A computational study of synaptic mechanisms of partial memory transfer in cerebellar vestibulo-ocular-reflex learning

There is a debate regarding whether motor memory is stored in the cerebellar cortex, or the cerebellar nuclei, or both. Memory may be acquired in the cortex and then be transferred to the cerebellar nuclei. Based on a dynamical system modeling with a minimal set of variables, we theoretically investigated possible mechanisms of memory transfer and consolidation in the context of vestibulo-ocular reflex learning. We tested different plasticity rules for synapses in the cerebellar nuclei and took robustness of behavior against parameter variation as the criterion of plausibility of a model variant. In the most plausible scenarios, mossy-fiber nucleus-neuron synapses or Purkinje-cell nucleus-neuron synapses are plastic on a slow time scale and store permanent memory, whose content is passed from the cerebellar cortex storing transient memory. In these scenarios, synaptic strengths are potentiated when the mossy-fiber afferents to the nuclei are active during a pause in Purkinje-cell activities. Furthermore, assuming that mossy fibers create a limited variety of signals compared to parallel fibers, our model shows partial memory transfer from the cortex to the nuclei.     

Cerebellar Cortex Granular Layer Interneurons in the Macaque Monkey Are Functionally Driven by Mossy Fiber Pathways through Net Excitation or Inhibition

The granular layer is the input layer of the cerebellar cortex. It receives information through mossy fibers, which contact local granular layer interneurons (GLIs) and granular layer output neurons (granule cells). GLIs provide one of the first signal processing stages in the cerebellar cortex by exciting or inhibiting granule cells. Despite the importance of this early processing stage for later cerebellar computations, the responses of GLIs and the functional connections of mossy fibers with GLIs in awake animals are poorly understood. Here, we recorded GLIs and mossy fibers in the macaque ventral-paraflocculus (VPFL) during oculomotor tasks, providing the first full inventory of GLI responses in the VPFL of awake primates. We found that while mossy fiber responses are characterized by a linear monotonic relationship between firing rate and eye position, GLIs show complex response profiles characterized by “eye position fields” and single or double directional tunings. For the majority of GLIs, prominent features of their responses can be explained by assuming that a single GLI receives inputs from mossy fibers with similar or opposite directional preferences, and that these mossy fiber inputs influence GLI discharge through net excitatory or inhibitory pathways. Importantly, GLIs receiving mossy fiber inputs through these putative excitatory and inhibitory pathways show different firing properties, suggesting that they indeed correspond to two distinct classes of interneurons. We propose a new interpretation of the information flow through the cerebellar cortex granular layer, in which mossy fiber input patterns drive the responses of GLIs not only through excitatory but also through net inhibitory pathways, and that excited and inhibited GLIs can be identified based on their responses and their intrinsic properties.     

Distribution of beta-adrenergic receptors in different cortical and nuclear regions of cat cerebellum, as revealed by binding studies

In addition to mossy fibers and climbing fibers, the cerebellum receives NE-containing fibers originating particularly from the locus coerulus complex. Since the neurotransmitter of the coeruleo-cerebellar afferents acts mainly on Purkinje cells through beta-receptors, experiments were performed in cats to study the regional distribution and properties of the beta-adrenoceptors at corticocerebellar level; moreover, attempts were made to identify also the presence of beta-adrenoceptor binding in the cerebellar nuclei underlying the different zones of the cerebellar cortex. (-)-[3H]Dihydroalprenolol, a very potent beta-adrenergic antagonist, was used to characterize the beta-adrenergic receptors. (-)-[3H]DHA bound specifically to membrane preparations from all the cortical and nuclear zones of the cerebellum. In particular, beta-adrenergic receptors showed a high density and affinity in the cerebellar cortex with no significant difference in the medial with respect to the intermediate-lateral cortical area. The cerebellar nuclei showed a lower density of beta-adrenoceptors with a comparable or slightly lower affinity with respect to the cerebellar cortex. However, no difference was observed between the fastigial nucleus and the interposite-dentate nuclei. Scatchard analysis of saturation data revealed the presence of a single population of high affinity binding sites in all the examined regions, while the Hill plots excluded the presence of cooperative effects among the binding sites. Attempts to differentiate in the cerebellum beta 1- and beta 2-receptors by using agents which act as selective beta 1 and beta 2 ligands indicated that (-)-[3H]DHA specific binding in cerebellar cortex and nuclei affects predominantly the beta 2 subtype of adrenoceptors. A comparison between results obtained from the cerebellar cortex and those obtained from the whole cerebral cortex was also made. The whole cerebral cortex showed a lower density but a higher affinity than the cerebellar cortex. Moreover, inhibition of (-)-[3H]DHA binding by selective beta 1 and beta 2 ligands indicated the prevalence of the beta 1 subtype of adrenoceptors at this level.     

