Cerebellum - Oxford Scholarship Jump to ContentJump to Main Navigation
The Synaptic Organization of the Brain$

Gordon M. Shepherd

Print publication date: 2004

Print ISBN-13: 9780195159561

Published to Oxford Scholarship Online: May 2009

DOI: 10.1093/acprof:oso/9780195159561.001.1

Show Summary Details
Page of

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (www.oxfordscholarship.com). (c) Copyright Oxford University Press, 2015. All Rights Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a monograph in OSO for personal use (for details see http://www.oxfordscholarship.com/page/privacy-policy). Subscriber: null; date: 27 August 2016



(p.271) 7 CEREBELLUM
The Synaptic Organization of the Brain




Oxford University Press

Abstract and Keywords

The cerebellum is a very distinct region of the brain, occupying a position immediately behind the tectal plate and straddles the midline as a bridge over the fourth ventricle. The basic functional design of the cerebellum is that of an interaction between two sets of different neuronal elements: those of the cortex and those in the centrally located cerebellar nuclei. The cerebellar cortex receives two types of afferents, the climbing fibers and the mossy fibers, and generates a single output system, the axons of Purkinje cells. The cerebellar nuclei receive collaterals from the climbing and mossy fibers and are the main targets for the Purkinje cell axons. The cerebellum as a whole is connected to the rest of the central nervous system by three large fiber bundles, the cerebellar peduncles. This chapter discusses the general organization of the cerebellum, covering its neuronal elements, synaptic connections, basic circuit organization, intrinsic membrane properties, synaptic actions, dendritic properties, and functional circuits.

Keywords:   brain regions, synaptic circuits, membrane properties, synaptic actions, dendritic properties, functional circuits

The cerebellum, a very distinct region of the brain, derives its name as a diminutive of the word “cerebrum.” To the ancient anatomists, this was a second, smaller brain in its own right. This is particularly explicit in the German language, where Kleinhim (“cerebellum”) translates literally into “small brain.” It occupies, in all vertebrates, a position immediately behind the tectal plate and straddles the midline as a bridge over the fourth ventricle. In addition, it is the only region of the nervous system to span the midline without interruption.

The cerebellum has undergone an enormous elaboration throughout evolution, in fact, more so than any other region of the central nervous system (CNS), including the cerebrum. On the other hand, the cerebellum has maintained its initial neuronal structure, almost invariant, throughout vertebrate evolution. Thus, its size but not its wiring has changed in evolution. As an example, the cerebellar cortex in a frog has an area approximately 12 mm2; that is, 4 mm wide (in the mediolateral direction) and 3 mm long (in the rostrocaudal direction). In humans, the cerebellar cortex is a single continuous sheet with an area of 50,000 cm2 (1,000 mm wide and an average of 50 mm long). This is 4 × 103 times more extensive than that of a frog (Braitenberg and Atwood, 1958). This cortex folds into very deep folia (Fig. 7.1), allowing this enormous surface to be packed into a volume of 6 cm × 5 cm × 10 cm. Because the cerebellar cortex is very long in the rostrocaudal direction, most of the foldings occur in that direction.

The basic functional design of the cerebellum is that of an interaction between two sets of quite different neuronal elements: those of the cortex and those in the centrally located cerebellar nuclei. The cerebellar cortex receives two types of afferents, the climbing fibers and the mossy fibers, and generates a single output system, the axons of Purkinje cells (Cajal, 1904). The cerebellar nuclei receive collaterals from the climbing and mossy fibers (Bloedel and Courville, 1981; Shinoda et al., 1992) and are the main targets for the Purkinje cell axons. The cerebellum as a whole is connected to the rest of the central nervous system by three large fiber bundles, the cerebellar peduncles.

The function of the cerebellum must be considered within the context of the rest of the nervous system because it is not a primary way station for sensory or motor function; that is, its destruction does not produce sensory deficits or paralysis. Nevertheless, (p.272)


Fig. 7.1. A: Drawing of the lateral view of the human brain showing the cerebellum. B: Midsagittal section of the cerebellum. C: Drawing of a single folium, showing the three layers of cerebellar cortex and the white matter.

lesions of the cerebellum are accompanied by well-defined and often devastating changes in the ability of the rest of the nervous system to generate even the simplest motor sequences used to attain motor goals. Indeed, the cerebellum is essential to the execution of specific movements as well as to placing motor sequences in the context of the total motor state of the individual at a given instant. Such a function is called motor coordination and relates to many different levels of brain function. It is not surprising then that the cerebellum has a complex neuronal organization and that it is vigorously connected with the rest of the brain. The enormous Purkinje cells are the sole link between the cerebellar cortex and the cerebellar nuclei. These neurons are the largest neuronal elements in the brain with respect to the number of synapses they receive and probably also with regard to the complexity of their integrative properties. In this chapter, we show how the role of the cerebellum in motor coordination arises from an interplay between the intrinsic excitability of the Purkinje and cerebellar nuclear cell membranes and from the crystal-like organization of the synaptic connectivity in the cerebellar cortex (Fig. 7.2).



Fig. 7.2. Geometric organization of the neuronal elements of the cerebellar cortex. Three planes of section through a cerebellar folium. A: Tangential plane (looking down on the cortical surface). B: Transverse (medial-to-lateral) plane. C: Sagittal (anterior-to-posterior) plane.


The cerebellar cortex is one of the least variable of CNS structures with respect to its neuronal elements (Cajal, 1904; Palay and Chan-Palay, 1974). In fact, a basic circuit present in all vertebrates is now well recognized as being composed of the Purkinje cell, the single output system of the cortex, and two inputs: (1) a monosynaptic input to the Purkinje cell, the climbing fiber, and (2) a disynaptic input, the mossy fiber–granule cell–Purkinje cell system.

(p.274) Because the Purkinje cell bodies are arranged in a single sheet, the Purkinje cell layer, the cerebellar cortex is divided into two main strata: (1) the level peripheral to the Purkinje cell layer, known as the molecular layer, and (2) the layer deep to the Purkinje cells (i.e., toward the white matter), the granular layer. Central to the granular layer is the white matter formed by the input and output nerve-fiber systems of this cortex (see Fig. 7.1B,C).


Climbing Fibers.

The two types of cerebellar afferents, the climbing fibers and the mossy fibers, represent opposite extremes among the afferents in the CNS. The climbing fibers originate from only one brainstem nucleus, the inferior olive. The main inputs to the inferior olive originate in the spinal cord, brainstem, cerebellar nuclei, and motor cortex. Olivary axons are long, fine (1–3 μm in diameter), and myelinated. They cross the brainstem at the level of the inferior olive, after which they course rostrally to enter the cerebellum primarily via the inferior cerebellar peduncle (a small contingent from the caudal portion of the inferior olive enters via the superior peduncle). Upon entering the cerebellar mass, they give off collaterals to the cerebellar nuclei and proceed toward the cerebellar cortex after branching into several fine fibers. The fibers lose their myelin as they penetrate through the granular layer before meeting with the Purkinje cell dendrites in the molecular layer (Fig. 7.3, CF). Each fiber branches repeatedly to “climb” along the entire Purkinje cell dendritic tree; thus, they were named climbing fibers by Ramón y Cajal. Each Purkinje cell receives only one climbing fiber. However, a given inferior olivary cell axon branches to form several climbing fibers. On average, about 10 climbing fibers are generated by a single inferior olivary cell.

Mossy Fiber–Parallel Fiber Pathway.

The other cerebellar afferents, the mossy fibers, originate from many CNS regions. Chief among these are the vestibular nerve and nuclei, spinal cord, reticular formation, cerebellar nuclei, and basilar pontine nuclei. The pontine nuclei receive input from much of the neocortex, making the cortico-ponto-cerebellar pathway one of the most massive in the brain. Mossy fibers enter through all three cerebellar peduncles (inferior, middle, and superior) and send collaterals to the deep cerebellar nuclei before branching in the white matter and synapsing on the granule cells (Chan-Palay, 1977; Shinoda et al., 1992). Thus, unlike the climbing fibers, mossy fibers do not synapse directly on Purkinje cells but rather on the small granule cells lying directly below them (Fig. 7.3B). This connectivity increases the number of Purkinje cells ultimately stimulated by one mossy fiber axon. Also, because mossy fibers branch profusely in the white matter, a given mossy fiber innervates several folia. The synapses between mossy fibers and granule cells occur as the fine branches of the mossy fibers twine through the granular layer. The contacts are made as the mossy fiber enlarges and generates tight knottings along its length. These portions of contact are called mossy fiber rosettes. One mossy fiber may have 20–30 rosettes (see Fig. 7.5B).

An integral part of the mossy-fiber input pathway is the granule cell axon, which completes the disynaptic input connection to the Purkinje cells. The axon of the granule cell, usually nonmyelinated, projects upward, past the Purkinje cell layer, into the molecular layer. On its way, it may form synapses with the dendritic trunk of Purkinje cells. In the molecular layer, the axon splits into two branches which take diametrically opposite directions, forming the shape of an uppercase “T” (see Fig. 7.3B). Fibers forming the horizontal (p.275)


Fig. 7.3. Drawing of the cerebellar afferent circuits and intrinsic neurons. A: The climbing fiber–Purkinje cell circuit. A fine branch of an axon from the inferior olivary nucleus (CF) climbs over the extensive arborization of the Purkinje cell (PC) dendritic tree; note the axon collaterals of the Purkinje cell axon. The Purkinje cell is viewed in profile here since it is drawn from a coronal section of the cerebellar cortex. B: In the glomeruli, activity in the mossy fibers (MF) excites granule cells (GrC), whose axons project toward the surface of the cortex where they bifurcate to form parallel fibers (PF); these in turn pass through many Purkinje cell dendrites with which they form excitatory synapses. C: In this drawing, the two afferent systems shown in A and B are combined and the two main types of intrinsic neurons are depicted: (1) the Golgi cells (GC), with cell bodies just below the Purkinje cell layer; and (2) the basket cells (BC), with cell bodies in the molecular layer. [Modified from Cajal, 1904, with permission.]

part of the T are found in all depths of the molecular layer. Because these fibers are precisely arrayed parallel to each other along the length of a folia, they have been named parallel fibers. These are perpendicular to the plane of the Purkinje cell dendrites (Fig. 7.3B), so that each Purkinje cell dendritic tree in humans may be intersected by as many as 200,000 parallel fibers (Braitenberg and Atwood, 1958).


Purkinje Cells.

As stated earlier, Purkinje cell axons provide the only output of the cerebellar cortex. These cells, which reach numbers as high as 15 × 106 in humans, were among the first neurons recognized in the nervous system (Purkinje, 1837) (Fig. 7.4A). Each cell has a large and extensive dendritic arborization, a single primary dendrite, a sphere-like soma (20–40 μm), and a long, slender axon that is myelinated when it leaves the granular layer. As the main Purkinje cell axon leaves the cortex, it gives off recurrent collaterals that ascend back through the granular layer to form plexi above and below the Purkinje somata and ultimately form synapses with Golgi and basket cells.

