The Cerebellum: Brain for an Implicit Self (11 page)

Purkinje cell outputs from the cerebellar cortex inhibit their target neurons with GABA as the transmitter. Because Purkinje cells provide ~73% of the total synapses of cerebellar nuclear neurons, including almost all of the somatic synapses of cerebellar nuclear neurons (
Palkovits et al., 1977
;
De Zeeuw and Berrebi, 1995
), the question arises as to how such inhibitory inputs accurately control spiking in the latter neurons. To answer this question, Gauck and Jaeger (
2000
) applied the dynamic clamp method, in which they injected a conductance waveform that simulated the synaptic input of several hundred GABA A-type inputs to a cerebellar nuclear neuron in
in vitro
slices. They found that the time of inducing individual spikes was controlled precisely by brief decreases in inhibitory conductance, these being the consequence of the synchronization of many inputs. They also showed that spike rate was controlled linearly by the discharge rate of inhibitory inputs.

Purkinje cells project recurrent axon collaterals and thereby inhibit each other. These collaterals extend to neighboring Purkinje cells within ~300 micrometers of the parent cell (
Hawkes and Leclerc, 1989
;
O’Donoghue and Bishop, 1990
). Axon
collaterals of Purkinje cells also inhibit basket cells, which, in turn, inhibit Purkinje cells. Therefore, Purkinje cells may be involved in a mixed reciprocally inhibitory network containing both Purkinje cells and basket cells.

Climbing fibers are a unique structure of the cerebellum with no homolog elsewhere in the CNS (Color Plate IX A–B). The major transmitter of climbing fibers is glutamate. Each Purkinje cell is innervated by one climbing fiber. This is a consequence of the postnatal elimination of multiple innervation, which, after birth in rats and mice, attains its maximum in one week and fades out in two weeks via its interaction with developing parallel fiber-Purkinje cell synapses (
Mariani and Changeux, 1981
;
Hashimoto and Kano, 2003
;
Scelfo and Strata, 2005
;
Hashimoto et al., 2009
). Each climbing fiber forms numerous synaptic contacts with the dendrites of a single Purkinje cell [~1,300 in proximal dendrites of rat Purkinje cells (
Strata, 2002
), but a much larger number,~26,000, is derived from the density ratio of climbing fiber to parallel fiber synapses (
Nieto-Bona et al., 1997
)].

The above arrangements for climbing fibers result in a particularly large EPSP in Purkinje cells superimposed with Ca
²⁺
spikes (
Llinas and Sugimori, 1980a
,
b
). Extracellular recording has revealed that Purkinje cells spontaneously generate two different types of spike: simple spikes (
Figure 15E
,
F
) and complex spikes (E, G). In intracellular recording, stimulation of parallel fibers cells elicits simple spikes (
Figure 15A
), whereas climbing fiber stimulation evokes complex spikes (
Figure 15B
). Simple spikes are actually Na
²⁺
spikes generated in the somatic region that spread passively into the dendrites, whereas complex spikes involve Ca
²⁺
spikes generated in dendrites. In
in vivo
conditions, simple spikes discharge spontaneously at a rate of 50–100 Hz, whereas complex spikes discharge at an irregular, low rate of ~1 Hz (
Thach, 1967
).

The unique role of climbing fibers in inducing synaptic plasticity in Purkinje cells is dealt with in
Chapter 7
. Because of the powerful depolarizing action accompanying Ca
²⁺
entry, it has been suggested that climbing fiber responses also play a critical role in cellular function. Indeed, in rat cerebellar slices, climbing fiber discharges occurring at physiological frequencies (0.4–10 Hz) substantially modified the frequency and pattern of simple spike discharges (
McKay et al., 2007
). Repetitive climbing fiber discharges converted a spontaneous pattern of simple spike discharges into a more natural nonbursting pattern that consisted of simple spike trains interrupted by short climbing fiber-evoked pauses or longer pauses associated with state transitions. These effects were reproduced by injecting currents simulating complex spike depolarizations in the presence of synaptic blockers.
Hence, these appeared to occur intrinsically—for example, by activation of Ca
²⁺
-dependent K
⁺
channels.

