Skip navigation


Glomeruli of Cerebellar Cortex: Computation by Extrasynaptic Inhibition?

Erik De Schutter

Theoretical Neurobiology, Born-Bunge Foundation, University of Antwerp, Universiteitsplein 1, B2610 Antwerp, Belgium.

Inhibition of cerebellar granule cells has prominent tonic and spillover components due to activation of extrasynaptic receptors. A recent study shows how extrasynaptic inhibition affects information flow through cerebellar cortex.

The cerebellar cortex contains more granule cells than neurons in the rest of the brain but comparatively little is known about their function compared to the more impressive Purkinje cell. Cerebellar granule cells are, however, a popular preparation for pharmacological studies of receptors and channels. A long series of studies of GABA receptors on granule cells has now led to some fundamental insights in how a specific anatomical specialization of cerebellar cortex, the glomerulus, may work [1]. Glomeruli are formed around the large axonal terminals of glutamatergic mossy fiber afferents (Fig. 1). Each terminal is contacted by dendrites from 50–60 distinct granule cells. In addition glomeruli contain the GABAergic synapses that inhibitory Golgi cells make with granule cells, and the glutamatergic contacts between mossy fibers and Golgi cells. The structure has a radius of ~ 2.5 µm and is enwrapped by glial sheathing. Granule cells have 1-8 dendrites, each participating in a different glomerulus.


The first indication that GABAergic inhibition of granule cells had unusual properties was the discovery of a strong tonic component [2, 3]. This tonic GABA current is much larger than that evoked by spontaneous events and can be completely blocked by bicuculline. Its fraction of GABA current increases from 5% in young to 99% in mature rats [3]. An attractive hypothesis to explain this phenomenon is spillover of GABA between neighboring Golgi to granule cell synapses belonging to the same glomerulus. The initiating event would still be action potential evoked GABA. But because this release happens at relatively distant synapses delays due to diffusion and summation of multiple events will filter out the synaptic transients. If due to spillover the tonic current is expected to be blocked by TTX or low external calcium. These manipulations partially block the tonic current in young animals and not at all in adult ones [1-3]. Therefore it is assumed that the ambient GABA concentration in glomeruli can activate some GABAA receptors.

But several lines of evidence support the existence of spillover also. First, two different kinds of synaptic currents can be recorded in a granule cell following Golgi cell stimulation. Besides release failures both fast responses with time courses similar to spontaneous events and much slower responses with slow decay are present [4]. Second, granule cells in cerebellum and cochlear nucleus are the only cells expressing the ?6 subunit of the GABAA receptor [5]. Receptors containing the ??? subunit combination have a 50-fold higher affinity for GABA than other GABAA receptors, and they do not desensitize upon prolonged presence of agonist [6]. The ? subunit is found exclusively in extrasynaptic locations on the dendrites and somata of granule cells [7]. Both the tonic and the evoked slow currents have the pharmacological profile of a receptor containing ?6 and ? subunits: furosemide-sensitive but diazepam- and neurosteroid-insensitive [1, 4]. Taken together this evidence strongly suggests that the slow evoked currents are due to activation of extrasynaptic ?6?2/3? receptors by spillover of GABA from synapses activated on dendrites from other granule cells participating to the same glomerulus. The diffusion boundaries caused by the glomerular sheath may further promote extrasynaptic interaction between granule cells.

From the beginning it was assumed that the extrasynaptic activation of GABAA receptors might have an important role in cerebellar processing but until recently little experimental evidence for such a notion existed. In mature animals, where most of the spontaneous GABAA current is tonic, block of all GABAA receptors by bicuculline leads to increased responses of granule cells to current injection [3]. Much progress has now been made by using furosemide to specifically block ?6 subunit containing GABAA receptors in cerebellar slices from adult animals [1]. Using this approach it is estimated that 97% of the charge evoked during Golgi cell stimulation flows through extrasynaptic GABAA receptors! This fraction comprises also the tonic current which accounts for 75% of the charge transfer. These measurements were, however, done at 29° and tonic inhibition may be much smaller at body temperature [8]. Specific block of extrasynaptic GABAA receptors by furosemide causes a leftward shift of the firing curve of granule cells, as expected for the removal of a shunting inhibition [9], but has no effect on the excitability of other neurons of cerebellar cortex.