The Purkinje cell dendrites extend densely above the Purkinje cell layer through the molecular layer toward the boundary of the cortex. The unusual arrangement of the Purkinje cell dendrites makes them at once the most conspicuous structural element in the cerebellar cortex and provides an important clue to its functional organization. The


Fig. 7.4. Golgi preparations of cerebellar neurons. A: Purkinje cell soma, axonic initial segment, and dendritic arbor in a sagittal plane of section. The extent of the dendritic tree is from the Purkinje cell layer and spreads rostrocaudually to reach the cerebellar surface. B: Molecular layer interneurons and stellate (sc) and basket (be) cells. The stellate cell is found in the upper three-fourths of the molecular layer. Their dendrites have few branching points and project in the same plane as the Purkinje cell dendrites. Basket cells are found deeper in the cortex, their dendrites project horizontally subtending over 180° of arc. The interneuron axons (not shown) project horizontally to the Purkinje cell layer where they contact Purkinje cell dendrites (see D below). C: Granule cell showing a soma with five emerging dendrites. Note that the dendrite ends in the form of a claw (dc) for contact with a mossy fiber and Golgi cell axons. D: Basket cell axon projects horizontally above and along the Purkinje cell layer in the same plane as Purkinje cell dendrites. Short projections of the basket axon descend about 30 μm into the Purkinje cell layer and each clasps a Purkinje cell soma. Scale: A = 100 μm; B = 20 μm; C = 5 μm; D = 50 μm. [Micrographs courtesy of Dean Hillman.]

(p.277) entire mass of tangled, repeatedly bifurcating branches is confined to a single plane, very much like a pressed leaf. Moreover, the planes of all the Purkinje cell dendrites in a given region are parallel, so that the dendritic arrays of the cells stack up in neat ranks; adjacent cells in a single plane form equally neat, but overlapping, files (see Fig. 7.2A). To a large extent, this orderly array determines the nature and number of contacts made with other types of cells. Thus, parallel fibers running perpendicular to the plane of the dendrites can intersect a great many Purkinje cells. Conversely, the dendrites of a typical human Purkinje cell may form as many as 200,000 synapses with afferent fibers—more than any other cell in the CNS.

The Purkinje cell is not merely a transmitter or repeater of information originating elsewhere. As we shall see, its output is determined by its synaptic interactions with other neurons, by their interactions with one another, and by its quite complex intrinsic membrane properties.


The basic circuit common to all cerebella contains only one excitatory intrinsic neuron, the granule cell (Fig. 7.3C, GrC). This basic circuit is augmented by three types of inhibitory interneurons: the Golgi cells of the granular layer (see Fig. 7.3C, GC) and the basket (see Fig. 7.3C, BC) and stellate cells of the molecular layer, which are elaborated progressively in evolution. We begin with the granule cells.

Granule Cells.

These are the smallest cells in the cerebellum, with an oval or a round soma 5–8 μm in diameter. They are densely packed in the granule cell layer, which occupies about one-third of the cerebellar mass. In fact, these cells are the most numerous in the CNS; there are about 5 × 1010 cerebellar granule cells in the human brain. Each cell has four or five short dendrites (each less than 30 μm long) that end in an expansion called a dendritic claw (Fig. 7.4C). Their thin (0.1–0.2 μm in diameter), ascending axon has varicosities where synapses are formed, before it bifurcates to form the parallel fibers (see earlier). After bifurcating, the parallel fiber may run for 6 mm (3 mm on each side) before coming to an end (see Fig. 7.3C).

Golgi Cells.

There are two sizes of Golgi cells: (1) large ones (somata 9–16 μm in diameter), which are found mainly in the upper part of the granular cell layer, and (2) smaller ones (somata 6–11 μm in diameter), which are found in the lower half of the granular layer. They have extensive radial dendritic trees (Fig. 7.2A) that extend through all layers of the cortex (Fig. 7.3C). They receive input from the parallel fibers in the molecular layer and from climbing and mossy fiber collaterals in the granular layer. Their axons branch repeatedly in the granular layer, where they terminate on granule cell dendrites in the cerebellar glomeruli (see later). There are approximately as many Golgi cells as Purkinje cells.

Basket and Stellate Cells.

These are both found in the molecular layer, receive input from parallel fibers, and may be considered to be members of a single class. The processes of both cell types are oriented transversely to the long axis of the folia (see Fig. 7.3C, BC).

(p.278) Basket cells are found in the deep parts of the molecular layer, near the Purkinje cell layer. Their dendrites ascend into the molecular layer, in some instances as far as 300 μm (Fig. 7.4). Their axons extend along the Purkinje cell layer at right angles to the direction of the parallel fibers. They may spread over a distance equal to 20 Purkinje cell widths and 6 deep and may contact as many as 150 Purkinje cell bodies. During its course, the horizontal segment of a basket cell axon sends off groups of collaterals that descend and embrace the Purkinje cell soma and initial segment (see Figs. 7.3C and 7.4D; see also later). As many as 50 different basket cells are thought to wrap their axon terminals around each Purkinje cell soma, forming a basket-like meshwork resembling that on a Chianti bottle (Hámori and Szentágothai, 1966). Basket cell axons also ascend to contact the Purkinje cell dendritic tree. There are about six times as many basket cells as Purkinje cells.

The stellate cells are generally found in the outer two-thirds of the molecular layer. The smallest stellate cells, in the most superficial regions of the molecular layer, have 5- to 9-μm-diameter somata, a few radial dendrites, and a short axon (see Fig. 7.4B, sc). Deeper stellate cells are larger, have more elaborate dendritic arborizations that radiate in all directions, and have varicose axons that can extend parallel to the Purkinje cell dendritic plane as far as 450 μm. There are about 16 times as many small stellate cells as Purkinje cells.


There are three cerebellar nuclei on each side of the midline; each receives input via Purkinje cell axons from the region of cortex directly above it and projects to specific brain regions. The most medial nucleus, the fastigial, receives input from the midline region of the cerebellar cortex, the vermis. It projects caudally to the pons, medulla, vestibular nuclei, and spinal cord and rostrally to the ventral thalamic nuclei. Lateral to the vermis are the newer parts of the cerebellar cortex, the paravermis, which projects to the interpositus nucleus (which itself is divided into anterior and posterior divisions), and the hemispheres, which project to the convoluted dentate nucleus. The latter two cerebellar nuclei project rostrally to the red nucleus and ventral thalamic nuclei and caudally to the pons, medulla, cervical spinal cord, and reticular formation. There is a pattern of innervation of the cerebellar nuclei within this broad radial organization whereby the rostrocaudal and mediolateral groups of Purkinje cell axons parcel each cerebellar nucleus into well-defined territories (Voogd and Bigaré, 1980).

The cells of the cerebellar nuclei are not uniform in size: cells of small, medium, and even large diameter (≈35 μm) are found. The large cells have 10–12 long dendrites (≈400 μm long) that radiate to encompass a sphere. There are a few small cells with short axons, but the majority have long axons that leave the nuclei. It is also useful to distinguish cerebellar nuclear cells according to whether they are GABAergic. Only the smaller cerebellar nuclear neurons are GABAergic and virtually all of these neurons project to the inferior olive, whereas almost no non-GABAergic neurons do so (De Zeeuw et al., 1989; Fredette and Mugnaini, 1991). The non-GABAergic neurons project to the other targets of the cerebellum, as described earlier. Thus, the non-GABAergic neurons carry cerebellar influences to the rest of the brain, whereas the GABAergic neurons provide feedback to the inferior olive, (p.279) one of the principal afferent sources to the cerebellum. We take up the function of this feedback circuit in Functional Circuits.

The cerebellar nuclei are not simply “throughput” stations; rather, the synaptic integration that takes place here is a fulcrum for cerebellum function. Indeed, it is here that information from the cerebellar cortex is integrated with direct input from the mossy and climbing fibers. (This is discussed in Basic Circuits.)

As in Purkinje cells, the intrinsic properties of the nuclear neurons are very important to their function (see Intrinsic Membrane Properties).


Over 100 years ago, Ramón y Cajal (1888) published his description of the cerebellum. In this study of Golgi-stained material, the synaptic connections were already indicated, as shown in Fig. 7.2, as were the directions of flow of impulses in this cortex. Electron microscopic studies have provided additional information about the type of synaptic connections and their fine structure (cf. Palay and Chan-Palay, 1974) and confirmed Cajal’s initial description. The synaptic connections among the elements in the cerebellum are discussed by layer, not by cell type, to highlight the local circuits at each level of the cerebellum. We begin with the granular layer.


Two cell types receive input here: the granule cells and the Golgi cells. The synapses onto granule cells take place in the cerebellar “glomeruli.” The rosettes, which occur along the fine branches at the terminals of mossy fibers, form the core of each glomerulus. Excitatory synapses (Gray’s type 1) are made between the rosettes and the interdigitating dendrites from as many as 20 granule cells. This can be seen in the electron micrograph in Fig. 7.5B, where a large mossy fiber presynaptic terminal (mf) is seen to be surrounded by several granular cell dendritic claws (dc). The presynaptic terminal contains spherical presynaptic vesicles of about 450 Å in diameter. Golgi cell axon terminals surround the rosettes, where they make inhibitory (Gray’s type 2) synapses onto the granule cell dendrites (Fig. 7.5B, ga). All are encapsulated by a glial lamella that marks the border of each glomerulus.

In the granular layer, Golgi cells receive excitatory (type 1) input from the mossy fibers. These synapses are formed on the Golgi cell dendrites and somata. Thus, mossy fiber volleys excite Golgi and granule cells. Climbing fibers also contact Golgi cells in the granular cell layer. Finally, Purkinje-cell recurrent axon-collateral varicosities and terminals make inhibitory (type 2) synapses on Golgi-cell dendritic trunks and primary branches.


The synapse formed in this region is between the basket-cell axon terminal and the Purkinje-cell soma and initial segment. As many as 50 basket-cell axon branches make intricate arborizations surrounding the somata, which form many axosomatic synapses; the electron micrograph in Fig. 7.5C illustrates basket cell axon (ba) contacts on the soma and initial segment of a Purkinje cell. Even though basket cell terminals cover (p.280)


Fig. 7.5. Electron micrographs of parallel fiber and basket axon synaptic relationships. A: Climbing fiber (cf) synapse with spines from a large adjacent Purkinje cell dendrite (Pcd). The contact is made on Purkinje cell spines (s) as the climbing fiber follows the main Purkinje cell dendrite. Note that glial projections surround the dendrite and synaptic spines. A stellate cell or Golgi cell dendrite is adjacent to the climbing fiber and is contacted by a parallel fiber. B: A mossy fiber rosette (mf) filled with synaptic vesicles and mitochondria. Surrounding the mossy fiber axon are numerous profiles from dendritic claws (dc) making synaptic contacts. Golgi axon boutons (ga) contact the dendritic claws. C: Initial axonal segment of Purkinje cell. The base of the soma has basket axonal contacts (ba). Basket axons are separated from the Purkinje axon by glia but contact each other forming the pinceau of contacts between axons at their tips below this region. Scale: A = 5 μm; B = 3 μm; C = 1 μm. [Micrographs courtesy of Dean Hillman and Suzanne Chen.]

