Implications of functional anatomy on information processing in the deep cerebellar nuclei

Abstract

The cerebellum has been implicated as a major player in producing temporal acuity. Theories of cerebellar timing typically emphasize the role of the cerebellar cortex while overlooking the role of the deep cerebellar nuclei (DCN) that provide the sole output of the cerebellum. Here we review anatomical and electrophysiological studies to shed light on the DCN's ability to support temporal pattern generation in the cerebellum. Specifically, we examine data on the structure of the DCN, the biophysical properties of DCN neurons and properties of the afferent systems to evaluate their contribution to DCN firing patterns. In addition, we manipulate one of the afferent structures, the inferior olive (IO), using systemic harmaline injection to test for a network effect on activity of single DCN neurons in freely moving animals. Harmaline induces a rhythmic firing pattern of short bursts on a quiescent background at about 8 Hz. Other neurons become quiescent for long periods (seconds to minutes). The observed patterns indicate that the major effect harmaline exerts on the DCN is carried indirectly by the inhibitory Purkinje cells (PCs) activated by the IO, rather than by direct olivary excitation. Moreover, we suggest that the DCN response profile is determined primarily by the number of concurrently active PCs, their firing rate and the level of synchrony occurring in their transitions between continuous firing and quiescence. We argue that DCN neurons faithfully transfer temporal patterns resulting from strong correlations in PCs state transitions, while largely ignoring the timing of simple spikes from individual PCs. Future research should aim at quantifying the contribution of PC state transitions to DCN activity, and the interplay between the different afferent systems that drive DCN activity.

Deep cerebellar neurons mirror the spinal cord’s gain to implement an inverse controller

“…However, physiologically, the earliest DCN recordings revealed the occurrence of shortlatency excitatory-post-synaptic-potentials (EPSP) as a result of harmaline injection into the IO (Llinas and Muhlethaler 1988). Furthermore, IO's ability to directly excite DCN has been confirmed by experiments involving direct electrical excitation of the IO (Rowland and Jaeger 2008); however, the complexity of the olivo-nuclear connection, which has an additional indirect pathway through PC inhibition, have made it difficult to rule out the contribution of PC's desinhibition casting doubts on the overall effect that IO has on DCN (Baumel et al 2009). …”

Specification and Development of GABAergic Interneurons

“…It has been estimated that they account for some 10-20% of all deep nuclear neurons (Baumel et al 2009;Aizenman et al 2003), i.e., some 5,000-10,000 cells (in the rat). There is compelling evidence that these cells also use glycine as a cotransmitter, but whether all of them do, or whether some deep nuclear interneurons use only glycine as transmitter -in analogy to the purely glycinergic Golgi cells mentioned above -is not known (Chen and Hillman 1993;Uusisaari and Knopfel 2010).…”

Rebuilding cerebellar network computations from cellular neurophysiology

“…Interestingly, the mossy fiber input can play a critical role in determining both the spike-to-spike variability and the UP/DOWN state of Purkinje cells, generating a complex blend of dynamics on multiple time-scales. Purkinje cell synchrony and the repercussion of PC firing on DCN neurons have been investigated in vivo (Jorntell and Ekerot, 2006;Baumel et al, 2009;Bengtsson and Jorntell, 2009) and novel information on PC functioning has been provided through the development of genetically expressed fluorescent proteins in these neurons (Akemann et al, 2009). The cells of DCN consists of diverse neuronal populations with distinct integrative properties (Uusisaari et al, 2007) and generate the sole cerebellar output.…”

“…The oscillations generated by the IO along the sagittal axis (climbing fiber-mediated) may collide with that conveyed by the cerebral cortex and diffusing along the transverse axis (parallel fiber-mediated), causing local resonance at the intersection of the parallel fiber and climbing fiber signals. DCN neurons may finally transfer temporal patterns resulting from strong correlations in PCs state transitions, while largely ignoring the timing of simple spikes from individual PCs (Baumel et al, 2009). PC correlations occur also in high-frequency bands (Akemann et al, 2009;Bengtsson and Jorntell, 2009).…”

“…A critical issue is that the convergence of Purkinje cell inhibition on DCN neurons remains to be demonstrated. Here, the discharge of DCN neurons in vivo is considered under the effect of harmaline (Baumel et al, 2009). Harmaline induced a rhythmic firing pattern of short bursts on a quiescent background at about 8 Hz, while other neurons become quiescent for long periods (seconds to minutes).…”