In regard to the function of climbing fibers in cerebellar circuits, recent studies have revealed unexpectedly that climbing fibers also excite interneurons in the cerebellar cortex via atypical transmission mechanisms, as explained in
Chapter 5
, “
Inhibitory Interneurons and Glial Cells in the Cerebellar Cortex
.” In brief, such transmission might be mediated by a spillover of glutamate released from climbing fiber terminals (
Szapiro and Barbour, 2007
), which may then spread to interneurons via volume transmission (
Agnati et al., 1995
). An alternative mechanism would be for climbing fibers to activate synaptically NG2+ glial cells, which could, in turn, excite interneurons (
Lin et al., 2005
).

The cerebellar cortex receives not only mossy fibers and climbing fibers, but also beaded fibers, which contain various amines, such as serotonin, norepinephrine, or histamine, or neuropeptides, such as angiotensin II or orexin (
Haines and Dietrichs, 1984
;
Haines et al., 1984
;
Airaksinen and Panula, 1988
;
King et al., 1992
;
Onat and Cavdar, 2003
;
Zhu et al., 2006
;
Ito, 2009
). The beaded fibers extend fine varicose axonal fibers sparsely throughout the granular and molecular layers to form direct contacts with Purkinje cells and other cerebellar neurons. These axonal fibers are often called the third type of cerebellar afferent. On the basis of their diffuse extensions, it is considered that this third type of afferent does not convey specific information to the cerebellar cortex. Rather, its role could be modulatory. Akin to stomatogastric ganglia (
Marder et al., 1986
), such neuromodulation would set the activity level or switch the operational mode of a cerebellar microcomplex (
Chapter 9
, “
Network Models
”) to match a behavioral demand (Schweighofer et al., 2004) (for further description, see
Chapter 6
).

The mossy fiber-granule cell-Purkinje cell pathway provides the core of cerebellar cortical neuronal circuits. Unipolar brush cells appear to amplify the mossy fiber-to-granule cell transmission, but their special need in the vestibulocerebellum is unclear. This pathway, together with climbing fiber and beaded fiber afferents, forms the skeleton of the cerebellar neuronal circuits. Other types of neurons and glial cells put flesh on this skeleton to achieve the elaborate functional mechanisms of the cerebellum.

The cerebellar cortex contains two types of inhibitory neurons in the molecular layer; basket and stellate cells. In physiological experiments, basket cells are not always distinguishable from stellate cells on the basis of their responses during recording; hence, they are often lumped together as “basket/stellate cells” or collectively called inhibitory interneurons in the molecular layer. The granular layer, on the other hand, contains large and small inhibitory neurons. Large neurons are the Golgi and Lugaro cells, whereas small neurons are “small Golgi” cells, “small-fusiform Lugaro” cells, and “small globular” neurons. The latter three types of neuron have recently been identified on the basis of their characteristic morphology and location. In this chapter we consider recent knowledge about these neurons. We also consider Bergmann glial cells and NG2+ glial cells as important elements of cerebellar neuronal circuits.

Basket cells are middle-sized neurons located deep in the molecular layer close to the Purkinje cell layer, whereas stellate cells are smaller cells dispersed through the molecular layer (Color Plate IV). Basket cells supply inhibitory synapses to the “bottleneck” of a Purkinje cell soma where they form a unique complex structure called a “pinceau.” It can be labeled specifically by monoclonal antibodies raised using Xenopus oocytes as immunological vectors (
Tigyi et al., 1990
). On the other hand, stellate cells supply inhibitory synapses to Purkinje cell dendrites. A bundle of parallel fibers forms a synaptic contact with not only the dendrites of Purkinje cells but also those of basket and stellate cells, which in turn supply GABA-mediated inhibitory synapses to Purkinje cells. The somata of Purkinje cells and stellate cells are immunopositive for GABA but not for glycine (
Reichenberger et al., 1993
). Basket and stellate cells mediate the feedforward inhibition of Purkinje cells supplemental to the direct parallel fiber-Purkinje cell pathway.

In the C
₃
forelimb zone, it has been shown that basket and stellate cells share the same receptive field with Purkinje cells located in the same microzone (see
Chapter 9
, “
Network Models
”) (
Ekerot et al., 1995
). Parallel fiber-basket/stellate cell synapses are mediated by both AMPA and NMDA receptors, which produce prolonged (for hundreds of milliseconds) EPSCs in response to the burst stimulation of parallel fibers, the latter being presumably induced by the spillover of the transmitter, glutamate (
Carter and Regehr, 2000
). Parallel fiber-basket/stellate cell synapses are also mediated by mGluR1a (
Karakossian and Otis, 2004
). As in Purkinje cells, the activation of mGluR1s induces slow EPSCs. During low-frequency transmission, also as in Purkinje cells, basket/stellate cells are predominantly activated via AMPA receptors, whereas mGluR1s are recruited during high-frequency transmission.

