Cerebellar long-term depression (LTD) is a form of synaptic plasticity, first described by Ito and co-workers, in which simultaneous activation of two excitatory inputs to a Purkinje neuron, parallel fibers (PFs) and climbing fibers (CFs), results in a sustained depression of PF synaptic drive. The purpose of this target article is not to assess the possible role of this synaptic alteration in motor learning, an issue which is addressed by other authors in this volume, nor is it to provide a detailed summary of the work on cerebellar LTD to this point (see Linden & Connor 1993; Crépel et al. 1993 for review) or to place cerebellar LTD within the context of other forms of persistent synaptic depression that occur within the mammalian brain (see Linden 1994b). Rather, it is to discuss results obtained using a very reduced preparation for the study of LTD, embryonic Purkinje neurons grown in culture and stimulated with exogenous excitatory amino acids, and to consider some advantages and limitations of this approach. Recent work using this preparation has suggested that three processes are necessary for the induction of cerebellar LTD: Ca influx through voltage-gated channels, Na influx through AMPA receptor-associated channels or voltage-gated Na channels, and protein kinase C activation - which is dependent upon activation of the metabotropic glutamate receptor mGluRl. In addition, input-specific induction of LTD has been demonstrated in this preparation Under conditions where both spontaneous and evoked neurotransmitter release are reduced or eliminated, indicating that postsynaptic alterations are sufficient to confer this important computational property.

Long-term depression (LTD) of synaptic transmission at parallel fibre-Purkinje cell synapses is thought to be a cellular substrate of motor learning in the cerebellum. This use-dependent change in synaptic efficacy is induced by conjunctive stimulation of Parallel fibres and climbing fibres. Researchers agree that the induction of LTD requires, as an initial step, a calcium influx via voltage-gated Ca2+ channels into a Purkinje cell, together with activation of ionotropic (AMPA) and probably metabotropic subtypes of glutamate receptors of this cell. Indeed, due to the lack of specific antagonist, the final demonstration of the contribution of metabotropic receptors in the LTD induction process, under functional conditions, remains unanswered. The debate is now focused on the second-messenger processes leading to LTD of synaptic transmission at parallel fibre-Purkinje cell synapses, after the calcium influx into the cell. All researchers agree that a calcium-dependent cascade of events including activation of protein kinase C is necessary for LTD induction. In contrast, the recruitment in the LTD induction of another cascade, also triggered by Ca2+, that is, through synthesis of nitric oxide and cyclic GMP, remains controversial.

On the other hand, growing evidence suggests that these chains of reaction underlying LTD might ultimately lead to a genuine change in the functional characteristics of AMPA receptors at the parallel fibre-Purkinje cell synapses.

The cellular basis of motor learning in the cerebellum has been attributed mostly to long-term depression (LTD) at excitatory parallel fiber (PF)-Purkinje cell (PC) synapses. LTD is induced when PFs are activated in conjunction with a climbing fiber (CF), the other excitatory input to PCs. Recently, by using whole-cell patch-clamp recording from PCs in cerebellar slices, a new form of synaptic plasticity was discovered. Stimulation of excitatory CFs induced a long-lasting (usually longer than 30 min) “rebound potentiation (RP)” of γ-amino-butyric acid A (GABAa)-receptor mediated inhibitory postsynaptic, currents (IPSCs). As in LTD, induction of RP requires transient elevation of intracellular calcium concentration ([Ca2+]1) due to activation of voltage-gated Ca2+ channels. Activity of inhibitory synapses seems also to be necessary for RP to occur. RP is mainly due to up-regulation of postsynaptic GABAa receptor function, since PC response to bath-applied exogenous GABA is also potentiated with a time course similar to RP. The difference in the time scale between the Ca2+ transients (10–30 sec) and the durations of RP (>30 min) strongly suggests that some intracellular biochemical machinery is involved. Pharmacological evidence suggests that protein kinases are involved in RP of inhibitory synapses and LTD of excitatory PF synapses. Besides the well-described LTD, RP could be a cellular mechanism that plays an important role in motor learning.