Morphological study of organotypic cerebellar cultures

Organotypic cerebellar cultures from 8-days-old (P8) mouse pups were studied following 11 days of in vitro (I IDIV) culturing. The cerebellar cytoarchitectonic structure was maintained in most parasagittal cerebellar cortical slice cultures (also containing the deep cerebellar nuclei). The two main extrinsic excitatory inputs (the climbing and the mossy fibers) seem to be replaced by other axonal types: in the molecular layer mostly by parallel fibers (for climbing fibers) and in the granular layer by intrinsic mossy fiber collaterals of local excitatory interneurons, the unipolar brush cells. However, in a few organotypic cultures, which (although preserving the trilaminar cerebellar cortical structure) were "granuloprival" but also contained some of the deep cerebellar nuclei, the participation of extracortical axons from the deep cerebellar nuclei in the replacement of the missing afferents is suggested.     

Immunogold electron microscopic demonstration of glutamate and GABA in normal and deafferented cerebellar cortex: correlation between transmitter content and synaptic vesicle size

Selective labeling of mossy fiber terminals and parallel fibers was obtained in rat cerebellar cortex by a glutamate antibody produced and characterized by Hepler et al. The high-resolution electron microscopic immunogold demonstration of this amino acid offered the possibility of determining the size and shape of synaptic vesicles in glutamate-positive mossy endings. Mossy terminals that stained with the glutamate antibody formed two distinct populations, one with spherical synaptic vesicles with an average diameter of 34.0 nm (more than 90% of all mossy fiber endings) and one with pleomorphic and smaller synaptic vesicles which had an average diameter of 28.5 nm. We present experimental evidence that the mossy terminals with large round vesicles are of extracerebellar origin, whereas those with small pleomorphic synaptic vesicles are endings of nucleocortical fibers. The presence of two distinct classes of gamma-aminobutyric acid (GABA)-containing axon terminals within cerebellar glomeruli has also been demonstrated; those originating from the cerebellar nuclei contain large (36.2 nm) synaptic vesicles, whereas the majority of GABA-stained axon terminals that are of local (cortical) origin contain small (29.1 nm) synaptic vesicles. It therefore appears that, at least in the case of glutamate and GABA, morphological characterization of the axon terminals based on the size and shape of synaptic vesicles is not a reliable indicator of their functional nature (i.e., whether they are excitatory or inhibitory); convincing evidence for the identity of the transmitter can be obtained only by electron microscopic immunostaining procedures. Our results also suggest the existence of both inhibitory and excitatory feedback from cerebellar nuclei to cerebellar cortex.     

Reliability of Computation in the Cerebellum

The mossy fiber-granule cell-parallel fiber-Purkinje cell system of the cerebellar cortex is investigated from the viewpoint of reliability of computation. It is shown that the effects of variability in the inputs to a Purkinje cell can be reduced by having a large number of parallel fibers whose activities are statistically independent. The mossy fiber-granule cell relay is shown to be capable of performing the required function of transforming the activity in a small number of mossy fibers into activity in a much larger number of parallel fibers, while ensuring that there is little correlation between the activities of individual parallel fibers. The effects of variability in the outputs of Purkinje cells may be reduced by redundancy and convergence schemes, as evidenced by the geometrical pattern of parallel fibers and Purkinje cells and the convergence of these cells onto their target neurons.     