Basket cells extend axons perpendicular to parallel fibers and cover an area containing ~10 × 7 rows of Purkinje cells, with a probable divergence number of ~50 (
Eccles et al., 1967
). Twenty to thirty basket cell axons may converge onto one Purkinje cell, although this figure is not particularly accurate. The middle band of parallel fiber-excited Purkinje cells flanked by side bands of basket/stellate cell-inhibited Purkinje cells constitutes a spatial pattern of lateral inhibition. This can be visualized in the cerebellar cortex following electrical stimulation of a selected number of parallel fiber bundles (
Coutinho et al., 2004
). However, because such a lateral inhibition pattern has not been observed to appear spontaneously, its significance under natural conditions remains uncertain.

Activation of basket/stellate cells induces powerful IPSPs in Purkinje cells (
Eccles et al., 1967
). Basket cells receive collaterals of climbing fibers and also those of Purkinje cell axons (
Palay and Chan-Palay, 1974
;
Jeager et al., 1988
). Indeed, climbing fiber responses evoked in Purkinje cells are accompanied by EPSPs in basket cells at about the same latency. These EPSPs in basket cells are followed by IPSPs with a delay of ~1 millisecond and constitute EPSP-IPSP responses, which display oddly the all-or-none property at their threshold stimulus intensity (
O’Donoghue et al., 1989
). The possible source of the so-observed unitary IPSP is a Purkinje cell and its recurrent axonal collaterals because it has been confirmed electronmicroscopically that each basket cell receives somatic inputs from only one Purkinje cell. On the other hand, EPSPs were shown to occur in basket cells at about the same latency as the climbing fiber responses evoked in Purkinje
cells, presumably being mediated by climbing fiber collaterals. However, no direct synaptic contact between climbing fibers and basket/stellate cells has been confirmed. Instead, it is now apparent that climbing fibers form synaptic contact with NG2+ glial cells via Ca
²⁺
-permeable AMPA receptors (
Lin et al., 2005
; see below). It is also apparent that climbing fiber signals activate interneurons in the molecular layer by a spillover of glutamate from their terminals, this being quite unlike typical synaptic transmission (
Szapiro and Barbour, 2007
).

Purkinje cells
in vivo
discharge spontaneously at a highly irregular rate (
Eccles et al., 1967
). This discharge is not caused by excitatory synaptic drives but, unexpectedly, by tonic influences from inhibitory interneurons (
Hausser and Clark, 1997
). Indeed, the pharmacological blockade of ionotropic and metabotropic glutamate receptors does not affect spontaneous Purkinje cell discharges, whereas the blockade of GABA
_A
receptors increases their rate and regularity. It thus appears that the tonic discharges of inhibitory interneurons modulate the spike discharges of Purkinje cells. Similar irregular discharges, and their dependence on tonic inhibition, have also been observed in inhibitory interneurons that receive tonic inhibition from themselves. Simulation studies using realistic models of Purkinje cell properties reproduce these irregular discharges and suggest that they are caused by endogenous tonic inhibitory current rather than excitatory current (De Schutter and Bower, 1994;
De Schutter, 1999
;
Jaeger and Bower, 1999
). The comprehensive functional meaning of this irregular spontaneous discharge in Purkinje cells and inhibitory interneurons is still unclear. A modeling study suggested, however, that the irregularity helps Purkinje cells and inhibitory interneurons to respond rapidly, sensitively, and linearly to external inputs (
Van Vreeswijk and Sompolinsky, 1996
).

Inhibitory interneurons in the molecular layer are reciprocally connected via inhibitory synapses (
Kondo and Marty, 1998
). They are also linked with each other through electrical synapses (
Mann-Metzer and Yarom, 1999
). A computer simulation suggested that reciprocal inhibition causes a 100–250 Hz oscillation in the activity of basket/stellate cells (
Maex and de Schutter, 2005
). Isope et al. (
2002
) drew attention to oscillations with characteristic frequencies between 150 Hz and 270 Hz. Interestingly, these were recorded far earlier on the cerebellar cortical surface by Edgar Adrian (1889–1977) (
1935
).