Cerebellum and Motor Learning
INTRODUCTION

“Correction and instruction are the way to life.” Proverbs 6:23

The cerebellum is a coordinative brain structure that functions in motor adaptation and learning. The organizational structure of the cerebellum integrates a variety of input that evaluates the disparity between intended action and the actual physical state of the body. This error detection and correction is important in skill acquisition and is strongly mediated by the modification of Purkinje cell activity. Climbing fibers appear to play an important role in this modification through the effects of long-term depression.
Figure 1- The Cerebellum (http://goo.gl/9WJKn)
Figure 1- The Cerebellum (http://goo.gl/9WJKn)

FUNCTIONAL ANATOMY


Overview
The cerebellum is located in the posterior cranial fossa, inferior to the occipital and temporal lobes and dorsal to the brainstem (Knierim, 1997). It is connected to the brainstem via the inferior cerebellar peduncle, middle cerebellar peduncle, and superior cerebellar peduncle. Transversely, the cerebellum is divided into the anterior, posterior, and flocculonodular lobes. Longitudinally, it is divided into the vermis, intermediate hemisphere, and lateral hemisphere (Squire et al., 2003). In functional terms, the cerebellum is separated into the vestibulocerebellum (flocculonodular lobe), spinocerebellum (vermis and intermediate hemisphere), and the cerebrocerebellum (lateral hemisphere). These cerebellar cortical structures project to corresponding cerebellar deep nuclei (Kandel, Schwartz, & Jessell, 2000).

Vestibulocerebellum
The vestibulocerebellum receives direct vestibular input from semicircular canals and otolithic organs and indirect visual input from the superior colliculi and striate cortex. These afferent projections make their way through the inferior cerebellar peduncle. The vestibulocerebellum does not synapse with a deep nucleus, but sends fibers directly to the lateral vestibular nucleus. This input on the vestibular nucleus influences muscle tone through descending vestibulospinal tracts that innervate extensor anti-gravity muscles and ocular muscles responsible for eye movements, including the vestibular ocular reflex (VOR). In various lesion studies, it is apparent that the vestibular cerebellum plays an important role in maintaining balance and posture (Kandel et al., 2000).


Spinocerebellum
Sensory information from the dorsospinocerebellar and ventrospinocerebellar tracts (DSCT and VSCT), vestibular nucleus, and reticular formation, project to the vermal and intermediate hemispheres of the cerebellum via the inferior cerebellar peduncle. Vermal regions connect with the fastigial nucleus and intermediate regions connect with the interposed nuclei (globose and emboliform). These nuclei send afferent axons to the vestibular nucleus, reticular formation, and red nucleus to influence the descending pathways of the vestibulospinal, reticulospinal, and rubrospinal tracts (Knierim, 1997).

Cerebrocerebellum
The cerebrocerebellum functions in the motor planning and timing of voluntary movements. Corticopontine input arrives through the middle cerebellar peduncle, and dentate nuclei output travels through the superior cerebellar peduncle to the ventrolateral thalamus, which projects back to the cerebrocortex (Knierim, 1997).

Figure 2- Functional Regions of the Cerebellum (http://goo.gl/5vjxx)
Figure 2- Functional Regions of the Cerebellum (http://goo.gl/5vjxx)


CIRCUTRY

There are two major types of afferent fibers that enter into the cerebellum: mossy fibers and climbing fibers. Both fibers send excitatory input to the deep cerebellar nuclei as well as projections to an inhibitory coritical loop that modulates the activity of these deep nuclei. In the cerebellar cortex, mossy fibers form a bulbous terminal where it synapses with granular cells. These granular cells ascend into the molecular layer of the cortex and split into parallel fibers, innervating a vast number of purkinje cells (Kandel et al., 2000). The excitatory input of these parallel fibers produces repetitive simple action potential spikes that excite purkinje cells. The result of Purkinje cell excitation is inhibition of the deep nuclei. These spikes appear to be correlated with movement activity, velocity, and direction (Squire et al., 2003). Climbing fiber branches in the cerebellar cortex wrap around 1-10 purkinje cells, each Purkinje cell being innervated by only one climbing fiber. This pathway also facilitates Purkinje cell firing, but with occasional complex action potential spikes (Kandel et al., 2000). In the bigger picture, these connections in the cerebellum can be understood as a “pattern classification device”, where raw input is processed into a meaningful response output. In the Marr-Albus model, mossy fibers innervate a select population of granule cells and overlapping input stimulates granule cells associated with a specific motor output (Boyden, Katoh, & Raymond, 2004).
Figure 3- Cerebellar Cortex (http://goo.gl/17c0U)
Figure 3- Cerebellar Cortex (http://goo.gl/17c0U)