(p.281) both the soma and the axon hillock of the Purkinje cells, only a few synapses with the typical structure of Gray’s type 2 (see Chap. 1) have been observed at the axon hillock level; however, a rather impressive morphological structure known as the pinso terminale may be found at this level (Cajal, 1888). This terminal portion is not a chemical synapse but is similar to the electrical inhibitory synapse in Mauthner cells. These synapses very effectively shut down the output of the cortex.


Climbing Fiber–Purkinje Cell Connection.

Among the afferent systems of central neurons, none is more remarkable in extent and power than the climbing fiber junction with Purkinje cells. This junction is unusual not only for its large coverage of a considerable portion of the Purkinje cell dendritic tree but also because, as we have seen, only one climbing fiber afferent contacts each Purkinje cell. The synapses are made between varicosities (2 μm across) on the climbing fiber and stubby spines on the soma and main dendrites of the Purkinje cell; as many as 300 synaptic contacts may be made between a climbing fiber and its Purkinje cell. Each climbing fiber varicosity may synapse with one to six spines. A climbing fiber terminal (cf) contacting a Purkinje cell spine(s) near a dendrite (Pcd) is shown in Fig. 7.5A. A dendrite (d) from a Golgi or stellate cell is adjacent to the climbing fiber terminal. The presynaptic vesicles are round and 440–590 Å in diameter. Morphologically, the presence of a climbing fiber synapse seems to exclude nearby parallel fiber–Purkinje cell contacts. The Purkinje cell dendrites can thus be divided into a central area covered by the climbing fibers and the more peripheral, spiny dendritic portion that is contacted by parallel fibers.

Parallel Fiber Connections.

In contrast to the climbing fibers, which mainly contact Purkinje cell dendrites, the parallel fibers terminate on the dendrites of all the neuronal elements in the cerebellar cortex, except for the granule cells. Thus, parallel fibers contact the dendrites of Purkinje cells, basket cells, stellate cells, and Golgi cells. On the Purkinje cells, parallel fibers synapse with the spines on the terminal regions of the Purkinje cell dendrites, called spiny branchlets. These are shown in Fig. 7.6A, where an antibody against calbindin has been used to reveal the great density of spines on the dendritic trees of two Purkinje cells. The synaptic junction is formed between the head of a spine and a globular expansion of the parallel fiber; the spine penetrates the swollen part of the fiber. The electron micrograph in Fig. 7.6B illustrates a Purkinje cell spiny branchlet (sb) with at least three spines (one is marked). A synapse with a parallel fiber is clearly seen on each of the three right-hand spines. The synaptic vesicles are spherical and 260–440 Å in diameter. A parallel fiber forms synapses with one of every three to five Purkinje cells that it traverses. Thus, most of the parallel fibers passing through the dendritic tree of a Purkinje cell will not form synapses. Nevertheless, there is such a large number of parallel fibers that as many as 200,000 synapses on one Purkinje cell dendrite may be formed in humans, by far the largest number of synaptic inputs to any central neuron. In addition, the ascending portion of the granule cell axon has varicosities that are presynaptic to spines on the lower dendrites of Purkinje cells.

Golgi cell dendrites receive excitatory synapses from the parallel fibers. These axodendritic synapses are by far the largest number of synapses onto Golgi cells. An (p.282)


Fig. 7.6. Cerebellar molecular layer synaptic relationships of parallel fibers (granule cell axons) with Purkinje cells and interneurons. A: Immunoreaction of a Purkinje cell showing the detail of spine density on spiny branchlets (calbindin antibody on a l-μm plastic section). The profiles of two Purkinje cell somata are seen with segments of the main dendritic arbor and numerous spiny branchlets. Emerging spines and profiles of spine heads dot the field, revealing the high density of Purkinje cell spine synapses with parallel fibers. Longitudinal sections of spiny branch-lets show that the interspace interval of spines along the dendrite is near the diameter of the spine head. Note that the larger main branches have few spines. B: Electron micrograph of a Purkinje cell spiny branchlet (sb) that is longitudinally sectioned and has spines (s) emerging from the dendritic shaft in contact with a parallel fiber bouton. Bergmann glial projections shroud the spine shaft and junctional site. C: Golgi cell dendrite (Gcd) with parallel fiber (b) synapses. Spine (s) emerges from the dendritic shaft with parallel fiber synapse on the head and the shaft. Parallel fiber boutons (b) synapse directly on the dendrite. Scale: A = 10 μm; B = 1 μm; C = 1 μm. [Micrographs courtesy of Dean Hillman and Suzanne Chen.]

(p.283) example is shown in Fig. 7.6C, where parallel fiber boutons (b) synapse directly onto a Golgi cell dendrite (Gcd) as well as with the head and shaft its dendritic spine(s).

Plasticity ofPurkinje Cell Connectivity.

In the Purkinje cell dendritic tree, the climbing fiber input is normally proximal to the parallel fiber input (Fox et al., 1967). When damage to the climbing fibers occurs in the adult animal, however, spines proliferate in large numbers on Purkinje cell smooth dendrites. These are promptly invaded by newly formed parallel fiber contacts (Sotelo et al., 1975), indicating a tug of war or a territoriality between the two systems. Also, destruction of the parallel fibers promotes multiple climbing fiber innervation (Mariani et al., 1977), indicating that a true competition for a Purkinje cell dendritic tree exists between parallel and climbing fiber afferents and even between climbing fiber afferents themselves. It also indicates that a single climbing fiber cannot provide all of the necessary input, because Purkinje cells become multiply innervated by climbing fibers after parallel fiber damage.

Quantitative studies have been made of the changes in the parallel fiber–Purkinje cell synapse localization after lesioning of the parallel fiber input. In one set of experiments, the parallel fibers were sectioned and the molecular layer was undercut (to destroy the granule cells) (Hillman and Chen, 1984). The number, size, and average contact area of the parallel fiber–Purkinje cell synapses were evaluated 2–3 weeks after the lesion and compared with control values from unlesioned animals. It was found that the number of parallel fibers contacting a Purkinje cell was reduced in relation to the extent of the lesion but that the area of synaptic contact of the surviving synapses was proportionately increased. Thus, there was a change in the position and size of the synapses in response to perturbations, but the total area of synaptic contact remained stable. Change in the size of the presynaptic boutons was not accompanied by a change in the presynaptic grid densities or the number of synaptic vesicles (Hillman and Chen, 1985a). This suggests that as the size of the boutons increased, there was a parallel increase in the morphological correlates of the neurotransmitter release machinery. Stabilization of the total synaptic area has also been seen in other areas of the CNS (see Hillman and Chen, 1985b).

Other Connectivity in the Molecular Layer.

In addition to Purkinje cells, the dendrites of stellate, basket, and Golgi cells receive inputs in the molecular layer (see Fig. 7.3C). Parallel fiber swellings make excitatory synapses onto stellate cell dendritic spines. The stellate cells in turn make inhibitory synapses into Purkinje cell dendritic shafts. The basket cells receive excitatory synaptic connections from climbing fibers and parallel fibers and are inhibited by Purkinje cell axon collaterals. Parallel and climbing fibers make the same en passant synapses with basket cell dendrites as with Purkinje cell dendrites.


Five different types of synaptic terminals have been distinguished on the basis of the characteristics of their membrane attachment and shape of synaptic vesicles. Both axo-somatic and axodendritic synapses are found. The presynaptic terminals are made by collaterals of the mossy and climbing fibers and by Purkinje cell axons (Palkovits et al., 1977). Purkinje cell axons have two or three branches that arborize extensively in the nucleus, describing a narrow cone. Synapses are formed at the terminals and at en passant thickenings along the length of the axon. Synapses are usually formed with (p.284) dendritic thorns or spines of nuclear cells, although some synapses are axosomatic. The thickenings and terminals have dispersed ovoid vesicles, which are usually found where they contact the dendrites of nuclear neurons.

There is both significant divergence and convergence in the Purkinje cell–cerebellar nuclei projection (Palkovits et al., 1977). For example, in cats, individual Purkinje cells are estimated to contact as many as 35 nuclear cells; however, the bulk of the synapses are made with only 3–6 nuclear cells. Conversely, there are about 26 Purkinje cells for each nuclear cell, and each Purkinje cell axon branches extensively. As a result, each nuclear cell may receive input from up to 860 different Purkinje cells.

Complex synaptic combinations such as serial and triadic synapses are found in the cerebellar nuclei (Hámori and Mezey, 1977), as is also seen in the retina (see Chap. 6) and thalamic nuclei (see Chap. 8). These synaptic arrangements imply a quite complex interaction between the afferents and nuclear cells. In these synapses, the first presynaptic element may be a Purkinje cell axon terminal, a brainstem afferent terminal (collateral of a climbing or mossy fiber), or an axon terminal that is probably from a collateral of a cerebellar nuclear projection cell. The second terminal in the sequence is from either an axon collateral of a projection neuron or a Golgi type II interneuron and is both post-synaptic to the first element and presynaptic to the third element. The third element of the triad is a dendrite of a cerebellar nuclear neuron, which receives synaptic input from the other two elements. Although such triadic synapses are a regular feature of the nuclei, they do not form as large a percentage of synapses as in some sensory systems.


There are three main circuits in the cerebellum: two circuits in the cortex, which include afferent fibers as shown in Fig. 7.3, and one circuit in the deep nuclei. They are diagrammed in Fig. 7.7.


The sequence of events that follows the stimulation of mossy fibers was first suggested by János Szentágothai at the Semmelweis University School of Medicine in Budapest: the stimulation of a small number of mossy fibers activates, through the granule cells and their parallel fibers, an extensive array of Purkinje cells and all three types of inhibitory interneurons. Subsequent interactions of the neurons tend to limit the extent and duration of the response. The activation of Purkinje cells through the parallel fibers is soon inhibited by the basket cells and the stellate cells, which are activated by the same parallel fibers. Because the axons of the basket and stellate cells run at right angles to the parallel fibers, the inhibition is not confined to the activated Purkinje cells; those on each side of the beam or column of stimulated Purkinje cells are also subject to strong inhibition. The effect of the inhibitory neurons is therefore to sharpen the boundary and increase the contrast between those cells that have been activated and those that have not.

At the same time, the parallel fibers and the mossy fibers activate the Golgi cells in the granular layer. The Golgi cells exert their inhibitory effect on the granule cells and thereby quench any further activity in the parallel fibers. This mechanism is one of negative feedback: through the Golgi cells, the parallel fiber extinguishes its own stimulus (see Fig. 7.7A). The net result of these interactions is the brief firing of a relatively large but sharply defined population of Purkinje cells.



Fig. 7.7. Diagram of the basic circuit in the mammalian cerebellum. A: This circuit includes all the elements making specific synaptic connections in the cerebellar cortex and nuclei. B and C: Simplified diagrams of cerebellar cortex showing the afferent circuits (B) and the intrinsic neurons (C). Abbreviations: BC, basket cell; cf, climbing fiber; CN, cerebellar nuclear cell; G, granule cell; GC, Golgi cell; mf, mossy fiber; PC, Purkinje cell; pf, parallel fiber; SC, stellate cell.


