). As can be seen in Fig. 2A the P-wave recorded at the surface of the corpus cerebelli half way between midline and auricle has about the same amplitude and latency when evoked by stimulation of the ipsi- or contralateral vestibular nerve. The preceding N3 potentials had latencies of 2 . 5 - 3.0 and 7 . 0 - 8.0 ms after stimulation of the ipsiand contralateral vestibular nerve, respectively (arrows in Fig. 2). This latency difference of 5 ms is too short to allow for a spread of activity along parallel fibers. With a conduction velocity of 0.2 m/s the volley in parallel fibers would need 7.5 ms to travel the distance of 1,500 gm across the cerebellar midline. Furthermore, the latency of the contralateral N3 potential recorded close to the midline was even longer (about 10 ms) than half way between midline and auricle on the contralateral side. Stimulation of the contralateraI vestibular nerve evoked an Nz potential with a latency of 5 - - 6 ms (Fig. i A) indicating a separate mossy fiber input to the contralateral corpus cerebelli, Cutting either the cerebellar midline or the contralateral cerebellar peduncle reduced the amplitude of both N and P-waves but in neither case abolished them. We, therefore, conclude that in addition to a possible spread of activity across the cerebellar midline via parallel fibers at least two other inputs contribute to the activation of con-

N. Dieringer and W. Precht: Timing of Cerebellar Output to Vestibular Inputs

tralateral PC: a cerebellar mossy fiber commissure and a brain stem commissure. The latter enters the cerebellum via the contralateral peduncle. This mossy fiber pathway crossing the midline in the brain stein does not seem to involve contralateral vestibular neurons, since these neurons are rarely activated in the intact frog [5]. In the cerebellar auricle a P-wave cannot be detected probably because of the different geometry compared with the corpus. On the ipsilateral side two negative field potentials were observed on the surface (Fig.2B). The first field potential had a similar short

81

latency (0.7 ms) as in the study of Precht and Llinfis [19]. The onset of the second negativity occurred after ca. 2.5 ms and the latency of its peak was slightly shorter than that of field potentials evoked by stimulation of the contralateral VIIIth nerve (Fig. 2B). The latencies of these evoked P-waves and N4 potentials indicate that PC on either side of the cerebellar corpus are activated after similar latencies of about 17 ms (Fig. 2A). Furthermore, the similarities of the amplitudes of the P-waves evoked at the same recording point after stimulation of the • and of the contralateral VIIIth nerves (Fig. 2 A) suggest that about equal numbers of PC in the corpus may respond to either stimulus.

B. Recordingsfrom S&gle Purkinje Cells

A

Fig. 2. Field potentials recorded in the molecular layer of the cerebellar corpus (A) and auricle (B) following stimulation of the vestibular nerve on the • and contralateral side respectively. A Stimulation on either side induces a short latency, small negative potential that is followed by a long lasting negative and positive (upward) potential. Note the similarity in latency and amplitude of the large P-waves generated by activation of Purkinje cells. The preceding negative potentials start earlier after ipsilateral than after contralateral stimulation (arrows). B In the molecular layer of the auricle only negative field potentials are observed. Again the field potentials evoked by ipsilateral stimulation start earlier than those evoked by contralateral stimulation. The peaks of these two potentials, however, have a very similar latency. Each record consists of an average of 16 responses. Calibration pulses indicate 1 ms, +0.2 mV

Stimulation of the VIIIth nerve evoked only simple spikes with a short and consistent latency. Evoked complex spikes had latencies of 80-120 ms. The majority of PC (about 70~) were activated from vestibular stimuli on either side. About equal numbers of PC on either side of the cerebellum (8 ~o) stopped firing for a brief period (50-100 ms) after stimulation. The onset of firing of PC measured in the cumulative frequency distribution (averages of 2 0 - 40 trials) were obtained in 5 different compartments of the posterior rim. As shown in Table 1 latencies were shortest in the ipsilateral auricle and longest around the midline. Only this latter difference was statistically significant (P < 0.005). In all other parts latencies were quite similar. Even in the ipsilateral auricular lobe the shortest latencies of PC were as long as 10 ms and neither antidromically driven PC nor short latency complex spikes were observed. These results are in good agreement with the values obtained by analysis of the evoked P and N 4 field potentials. However, there is a considerable delay of some 10 ms between the onset of the N3 potential and the onset of firing of PC. A similar delay was also observed for spinal input in the cerebellum [4]. In both cases this delay can be explained by temporal and spatial summation of the synaptic input. Intracellular records (Fig.3B) showed small, slowly rising EPSPs. The onset of these EPSPs occurred about 5 - 1 0 ms