Inhibitory interneurons in the cortex regulate the inhibitory loop. Golgi cells have an autonomic gain control on purkinje cells through a negative feedback mechanism where parallel fibers signal Golgi cells to inhibit the mossy fiber- granule cell synapse. Stellate and basket cells are involved in negative feed forward mechanisms that reduce fluctuations in excitatory purkinje cell input. In contrast to Golgi cells, unipolar brush cells in the vermal and intermediate hemispheres amplify mossy fiber input via positive feedback. This is thought to play a role in short-term storage of body orientation. Nonlaminar afferents is another source of circuit control that distributes its influence to all regions of the cerebellar cortex and deep nuclei (Squire et al., 2003).

Small loops between the cerebellar cortex and deep nuclei, known as microscopic modules, regulate the activity of specific preferred muscle synergies. This activity is constantly refined for the task for appropriate duration and intensity. The interaction of several microscopic modules creates a macroscopic module. Divergences of fibers that connect microscopic modules initiate an amplification process. From the deep cerebellar nuclei, efferent fibers are sent to an output population of neurons that also project reciprocal fibers back to the cerebellum. The high degrees of connectivity in the cerebellum support the function of this structure in dynamical coordination. The presence of recurring networks provides a basis for learning where basins of attraction are formed. These basins of attraction compete with one another and are ultimately modulated by inhibition from purkinje cells, which are constantly modified by climbing fibers (Squire et al., 2003).


CLIMBING FIBERS

Circuit Unity
Climbing fibers originate from the inferior olive and are organized into subdivisions that are linked to corresponding zones in the cerebellar cortex and cerebellar/vestibular nuclei. The loops between these zones unify these structures and provide constructive feedback. Certain olivary sub nuclei are somatotopically organization, and this is maintained in climbing fiber microzones. Furthermore, climbing fibers from the same sub nuclei are electrically linked together by gap junctions. Inferior olive climbing fiber output is highly innervated by excitatory sensory input and inhibitory cerebellar nuclear input (Squire et al., 2003).
Figure 4- Zonal Organization and continuity between A) cerebellar cortex, B) cerebellar deep nuclei, and C) inferior olive sub nuclei (http://goo.gl/8FpX0)
Figure 4- Zonal Organization and continuity between A) cerebellar cortex, B) cerebellar deep nuclei, and C) inferior olive sub nuclei (http://goo.gl/8FpX0)

Synaptic Plasticity
Climbing fibers form several axodendritic synapses on Purkinje cell spines. Because of the modulating effects of spiny neurons in other areas of the brain such as the basal ganglia, it is thought that calcium channels in neuron spines are responsible for long-term influences on neuron connectivity (Andrews, Leapman, Landis, & Reese, 1988). Synaptic plasticity can occur due to either long-term potentiation (LTP) or long-term depression (LTD). LTP sensitizes postsynaptic receptors whereas LTD desensitizes them. In the cerebellum, LTD is the main mechanism active in the process of motor learning (Squire et al., 2003).

There are three factors at play in the cerebellum that direct LTD for motor learning:
1. Postsynaptic Factor
Parallel fibers release glutamate, activating AMPA and mGluR1 receptors on the Purkinje cell dendrite spines. Activated AMPA opens channels to create a depolarizing current out of the spine into the dendrite. This depolarization current combines with other currents to activate a plateau potential, eventually leading to Purkinje cell discharge. Depolarization currents move back to the dendrite spine and contribute to the localized chemical changes involved in factor 2 (Squire et al., 2003).
2. Synapse-Specific Factor
Activated mGluR1 receptors initiate second messenger pathways that result in calcium influx (Squire et al., 2003). High voltage activated (HVA) calcium channels (mainly P/Q-type) and low voltage activated (LVA) calcium channels (T-type) mediate calcium influx. The small simple spikes produced by parallel fibers is thought to be linked to LVA calcium channels while the large complex spikes from climbing fibers is thought to be related to HVA calcium channels. LVA calcium channel activation alone may contribute to LTP and HVA calcium channels to LTD (Isope & Murphy, 2005). The calcium influx from P-type currents is an essential part of cerebellar functioning. Mutations in the P-type channels have been linked to neurologic disorders (Lorenzon, Lutz, Frankel, & Beam, 1998). Calcium triggers another secondary messenger pathway that desensitizes AMPA receptors via phosphorylation and makes this synapse eligible for LTD (Squire 2003). This phosphorylation process appears to be linked to protein-tyrosine phosphatase PTPMEG (Kina et al., 2007).
3. Training Signal
Meanwhile, the Purkinje cell discharge in factor 1 combined with other Purkinje cell discharges to form a macroscopic module that eventually became a composite movement command. The resulting movement generates sensory information and activates climbing fibers when an error is detected. Climbing fibers activate PKC, which prevents the dephosphorylation of AMPA receptors (Squire et al., 2003).