In the normal adult cerebellum, a one-to-one relationship exists between a climbing fiber and a given Purkinje cell (i.e., each Purkinje cell receives one climbing fiber); however, each olivary axon branches to provide climbing fibers to approximately 10 Purkinje cells. The branching patterns of olivocerebellar axons are not random, but rather the branches of an individual axon predominantly remain within a relatively narrow plane that is aligned to the rostrocaudal axis (Sugihara et al., 2001). Moreover, neurons from the same region of the inferior olive tend to project to the same rostrocaudally oriented strip of cerebellar cortex. Thus, the projection pattern of the olivocerebellar pathway divides the cerebellar cortex into a series of parasagittally oriented zones. Interestingly, the projection pattern of the olivocerebellar pathway is largely in register with corticonuclear (Purkinje cell axons to deep cerebellar nuclei) and cerebellar nucleo-olivary projections, such that a series of reentrant loops are formed. For example, climbing fibers from the principal nucleus of the inferior olive project to the lateral part of the cerebellar hemisphere and also send collaterals to the dentate nucleus. In turn, the dentate is targeted by Purkinje cells of the lateral part of the hemisphere, and its GABAergic cells project back to the principal olivary nucleus.

Although climbing fibers have Purkinje cells as their primary targets, they also activate other neurons of the cerebellar cortex. For example, they activate Golgi cells, which will inhibit the input through the mossy fibers (see Fig. 7.7A). Thus, when climbing fibers fire, their Purkinje cells are dominated by this input. The climbing fiber input to basket and stellate cells sharpens the area of activated Purkinje cells.

An additional feature of the anatomy of the olivocerebellar system is of particular note with regard to its action on the cerebellum: olivary neurons are electrotonically coupled by gap junctions (Llinás et al., 1974; Sotelo et al., 1974; Llinás and Yarom, 1981). In fact, immunoflourescence and mRNA studies indicate that the inferior olive has one of the highest densities of connexin 36 (Condorelli et al., 1998; Belluardo et al., 2000), the protein from which neuronal gap junctions are usually formed (Rash et al., 2000). This electrotonic coupling is thought to allow olivary neurons to synchronize their activity. Interestingly, most of these gap junctions occur between dendritic spines that are part of complex synaptic arrangements known as glomeruli. Olivary glomeruli, in addition to the gap-junction–coupled dendritic spines, contain presynaptic terminals, whose function is thought to be to control the efficacy of the electrotonic coupling.


Electrical activation of mossy fiber inputs to the cerebellar system generates an early excitation in the cerebellar nuclei because the collaterals terminate directly on the cerebellar nuclear cells (see Fig. 7.7A). The same information then proceeds to the cerebellar cortex, which in turn produces an early excitation of Purkinje cells to be translated into inhibition at the cerebellar nucleus. This inhibition is followed by a prolonged increase in excitability of the cerebellar nuclear cells. The increased excitability is the result of two actions: (1) disinhibition due to reduced Purkinje cell activity, which in turn results from the inhibitory action of basket and stellate cells after the initial activation of Purkinje cells, and (2) cerebellar nuclear cell intrinsic properties (see later). The Purkinje cell inhibition is also due indirectly to the inhibitory action of the Golgi (p.287) interneuron, which, by preventing the mossy fiber input from reaching the molecular layer, reduces the excitatory drive to Purkinje cells. The cerebellar nuclear projection neurons themselves send axon collaterals to cortical inhibitory interneurons including basket cells, which thus provide recurrent inhibition of the cerebellar nuclear neurons, as seen in spinal motoneurons (see Chap. 3).


In Chap. 2, it was emphasized that the functional characteristics of a neuron are the outcome of a complex interplay between its intrinsic membrane properties and its synaptic interactions. In no part of the brain is this exemplified more vividly than in the cerebellum. Indeed, as already mentioned in Chap. 2, the Purkinje cell is one of the best known models for demonstrating these properties. Because of this importance, we consider the intrinsic membrane properties separately in this section before addressing the synaptic actions of the system.


The intrinsic membrane properties of cells may be considered independent of synaptic input, although interaction of synaptic potentials with intrinsic membrane properties shapes the activity of the cell. Intrinsic properties are usually studied by determining the response to direct activation, that is, to depolarizing or hyperpolarizing current injected into the cell, usually into the soma. Purkinje cell electrical activity may be recorded under in vivo or in vitro conditions; however, because the most reliable recordings are obtained in vitro, our understanding of the electrical properties of the mammalian Purkinje cell membrane has come mainly from studies of cerebellar slices (Llinás and Sugimori, 1978, 1980a,b). Antidromic activation of a Purkinje cell is characterized by a large spike having an initial segment–soma dendritic (IS-SD) break that is in many ways similar to that obtained in vivo from motoneurons and other central neurons.

Direct stimulation of Purkinje soma via the recording microelectrode demonstrates that these cells fire in a way that is quite different from that seen in other neurons. Indeed, square current pulses lasting about 1 sec (Fig. 7.8A) produce, at just threshold depolarization, a repetitive activation of the cell. That is, with long pulses, the neuron fires, but a single isolated spike cannot be generated by this type of stimulation. This burst of activity is produced by a low-threshold, sodium-dependent conductance that does not inactivate within several seconds and serves to trigger the fast action potentials. This sodium conductance is different from that responsible for the fast action potentials seen in virtually all nerve cells: it is activated at a lower voltage and does not inactivate. With increased stimulation, the onset of the repetitive firing moves earlier. Also at the end of the initial pulse of firing, a reduction in the amplitude of the spikes is followed by a rhythmic bursting, as marked by arrows in Fig. 7.8B.

Pharmacological studies in cerebellar slices have shown that the fast action potentials and the bursting responses have different ionic mechanisms. Removal of extracellular sodium or the application of tetrodotoxin (TTX, a sodium-conductance blocker) to the bath causes a complete abolition of the fast action potentials seen in Fig. 7.8A and B but leaves a late, slow-rising burst potential intact, as shown in Fig. 7.8C. This slow bursting of Purkinje cells has been found to be generated by a voltage-activated (p.288)


Fig. 7.8. Intrinsic properties of mammalian Purkinje cells recorded in vitro. A: A prolonged, threshold current pulse injected into the soma of a Purkinje cell elicits a train of action potentials after an initial local response (arrow). B: Increased current strength elicits high-frequency firing and oscillatory behavior (arrows). C: After addition of TTX to the bath, the fast action potentials are blocked, and the slowly rising action potentials underlying the oscillations seen in B are revealed. A slow afterdepolarization may also be seen. D: Addition of cobalt chloride (Co2+) to the TTX perfusate removes all electroresponsiveness, indicating that the slow action potentials in C were calcium dependent. [Modified from Llinás and Sugimori, 1980a, with permission.]

calcium conductance followed by a calcium-dependent potassium conductance increase. We know the spikes are calcium dependent because they are seen in the absence of sodium and because they are blocked by the removal of calcium from the extracellular medium or by ions that block the slow calcium conductance (cobalt, cadmium, manganese), as shown in Fig. 7.8D. When the calcium in the bathing solution is replaced by barium, the afterhyperpolarization is reduced and the bursting response is converted into a prolonged single action potential. This demonstrates the presence of a calcium-activated potassium conductance, because it is known that barium does not activate the calcium-activated potassium conductance. All electroresponsiveness is gone after calcium and sodium blockade, as shown by the application of both TTX and cobalt to the extracellular medium (see Fig. 7.8D). Thus, at the somatic level, Purkinje cells have not one, but three, main mechanisms for spike generation: (1) a sodium-dependent spike similar to that seen in other cells, which is blocked by the absence of extracellular sodium or by the application of TTX; (2) a low-threshold, noninactivating sodium spike; and (3) a calcium-dependent action potential, which has a slow rising time and a rather rapid return to baseline.

The distribution and properties of voltage-gated channels in the dendrites are discussed later (see Dendritic Properties).


The electrical properties of cerebellar nuclear neurons were first studied in detail in in vitro preparations (Jahnsen, 1986a,b; Llinás and Mühlethaler, 1988b). Like Purkinje (p.289) cells, cerebellar nuclear cells have a collection of ionic conductances that give them complex firing abilities. Cerebellar nuclear cells have a noninactivating sodium conductance similar to that described in Purkinje cells, in addition to the usual sodium-and potassium-dependent conductances that generate fast action potentials. The firing of cerebellar nuclear cells depends on their resting potential. If a cell is depolarized with a current pulse from the resting potential, as in Fig. 7.9A, the cell fires a train of


Fig. 7.9. Intrinsic properties of cerebellar nuclear neurons. A: A depolarizing current injection from resting potential elicits tonic firing. B: When the same strength current pulse is delivered from a hyperpolarized membrane level, an all-or-none burst response is elicited. C: Hyper-polarizing current injection from the resting potential elicits a strong rebound burst of action potentials from a slow depolarization. D: Response to current injection from a hyperpolarized level (resting potential marked by broken line). E: Addition of TTX to the perfusate blocked the fast action potentials, revealing a slowly rising, prolonged depolarization and afterdepolarization; these responses were then blocked by addition of Co2+ to the bath. [Modified from Llinas and Muhlethaler, 1988b, with permission.]

(p.290) action potentials. However, if the same current pulse is injected when the cell is held hyperpolarized from the resting potential, all-or-nothing bursts are seen, as shown in Fig. 7.9B and D. Also, if a hyperpolarizing current pulse is injected into a cerebellar nuclear neuron, a burst of action potentials is seen at the end of the current injection (Fig. 7.9C).

This “rebound response” following hyperpolarization is important in cerebellar nuclear cell function. This is easily understood because Purkinje cells are inhibitory and generate inhibitory postsynaptic potentials (IPSPs) in cerebellar nuclear cells. The ionic basis for these burst responses was determined by pharmacological studies. Thus, after eliminating the fast sodium conductance by the application of TTX, the fast action potentials seen in Fig. 7.9D are blocked and a slowly rising spike is elicited from the hyperpolarized membrane potential (Fig. 7.9E). The threshold for these spikes is lower than that for the fast sodium-dependent action potentials; they are therefore called low-threshold spikes (LTSs). They are calcium dependent because they are blocked after the addition of cobalt or the removal of calcium from the bath and are insensitive to TTX. The presence of an LTS is probably of major functional significance in cerebellar nuclear neurons because following climbing fiber activation of Purkinje cells, such bursts can easily be elicited following the powerful IPSPs produced by this input (see Fig. 7.18; Llinás and Mühlethaler, 1988b).



