Report 2003

Report 2004

Report on the acitivities of the European CEREBELLUM consortium

Good progress was made in the first year on all workpackages and several important discoveries relevant to understanding how the cerebellum operates were made. A system for the normalization of spatial distribution data in anatomical sections was developed and applied to the study of plasticity of mossy fiber projections following peripheral nerve lesions. An extensive study was made of coupling between olivary neurons and the presence of oscillatory activity in the olivary network. Use of a transgenic animal lacking gap junction proteins forces us to reevaluate the theoretical predictions of the importance of gap junctions in olivary oscillations. It was demonstrated that spillover of glutamate plays an important role in AMPA receptor-mediated transmission at the mossy fiber synapse and these effects were modeled. Another transgenic animal demonstrated the importance of NMDA receptors in granule cells for the induction of long-term potentiation of their mossy fiber synapses. We have recorded for the first time from individual granule cells in vivo. A modeling study demonstrates that inhibition can have both multiplicative and additive effects depending on voltage noise. Anatomically correct network models of the granular layer and of the molecular layer were constructed. The firing characteristics of stellate cells were studied and were found to be quite complex. Modeling studies of the molecular layer network predict that these cells may synchronize nearby Purkinje cells, while common parallel fiber excitation does not cause any synchronization. A system for distributed non-curated databases for sharing the experimental data was developed and will be implemented in the next period. A successful cerebellum meeting was organized.

Overview

Coordinate system for anatomical data

Anatomical data are typically collected over many different studies in different brain regions. The analysis of a data set is often done in a local coordinate system suitable to the brain region under study. While this facilitates reporting of the original data it greatly impedes integration and comparison of multiple anatomical data sets obtained in different studies. We have developed and tested the use of an internal coordinate systems for mapping of anatomical data together with appropriate graphical tools for data visualization (Bjaalie, Nature Neuroscience Reviews 2002). Following 3-D reconstruction of the region in question, one applies a bounding box with a specific orientation and size determined by local landmarks and boundaries (midline, fiber tracts and surfaces, architectonic boundaries). Using linear warping, data sets from different animals can be analysed and compared in a common coordinate system. Data can be combined across animals and inspected using a JAVA viewer

Local organization of the inferior olive

Several theoretical studies have demonstrated that electrotonic coupling through gap junctions are essential to generate stable oscillations of olivary neuron. Synchrony and oscillations are thought to be important in generating synchronized climbing fiber activity in the cerebellar cortex. We have characterized olivary electrotonic coupling in normal animals and estimate that each olivary neuron is directly coupled to about 50 neurons. We studied synchronization of oscillating pairs of olivary neurons and found zero phase difference between oscillations 50% of the cases and phase differences of 20-50 ms in 25% of the cases. The latter agree with the observation of propagating waves of activity within an oscillating patch during optical imaging.

While these physiological and imaging data support the previous theoretical studies, our study of transgenic mice contradicts them. We studied transgenic mice that lack connexin36 (Cx36), which appears necessary for functional olivary gap junctions. These mice have normal motor behavior and also normal climbing fiber activity and synchrony. Moreover transgenic animals showed normal reaction to the drug harmaline which induces high frequency olivary oscillations. In conclusion, the present data does not support the notion that electrotonic coupling by gap junctions underlies synchronization of olivary spike activity or that these gap junctions are essential for normal motor performance.

In vivo activity of granule cells

Because of their small size and high density it has not been possible to date to record from granule cells extracellularly. Consequently, no data are available on the normal in vivo firing rates of the cells, which constitute the largest number of neurons in the brain, or of their axon, the parallel fibers. Such data is nevertheless essential for modeling studies.

We can now routinely record from cerebellar granule cells using the patch clamp technique in anesthetized rats. Cell-attached recordings and whole-cell recordings have shown that these neurons are usually not spontaneously active in vivo, and thus require mossy fiber synaptic input to fire. Excitatory spontaneous synaptic input exhibited periods of rhythmicity within the theta range (5-13 Hz), suggesting a role in the timing of granule cell output. Such periods of rhythmicity were discontinuous but could last for several minutes.

These results predict a very sparse coding in the granular layer.

Spillover mechanisms in the glomerulus

The glomerulus is a typical anatomical feature of the cerebellar granular layer, but its function is poorly understood. Many forms of local synaptic interaction are possible, including heterosynaptic activation due to local build-up of neurotransmitter concentrations in this tightly wrapped structure or postsynaptic interactions on single granule cell dendrites. Previous studies have mainly focused of spillover of GABA molecules, in the present work we have focused on the spillover of glutamate (DiGregorio et al., Neuron 2002). It was observed that AMPA receptor-mediated transmission at the mossy fiber synapse has two kinetically distinct components. A fast current component arises from direct release of glutamate at active zones, while a slower component arises from spillover of glutamate released on to neighboring granule cells. The spillover component is a major determinate of transmission, contributing half the charge during a single synaptic event. The spillover was simulated using the diffusion equation on a model of stochastic release at multiple active zones on the mossy fiber terminal.

Development of a non-curated distributed database system

Neuroscience is generating vast amounts of highly diverse data which is of potential interest to researchers beyond the labs in which it is collected. At present, the only widely used forms of dissemination are figures and tables in published papers which suffer from inaccessibility and the loss of machine readability. They may also present only an averaged or otherwise selected subset of the available data. Numerous database projects are in progress to address these shortcomings. They employ a variety of architectures and philosophies, each with its own merits and disadvantages. One axis on which they may be distinguished is the degree of top-down control, or curation, involved in data entry. We have developed a new approach in which there is no curation, minimal standardization, and a wide degree of freedom in the form of records used to document data (Cannon et al., Network2002). Such a scheme has advantages in the ease of database creation and in the equitable assignment of perceived intellectual property by keeping the control of data in the hands of the experts who collected it. It does, however, require a more sophisticated infrastructure than conventional databases since the software must be capable of organizing diverse and differently documented data sets in an effective way. Several components of a software system to provide this infrastructure are now in place.