Neurotransmitters, neuropeptides and calcium binding proteins in developing human cerebellum: a review

Many endogenous neurochemicals that are known to have important functions in the mature central nervous system have also been found in the developing human cerebellum. Cholinergic neurons, as revealed by immunoreactivities towards choline acetyltransferase or acetylcholinesterase, appear early at 23 weeks of gestation in the cerebellar cortex and deep nuclei. Immunoreactivities gradually increase until the first postnatal month. Enkephalin is localized in the developing cerebellum, initially in the fibers of the cortex and deep nuclei at 16–20 weeks and then also in the Purkinje cells, granule cells, basket cells and Golgi cells at 23 weeks onward. Another neuropeptide, substance P, is localized mainly in the fibers of the dentate nucleus from 9 to 24 weeks but substance P immunoreactivity declines thereafter. GABA, an inhibitory neurotransmitter of the central nervous system, starts to appear at 16 weeks in the Purkinje cells, stellate cells, basket cells, mossy fibers and neurons of deep nuclei. GABA expression is gradually upregulated toward term forming networks of GABA-positive fibers and neurons. Catecholaminergic fibers and neurons are also detected in the cortex and deep nuclei at as early as 16 weeks. Calcium binding proteins, calbindin D28K and parvalbumin, make their first appearance in the cortex and deep nuclei at 14 weeks and then their expression decreases toward term, while calretinin appears later at 21 weeks but its expression increases with fetal age. The above findings suggest that many neurotransmitters, neuropeptides and calcium binding proteins (1) appear early during development of the cerebellum; (2) have specific temporal and spatial expression patterns; (3) may have functions other than those found in the mature neural systems; and (4) may be able to interact with each other during early development.     

Plausible function of Golgi cells in the cerebellar cortex

The function of Golgi cells in the cerebellar cortex is quantitatively examined in consideration of the nonlinear input-output characteristics and convergence and divergence numbers of cells. It is strongly suggested that the two signal paths to Golgi cells have different function. The feed-forward path will have the same function as assumed in the previous theories of the cerebellar cortex, that is, to keep the firing rate of granule cells approximately constant over considerable variation in the firing rate of mossy fibers. The feedback path will, on the other hand, have a new function which has not been assumed in the previous theories. The function is to cause oscillation of the firing rate of granule cells for stationary mossy fiber inputs. The assumption of the new function enables us to explain cerebellar function to keep stationary posture.     

Learned movements elicited by direct stimulation of cerebellar mossy fiber afferents

Definitive evidence is presented that the conditioned stimulus (CS) in classical conditioning reaches the cerebellum via the mossy fiber system. Decerebrate ferrets received paired forelimb and periocular stimulation until they responded with blinks to the forelimb stimulus. When direct mossy fiber stimulation was then given, the animals responded with conditioned blinks immediately, that is, without ever having been trained to the mossy fiber stimulation. Antidromic activation was prevented by blocking mossy fibers with lignocaine ventral to the stimulation site. It could be excluded that cerebellar output functioned as the CS. Analysis of latencies suggests that conditioned responses (CRs) are not generated by mossy fiber collaterals to the deep nuclei. Hence, the memory trace is probably located in the cerebellar cortex.     

Specificity of a target cell-derived stop signal for afferent axonal growth.

With a novel model culture system in which afferents are co-cultured with purified populations of target neurons, we have demonstrated that a target cell within the central nervous system (CNS), the cerebellar granule neuron, poses a "stop-growing signal" for its appropriate afferents, the mossy fibers. To ask whether this stop signal is afferent specific, we co-cultured granule neurons with another cerebellar afferent system, the climbing fibers from the inferior olivary nuclei, which normally contact Purkinje neurons, and with retinal ganglion cell afferents, which never enter the cerebellum. Granule neurons do not pose a stop signal to either of these afferents. In contrast to pontine mossy afferents that grow well on laminin and showed reduced outgrowth on granule neurons, both olivary and retinal fibers displayed similar growth on laminin alone or on granule neurons. In addition, each afferent showed different degrees of fasciculation and growth cone morphology on laminin. Thus, the growth arrest signal sent by granule neurons is specifically recognized by their appropriate afferents. Moreover, these three types of afferents exhibit varying growth patterns on the same noncellular and cellular substrates, implicating distinct molecular characteristics of growth regulation for different classes of neurons that would contribute to specificity of synapse formation.     

Cerebellar cortex: its simulation and the relevance of Marr's theory.