But how will a changed excitability of granule cells affect information transfer in cerebellar cortex? This is not easy to predict as additional properties of the circuitry need to be taken into account. An increase in granule cell activity due to block of extrasynaptic inhibition will also enhance Golgi cell activity through excitatory contacts made by parallel fibers [10], which may lead to increased synaptic GABA release. Besides its inhibitory effect on granule cells this may also activate GABAB receptors on mossy fibers. It has been shown that such receptors are activated by GABA spillover and reduce evoked mossy fiber responses in granule cells at low stimulation frequencies [8]. Multiple effects on Purkinje cells are possible as they are both directly activated by increased parallel fiber activity and inhibited by stellate/basket cells which also receive parallel fiber input. In their recent study Hamann et al. measured the effect of blocking extrasynaptic GABAA receptors with furosemide on the input and output elements of the pathway [1]. The enhanced granule cell excitability increases the number of spikes they fire in response to mossy fiber stimulation on average by 100%. Purkinje cells increase their firing frequency in response to mossy fiber stimulation also by about 100%. This is due to an increase in the size of evoked EPSPs, corresponding to a larger number of co-activated parallel fiber synapses. In conclusion, blocking extrasynaptic inhibition increases the flow of neural activity through cerebellar cortex or, conversely, the tonic inhibition normally present reduces this flow.

What may be the functional impact of such a reduced transmission of activity due to extrasynaptic inhibition? Hamann et al. [1] refer to the seminal work of David Marr on cerebellar motor learning [11] to suggest that decreasing the number of granule cells activated by mossy fiber input increases the storage capacity of the cerebellum. This is actually a simplification of what Marr really wrote, i.e. that Golgi cell inhibition should keep “the numbers of active parallel fibres ... reasonably small over quite large variation in the number of active mossy fibres” [11]. In other words, there should be a dynamic component to the inhibition of granule cell activity, stronger when many mossy fibers are active and weaker when few are firing. This was called automatic gain control by Albus [12]. The anatomy seems to favor such a role for Golgi cells as the combined direct mossy fiber and indirect parallel excitation makes them sensitive both to input to the granular layer and the resulting activity of granule cells. But at the physiological level it is more difficult to reconcile their properties with a gain control function [13]. Recordings in vivo show that Golgi cells typically fire a few accurately timed spikes in response to natural stimulation [10], followed by a long pause due to afterhyperpolarization [14]. Moreover, spike evoked inhibition by Golgi cells in vivo is strong enough to cause the theoretically predicted [15] synchronization of Golgi cell activity along the parallel fiber beam [16, 17].

The Golgi cell firing pattern may not be suitable for gain control as it would promote rebound firing by granule cells during the Golgi cell afterhyperpolarization [13] unless one can assume mechanisms which prolong the effect of each Golgi cell spike. While the synaptic GABAA channels already have relatively slow kinetics, the much slower spillover mechanism seems quite suitable to enhance a gain control function. Increased Golgi cell activity through either direct mossy fiber activation or indirect parallel fiber activation will thereby reduce excitability of all granule cells participating in a glomerulus. This would provide for rather a slow gain control mechanism which may not prevent fast swings in granule cell activity [13].

But how does the stronger TTX resistant tonic inhibition fit into this picture? It could provide a constant baseline, raising the threshold for spike initiation in granule cells. But if this is desired it seems more straightforward to reduce the intrinsic excitability of granule cells, as is the case in transgenic mice that lack the ?6 and ? subunits and have therefore no tonic current [18]. A more attractive concept is that tonic inhibition itself is regulated somehow. This could occur at the receptor side, for example by phosphorylation which depends on ? subunit expression [19], or by changing the baseline GABA concentration in the glomerulus. A reduced tonic inhibition may explain the evidence for spike mediated inhibition in vivo [13, 17] which is difficult to reconcile with only 3% of the GABAA current being spike mediated as found in vitro [1]. If tonic inhibition is regulated it could have additional effects beyond controlling information flow through the cerebellum. Long-term potentiation (LTP) of mossy fiber to granule cell synapses can only be reliably evoked in vitro when inhibition is blocked [20]. Any mechanism that reduces tonic inhibition will enhance LTP of this synapse. Of particular interest with respect to computational functions of the glomerulus is if extrasynaptic inhibition can be regulated separately in each individual glomerulus.