One of the most powerful synaptic junctions in die CNS is that between the climbing fiber afferent and the dendrites of a Purkinje cell. It has been called a distributed synapse because a single presynaptic fiber makes contact with the postsynaptic cell at many points (≈300) throughout the Purkinje cell dendritic tree, and thus the synapse is distributed over a large surface area. This is in contrast to more typical synapses, such as between a Ia terminal and motoneuron, where there are only a few, relatively localized points of contact (see Chap. 3). Eccles et al. (1966a) demonstrated electrophysiologically that stimulation of the inferior olive produces a powerful activation of the Purkinje cell. This synaptic excitation is characterized by an all-or-nothing burst of spikes that shows little variability from one activation to the next. These are called complex spikes. Several complex spikes recorded from an isolated preparation are superimposed in Fig. 7.10A and B. It is now known that the spikes on the broad EPSP are produced in the dendrites by a voltage-activated calcium conductance (see later) and at the somatic and axonic levels by the usual Hodgkin-Huxley sodium and potassium conductances (Llinás and Sugimori, 1978, 1980b).

Climbing fiber responses in Purkinje cells may be elicited by placing a stimulating electrode in the white matter near the midline. This “juxtafastigial” stimulation activates inferior olivary axons in the white matter. Following a juxtafastigial stimulus, climbing fiber synapses are activated simultaneously and produce a very large unitary EPSP in the postsynaptic dendrite. This unitary synaptic potential usually has an amplitude of 40 mV and lasts 20 msec. The all-or-nothing character of the climbing fiber–evoked EPSP contrasts with the usual graded nature of EPSPs within the CNS and reflects the singular climbing fiber innervation of a Purkinje cell. If the Purkinje (p.291)


Fig. 7.10. Climbing fiber activation of mammalian Purkinje cells in vitro. A: All-or-none complex spikes in a Purkinje cell evoked by white matter stimulation are superimposed. B: In another Purkinje cell, five threshold white-matter stimuli (arrow) evoke very uniform complex spikes on four occasions. C: If threshold stimuli are delivered when the cell is hyperpolarized (to prevent action potential firing), the all-or-none climbing fiber EPSP may be seen. D: Reversal of climbing fiber EPSP. Notice that, as expected in a distributed synapse, the reversal is biphasic, with the early portion of the potential reversing at lower levels of injected current than the late part; this may be seen at 18, 22, and 28 nA. E: Plot of the voltage current relation for the EPSP reversal shown in D. [Modified from Llinás and Nicholson, 1976, and Llinás and Mühlethaler, 1988a, with permission.]

cells are hyperpolarized far enough to prevent the cell from spiking, the all-or-nothing character of the EPSPs may be seen (see Fig. 7.10C).

Under conditions in which the sodium- and calcium-dependent spikes are prevented, the chemical nature of the synapse may be studied in detail and its distributed character clearly demonstrated. Depolarization of the soma or dendrite can produce a reduction in amplitude and an actual reversal of the sign of the climbing fiber EPSP, as shown (p.292) in Fig. 7.10D. A large increase in the EPSP amplitude is seen when the membrane potential is moved in the hyperpolarizing direction (lower traces in D). The reversal in sign (shown in Fig. 7.10D, 22.1 nA) is then the necessary and sufficient evidence to indicate that a synaptic junction is chemical in nature (see Chap. 2).

The fact that different parts of the EPSP (the peak and falling phase) reverse at different levels of depolarization (see biphasic reversal at 22.1 nA in Fig. 7.10D) indicates that the synapse occurs at multiple sites with different distances from the site of recording in the soma (Llinás and Nicholson, 1976). Because a current point source, a microelectrode, is used to change the membrane potential, the potential change along the dendrite is maximum near the site of impalement and decreases with distance. Because the synapses closest to the site of recording generate most of the rising phase of the recorded EPSP, this component is the first to reverse. Those synapses located at a distance generate the slowest components (owing to the cable properties of the dendrites) and are less affected by the current injection. Recordings similar to those obtained in vitro can also be obtained in vivo.

Activation of the climbing fiber afferents generates a burst of action potentials at the Purkinje cell axon. The frequency of this response is high, generally 500/sec. Indeed, it is higher normally than that seen after parallel fiber stimulation, suggesting that one of the possible functions of the climbing fiber system is to produce a discharge of distinct bursts of action potentials. As discussed later, climbing fiber activation also produces very sharp IPSPs in the target neurons of Purkinje cells.


As discussed earlier, mossy fiber inputs activate Purkinje cells via the parallel fiber–Purkinje cell synapse (Eccles et al., 1966b). Early investigators named these responses simple spikes. Purkinje cell responses to spontaneous activity in the parallel fibers are illustrated in Fig. 7.11A; notice that during this recording period, two complex spikes were also recorded. This circuit can be activated by direct parallel fiber stimulation of the cerebellar surface or via the mossy fiber–granule cell–parallel fiber pathway from white matter stimulation. Both types of stimulation generate short-latency EPSPs in Purkinje cells.

This postsynaptic potential differs from that generated by the climbing fiber in two ways. First, it is graded as shown by the response to juxtafastigial stimuli of increasing intensity (Fig. 7.11B, C). Second, it is generally followed by an IPSP (see trace, Fig. 7.11B). The IPSP is generated by activation of the inhibitory interneurons of the molecular layer. The parallel fiber synaptic depolarization can generate action potentials at the somatic level as well as dendritic calcium spikes if the stimulus is large enough (see later). Because parallel fiber activation of Purkinje cells is followed by a disynaptic inhibition, this synaptic sequence is reviewed in detail in conjunction with the inhibitory systems in the next section.


Inhibitory neurons are organized in the cerebellar cortex into two main categories: those that reside in the molecular layer (basket and stellate cells), and those that reside in the granular layer (Golgi cells).



Fig. 7.11. Mossy fiber activation of mammalian Purkinje cells in vitro. A: Spontaneous activity in the mossy fiber–parallel fiber system gives rise to fast, simple spikes in Purkinje cells, which are in contrast to the two broad, climbing fiber–evoked complex spikes in the trace. B: White matter stimulation of increasing strength evoked graded EPSP–IPSP sequences due to mossy fiber–parallel fiber activation. C: When such stimulation is delivered during hyperpolarizing pulses of increasing amplitude (middle trace), the parallel fiber–mediated EPSP may be seen (top trace); the bottom trace illustrates the graded nature of the synaptic potential. [Modified from Llinás and Mühlethaler, 1988a, with permission.]

Granular Layer.

In the granular layer, the main inhibitory system is the Golgi cell axonic plexus. This plexus releases GABA, inhibiting granule cell dendrites within the granule cell glomerulus. Indeed, while mossy fibers activate the terminal dendritic claws of the granule cells, the Golgi cell axons also distribute their contacts on the dendrites of the granule cells and act to counter the synaptic action of the mossy fibers by the (p.294) release of GABA. The inhibition that ensues is so powerful as to totally block parallel fiber activity in the cerebellar cortex (Eccles et al., 1966d).

Molecular Layer.

In the molecular layer, inputs from climbing fibers and parallel fibers represent the two types of excitatory afferents terminating on a Purkinje cell. The Purkinje cell also receives input from three inhibitory systems: the basket cell, the stellate cell, and the catecholamine system, which arises from the locus coeruleus (Bloom et al., 1971; Pickel et al., 1974). Activation of the basket cells generates a graded inhibition at each side of the activated bundle of parallel fibers (Andersen et al., 1964; Eccles et al., 1966b,c). This can be seen clearly when recordings are made lateral to the beam of stimulated parallel fibers. In this case, at low stimulus intensity, only the IPSP is seen; however, if the stimulus intensity is increased, more Purkinje cells are excited by the parallel fibers and an EPSP–IPSP sequence is seen (Fig. 7.12A). The basket cell IPSP is generated by a membrane conductance increase to chloride, most probably by the release of GABA (see later). The second inhibitory system is that represented by the stellate cells, which synapse mainly on Purkinje cell dendrites. Electrophysiologically, they have the same pattern of inhibition as that of basket cells.

Monoaminergic Inhibition.

The third inhibitory system in the molecular layer is that of the locus coeruleus; its catecholamine-mediated inhibition generates a large, prolonged hyperpolarization in Purkinje cells (Hoffer et al., 1973). Although intriguing questions arise about the function of this system, it is possible (because of its rather widespread character) that it is related to the general state of wakefulness of the animal rather than to specific cerebellar functions. Indeed, morphologically, the system consists of rather thin filamentous afferents that reach the cerebellar cortex and bifurcate widely to cover not only the neuronal elements but probably also the vascular system (Bloom et al., 1971).


Perhaps one of the most surprising findings in the physiology of the cerebellum is the fact that the only output of the cerebellar cortex, the Purkinje cells, exercises an inhibitory input onto the cerebellar nuclear neurons (Ito et al., 1964). This finding indicates that the cerebellar cortex is the most sophisticated inhibitory system in the brain, not only because of its refinement of connectivity and the integrative ability of these neurons but also because of the extent of information reaching the cerebellar cortex. Indeed, there are as many neurons in the cortex (≈5 × 1010) as there are neurons in the rest of the brain. The powerful GABAergic inhibition of the Purkinje cells on the cerebellar nuclei also demonstrates the rich biochemistry of the system (Obata et al., 1967). The Purkinje cells project in a radial pattern onto the nuclei as discussed previously (see Neuronal Elements).

Electrical stimulation of the cerebellar white matter can elicit quite complex sequences of EPSPs and IPSPs in cerebellar nuclear cells. Here, we consider the simplest case—where white matter stimulation is limited to the Purkinje cell axons. In this case, only IPSPs are recorded. The records shown in Fig. 7.12B were made from a cerebellar nuclear cell in a cerebellum-brainstem preparation isolated from adult guinea pig (Llinás and Mühlethaler, 1988b). In the example shown in Fig. 7.12C, several (p.295)


Fig. 7.12. Inhibitory synaptic potentials in Purkinje cells and cerebellar nuclear cells. A: Here, the stimulating electrode was placed on the cerebellar surface lateral to the recorded Purkinje cell because under such conditions, powerful IPSPs could be recorded in the Purkinje cell. As the stimulus intensity was increased (lower traces), the band of activated parallel fibers became wider, and finally the parallel fibers synapsing on the recorded Purkinje cell were themselves activated; thus an EPSP preceded the IPSP. B: IPSPs recorded in a cerebellar nuclear cell. Stimulation of the white matter between the cerebellar cortex and nuclei may elicit graded EPSPs and IPSPs. For particular locations and amplitudes of stimulation, IPSPs may be elicited in the absence of an early EPSP, as shown here. These IPSPs are very regular, often triggering rebound firing of the cell, as seen here. C: That these large potentials are synaptic potentials is shown by their reversal upon injection of a hyperpolarizing current.

IPSPs were elicited; it can be seen that their onset and amplitude are very reliable (four traces are superimposed) and that they can be easily reversed in sign by current injection, as in this example. The response of cerebellar nuclear cells to white matter stimulation is not always so straightforward, as is discussed later (see Functional Circuits).


In addition to the excitatory action of climbing and parallel fibers on Purkinje cells (see earlier) and their intrinsic roles in Purkinje cell integration, other functions related to their temporal interaction have been proposed. Ito et al. (1982) reported that simultaneous low-frequency activation (1–4 Hz) of these two inputs such that climbing fibers precede parallel fiber activation (induction) reduces subsequent parallel fiber action on Purkinje cells when both inputs are again stimulated (expression). Thus, following such pairing, the parallel fiber EPSP or EPSC amplitude is reduced by 20%–50%; this effect is maximal after 5–10 min. It lasts as long as it has been studied, usually 1–2 hr, and is called long-term depression (LTD). Comparable phenomena induced by low-frequency stimulation have been found in other regions of the brain (cf. Chaps. 10–12).