Marr's theory of the cerebellar cortex as an associative learning device is one of the best examples of a theory that directly relates the function of a neural system to its neural structure. However, although he assigned a precise function to each of the identified cell types of the cerebellar cortex, many of the crucial aspects of the implementation of his theory remained unspecified. We attempted to resolve these difficulties by constructing a computer simulation which contained a direct representation of the 13,000 mossy fibres and the 200,000 granule cells associated with a single Purkinje cell of the cerebellar cortex, together with the supporting Golgi, basket and stellate cells. In this paper we present a detailed explanation of Marr's theory based upon an analogy between Marr's cerebellar model and an abstract model called the associative net. Although some of Marr's assumptions contravene neuroanatomical findings, we found that in general terms his conclusion that each Purkinje cell can learn to respond to a large number of different patterns of activity in the mossy fibres is substantially correct. However, we found that this system has a lower capacity and acts more stochastically than he envisaged. The biologically realistic simulated structure that we designed can be used to assess the computational capabilities of other network theories of the cerebellum.     

Retrograde degeneration of the neurons of the cerebellar nuclei after partial removal of the associative cerebral cortex

Retrograde degeneration of the cerebellar nuclei cells has been studied after partial ablation of the associative parietal cerebral cortex in the cat. The material is stained after Nissl. Retrogradely degenerated and normal cells are counted. The "ghost-cells" in the cerebellar nuclei indicate that a direct axonal connection exists between some neurons and the cerebral cortex operated, while the cells that are at other stages of degeneration are, perhaps, connected with this part of the cortex by means of axonal collaterals.     

The projection of spinocerebellar neurons from the sacrococcygeal region of the spinal cord in the cat. An experimental study using anterograde transport of WGA-HRP and degeneration.

The projection from the sacro-coccygeal region of the spinal cord to the cerebellum was studied by two different techniques in the cat. In five cats wheat germ agglutinin-horseradish peroxidase conjugate (WGA-HRP) was injected caudal to a preceding unilateral cordotomy at the sacral level, aimed at interrupting the spinocerebellar tracts on one side completely, and the distribution of WGA-HRP labeled mossy fibers and mossy fiber terminals was studied in the cerebellum. In three additional cats, degenerating fibers were examined in Fink-Heimer stained sections following unilateral transection of the lateral and ventral funiculi at L7 or S3 level. In the WGA-HRP experiments the labeled mossy fiber terminals were located bilaterally in lobules I-V. Most of them were found in the anterior part of lobule II. In addition, labeled terminals were observed in sublobule VIIIB and in pars copularis of the paramedian lobule, contralateral to the cordotomy. The terminals in the anterior lobe were concentrated in longitudinal zones parallel to the mid sagittal plane. In lobule II, the terminals were most abundant in the superficial, apical parts of the folia. Some presumed terminals were also seen in the cerebellar nuclei. Labeled fibers were found contralateral, but not ipsilateral to the cordotomy in the superior and inferior cerebellar peduncles, as well as in the spinal cord rostral to the cordotomy. The results of the degeneration experiments were the same as those of the WGA-HRP experiments with regard to the detailed projections in the cerebellar cortex. This is strong support against the possibility that WGA-HRP labeled cerebellar mossy fiber terminals, following WGA-HRP injections in the spinal cord, would represent terminals of collaterals of retrogradely labeled neurons. It also lends strong support in favour of WGA-HRP as a reliable anterograde tracer for studying cerebellar cortical projections of spinocerebellar neurons in the cat.     

Further consideration on pattern separability in a random neural net with inhibitory connections

A two-layer random neural net with inhibitory connections composing of threshold elements has been regarded as a model of the cerebellar cortex. Many properties of pattern separation with the model have been disclosed through consideration on the degree of pattern separation. However, we have not shown yet that the degree of pattern separation is given by some different functions which are decided by the relation between the firing rates of input patterns. The present study is intended to reveal that the functions of the degree of pattern separation are synthesized with some different partial functions, and they are differently given on the relation between the firing rates of input patterns. Simultaneously, it is proved that the number of the functions also depend on the number of connections between two layers in the model. We also disclose the properties of the degree of pattern separation, and give some suggestions on the sizes of the firing rates of mossy fibers and granule cells under the knowledge about them.     

Genome-Wide Divergence of DNA Methylation Marks in Cerebral and Cerebellar Cortices

Background

Emerging evidence suggests that DNA methylation plays an expansive role in the central nervous system (CNS). Large-scale whole genome DNA methylation profiling of the normal human brain offers tremendous potential in understanding the role of DNA methylation in brain development and function.