  1. Hamann, M., Rossi, D. J. and Attwell, D. (2002). Tonic and spillover inhibition of granule cells control information flow through cerebellar cortex. Neuron 33, 625-633.
  2. Wall, M. J. and Usowicz, M. M. (1997). Development of action potential-dependent and independent spontaneous GABAA receptor-mediated currents in granule cells of postnatal rat cerebellum. Eur. J. Neurosci. 9, 533-548.
  3. Brickley, S. G., Cull-Candy, S. G. and Farrant, M. (1996). Development of a tonic form of synaptic inhibition in rat cerebellar granule cells resulting from persistent activation of GABAA receptors. J. Physiol. 497, 753-759.
  4. Rossi, D. J. and Hamann, M. (1998). Spillover-mediated transmission at inhibitory synapses promoted by high affinity alpha6 subunit GABAA receptors and glomerular geometry. Neuron 20, 783-795.
  5. Pirker, S., Schwarzer, C., Wieselthaler, A., Sieghart, W. and Sperk, G. (2000). GABA(A) receptors: immunocytochemical distribution of 13 subunits in the adult rat brain. Neurosci. 101, 815-850.
  6. Saxena, N. C. and Macdonald, R. L. (1996). Properties of putative cerebellar gamma-aminobutyric acid A receptor isoforms. Molec. Pharmacol. 49, 567-579.
  7. Nusser, Z., Sieghart, W. and Somogyi, P. (1998). Segregation of different GABAA receptors to synaptic and extrasynaptic membranes of cerebellar granule cells. J. Neurosci. 18, 1693-1703.
  8. Mitchell, S. J. and Silver, R. A. (2000). GABA spillover from single inhibitory axons suppresses low-frequency excitatory transmission at the cerebellar glomerulus. J. Neurosci. 20, 8651-8658.
  9. Holt, G. R. and Koch, C. (1997). Shunting inhibition does not have a divisive effect on firing rates. Neural Comput. 9, 1001-1013.
  10. Vos, B. P., Volny-Luraghi, A. and De Schutter, E. (1999). Cerebellar Golgi cells in the rat: receptive fields and timing of responses to facial stimulation. Eur. J. Neurosci. 11, 2621-2634.
  11. Marr, D. A. (1969). A theory of cerebellar cortex. J. Physiol. 202, 437-470.
  12. Albus, J. S. (1971). A theory of cerebellar function. Math. Biosci. 10, 25-61.
  13. De Schutter, E., Vos, B. P. and Maex, R. (2000). The function of cerebellar Golgi cells revisited. Prog. Brain Res. 124, 81-93.
  14. Dieudonné, S. (1998). Submillisecond kinetics and low efficacy of parallel fibre-Golgi cell synaptic currents in the rat cerebellum. J. Physiol. 510, 845-866.
  15. Maex, R. and De Schutter, E. (1998). Synchronization of Golgi and granule cell firing in a detailed network model of the cerebellar granule cell layer. J. Neurophysiol. 80, 2521-2537.
  16. Maex, R., Vos, B. P. and De Schutter, E. (2000). Weak common parallel fibre synapses explain the loose synchrony observed between rat cerebellar Golgi cells. J. Physiol. 523, 175-192.
  17. Vos, B. P., Maex, R., Volny-Luraghi, A. and De Schutter, E. (1999). Parallel fibers synchronize spontaneous activity in cerebellar Golgi cells. J. Neurosci. 19, RC6: 1-5.
  18. Brickley, S. G., Revilla, V., Cull-Candy, S. G., Wisden, W. and Farrant, M. (2001). Adaptive regulation of neuronal excitability by a voltage-independent potassium conductance. Nature 409, 88-92.
  19. Cherubini, E. and Conti, F. (2001). Generating diversity at GABAergic synapses. Trends Neurosci. 24, 155-162.
  20. D’Angelo, E., Rossi, P., Armano, S. and Taglietti, V. (1999). Evidence for NMDA and mGlu receptor-dependent long-term potentiation of mossy fiber-granule cell transmission in rat cerebellum. J. Neurophysiol. 81, 277-287.