MOTOR LEARNING THEORIES

According to Charles Sherrington, “to move things is all that mankind can do.” However, meaning cannot be applied to movement without the corrective process of learning. The body’s dynamical motor system is constantly under revision, and with these revisions man becomes more skillful- his responses to the world become more optimally fitting. As aforementioned, long-term depression from climbing fiber input appears to play an important role in this motor learning process. While this is supported by many studies, it is also clear that several other factors contribute to plasticity changes in the cerebellum that contribute to motor learning.

A Model for Cerebellar Motor Learning
In a model proposed by Paul Fitts and Michael Posner, there are three stages to motor learning. The initial stage relies heavily on sensory feedback, the associative stage involves cognitive control, and the automatic stage involves less attention and error, and more transfer to different contexts. The cerebellum contributes to the progression down these three stages in processes of skill acquisition/refinement and skill automation/retrieval (Saywell & Taylor, 2008).

During skill acquisition and retrieval, sensory information is used for pattern recognition. This allows for a feed-forward mechanism that predicts future sensory input and prepares the body for movement with minimal errors. The cerebellum also acts as a sieve for sensory information, removing unrelated stimuli and placing emphasis on relevant stimuli. Visual information is a particularly significant stimulus in observational learning. In a study with a rat subject group that had a cerebellar lesion and a control that did not, the lesion group showed no evidence of observational learning while the control did. The control rats were then given a cerebellar lesion and were still able to retrieve this learned information, though unable to modify it (Saywell & Taylor, 2008).

Skill automation and retrieval is based on relative excitatory and inhibitory input from the cerebellar cortex and deep nuclei (respectively) to the motor cortex (Saywell & Taylor, 2008). The variations in these inputs create attractor wells for self-organizing movement. These self-organizing movements (directed by macroscopic modules) must be specific enough to be appropriate for a particular context, but also capable of an active process of generalization for transfer. This process is non-linear and involves plastic mechanisms at several dimensions (Boyden et al., 2004).

Vestibular Ocular Reflex and Multiple Plastic Mechanisms
The vestibular ocular reflex (VOR) is one example of a movement that is affected by plastic learning adaptations. The VOR coordinates eye movement in the opposite direction of head movement to maintain fixation on an external object. The gain (head to eye velocity ratio) of the VOR can be altered by image motion. Motion in the same direction as head motion will decrease gain whereas movements in opposite directions will increase gain. The Marr-Albus-Ito hypothesis states that memory of learned gain modifications are the result of climbing fibers that detect retinal slip and produce a LTD effect on the vestibular parallel fiber synapses. On the other hand, Miles and Lisberger proposes that the vestibular nucleus is the site of synaptic plasticity, and changes in purkinje cell firing is due to the effernce copy of the altered eye movement that is initiated by the vestibular nucleus (Boyden et al., 2004).
Figure 5- (top) Schematic of possible sites for neural plasticity indicated by lightning bolts. (bottom) Marr-Albus model. (http://goo.gl/OzSfe)
Figure 5- (top) Schematic of possible sites for neural plasticity indicated by lightning bolts. (bottom) Marr-Albus model. (http://goo.gl/OzSfe)

In support of the Miles-Lisberger hypothesis, the learning-related changes in the vestibular nuclei are not correlated with learning-related changes in the purkinje cells. Also, in subjects with cerebellar lesions, learned VOR changes are still expressed, proposing that the memory storage of these adjustments is external to the cerebellum. On the other hand, clear evidence supports the Marr-Albus hypothesis and the role of the climbing fibers in LTD adaptations. When nitric oxide (NO), GluRd2, or protein kinase C (PKC) is blocked in purkinje cells, the LTD pathway is inhibited. The resulting deficits in motor learning propose that climbing fiber input is fundamental. While this may be true, the multifunctional effects of these molecules also suggests that their inhibition may have affected some other area involved in the learning process. Other studies reinforce the significance of climbing fiber motor learning input in high frequency stimulation when purkinje cells are unable to discriminate stimuli (Boyden et al., 2004).

