Spatio-temporal aspects of information processing in the cerebellar cortex

The cerebellum is a phylogenetically ancient part of the vertebrate brain, which is crucial for accurately timing complex motor tasks. Accumulating evidence suggests that the cerebellum is also involved in other tasks, such as sensory processing and cognitive tasks. The role of the cerebellum in this diverse set of tasks is heatedly debated. The cerebellum has been suggested to play a secondary role of coordination, control or timing of tasks which can be performed – albeit imperfectly – in its absence. The simple, uniform anatomy of the cerebellum suggests that it performs one type of computation, which is then used by a variety of other brain areas for a variety of different tasks. A common theme uniting many of the tasks the cerebellum has been implicated in is temporal coordination. Cerebellar lesions lead to “asynergy” – a decomposition of complex motor tasks into their individual components, and a loss of temporal coordination between these components on the 10‐500 millisecond range. The ability to estimate the duration of sensory stimuli is also compromised, on the same time scale. One mechanism which has been proposed for the ability of the cerebellum to support timing has been a clock mechanism in the inferior olive. Neurons of the inferior olive can generate subthreshold oscillations of their membrane potential. These oscillations generate in turn action potentials that are then transmitted to other brain areas and may thus support timing. This mechanism suffers from several limitations. Most severely, it is unclear how oscillations at 10 Hz, typical of the inferior olive, may underlie timing on intervals as short as 10 milliseconds, significantly shorter than the cycle duration. My PhD thesis addresses the question whether – and how – the cerebellum and inferior olive may support timing on the entire range implicated by behavioural studies. The work is composed of three parts. The first part deals with neuronal activity in the cerebellar cortex of awake, behaving rats that result from oscillations in the inferior olive. The second part I study the source simple spikes in Purkinje cells, the only output neurons of the cerebellar cortex, using a combination of electrophysiology and voltage‐sensitive dye imaging. In the third part, I propose a model for the olivo‐ cerebellar circuit as a temporal pattern generator. In the first results chapter I demonstrate for the first time the ability of the inferior olive to support timing on a time range shorter than the cycle duration. Inferior olivary neurons innervate the cerebellar cortex, and especially Purkinje cells. The dendrite of each Purkinje cell is engulfed by the ii climbing fibre axon of a single inferior olive neuron. Activity in the climbing fibre is translated into at least one Purkinje cell output spike, and a prolonged depolarisation and calcium influx into its dendrite (a “complex spike”). In the awake, behaving animal it is impossible to record activity directly from the inferior olive. Inferior olivary output can nonetheless be monitored by recording complex spike activity from Purkinje cells. To monitor inferior olive output, I used rats implanted with chronic arrays of 32 tungsten electrodes in the cerebellar cortex. Activity was monitored simultaneously in many sites in the cerebellar cortex, over an area of 1 x 2 mm. To enhance and regularise olivary output, rats were injected with harmaline, a mono‐amine oxidase inhibitor that acts directly on olivary neurons and enhances their propensity to oscillate. I employed advanced signal analysis methods to follow the phase of oscillations on the different electrodes. Past reports have mainly demonstrated zero‐phase synchronisation between complex spikes of different Purkinje cells. In this chapter, I demonstrate that electrode pairs in the cerebellar cortex can oscillate at the same frequency but with a constant, non‐zero phase difference. These phase differences generate the time intervals shorter than a cycle duration necessary for temporal coordination. I also demonstrate that oscillation frequency is modulated synchronously in the entore recorded region on slow time scales, and that frequency elevation is correlated with the behavioural state of the rat. During frequency changes, no significant change occurs in the phase difference between electrode pairs. The invariance of phase differences to frequency changes allows the olivo‐cerebellar circuit to generate temporal patterns that may then be replayed at different speeds without distortion. This ability may be crucial for time warping of complex motor tasks. In the second results chapter, I demonstrate that Purkinje cell simple spikes are intrinsically generated, and can therefore not carry the timing signal of the olivo‐cerebellar circuit. Each Purkinje cell receives inputs from about 200,000 parallel fibres. The prevalent dogma is that parallel fibre activity underlies Purkinje cell simple spikes. To study the relationship between Purkinje cell simple spikes and the network activity, I performed extracellular recordings of Purkinje cell activity simultaneously with voltage‐sensitive dye imaging of the surrounding cerebellar cortical area, in anaesthetised rats and guinea pigs. Parallel fibres were electrically stimulated as a control for the voltage‐sensitive dye signal. This direct stimulation yielded a significant, easily identifiable signal in the fluorescent signal (1% change). To study the relationship between parallel fibre activity and Purkinje cell firing, I collected data simultaneously from the imaging and electrophysiology, and averaged the imaging signal around simple spike times. Such averaging has previously been used to uncover structure of neuronal activity in the visual cortex. Despite the quality of the evoked optical signal and the massive averaging, no spatio‐ temporal structure was revealed in the average activity preceding simple spikes. This supports the view iii that simple spikes are intrinsically generated, and are largely unrelated to the ongoing activity in the parallel fibres. This suggests that the timing signal of the olivo‐cerebellar system has a source other than Purkinje cell simple spike firing. In the third results chapter I present a model of olivo‐cerebellar temporal pattern generation. This model runs against the prevalent dogma, according to which the cerebellar cortex determines cerebellar output, while the inferior olive is merely an internal “teacher”. The model suggests, rather, that the cerebellar output is governed by the temporal patterns generated by inferior olive oscillations. The cerebellar cortex, in contrast, is suggested to be an orchestrator of the inferior olive, by dynamically changing the coupling within the inferior olive to generate different output patterns. This model incorporates a variety of physiological observations hitherto unaccounted for, and provides several clear predictions that may easily support or refute it by simply‐designed experiments. In conclusion, my thesis contributes both positive findings (phase differences in olivary activity) and negative findings (simple spikes are independent of parallel fibre activity), that together help to establish a new and testable model of olivo‐cerebellar function.