The order and temporal sequence for the generation of this depression were initially proposed on theoretical grounds by Albus (1971) as the basis for his hypothesis that the cerebellar cortex may be the seat of motor learning. Ito et al. (1982) interpreted their results as a confirmation of Albus’s theory, but this is a matter of controversy. The phenomenon has since been studied largely in cerebellar slices, dispersed Purkinje cells, and more reduced preparations (Narasimhan and Linden, 1996). With these in vitro preparations, the cellular mechanism underlying this form of “memory” has become (p.296) an area of active research (Linden and Connor, 1993) and discussion (Llinás and Welsh, 1993).

An important issue with LTD has been its apparent specificity. Because climbing fiber activation stimulates the entire dendritic tree, the specificity is determined by the parallel fiber synapses. That is, only Purkinje cells that respond to those parallel fibers that were coactivated with climbing fiber input during the induction phase would presumably show a decrease in parallel fiber activation. This would mean that the input from a small group of granule cells would be selectively depressed, modifying the “computational” power of each Purkinje cell.

However, it has been shown that the opposite order of stimulation—parallel fiber activation preceding climbing fiber activation of Purkinje cells—can also lead to LTD (Chen and Thompson, 1995) and that parallel fibers in their own right can also activate such a process (Hartell, 1996). Indeed, parallel fibers alone can activate calcium entry on the spines of Purkinje cells (Denk et al., 1995). Thus, a new hypothesis as to how LTD may relate to motor function must be developed because Albus’ learning hypothesis was quite specific on the nature and order of climbing fiber–parallel fiber interaction.

From a molecular biological point of view, it has been proposed that LTD induction is associated with activation of voltage-gated calcium channels following climbing fiber activity and of metabotropic glutamate receptors (mGluRl) and AMPA glutamate receptors following parallel fiber activity. Climbing fiber activation of Purkinje cells leads to the opening of voltage-gated calcium channels and the generation of calcium spikes in the dendrites. The resulting increased intracellular concentration of calcium is necessary for LTD induction. Direct Purkinje cell depolarization can be substituted for climbing fiber activation. Activation of the metabotropic glutamate channels leads to phospholipase C–mediated production of diacylglycerol and inositol-1,4,5-triphos-phate. The AMPA receptors are linked to Na+-selective channels and sodium entry through the AMPA channels is necessary for the induction of LTD (Linden et al., 1993). Ionophoresis of glutamate can be substituted for parallel fiber stimulation. The ultimate expression of LTD is thought to be due to desensitization of AMPA receptor function (Linden, 1994). Finally, another issue has come up; LTD may in fact be a neuroprotective mechanism to control possible damage of the Purkinje cell dendritic tree by excess calcium entry during high-level activation (Llinás et al., 1997).

We can summarize what is known about the induction of LTD in the cerebellar cortex as follows. (1) Climbing fiber stimulation leads to increased intracellular concentration of calcium through voltage-gated channels and to increased cGMP, possibly through nitric oxide and guanylate cyclase. Parallel fibers also increase, in their own right, calcium concentration in these dendrites. (2) Parallel fiber activation leads to activation of metabotropic glutamate receptor–linked channels, which in turn leads to increased diacyglycerol and inositol-l,4,5-triphosphate. (3) Parallel fiber activation leads to activation of glutamate receptors and inflow of sodium and calcium via the ligand-dependent channels and of calcium via voltage-gated channels. (4) The expression of LTD is through desensitization of specific, parallel fiber–activated Purkinje cell AMPA receptors. The physiological role of LTD and its mode of generation remain matters of debate. Indeed, placing LTD in the context of cerebellar function awaits further studies carried out under physiological conditions.



In the cerebellar cortex, as is the case throughout most of the CNS, glutamate appears to be the major excitatory transmitter. Supporting its role as a neurotransmitter in the cerebellum, its release is dependent on calcium, antagonized by increases in magnesium, and stimulated by membrane depolarization caused by elevated levels of potassium (Sandoval and Cotman, 1978).

There is strong evidence that glutamate is the neurotransmitter of granule cells. Glutamate depolarizes Purkinje cells when applied ionophoretically to the dendrites (Krnjevic and Phillis, 1963; Curtis and Johnston, 1974; Sugimori and Llinás, 1981). Further, naturally occurring L-glutamate is more potent than the D-glutamate isomer (Chujo et al., 1975; Crepel et al., 1982). In frog Purkinje cells, the reversal potential of the glutamate-elicited EPSP is close to that for parallel fiber–evoked EPSPs (Hackett et al., 1979). Moreover, neurochemical studies have shown that the glutamate content is lower than normal in cerebella in which the number of granular cells has been reduced by X-irradiation (Valcana et al., 1972; McBride et al., 1976), virus infection (Young et al., 1974), or mutation (Hudson et al., 1976; Roffler-Tarlov and Turey, 1982). Also, compared with control values, glutamate uptake is reduced in synaptosomal preparations from cerebella in which the granule cell number has been reduced (Young et al., 1974; Rohde et al., 1979). Immunocytochemical studies have demonstrated high levels of glutamate immunoreactivity in parallel fiber terminals (Somogyi et al., 1986), which decrease under conditions that induce transmitter release (Ottersen et al., 1990a; Ottersen and Laake, 1992). Last, glutamate receptors are found on Purkinje cell dendrites as well as on other cells (basket, stellate, and Golgi) whose dendrites are contacted by parallel fibers (Petralia and Wenthold, 1992).

Glutamate also appears to be the neurotransmitter of the large majority of mossy fibers (Ottersen, 1993). Mossy fiber terminals are enriched in glutamate (Somogyi et al., 1986), and glutamate receptors are present on the postsynaptic granule cells (Gallo et al., 1992; Petralia and Wenthold, 1992). Moreover, CNQX, an AMPA antagonist, blocks granule cell responses to mossy fiber stimulation (Garthwaite and Brodbelt, 1989). However, mossy fibers originate from a number of brain regions, and at least some mossy fibers may use transmitters other than or in addition to glutamate (see Mossy Fiber).

There has been considerable debate regarding the neurotransmitter of the climbing fibers. Early biochemical results suggested that aspartate rather than glutamate was the neurotransmitter (Wiklund et al., 1982). Homocysteate was also put forward as a candidate because selective lesion of the climbing fibers abolished its release from cerebellar slices (Vollenweider et al., 1990). However, climbing fiber terminals show high immunoreactivity for glutamate but not for aspartate or homocysteate (Zhang et al., 1990; Zhang and Ottersen, 1993). Furthermore, homocysteate staining in the cerebellar molecular layer has been localized to the glial processes surrounding the Purkinje cell dendrites (Grandes et al., 1991). Thus, the evidence to date points to glutamate being the transmitter of the climbing fibers.


GABA is the major inhibitory neurotransmitter of the cerebellum. In the cortex, it is utilized by Purkinje cells and all three local inhibitory interneurons (basket, stellate, (p.298) and Golgi cells). In the cerebellar nuclei, the cells that give rise to the nucleo-olivary projection are GABAergic (De Zeeuw et al., 1989; Fredette and Mugnaini, 1991).

The inhibitory nature of Purkinje cells was first demonstrated in Deiters’ nucleus. Ionophoretic application of GABA hyperpolarizes Deiters’ neurons (Obata et al., 1967), a target of Purkinje cell axons. IPSPs following Purkinje cell activation, as well as GABA-induced potentials, reverse near the same membrane potential and are mediated by an increased conductance to chlorine (Obata et al., 1970; ten Bruggencate and Engberg, 1971). Picrotoxin, which blocks the chloride channel associated with the GABA-A receptor, blocks both Purkinje cell IPSPs and GABA potentials in Deiters’ neurons. A reduction in the GABA-synthesizing enzyme glutamic acid dehydrogenase (GAD) in the interpositus nucleus is associated with destruction of the cerebellar hemisphere of the same side. Immunocytochemical studies have associated GAD activity with Purkinje cell axon terminals (Fonnum et al., 1970). In fact, GAD activity in Purkinje cell axon terminals is very high; 350–1,000 mM GABA can be synthesized per hour (Fonnum and Walberg, 1973).

Basket cell inhibition of Purkinje cell electrical activity is blocked by the application of agents known to block GABA receptors, such as bicuculline or picrotoxin. This effect has been demonstrated in several ways, involving a reduction in the ability of basket cell activation to (1) depress Purkinje cell spontaneous firing (Curtis and Felix, 1971), (2) depress Purkinje cell antidromic field potentials (Bisti et al., 1971), or (3) elicit IPSPs in Purkinje cells (Dupont et al., 1979). Also, ionophoretic application of GABA inhibits Purkinje cell spontaneous activity (Kawamura and Provini, 1970; Okamoto et al., 1976; Okamoto and Sakai, 1981). Moreover, basket cells take up radioactive GABA (Sotelo et al., 1972; Ljungdahl et al., 1978). Immunocytochemical studies have demonstrated the presence of GAD in basket cell terminals around Purkinje cell somata (McLaughlin et al., 1974; Chan-Palay et al., 1979; Oertel et al., 1981). Antibodies against GABA itself also strongly stain basket cells (Ottersen, 1993).

Stellate cells produce IPSPs in Purkinje cells that reverse around −75 mV and are blocked by bicuculline and picrotoxin, suggesting that they are mediated by GABA-A receptors (Midtgaard, 1992). However, these results by themselves do not exclude taurine as the neurotransmitter, for which there also is some evidence (Frederickson et al., 1978). However, immunocytochemical studies have directly demonstrated the presence of GABA in stellate cells, as well as a GABA reuptake system (Ottersen, 1993).

Antibodies against GAD or GABA label Golgi cells, including their synaptic terminals, indicating their GABAergic nature (Ottersen, 1993). However, most Golgi cells (≈70%) are also labeled by glycine, and the labeling for both GABA and glycine is reduced following induction of synaptic activity with elevated potassium concentrations (Ottersen et al., 1990b). Moreover, granule cells, the targets of Golgi cell axons, have both GABA and glycine receptors (Triller et al., 1987; Somogyi et al., 1989).


In addition to the inhibition produced by local circuit neurons, elements of the cerebellar cortex (in particular, the Purkinje cells) may be inhibited by release of norepinephrine following activation of the locus coeruleus (see Foote et al., 1983). This form of inhibition, first demonstrated by Bloom and collaborators (Siggins et al., 1971b), suggests that Purkinje cell excitability may be depressed for protracted periods by the (p.299) release of norepinephrine from terminals arising from the brainstem neurons. Rather than synapsing at specific points, the terminals seem to be widespread within the cortex. Their activation apparently produces a widespread release of catecholamines that hyperpolarize the Purkinje cells. Such hyperpolarization seems to be mimicked by application of cyclic adenosine-3′,5′-monophosphate (cAMP) (Siggins et al., 1971a,c), and norepinephrine may function via the activation of an electrogenic sodium pump similar to those in other central neurons (Phillis and Wu, 1981). Indeed, the possibility that an electrogenic sodium pump may be activated by norepinephrine is indicated, because the hyperpolarization is accompanied by a decreased ionic conductance change (Siggins et al., 1971c).