Both VOR motor learning hypotheses need to be considered for a complete picture of the adaptations that are occurring. Multiple plastic mechanisms are at work, and as a composite they produce learned movement (Boyden et al., 2004). These mechanisms are seen in VOR motor learning properties:
1. Regulation of Movement Dynamics
The cerebellar cortex plays a role in timing information while the deep cerebellar nuclei influences movement amplitude (Boyden et al., 2004).
2. Consolidation of Memories
The cerebellum stores short-term learned motor changes, which is transferred to the vestibular nuclei in long-term training (Boyden et al., 2004).
3. Bidirectional Changes in Movement Amplitude
Increases and decreases of VOR gain are dependent on different plasticity mechanisms. This can be explained by competing LTD (climbing fiber initiated) and LTP (parallel fiber initiated) processes (Boyden et al., 2004). The notion of Purkinje cell LTP is supported by studies that showed increased parallel fibers density in rats that performed acrobatic movements compared to voluntary exercise and inactive groups (Kleim et al., 1998).


CONCLUSION

The cerebellum plays a key role in the motor learning process through a combination of multiple plastic mechanisms, including long-term depression from climbing fiber input. An understanding of the cerebellar pathways and functional anatomy is foundational for developing theories in motor learning. From our understanding of motor learning, there are several practical applications that can be made for medical and healthcare professionals in treating cerebellar disorders. Since the cerebellum is largely involved in procedural memory, a clinician should focus on breaking movement down into several declarative tasks. Physically demonstrating tasks should be used to engage the cerebellum in observational learning. Furthermore, targeting activities related to previous skill sets can help generalize those motor patterns in a new context (Saywell & Taylor, 2008).


ABBREVIATIONS

AMPA, a-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid; DSCT, dorsal spinal cerebellar tract; GluRd2, ionotropic glutamate-like receptor 2; HVA, high voltage activated; LTD, long-term depression; LTP, long-term potentiation; LVA, low voltage activated; mGluR1, metabotropic glutamate receptor; NO, nitric oxide; PKC, protein kinase C; PTP, protein-tyrosine phosphatase; VOR, vestibular ocular reflex; VSCT, ventral spinal cerebellar tract.


GLOSSARY
  • Basket Cells- interneuron in the molecular layer of the cerebellar cortex that have feed forward GABAergic inhibitory input on the purkinje cell bodies.
  • Cerebellar Cortex- Outer surface of the cerebellum consisting of the granular layer, Pukinje cell layer, and molecular layer.
  • Cerebrocerebellum- Cerebellar functional region that includes the lateral cortical hemisphere.
  • Complex Action Potential Spike­- large Purkinje cell depolarization produced by climbing fibers. Slow temporal stimulation (1/sec).
  • Climbing Fiber- One of two inputs to the cerebellum. Has 2 fibers, one that projects directly to the deep nuclei, and another that projects to the Purkinje cell dendrites and has long-term depressing effects.
  • Declarative Memory- factual recall.
  • Deep Cerebellar Nuclei- Receives excitatory input from direct climbing fiber/ mossy fiber projections and inhibitory input from Purkinje cells. Consists of the fastigial nuclei, interposed nuclei, and dentate nucleus.
  • Gain- Eye rotation to head rotation ratio.
  • Granular Cells- located in the granular layer of the cerebellar cortex. Receive input from mossy fibers and project into the molecular layer where they split into parallel fibers.
  • Inferior Olive- nucleus in the midbrain that sends climbing fibers to the cerebellum. Consists of sub nuclei that are organized into zones.
  • Long Term Depression- Desensitization (weakening) of a synapse.
  • Long Term Potentiation- Sensitization (strengthening) of a synapse.
  • Macroscopic Module- A combination of several microscopic modules that interact with one another
  • Microscopic Module- Cerebellar cortex- deep nuclei loops that correspond with a specific synergistic movement.
  • Mossy Fibers- Along with climbing fibers, these fibers send projections to the deep nuclei and the cortex. Within the cortex, mossy fibers innervate with granular cells.
  • Nonlaminar Afferents- Sends direct projections all over the cerebellar cortex and deep nuclei that influence circuit control.
  • Plasticity- Neuron synapse adaptability.
  • PTPMEG- A phosphatase enzyme that desensitizes AMPA
  • Procedural Memory- Sequential task recall
  • Purkinje Cell- located in the cerebellar cortex. Receives excitatory input from mossy fibers and climbing fibers. Has inhibitory input on deep cerebellar nuclei.
  • Simple Action Potential Spike­- Small Purkinje cell depolarization that is induced by parallel fibers. Rapid temporal stimulation.
  • Spinocerebellum- Cerebellar functional region that includes the vermis and intermediate hemisphere.
  • Stellate Cell- Along with basket cells, stellate cells have feed forward GABAergic inhibitory input on Purkinje cells (dendrites). Stellate have more localized innervations.
  • Unipolar Brush Cells- Facilitates mossy fiber input in vermal and intermediate zones.
  • Vestibular Ocular Reflex- Reflex where eyes remain fixed during head rotation.
  • Vestibulocerebellum- Cerebellar functional region that includes the flocculonodular lobe.
  • Voltage Activated Calcium Channels (HVA and LVA)- Calcium channels that open based on different voltage stimuli and allow calcium influx for LTD.



