The role of the cerebellar cortex in motor learning was investigated by comparing the paramedian lobule of adult rats given difficult acrobatic training to that of rats that had been given extensive physical exercise or had been inactive. The paramedian lobule is activated during limb movements used in both acrobatic training and physical exercise. Acrobatic animals had greater numbers ofsynapses per Purkin e cell than animals from the exercise or inactive groups. No significant difference in synapse number or size between the exercised and inactive groups was found. This indicates that motor learning required of the acrobatic animals, and not repetitive use of synapses during physical exercise, generates new synapses in cerebellar cortex. In contrast, exercise animals had a greater density ,f blood vessels in the molecular layer than did either the acrobatic or inactive animals, suggesting that increased Synaptic activity elicited compensatory angiogenesis.Although many aspects of experience can alter synaptic connectivity (1-5), it has-been difficult'to relate unequivocally these changes to learning and memory because the morphological effects of leaning could not be isolated from those of behaviors. required to perform the task. For example, maze training (3) and forelimb reach-training (5) can alter neuronal morpholoy, but substantia trepetition of movements is required fo arning these, tasks'. Thus it is not possible to ascrib~e te morphological effects to learning per se.The cerebellar cortex may be particularly appropriate for testing hypotheses about synaptic plasticity because empirical evidence has implicated cerebellar cortex in motor skill learning (6,7,36), and there is some indication that synapse formation underlies cerebellar cortical learning, as suggested by dendritic-field changes in Purkinje cells of rodents and monkeys exposed to challenging. sensory-motor environments (8-10). Synaptogenesis in adult rat cerebellar cortex also occurs when afferents are cut (11,12). Furthermore, theorists have noted the suitability of cerebellar cortex for motor learning, with its convergence of two afferent systems conveying extensive somatic and cerebral state information upon the Purkinje cell, a single-output neuron that modulates motor activity (13)(14)(15)(16)(17).The results of the present study show that learning, as opposed to the motor activity necessary for learning a complex motor task, is responsible for synapse formation in the cerebellar cortex. We report a dissociation oflearning and motor activity, in which animals provided with complex visuomotor learning and minimal motor activity (acrobatic training)-form substantial numbers of new synapses in cerebellar cortex, whereas animals given extensive locomotor exercise with minimal opportunities for learning (repetitive exercise) formed new blood vessels but formed no more new synapses than animals in an inactive control group. MATERIAL AND METHODSAnimals and Training. Thirty-eight adult Long-Evans hooded female rats, kept in small groups...

This study compared the morphology of cerebellar cortex in adult female rats exposed for 1 month to repetitive exercise, motor learning, or an inactive condition. In the exercise conditions, rats that were run on a treadmill or housed with access to a running wheel had a shorter diffusion distance from blood vessels in the molecular layer of the paramedian lobule when compared to rats housed individually or rats that participated in a motor skill learning task. Rats taught complex motor skills substantially increased the volume of the molecular layer per Purkinje neuron and increased blood vessel number sufficiently to maintain the diffusion distance. These results dissociate angiogenesis associated with increased neuropil volume (as seen in the motor learning group) from angiogenesis associated with increased metabolic demands (as seen in the exercise groups). While the volume fraction of mitochondria did not differ among groups, the mitochondrial volume fraction per Purkinje cell was significantly increased in the motor skill rats. This appears to parallel the previously reported increase in synapses and associated neuropil volume change.

Rats reared from weaning in a complex environment have an increase in 1) glial surface area, 2) capillary volume, and 3) the number of synapses, per neuron. In that paradigm it has not been possible to determine whether the glial increase more closely correlates with the increase in synaptic numbers or with angiogenesis. More recently we have found that rats that exercised had an increase in the density of capillaries without an increase in the synaptic numbers, whereas rats that learned new motor skills had a greater number of synapses per neuron without an increase in the density of capillaries. Those findings provided the opportunity to investigate whether changes in glial volume in the cerebellum correspond to changes in the number of synapses or in capillary volume. Glial area fraction estimates were obtained using point counts on electron micrographs from the previous studies. The skill learning group had a greater volume of molecular layer per Purkinje cell, and also a greater volume of glia per Purkinje cell, than rats in either an inactive group or rats in two exercise groups. No significant differences were found in glial volume per synapse and glial volume per capillary across groups, although there was a tendency for glial volume per capillary to be lower in the exercise groups. The data indicate that glial volume correlates with synaptic numbers and not with capillary density.

The cellular and subcellular localization of muscarinic receptor proteins m1 and m2 was examined in the neostriatum of macaque monkeys by using light and electron microscopic immunocytochemical techniques. Double-labeling immunocytochemistry revealed m1 receptors in calbindin-D28k--positive medium spiny projection neurons. Muscarinic m1 labeling was dramatically more intense in the striatal matrix compartment in juvenile monkeys but more intense in striosomes in the adult caudate, suggesting that m1 expression undergoes a developmental age-dependent change. Ultrastructurally, m1 receptors were predominantly localized in asymmetric synapse-forming spines, indicating that these spines receive extrastriatal excitatory afferents. The association of m1-positive spines with lesion-induced degenerating prefronto-striatal axon terminals demonstrated that these afferents originate in part from the prefrontal cortex. The synaptic localization of m1 in these spines indicates a role of m1 in the modulation of excitatory neurotransmission. To a lesser extent, m1 was present in symmetric synapses, where it may also modulate inhibitory neurotransmission originating from local striatal neurons or the substantia nigra. Conversely, m2/choline acetyltransferase (ChAT) double labeling revealed that m2-positive neurons corresponded to large aspiny cholinergic interneurons and ultrastructurally, that the majority of m2 labeled axons formed symmetric synapses. The remarkable segregation of the m1 and m2 receptor proteins to projection and local circuit neurons suggests a functional segregation of m1 and m2 mediated cholinergic actions in the striatum: m1 receptors modulate extrinsic glutamatergic and monoaminergic afferents and intrinsic GABAergic afferents onto projection neurons, whereas m2 receptors regulate acetylcholine release from axons of cholinergic interneurons.

No abstract