Interest in the role of nitric oxide (NO) in the nervous system began with the demonstration that glutamate receptor activation in cerebellar slices causes the formation of a diffusible messenger with properties similar to those of the endothelium-derived relaxing factor. It is now clear that this is due to the Ca2+/calmodulin-dependent activation of the enzyme NO synthase, which forms NO and citrulline from the amino acid L-arginine. The cerebellum has very high levels of NO synthase, and although it has low levels of guanylyl cyclase, cerebellar cyclic guanosine monophosphate (cGMP) levels are an order of magnitude higher than in other brain regions. A transcellular metabolic pathway is also present in the cerebellar cortex to recycle citrulline back to arginine. The NO formed binds to and activates soluble guanylyl cyclase to elevate cGMP levels in target cells. Studies employing NADPH-diaphorase, a selective histochemical marker for NO synthase, together with immunohistochemistry, in situ hybridization and biochemical studies have indicated that NO production occurs in granule and basket cells in the cerebellar cortex, whereas cGMP formation appears to occur largely in other cells, including Purkinje cells. Given that a long-term depression of AMPA currents can be seen in isolated Purkinje cells, this anatomical localization suggests that NO cannot play an essential role in the induction of this form of synaptic plasticity.

In vitro studies using cerebellar slices or cultures, as well as in vivo studies, have demonstrated that NO formation and release occur in the cerebellar cortex in response to NMDA receptor activation. Using intracerebellar microdialysis in awake-behaving animals, we have found that this is associated with a large increase in extracellular cGMP levels in the cerebellar cortex. A similar increase in cGMP efflux is seen in response to activation of AMPA or metabotropic glutamate receptors, or to activation of the climbing fiber input. The increase in extracellular cGMP required the Ca2+-dependent activation of NO synthase, and was potentiated by inhibition of phosphodiesterases or organic anion transport. These results suggest a possible role for cGMP as an intercellular messenger in the cerebellar cortex.

In summary, stimuli that elevate Ca2+ levels in granule or in basket cells will activate NO synthase, which by binding to its receptor-soluble guanylyl cyclase – causes an increase in cGMP. cGMP may then act through protein kinases, phosphodiesterases, and ion channels or receptors to affect cerebellar function.

This article reviews models of the cerebellum and motor learning, from the landmark papers by Marr and Albus through those of the present time. The unique architecture of the cerebellar cortex is ideally suited for pattern recognition, but how is pattern recognition incorporated into motor control and learning systems? The present analysis begins with a discussion of exactly what the cerebellar cortex needs to regulate through its anatomically defined projections to premotor networks. Next, we examine various models showing how the microcircuitry in the cerebellar cortex could be used to achieve its regulatory functions. Having thus defined what it is that Purkinje cells in the cerebellar cortex must learn, we then evaluate theories of motor learning. We examine current models of synaptic plasticity, credit assignment, and the generation of training information, indicating how they could function cooperatively to guide the processes of motor learning.

The persistence of many contrasting notions of climbing fiber function after years of investigation testifies that the issue of climbing fiber contributions to cerebellar transactions is still unresolved. The proposed capabilities of the climbing fibers cover an impressive spectrum. For many researchers, the climbing fibers signal errors in motor performance, either in the conventional manner of frequency modulation or as a single announcement of an “unexpected event”. More controversial is the effect of these signals on the simple spike modulation of Purkinje cells. In some hands, they lead to a long-term depression of the strength of parallel fiber synapses, while, in other hands, they lead to a short-lasting enhancement of the responsiveness of Purkinje cells to mossy fiber inputs or contribute to the often-seen reciprocal relation between complex and simple spike modulation. For still other investigators, the climbing fibers serve internal timing functions through their capacity for synchronous and rhythmic firing. The above viewpoints are presented in the spirit of trying to reach some consensus about climbing fiber function. Each point of view is introduced by summarizing first the key observations made by the respective proponents; then the issues of short-lasting enhancement, reciprocity between complex and simple spikes, and synchrony and rhythmicity are addressed in the context of the visual climbing fiber system of the vestibulocerebellum.