There is also evidence for dopaminergic cerebellar afferents projecting to the cerebellar nuclei and to the Purkinje and granular cell layers of the cortex (Simon et al., 1979). The raphe nuclei, which synthesize and release serotonin, project fibers to all parts of the cerebellar nuclei and cortex (Takeuchi et al., 1982). These terminate at mossy fiber rosettes diffusely throughout the granular layer; in the molecular layer, they bifurcate like parallel fibers and synapse with the intrinsic neurons (Chan-Palay, 1977). In the molecular and granular layers, beaded fibers with fine varicosities have been labeled with serotonin-specific antibodies (Takeuchi et al., 1982).


Although the majority of mossy fibers appear to use glutamate as a neurotransmitter, some mossy fibers appear to use other neurotransmitters. For example, among the peptides, somatostatin-immunoreactive fibers have been shown to enter the cerebellum (Inagaki et al., 1982); these probably end as mossy fibers. Acetylcholine (ACh) is present in some mossy fiber terminals isolated as synaptosomes (Israel and Whittaker, 1965), and some mossy fibers contain acetylcholinesterase (Phillis, 1968). Choline acetyltransferase (ChAT), the enzyme for ACh synthesis, has been demonstrated in mossy fibers and glomeruli through immunocytochemical methods (Kan et al., 1978, 1980). Subsequent labeling studies have suggested that ChAT-positive mossy fibers may be largely restricted to cerebellar regions that receive vestibular input (Barmack et al., 1992). However, a role for ACh as a transmitter has not been supported by pharmacological or physiological studies, which have shown that while cholinergic antagonists fail to block mossy fiber evoked activity, glutamate blockers are effective, even in the vestibular-related areas of the cerebellum (Crepel and Dhanjal, 1982; Rossi et al., 1995).



That dendrites are capable of electroresponsive activity and are not simple, passive cables was first shown in Purkinje cells. The earliest recordings indicating that the dendrites are active were made from alligator cerebellum. Here, intradendritic recordings revealed large dendritic spikes in response to parallel fiber stimulation (Llinás and Nicholson, 1971). The injection of hyperpolarizing current allowed these spikes to be dissected into several all-or-none components. From these early studies, it was proposed that there are several “hot spots” in the dendrites that are capable of spike generation (p.300) and that dendritic spikes travel toward the soma in a discontinuous manner. Subsequent intradendritic recordings from pigeon Purkinje cells showed that dendritic spikes are calcium dependent (Llinás and Hess, 1976). It was not until cerebellar slice preparation was used, however, that the dendritic properties of Purkinje cells were revealed in all their complexity.

The types of spontaneous action potentials that may be seen at different levels in a mammalian Purkinje cell soma and dendrites are illustrated for an in vitro experiment in Fig. 7.13. Typical bursts, consisting of fast sodium spikes and a terminal, slower rising calcium spike, are seen at the somatic level (Fig. 7.13B). Recordings obtained at different levels in the dendritic tree are shown in Fig. 7.13C–E. The decrease in amplitude of the fast spike that occurs as recordings are made farther from the soma indicates clearly that the fast sodium action potentials seen at the soma do not actively invade the dendrites. Rather, they are electrotonically conducted and can be detected only to about mid-dendritic level, their amplitude decrements rather quickly with distance from the soma.

The bursting calcium-dependent spike, on the other hand, is large and rather prominent in the upper dendrites, indicating a differential distribution for sodium and calcium


Fig. 7.13. Composite illustration of recordings made from different regions of a Purkinje cell in vitro. A: Drawing of typical mammalian Purkinje cell. B: Fast action potentials dominate this recording, with slower membrane oscillations. C–E: As the electrode moves away from the soma, (1) the amplitude of the fast, Na-dependent action potentials progressively decreases until they are not seen in the most distal branches; and (2) the slow, prolonged, Ca2+-dependent action potentials increase in amplitude and become distinct in the distal branches. Although the dendritic spikes are discontinuously propagated toward the soma, the somatic spikes do not actively invade the dendrites. [Modified from Llinás and Sugimori, 1980a, with permission.]


Fig. 7.14. Purkinje cell dendritic recording in the presence of TTX. Short depolarizing pulses elicit Ca-dependent plateau potentials and Ca2+ spikes. As the current amplitude is increased, the plateau responses increase in duration, and full spike bursts are generated. B: The calcium dependence of both the plateau and the spike bursts is demonstrated by their complete abolition after Cd2+ has been added to the TTX bathing solution.

conductances. Furthermore, direct stimulation of dendrites after the application of TTX, as shown in Fig. 7.14A, produces two types of calcium-dependent electroresponsiveness. A small stimulus can generate a plateau-like response and a burst of action potentials. Because both responses can be blocked by cobalt, cadmium, or D600 (see Fig. 7.14B), it must be concluded that the dendrites of the Purkinje cell are capable of generating calcium-dependent spikes, which may be of either a prolonged plateau form or clear, all-or-nothing action potentials.

The Purkinje cells thus demonstrate the following set of voltage-dependent ionic conductances. As discussed earlier, in the soma there are (1) a rapid, inactivating Hodgkin-Huxley sodium current that generates a fast spike; (2) a fast voltage-activated potassium current that generates the afterhyperpolarization following a fast spike; (3) a calcium-activated potassium conductance, and (4) a noninactivating, voltage-activated sodium conductance capable of generating repetitive firing of the Purkinje cell following prolonged depolarization. At the dendritic level, on the other hand, excitability seems to be due mainly to a voltage-activated calcium conductance increase. This conductance may generate a low plateau potential or calcium spikes (Fig. 7.14A), and the spikes may be followed by an increase in both voltage-activated and calcium-activated potassium conductances.

It is therefore clear that the complex electrical responses observed in these cells after direct stimulation or activation of climbing or parallel fibers are largely due to the electroresponsive properties of the cells themselves.


Optical probes have been used to mark the spatial distribution of voltage-sensitive ionic channels in Purkinje cells. The sodium conductance is restricted to the soma and axon as visualized by using fluorescently labeled TTX (Sugimori et al., 1986).

Mapping of the distribution of an increase in intracellular calcium concentration ([Ca2+]i) during spontaneous and evoked Purkinje cell activity allows visualization of the probable location of calcium channels in the somatodendritic membrane. This has been done in experiments using Arsenazo III absorption (Ross and Werman, 1987) and Fura-2 as a calcium indicator. Experiments using the fluorescent Ca2+ indicator Fura-2 have shown that during spontaneous bursting, the [Ca2+]i increases first in the fine dendritic branches, where the increase is also the largest (Tank et al., 1988). The [Ca2+], is later seen to increase in the dendritic trunk, and by this time it has begun to subside in the fine dendrites. The [Ca2+]i in the soma increases very little. This temporal sequence of increased calcium activity, first in the distal and then in the proximal dendrites, supports the electrophysiological description of the two calcium conductances—the low-threshold plateau and all-or-none calcium-dependent dendritic spikes (see Fig. 7.13). The presence of voltage-activated calcium channels in the spiny branchlets provides a mechanism whereby parallel fiber EPSPs can be enhanced by slow local increases in calcium conductance. In contrast, when the synaptic activity is in the larger dendritic branches, full calcium-dependent dendritic spikes can be generated in the main dendritic tree. Climbing fiber synapses tend to depolarize the main dendritic tree, producing full dendritic spikes. If a cell loaded with Fura-2 is depolarized by somatic current injection, the increased [Ca2+]i in the dendrites is not uniform. Rather, there are well-localized areas of marked increases, supporting the earlier hypothesis of “hot spots” of calcium influx (Llinás and Nicholson, 1971). Thus, the distribution of calcium channels over the dendritic tree is a critical element in the fine tuning of the electrophysiological sophistication of this most remarkable cell.

Unlike the climbing fiber input, which produces a widespread activation of the Purkinje cell dendritic tree, parallel fibers have the ability to excite small compartments of the dendritic tree. Given the complexity of its dendritic tree, the Purkinje cell has tremendous computational power when activated by the parallel fiber system. Indeed, the actual number of possible functional states will depend on the nature and, therefore, on the number of independent dendritic compartments activated. In vitro experiments with the calcium-sensitive dye Calcium Green have suggested that the smallest functional units of neuronal integration may in fact be the individual spines (Denk et al., 1995). Figure 7.15 shows an example of independent activation of single spines in a Purkinje cell following parallel fiber activation. The cell was filled with Calcium Green using a whole-cell patch electrode and imaged using two-photon fluorescence laser scanning microscopy. At low magnification the complete dendritic tree is shown (Fig. 7.15A), whereas at higher magnifications the individual spines of the spiny branch-lets are clearly resolved (Fig. 7.15B-D). Trains of electrical microstimulation pulses applied to a restricted parallel fiber group evoked well-resolved EPSCs recorded at the soma by the patch electrode (Fig. 7.15E) and produced activation of individual dendritic spines, as measured by taking the difference in fluorescence intensity between the resting and stimulated conditions (Fig. 7.15F). Because there are approximately 107 (p.303)


Fig. 7.15. Single spine activation via parallel fiber stimulation. After Calcium Green diffusion from the patch electrode, a complete Purkinje cell dendritic tree (A) is shown. Functional imaging was performed in a different cell, which is shown at three magnifications (B-D). Note that single spines are well resolved. A-D: Maximal value projections of a stack of optical sections. In (C) and (D), the spine activated in E is indicated by arrows. A train of low-amplitude parallel fiber stimuli generated small subthreshold synaptic currents (E) at the soma. In (F), difference images (stimulated minus resting; Δ) and the resting fluorescence level (Σ) taken at four different depths show the single spine calcium response produced by the parallel fiber stimuli. [Modified from Denk et al., 1995.]

(p.304) Purkinje cells in humans, each of which has over 100,000 spines, the number of computational events implementable from a neuronal point of view for the output layer of the cerebellar cortex exceeds 1012.


We have seen that the climbing fiber and the mossy fiber–granule cell–parallel fiber pathways are the two main types of afferents to the cerebellum as a whole and to the Purkinje cells in particular. These afferent systems differ dramatically in their interactions with the Purkinje cells. For example, the Purkinje cell and its climbing fiber afferent have a one-to-one relationship, whereas the relationship between the Purkinje cell and the mossy fiber–parallel fiber system can be characterized as many-to-many. Moreover, the directionality of the parallel fibers imparts a mediolateral orientation to Purkinje cell activation by the mossy fiber–parallel fiber system, whereas the climbing fiber system, as we shall see, is organized to produce synchronous activation of specific groupings of Purkinje cells, groupings that often have a rostrocaudal orientation. Their electrophysiological and anatomical differences lead to distinct functional roles for these two systems, which we discuss later.