Methodology/Significant Findings

Using methylation-sensitive SNP chip analysis (MSNP), we performed whole genome DNA methylation profiling of the prefrontal, occipital, and temporal regions of cerebral cortex, as well as cerebellum. These data provide an unbiased representation of CpG sites comprising 377,509 CpG dinucleotides within both the genic and intergenic euchromatic region of the genome. Our large-scale genome DNA methylation profiling reveals that the prefrontal, occipital, and temporal regions of the cerebral cortex compared to cerebellum have markedly different DNA methylation signatures, with the cerebral cortex being hypermethylated and cerebellum being hypomethylated. Such differences were observed in distinct genomic regions, including genes involved in CNS function. The MSNP data were validated for a subset of these genes, by performing bisulfite cloning and sequencing and confirming that prefrontal, occipital, and temporal cortices are significantly more methylated as compared to the cerebellum.

Conclusions

These findings are consistent with known developmental differences in nucleosome repeat lengths in cerebral and cerebellar cortices, with cerebrum exhibiting shorter repeat lengths than cerebellum. Our observed differences in DNA methylation profiles in these regions underscores the potential role of DNA methylation in chromatin structure and organization in CNS, reflecting functional specialization within cortical regions.     

Gangliosides in the human brain development and aging.

In this study, brain gangliosides in prenatal and postnatal human life were analyzed. Immunohistochemically, the presence of "c"-pathway of gangliosides (GQ1c) in embryonic brain was only recorded at 5 weeks of gestation. Biochemical results indicated a twofold increase in human cortex ganglioside concentration between 16 and 22 weeks of gestation. The increasing ganglioside concentration was based on an increasing GD1a ganglioside fraction in all regions analyzed except cerebellar cortex, which was characterized by increasing GT1b. In this developmental period, GD3 was found to be localized in the ventricular zone of the cortical wall. After birth, GD1b ganglioside in neuropil of granular cell layer corresponding to growing mossy fibers was expressed in cerebellar cortex. Between birth and 20/30 years of age, a cerebral neocortical difference of ganglioside composition was observed, characterized by lowest GD1a in visual cortex. Analyzing the composition of gangliosides in cortical regions during aging, they were observed to follow region-specific alterations. In frontal cortex, there was a greater decrease in GD1a and GM1 than in GT1b and GD1b, but in occipital (visual) cortex there was no change in individual gangliosides. In hippocampus, GD1a moderately decreased, whereas other fractions were stable. In cerebellar cortex, GD1b and GT1b fractions decreased with aging.     

Specificity of a target cell-derived stop signal for afferent axonal growth

With a novel model culture system in which afferents are co-cultured with purified populations of target neurons, we have demonstrated that a target cell within the central nervous system (CNS), the cerebellar granule neuron, poses a ?stop-growing signal”? for its appropriate afferents, the mossy fibers. To ask whether this stop signal is afferent specific, we co-cultured granule neurons with an other cerebellar afferent system, the climbing fibers from the inferior olivary nuclei, which normally contact Purkinje neurons, and with retinal ganglion cell afferents, which never enter the cerebellum. Granule neurons do not pose a stop signal to either of these afferents. In contrast to pontine mossy afferents that grow well on laminin and showed reduced outgrowth on granule neurons, both olivary and retinal fibers displayed similar growth on laminin alone or on granule neurons. In addition, each afferent showed different degrees of fasciculation and growth cone morphology on laminin. Thus, the growth arrest signal sent by granule neurons is specifically recognized by their appropriate afferents. Moreover, these three types of afferents exhibit varying growth patterns on the same noncellular and cellular substrates, implicating distinct molecular characteristics of growth regulation for different classes of neurons that would contribute to specificity of synapse formation. © 1992 John Wiley & Sons, Inc.     

BDNF and NT4/5 promote survival and neurite outgrowth of pontocerebellar mossy fiber neurons.