Table 1. Onset of firing in Purkinje cells Location in cerebellum

•

auricle

Number of cells

20

Latency (ms)

14.6 • 4.0

•

corpus

midline

contra, corpus

contra. auricle

18

10

22

23

18.6 _+3.4

21.3 • 2.6

19.4 • 3.9

19.3 • 3.0

82

Pflfigers Arch. 380 (1979) Fig. 3

0

J

3 mV

B

mV

1 j

20msec

2:inl I,Ii~i

_ ll.llll

1-

lIlll_l~mv

----r ~ I T F I T F I ! 200msec

TITi[-~

10

2 C l

2

11111111111111111111 I[111ti111111111111 2o

earlier than the extracellularly measured spike discharge. Intracellular recordings from PC that showed suppression of resting discharge following stimulation did not reveal clear IPSPs even in cells with low resting potential and injury discharge. Similar responses occur to a larger degree in PC of chronically hemilabyrinthectomized frogs [6]. Also in that preparation IPSPs were never observed. Subsequent to impalement of the cell some neurons stopped firing for a brief period after the stimulus (Fig. 3E), others only stopped when their firing rate was artificially reduced by injection of hyperpolarising current.

C. Responses to Natural Phasic Inputs To approximate the electrically induced synchronized activity in mossy fibers b y a more natural stimulus pattern we rotated the frog on a turntable horizontally at constant velocities and then stopped rotation rapidly (100~ No significant differences were found in the latencies of the onset of these postrotatory responses between type I, II and III PC. These latencies (96.5 + 16.3 ms) were significantly (P < 0.005) longer than those in type I neurons in the vestibular nucleus (70.0 _+11.6 ms). One can therefore assume, that the cerebellar loop is too slow to affect the onset of evoked reflex activity.

,

A

-

JmV

Responses of Purkin3e cells to stimulation of the VIlIth nerve. A and B Extra- and subsequent intracellular record from a Purkinje ceil responding with simple and complex (asterix) spikes. In B the response was recorded with two

different sweep speeds and gains (trace ] and 2) to allow comparison of the evoked EPSPs from parallel and climbing fibers. Note the small and slowlyrising EPSP induced by parallel fiber activity. C Extracellularrecord taken after withdrawal of the electrodefrom the neuron (calibration as in BI). D and E: Extra- and intracellular records from another Purkinje cell. The spontaneous activity of the neuron (26.7 Hz extra-; 46.2 Hz intracellularly) is suppressed after stimulation. Note the absence of a hyperpolarising potential in the intracellular record (E). Traces A, C, D1 and El: high gain A.--C. (300 ms time constant) and traces B, D2 and E2 low gain D.--C. coupled amplifiers

Discussion

The results of the present study show that vestibular signals reach the cerebellar granule cells via at least three different pathways: (1) primary afferent fibers terminate mainly in the ipsilateral auricular lobe; (2) second order vestibular neurons synapse in the ipsilateral and after crossing the cerebellar midline in the contralateral gr~inular layer; (3) second and higher order vestibular neurons reach the cerebellar granule layer after crossing the midline in the brain stem. These results are in good agreement with recent anatomical studies suggesting that vestibular fibers terminate as mossy fibers in the amphibian cerebellum [7, 1 0 - 1 2 , 18]. N o short latency complex spikes were evoked as was previously reported for a few PC in the bullfi'og [14]. This discrepancy could either be due to a small sampling number in this study or due to a species difference. However, a similar negative result was obtained by Cochan and Hackett [3] in their study of the much larger auricular lobe of the bullfrog's tadpole. Therefore, it may well be possible that these few short latency climbing fiber responses were evoked by current spread exciting intracranial climbing fibers en route to the cerebellum. Hillman's anatomical studies [11], however, indicate that there may indeed, be a few vestibular climbing fibers, reaching the auricular lobe. F r o m a functional point of view this scarce input may well be neglected.