Let us first consider the climbing fiber system. As a result of the electrotonic coupling between inferior olivary neurons and the topography of the olivocerebellar projection, this system generates synchronous (on a millisecond time scale) complex spike activity in rostrocaudal bands of Purkinje cells (Fig. 7.16B). These bands are normally only about 250 μm wide in the mediolateral direction but can be several millimeters long in the rostrocaudal direction and may extend down the walls of the cerebellar folia and across several lobules (Sugihara et al., 1993; Yamamoto et al., 2001). It is important to realize that the banding structure shown in Fig. 7.16B is an average from a long (20-min) recording and that the moment-to-moment synchrony distribution is dynamically controlled by afferents to the inferior olive (Llinás, 1974). In fact, instead of providing the primary drive for activity in the olivocerebellar system, the major role of olivary afferents may be to determine the pattern of “effective” electronic coupling between olivary neurons and thereby the distribution of synchronous complex spike activity across the cerebellar cortex. This idea is supported by results showing that spontaneous climbing spike activity persists following the block of glutamatergic and GABAergic input to the inferior olive (Lang, 2001, 2002).

The role of GABAergic and glutamatergic olivary afferents in shaping the patterns of olivocerebellar synchrony has been investigated using multiple electrode recordings of complex spike activity (Llinas and Sasaki, 1989; Lang et al., 1996; Lang, 2001, 2002) and voltage-sensitive dye imaging of inferior olivary activity (Leznik et al., 2002). The effect of neurotransmitter release within olivary glomeruli was proposed to increase the conductance of the membrane adjacent to the gap junctions. This would shunt any current flowing between olivary cells, thus decoupling their activity (Llinás, 1974). Evidence supporting this hypothesis was obtained by making multiple electrode recordings of complex spike activity and comparing the patterns of synchrony before and after elimination of GABAergic activity in the inferior olive (Lang et al., 1996). GABAergic activity was blocked either with microinjections of picrotoxin into the inferior olive or by destroying the cerebellar nuclei, the source of the GABAergic projection. (p.305)


Fig. 7.16. Complex spike synchrony patterns revealed by multiple electrode recording. A: (Top) Schematic of rat cerebellum showing the placement of an array of microelectrodes on lobule crus 2a. Each electrode records the complex spikes from a single Purkinje cell. (Bottom) Synchrony scale for plots in B and C. B: Distribution of synchronous complex spike activity with respect to the activity of reference cell M under control conditions. Each circle represents the location of an electrode in the recording array, where left and right correspond to lateral and medial on crus 2a and top and bottom correspond to rostral and caudal. The area of a circle is proportional to the level of synchronous activity between the cell at that location and the selected reference cell M. Synchrony is defined as the normalized cross correlation coefficient at 0 ms time lag as calculated from the spike trains of the two cells using a time bin of 1 ms. Note how cells showing high levels of synchrony with cell M form a column or band that roughly runs from the top to the bottom of the plot (i.e., rostral to caudal). Scale bar indicates the spacing of the electrodes. C: Distribution of synchronous complex spike activity after a lesion the cerebellar nuclei and therefore loss of GABAergic activity within the inferior olive. Note the higher synchrony level compared with B and the more uniform distribution.

Figure 7.16C shows the widespread distribution of synchronous complex spike activity that follows the loss of GABAergic input to the inferior olive. Note that the banding pattern seen under control conditions has been replaced by a uniform distribution. In contrast, it was shown that blocking glutamatergic input to the inferior olive actually accentuates the banding pattern (Lang, 2001). Thus, GABAergic and glutamatergic olivary afferents act in a complementary fashion to shape the exact pattern of synchronous complex spike activity across the cerebellum.

Optical recordings from in vitro inferior olivary slices treated with voltage-sensitive dyes have provided direct visualization of the GABAergic modulation of inferior olivary cell coupling. As shown in Fig. 7.17A, B (Control), an electrical stimulus delivered to the surrounding white matter results in coherent oscillatory activity in small, discrete clusters of inferior olivary neurons. This is presumably a result of electrotonic coupling via gap junctions. After the application of picrotoxin, an identical stimulus (p.306) generated a much stronger optical signal (Fig. 7.17A, B, Picrotoxin). This enhanced signal is not due to an increase in the responses of individual cells to the stimulus. Indeed, intracellular recordings show that the responses of individual cells before and after picrotoxin application are similar (Fig. 7.17C). Rather, the large increase in the dye signal reflects a more coherent population response, as a result of the greater efficacy of electrotonic coupling after blocking GABAergic synapses with picrotoxin. These examples show that pharmacological manipulations can dramatically alter the patterns of synchronous activity in the olivocerebellar system. Changes in synchrony patterns have also been demonstrated to be associated with movements made by animals performing a motor task (Welsh et al., 1995), which points to their significance for normal cerebellar function.


Fig. 7.17. Effects of picrotoxin on intracellularly and optically recorded oscillations in the inferior olive measured using a voltage-sensitive dye. A: A frame of imaged ensemble neuronal oscillating clusters in control conditions (left) and after addition of 20 μM picrotoxin (right) to block GABAergic activity. In the control condition, several representative clusters in the bottom right of the inferior olive are outlined in black. The clusters were defined as areas with pixels above a selected threshold value (0.007%). After picrotoxin an additional threshold level (0.014%) was added to delineate the areas with the highest response. B: Higher magnification view of boxed areas in A. Note how picrotoxin significantly increased the size of the clusters by merging several smaller discrete areas into a continuous larger area. C: Intracellular recording from an inferior olivary neuron showing spontaneous subthreshold oscillations before (left) and after (right) bath application of 20 μM picrotoxin. [Modified from Leznick et al., 2002.]

(p.307) To understand more fully the functional significance of the olivocerebellar circuit, we must consider what effect this activity has on the cerebellar nuclear cells. This is the case because the ultimate role of the cortex is to help determine the firing of cerebellar nuclear cells. We consider this below and then finish by adding the mossy fiber–granule cell–parallel fiber circuitry to the picture.

The activity of the cerebellar nuclei is regulated in three ways: (1) by excitatory input from collaterals of the cerebellar afferent systems, (2) by inhibitory inputs from Purkinje cells activated over the mossy fiber pathways, and (3) by inputs from Purkinje cells activated by the climbing fiber system. The effect of these inputs on cerebellar nuclear cells is shown by the intracellular recording of the response of these neurons to white matter stimulation (Fig. 7.18, right). The stimulus activates a variety of axons that are running through the white matter (Fig. 7.18, left), and as a result the response of these cells has five parts as shown in the figure: (1) an initial EPSP due to antidromic activation of the mossy fiber collaterals to the nuclear cell (1 in Fig. 7.18); (2) an IPSP, which results from direct excitation of Purkinje axons projecting to the nuclear cell (2 in Fig. 7.18); and (3 and 4) a second EPSP-IPSP sequence (3 and 4 in Fig. 7.18) with a latency of 3–4.5 msec. The EPSP results from climbing fiber collateral activation of the cerebellar nuclear cells, and the IPSP is generated as a result of climbing fiber activation of Purkinje cells that in turn project onto the cerebellar nucleus. Finally, (5) the second IPSP is terminated by a rebound response, which is due to the intrinsic membrane properties of the cerebellar nuclear cells themselves


Fig. 7.18. Response of cerebellar nuclear cells to white matter stimulation. A: Drawing of elements activated after white matter stimulation. B: White matter stimulation activates mossy fibers, climbing fibers, and Purkinje cell (PC) axons. The first response (1), a graded EPSP, is due to activation of the mossy fiber collaterals; the second (2), a small IPSP, is due to direct stimulation of Purkinje cell axons. The third response (3), a graded EPSP, is due to activation of climbing fiber collaterals. Finally (4), the powerful IPSP and smaller IPSPs follow climbing fiber activation of Purkinje cells. Although the cell is at the resting potential, the hyperpolarization is often sufficient to elicit a rebound response in the cerebellar nuclear cell (5).


Fig. 7.19. Diagram of the circuit involved in the production of rhythmic activity in the olivocerebellar system. (1) Rhythmic activity in the inferior olivary neurons is transmitted to the Purkinje cells (PC), where it is transformed to complex spikes (2) to the cerebellar nuclear projecting cells (white somata) and inhibitory cells (filled soma) eliciting EPSPs. Complex spikes trigger high-frequency firing of Purkinje cell axons that impinge on the cerebellar nuclear cells with powerful IPSPs and rebound firing (3). Thus bursts of spikes are transmitted to the rest of the nervous system including the cerebellum (as mossy fibers). The cerebellar nuclear cells projecting to the inferior olive (IO) are inhibitory and synapse in the glomeruli. (Filled synaptic terminals are inhibitory, and open synaptic terminals are excitatory.)

(Fig. 7.18). Thus, the response in Fig. 7.18 is a combination of the properties of the synaptic circuit and the intrinsic properties of the Purkinje and cerebellar nuclear cells.

We can now consider the effect of synchronous olivocerebellar activity on the output of the cerebellar nuclei. In this regard, it is of particular interest that punctate and rather powerful synaptic EPSP-IPSP sequences are often followed by a rebound spike burst, as is seen in Fig. 7.18 (right), because this type of EPSP-IPSP sequence is likely to occur as a result of synchronous olivocerebellar activity. (Remember that there is a large convergence of Purkinje axons onto individual cerebellar nuclear cells.) This means that if a sufficient number of inferior olivary neurons, having a common rhythmicity, are activated synchronously, a large and equally synchronous activation of Purkinje cells will occur. This should in turn produce a large IPSP followed by a rebound burst response in the cerebellar nuclear cells. In fact, this is what occurs when harmaline, a tremorigenic agent known to act directly on the inferior olive (de Montigny and Lamarre, 1973; Llinás and Volkind, 1973; Llinás and Yarom, 1986), is administered. Harmaline activation of the inferior olive produces alternating inhibition and rebound activation in cerebellar nuclear cells (Fig. 7.19). This activity has been demonstrated in vitro and probably occurs in vivo, as indicated by the increased rhythmicity and synchrony of complex spike activity observed in multiple electrode recordings following systemic harmaline injection (Llinás, 1985; Llinás and Sasaki, 1989; (p.309) Yamamoto et al., 2001). The behavioral consequence of these synchronous bursts from the cerebellar nuclei is a phase-locked tremor.

In contrast to the punctate nature of cerebellar activation by the olivocerebellar system, the mossy fiber—parallel fiber system provides a continuous and very delicate regulation of the excitability of the cerebellar nuclei, brought about by the tonic activation of simple spikes in Purkinje cells, which ultimately generates the fine control of movement known as motor coordination. The fact that the mossy fibers inform the cerebellar cortex of both ascending and descending messages to and from the motor centers in the spinal cord and brainstem gives us an idea of the ultimate role of the mossy fiber system: it informs the cortex of the place and rate of movement of limbs and puts the motor intentions generated by the brain into the context of the status of the body at the time the movement is to be executed. Moreover, through its effects on the inhibitory GABAergic cerebellar nuclear cells, which project back to the inferior olive, it helps shape the pattern of coupling among olivary cells and hence the synchrony distribution in the upcoming olivocerebellar discharge. (p.310)