The neurotrophins nerve growth factor (NGF), brain-derived neurotrophic factor (BDNF), neurotrophin-3 (NT3), and NT4/5 are all found in the developing cerebellum. Granule cells, the major target neurons of mossy fibers, express BDNF during mossy fiber synaptogenesis. To determine whether neurotrophins contribute to the development of cerebellar afferent axons, we characterized the effects of neurotrophins on the growth of mossy fiber neurons from mice and rats in vitro. For a mossy fiber source, we used the basilar pontine nuclei (BPN), the major source of cerebellar mossy fibers in mammals. BDNF and NT4/5 increased BPN neuron survival, neurite outgrowth, growth cone size, and elongation rate, while neither NT3 nor NGF increased survival or outgrowth. In addition, BDNF and NT4/5 reduced the size of neurite bundles. Consistent with these effects, in situ hybridization on cultured basilar pontine neurons revealed the presence of mRNA encoding the TrkB receptor which binds both BDNF and NT4/5 with high affinity. We detected little or no message encoding the TrkC receptor which preferentially binds NT3. BDNF and NT4/5 also increased TrkB mRNA levels in BPN neurons. In addition to previously established functions as an autocrine/paracrine trophic factor for granule cells, the present results indicate that cerebellar BDNF may also act as a target-derived trophic factor for basilar pontine mossy fibers.     

Structural abnormalities in neurons are sufficient to explain the clinical disease and fatal outcome of experimental rabies in yellow fluorescent protein-expressing transgenic mice

Under natural conditions and in some experimental models, rabies virus infection of the central nervous system causes relatively mild histopathological changes, without prominent evidence of neuronal death despite its lethality. In this study, the effects of rabies virus infection on the structure of neurons were investigated with experimentally infected transgenic mice expressing yellow fluorescent protein (YFP) in neuronal subpopulations. Six-week-old mice were inoculated in the hind-limb footpad with the CVS strain of fixed virus or were mock infected with vehicle (phosphate-buffered saline). Brain regions were subsequently examined by light, epifluorescent, and electron microscopy. In moribund CVS-infected mice, histopathological changes were minimal in paraffin-embedded tissue sections, although mild inflammatory changes were present. Terminal deoxynucleotidyltransferase-mediated dUTP-biotin nick end labeling and caspase-3 immunostaining showed only a few apoptotic cells in the cerebral cortex and hippocampus. Silver staining demonstrated the preservation of cytoskeletal integrity in the cerebral cortex. However, fluorescence microscopy revealed marked beading and fragmentation of the dendrites and axons of layer V pyramidal neurons in the cerebral cortex, cerebellar mossy fibers, and axons in brainstem tracts. At an earlier time point, when mice displayed hind-limb paralysis, beading was observed in a few axons in the cerebellar commissure. Toluidine blue-stained resin-embedded sections from moribund YFP-expressing animals revealed vacuoles within the perikarya and proximal dendrites of pyramidal neurons in the cerebral cortex and hippocampus. These vacuoles corresponded with swollen mitochondria under electron microscopy. Vacuolation was also observed ultrastructurally in axons and in presynaptic nerve endings. We conclude that the observed structural changes are sufficient to explain the severe clinical disease with a fatal outcome in this experimental model of rabies.     

Somatic mosaicism of expanded CAG repeats in brains of patients with dentatorubral-pallidoluysian atrophy: cellular population-dependent dynamics of mitotic instability.

Dentatorubral-pallidoluysian atrophy (DRPLA) is an autosomal dominant neurodegenerative disease caused by unstable expansion of a CAG repeat in the DRPLA gene. We performed detailed quantitative analysis of the size and the size distribution (range) of the expanded CAG repeats in various regions of the CNS of eight autopsied patients with DRPLA. Expanded alleles (AE) showed considerable variations in size, as well as in range, depending on the region of the CNS, whereas normal alleles did not show such variations, which indicates the occurrence of somatic mosaicism of AE in the CNS. The AE in the cerebellar cortex were consistently smaller by two to five repeat units than those in the cerebellar white matter. Moreover, the AE in the cerebral cortex were smaller by one to four repeat units than those in the cerebral white matter. These results suggest that the smaller AE in the cerebellar and cerebral cortices represent those of neuronal cells. The ranges of the AE in the cerebral cortex, cerebral white matter, and cerebellar white matter showed considerable variation ranging from 9 to 23 repeat units, whereas those in the cerebellar cortex showed little variance and were approximately 7 repeat units. The ranges of the AE in the cerebral cortex, cerebral white matter, and cerebellar white matter were much broader in patients with higher ages at death than they were in patients with lower ages at death, raising the possibility that the range of AE increases with time, as the result of mitotic instability of AE.