Although there is increasing agreement that the cerebellum plays an important role in motor learning, the basic substance of what constitutes motor learning has been difficult to define. Unless motor learning is somehow radically different from other forms of learning, it must involve relatively simple stimulus-stimulus and stimulus-response associations. All forms of learning, including purely sensory associations and cognitive learning as well as motor learning, effect changes in behavior. However, a singular characteristic of motor learning is that it adjusts joint and limb mechanics by altering the neural input to muscles through practice and mental rehearsal. The hypothesis proposed here is that the cerebellum plays an important role in motor learning by forming and storing associated muscle activation patterns for the time-varying control of limb mechanics. By modulating the cocontraction of agonist-antagonist muscles through adjustments in the timing and amplitude of muscle activity, the viscoelastic properties of joints can be appropriately regulated throughout Movement and adapted for transitions between postures and movements. Optimal control of joint viscoelastic properties cannot be achieved by online corrections initiated by reflex feedback because of the delays and consequent instabilities incurred. Instead, strategies for optimizing muscle activation patterns or synergies must be learned from the temporal association of proprioceptive stimuli signaling muscle lengths and forces and the rates of changes in these parameters, with reinforcement occurring when the movement achieves its objective. Such strategies would involve varying degrees of cocontraction or reciprocal inhibition of agonist-antagonist muscles that ultimately contribute to joint and limb stiffness. Evidence from neural recordings and clinical and experimental lesion studies are presented, suggesting that the cerebellum uses teleceptive and proprioceptive feedback as feedforward conditioned stimuli for specific muscle activation patterns contributing to joint stiffness (i.e., agonist-antagonist muscle synergies) for particular tasks and postures. A wide variety of observations are thought to be consistent with such a role for the cerebellum, but ultimately additional experiments could confirm or disconfirm this hypothesis.

Brindley proposed that we initially generate movements “consciously”, under higher cerebral control. As the movement is practiced, the cerebellum learns to link within itself the context in which the movement is made to the lower level movement generators. Marr and Albus proposed that the linkage is established by a special input from the inferior olive, which plays upon an input-output element within the cerebellum during the period of the learning. When the linkage is complete, the occurrence of the context (represented by a certain input to the cerebellum) will trigger (through the cerebellum) the appropriate motor response. The “learned” movement is distinguished from the “unlearned” conscious movement by its now being automatic, rapid, and stereotyped. The idea is still controversial, but has been supported by a variety of animal studies and, as reviewed here, is consistent with the results of a number of human PET and ablation studies. I have added to the idea of context-response linkage what I think is another important variable: novel combinations of downstream elements. With regard to the motor system and the muscles, this could explain how varied combinations of muscles may become active in precise time-amplitude specifications so as to produce coordinated movements appropriate to specific contexts. In this target article, I have further extended this idea to the premotor parts of the brain and their role in cognition. These areas receive influences from the cerebellum; they are active both in planning movements that are to be executed and in thinking about movements that are not to be executed. From recent evidence, the cerebellar output extends even to what has been characterized as the ultimate frontal planning area, the “prefrontal” cortex, area 46. The cerebellum thus may be involved in context-response linkage, and response combination even at these higher levels. The implication would be that, through practice, an experiential context would automatically evoke a certain mental action plan. The plan would be in the realm of thought, and could – but need not – lead to execution. The specific cerebellar contribution would be one of the context linkage and the shaping of the response, through trial and error learning. The prefrontal and premotor areas could still plan without the help of the cerebellum, but not so automatically, rapidly, stereotypically, so precisely linked to context, or so free of error. Nor would their activities improve optimally with mental practice.