SUGGESTED READINGS




QUIZ

True/False (1-3)
1. T/F Purkinje cells are influenced by LTP and LTD.
2. T/F Nonlaminar afferents innervate everywhere in the cerebellum.
3. T/F Consolidation of memories involves transfer of short-term motor memory from the cerebellum to the motor cortex.

Multiple Choice (4-6)
4. Order the following molecules in order that they take place during LTD:
a. Calcium, PTPMEG, PKC, mGluR1
b. PTPMEG, PKC, mGluR1, Calcium
c. mGluR1, Calcium, PKC, PTPMEG
d. mGluR1, Calcium, PTPMEG, PKC
5. Which would be least affected by a spinocerebellar lesion?
a. Proximal Limb Control
b. Balance
c. Planned movements
d. VOR
6. What is a useful strategy for treating cerebellar patients?
a. Stretching motor learning capabilities with procedural tasks
b. Capitalizing on previous skill sets
c. Verbally explaining, not enacting tasks
d. Inducing LTP through transmagnetic stimulation

Short Answer (7-10)
7. Describe the 3 factors involved in LTD.
8. How does cerebellar function address the 3 stages of motor learning?
9. Argue a case for either the Marr-Albo-Ito hypothesis or the Miles-Lisberger hypothesis. Then argue against it and explain why a multi plasticity model is necessary.
10. How is the cerebellum is a “pattern classification device”?

ANS: T, T, F, d, c, b

REFERENCES

  1. Squire LR, Bloom FE, McConnell SK, Roberts JL, Spitzer NC, Zigmond MJ. Cerebellum. In: Fundamental Neuroscience (2nd ed.). San Diego: Elsevier Science, 2003.
  2. Kandel ER, Schwartz JH, Jessell TM. The Cerebellum. In: Principles of Neural Science (4th ed.). New York: Appleton & Lange, 2000.
  3. Knierim J. (1997). Cerebellum. Neuroscience Online [Online]. The University of Texas Health Science Center at Houston (UT Health). http://neuroscience.uth.tmc.edu/s3/chapter05.html [10 Dec. 2012].
  4. Boyden ES, Katoh A, Raymond JL. Cerebellum-Dependent Learning: The Role of Multiple Plasticity Mechanisms. Annual Review of Neuroscience 27:581-609, 2004.
  5. Andrews BS, Leapman RD, Landis DMD, Reese TS. Activity-dependent accumulation of calcium in Purkinje cell dendritic spines. Proceedings of the National Academy of Sciences of the United States of America 85:1682-1685, 1988.
  6. Isope P, Murphy TH. Low threshold calcium currents in rat cerebellar Purkinje cell dendritic spines are mediated by T-type calcium channels. The Journal of Physiology 562.1:257-269, 2005.
  7. Lorenzon NM, Lutz CM, Frankel WN, Beam KG. Altered Calcium Channel Currents in Purkinje Cells of the Neurological Mutant Mouse leaner. The Journal of Neuroscience 18(12):4482-4489, 1998.
  8. Kina S, Tezuka T, Kusakawa S, Kishimoto Y, Kakizawa S, Hashimoto K, Ohsugi M, Kiyama Y, Horai R, Sudo K, Kakuta S, Iwakura Y, Iino M, Kano M, Manabe T, Yamamoto T. Involvement of protein-tyrosine phosphatase PTPMEG in motor learning and cerebellar long-term depression. European Journal of Neuroscience 26:2269-2278, 2007.
  9. Saywell N, Taylor D. The role of the cerebellum in procedural learning- Are there implications for physiotherapists’ clinical practice?. Physiotherapy Theory and Practice 24(5):321-328, 2008.
  10. Kleim JA, Swain RA, Armstorng KA, Napper RMA, Jones TA, Greenough WT. Selective Synaptic Plasticity Within the Cerebellar Cortex Following Complex Motor Skill Learning. Neurobiology of Learning and Memory 69:274-289