N. Dieringer and W. Precht: Timing of Cerebellar Output to Vestibular Inputs

As for the somesthetic primary afferent input projecting directly to the ipsilateral cerebellar granule layer [21] excitation of primary afferent vestibular fibers is not strong enough by itself to activate many PC. However, this input can clearly facilitate the output of auricular PC as can be seen from the shorter latencies of onset of firing in PC of this region and from the early onset of the evoked EPSPs. Due to the slow rise of these EPSPs considerable temporal and spatial summation has to occur in order to activate a given PC. Thus it is the subsequent input from second order and polysynaptic fibers that finally provokes firing in PC. The delay between the onset of the N 3 field potentials and the onset of firing in PC is thus accounted for by the slowly rising EPSPs induced in PC. The majority of PC encountered (70 %) were activated from the VIIIth nerve on either side. A similar high degree of convergence was found with natural stimulation [1, 16]. Most of this convergence seems to occur already at the level granule cells. Even though spread of input across the cerebellar midline via parallel fibers cannot entirely be ruled out, its contribution seems to be small, since in the cerebellar midline the Na potential started later than expected of a parallel fiber volley set up in the ipsilateral auricular lobe. Intracellular recordings from PC that were suppressed in their resting activity after stimulation of the VIIIth nerve failed to demonstrate IPSPs during the period of suppression. In some cases suppression was only observed when the spontaneous activity of the cell was reduced by artificial hyperpolarisation. We therefore assume that most of this suppression is due to a disfacilitation rather than generated by inhibitory interneurons which are rare in the cerebellar cortex of the frog [13]. Admittedly, remote inhibition cannot entirely be excluded. This interpretation is supported by observations in chronically hemilabyrinthectomized frogs, where a considerably larger number of PC was suppressed without clear underlying IPSPs [6]. In the latter preparation vestibular nerve stimulation induced in a larger number of vestibular neurons inhibition than in control animals. Assuming, that at least part of these inhibited vestibular neurons contribute to the cerebellar input, some of the disfacilitation in PC would be accounted for by the reduction of input from these neurons. Even though latencies as well as response patterns were highly variable from cell to cell as well as in a given PC to stimulation of the ipsi- or the contralateral VIIIth nerve, the response latencies do not fit into the concept of a delay-line expanding across the cerebellar midline. Instead, the onset of firing in PC located in either corpus cerebelli occurs on the average about simultaneously and the amplitudes of the outputs are likewise similar. In the auricular lobe, however, onset of

83

firing of PC is shorter after stimulation of the ipsilateral VIIIth nerve and the number of PC receiving convergent inputs from both labyrinths is smaller than in the corpus cerebelli. Most of the PC axons terminate in the homolateral cerebellar nuclei and in the brain stem [22]. Part of these PC inhibit vestibular neurons [6, 17]. At present the detailed functional connectivity of PC projection is unknown so that we have to refrain from detailed speculations as to the mechanism of action of this apparent symmetrical output. In general it may be said, however, that this cerebellar output occurs too late to influence the onset of vestibularly, somatically or visually (unpubl. observations) evoked reflex activity mediated by or controlled in the brain stem. Only the later part of the reflex activity and its duration can be controlled by the bilateral cerebellar output. During a vestibularly evoked head movement for instance, this later part of reflex activity will of course be influenced by additional vestibular and somesthetic inputs provoked by the movement itself. If the initiated reflex activity would tend to overcompensate, the head would only reach a stable position after a few oscillations. A bilateral simultaneous cerebellar output could dampen such possible oscillations, if the timing of its output would be matched with the timing of the movementprovoked secondary sensory inputs.