We ask what cerebellum and basal ganglia “do,” arguing that cerebellum tunes motor schemas and their coordination. We argue for a synthesis of models addressing the real-time role and error signaling roles of climbing fibers. “Synthetic PET” bridges between regional and neuro-physiological studies, while “synaptie eligibility” relates the neurochemis-try of learning to neural and behavioral levels. [CRÉPEL et al.; HOUK et al.; KANO; LINDEN; SIMPSON et al.; SMITH; THACH; VINCENT]

In this commentary, I review the articles by Crépel et al., Linden, and Vincent to compare and contrast the similarities and differences between hippocampal LTP and cerebellar LTD. In particular, the role of glutamate receptors in the induction of plasticity, the participation of different second messenger pathways, and the mechanisms involved in expression/maintenance of these forms of synaptic plasticity are discussed. [CRÉPEL et al.; LINDEN; Vincent]

The present commentary considers the question of what must be learned in different types of motor skills, thereby limiting the question of what should be adjusted in the APG model in order to explain successful learning. It is concluded that an open loop model like the APG might well be able to describe the learning pattern of motor skills in a stable, predictable environment. Recent research on saccadic plasticity, however, illustrates that motor skills performed in an unpredictable environment depend heavily on sensory (mostly visual) feedback, [HOUK et al.]

The role of nitric oxide in cerebellar synaptic plasticity is controversial. Two recent papers on nitric oxide-(NO) and synaptic plasticity are discussed which use new methods for investigating the actions of NO and suggest that under certain conditions it is an essential anterograde messenger. In the light of current thinking about the diversity of expression of hippocampal LTP, should a presynaptic contribution to cerebellar LTD be re-examined? If NO acts orthogradely, might its product cGMP have a retrograde action? [CRÉPEL et al.; LINDEN; VINCENT]

The primary assumption made in this series of target articles is that the cerebellum is directly involved in motor control. However, in my opinion, there is ample and growing experimental evidence to question this classical view, whether or not learning is involved. I propose, instead, that the cerebellum is involved in the control of data acquisition for many different sensory systems, [CRÉPEL et al., HOUK et al., SMITH, THACH]

Long-term modifications of synaptic efficacy at excitatory synapses are considered as a putative cellular substrate for learning and memory processes. Cerebellar long-term depression (LTD) occurring at parallel fibers (PF)-Purkinje neurons (PN) synapses is thought to repre-sent a cellular model of motor learning. Crépel et al. and Linden describe this phenomenon by utilizing different tissue preparations. [CRÉPEL et al., LINDEN]

The theories of cerebellar function presented in this BBS special issue cannot be reconciled with the established induction properties of cerebellar LTD. At the same time, the authors presenting their research on cerebellar LTD do not appear very interested in function.

The adjustable pattern generator (APG) model addresses physiological detail in a manner that renders it eminently testable. However, the problem for which the APG was developed, namely, limb control, may be computationally too complex for this purpose. Instead, it is proposed that recent empirical and theoretical advances in understanding the role of the cerebellum in low-level saccadic control could be used to refine and extend the APG. [HOUK et al.]

This study examines how a Purkinje cell receives its appropriate olivary error signal during the learning of compound movements. We suggest that the Purkinje cell only reinforces those target pyramidal cells which already participate in the movement, subsequently reducing any repeated error signal, such as its own climbing fiber input, [simpson et al.; smith]

The cerebellum probably obeys the rules of sensorimotor integration common in the nervous system. One such a rule is formulated: the nervous system organizes spatial frames of reference for the sensorimotor apparatus and produces voluntary movements by shifting their origin points. We give examples of spatial frames of reference for different single- and multi-joint movements including locomotion and also illustrate that the process of motor development and learning may depend critically on the formation of appropriate frames of reference and the organism's ability to manage them. We suggest that a solution to the problem of sensorimotor integration may not be trivial and may actually change the mental and experimental paradigms used in the understanding of the cerebellum and other brain structures, [HOUK et al.; SIMPSON et al.; SMITH; TIIACH]

A major theme of the systems physiologists is the critical timing function of the cerebellum. However, the biophysicists do not appear to directly address the biophysical basis of the adaptive timing competence implicated in the physiological and behavioral data. Thus, the bridge between the macroscopic and microscopic data bases seems to be incomplete in a critical area. We report successful results from an attempt to add the missing part of the bridge. It comes in the form of a model of how the second messenger system activated by parallel fiber inputs - the mGluR channel - can literally bridge the temporal gap between CS and CR, both in standard conditioning tasks and more generally, [CRÉPEL et al.; HOUK et al.; KANO; LINDEN; SIMPSON et al.; SMITH; THACH; VINCENT]