References 1. Blanks, R. H. I., Precht, W., Giretti, M. L.: Response characteristics and vestibular receptor convergence of frog cerebellar Purkinje cells. A natural stimulation study. Exp. Brain Res. 27, 181-201 (1977) 2. Braitenberg, V.: Functional interpretation of cerebellar histology. Nature 190, 539-540 0961) 3. Cochran, S. L., Hackett, J. T. : The source of climbing fibers to the vestibulo-cerebellum of the tadpole. Soc. Neurosci. Abs. III, VII Ann. Meeting 163 (1977) 4. Dieringer, N. : Responses of Purkinje cells in the cerebellum of the grassfrog (Rana temporaria) to somatic and visual stimuli. J. Comp. Physiol. 90, 409-436 (1974) 5. Dieringer, N., Precht, W.: Mechanisms of compensation for vestibular deficits in the frog. I. Modification of the excitatory commissural system. Exp. Brain Res. 36 (1979a) 6. Dieringer, N., Precht, W. : Mechanisms of compensation for vestibular deficits in the frog. II. Modification of the inhibitory pathways. Exp. Brain Res. 36 (1979b) 7. Fuller, P. M. : Projections of the vestibular nuclear complex in the bullfrog. Brain, Behav. Evol. 10, 157-169 (1974) 8. Freeman, J. A. : The cerebellum as a timing device: An experimental study in the frog. In: Neurobiology of Cerebellar Evolution and Development (R. Llinfis, ed.), pp. 397--420. Chicago: Am. Med. Assn. 1969 9. Freeman, J. A., Nicholson, C. N. : Spacetime transformation in the frog cerebellum through an intrinsic tapped delay-line. Nature 226, 640-642 (1970) 10. Gregory, K. M. : Central projections of the eight nerve in frogs. Brain, Behav. Evol. 5, 7 0 - 8 8 (1972)

84 11. Hillman, D. E. : Light and electron microscopical study of the relationships between the cerebellum and the vestibular organ of the frog. Exp. Brain Res. 9, 1 - 1 5 (1969) 12. Larsell, O. : Comparative anatomy and histology of the cerebellum from myxinoids through birds (J. Jansen, ed.). Minneapolis: Univ. Minnesota Press 1967 13. Llin~s, R.: Cerebellar physiology. In: Frog neurobiology (R. Llinas and W. Precht, eds.). Berlin, Heidelberg, New York: Springer 1976 14. Llin/~s, R., Precht, W., Kitai, S. T. : Climbing fiber activation of Purkinje cell following primary vestibular afferent stimulation in the frog. Brain Res. 6, 371-375 (1967) 15. Llinfis, R., Bloedel, J. R., Hillman, D. E. : Functional characterization of the neuronal circuitry of the frog cerebellar cortex. J. Neurophysiol. 32, 847- 870 (1969) 16. Llin/ts, R., Precht, W., Clarke, M. : Physiological responses of frog vestibular fibers to horizontal angular rotation. Exp. Brain Res. 13, 408--431 (1971) 17. Magherini, P. C., Giretti, M. L., Precht, W. : Cerebellar control of vestibular neurons of the frog. Pfltigers Arch. 356, 99-109 (1975)

Pfliigers Arch. 380 (1979) 18. Mehler, W. R.: A comparative anatomical survey of the vestibular nuclear complex in submammalian vertebrates. In: Progress in Brain Research. Vol. 37 (A. Brodal and O. Pompeiano, eds.), pp. 55-67. Amsterdam: Elsevier 1972 19. Precht, W., Llings, R. : Functional organization of the vestibular afferents to the cerebellar cortex of the frog and cat. Exp. Brain Res. 9, 30-52 (1969) 20. Precht, W., Richter, A., Ozawa, S., Shimazu, H. : Intracellular study of frog's vestibular neurons in relation to the labyrinth and spinal cord. Exp. Brain Res. 19, 377-393 (1974) 21. Rushmer, D. S.: Electrophysiological evidence for primary afferent connections in the frog cerebellum. Brain Res. 18, 560564 (1970) 22. Stern, T. A., Rubinson, K. : Effeerent projections of cerebellar cortex of Rana pipiens. Anat. Rec. 169, 438 (1971)

Received February 3, 1979