Postnatal development of Rattus norvegicus B. cortical neurons and the influence of trauma]

The differentiation of the pyramid neurons during ontogenesis and the effects of a trauma on the process of neuron differentiation and synaptogenesis were investigated in the rat cerebral cortex after Golgi-impregnation. 1. There are temporal differences in the differentiation of the processes of cortical neurons. Axons differentiate earlier than dendrites, apical dendrites earlier than basal ones. 2. Varicosities in the processes of cortical neurons during the early postnatal period are regarded as a feature of growth processes. 3. The appearance of the dendritic spines is an important process in the ontogenetic and phylogenetic development of cortical neurons. 4. The different cortical layers show a different degree of differentiation during development. The deeper layers precede the upper layers in the process of differentiation. 5. Ingrowing afferents have an essential influence on the differentiation - especially on the differentiation of the dendritic postsynaptic structures. 6. The cortex of 6 month old rats shows no principle differences in comparison with 24 days old animals. It is concluded that the visible differentiation processes of cortical neurons are nearly finished 24 days post partum. 7. As an effect of the trauma a considerable loss of dendritic spines of layer III and V pyramid cells is found in addition to general degeneration features. 8. The following factors are thought to be responsible for the loss of spines: (i) transneuronal processes with spine degeneration and subsequent phagocytosis of the synaptosome. (ii) destruction of differentiation furthering afferents results in differentiation defects of the neuron with the failure of further postsynaptic differentiation (spines).     

Long-term effects of parallel fiber loss in the cerebellar cortex of the adult and weanling rat

Short- and long-term effects of parallel fiber deafferentation of adult and weanling cerebellar cortex were investigated following parasagittal transections of the lateral cerebellar hemisphere. Short-term electron microscopic examination revealed that parallel fibers undergo rapid electron-dense degeneration within 5 days of axotomy. These axons were the only neuronal elements immediately affected by the lesion. The continued maintenance of Purkinje cell terminal branchlets and stellate cell dendrites is dependent upon the presence of an adequate parallel fiber milieu. Morphological evidence is provided which suggests that Purkinje cell dendritic spines may be phagocytically removed by Bergmann glial cells following parallel fiber loss. Although a marked decrease was reported in the number of spines projecting from terminal branchlets following deafferentation of both adult and weanling rats, these data suggest that some spines are capable of increasing their length. The elongation of these spines may represent a form of dendritic plasticity. No evidence was found to suggest that deafferentated terminal branchlets are receptive to forming heterologous synaptic contacts. The primary response to parallel fiber deafferentation for both the adult and weanling cerebellum therefore appears to be transneuronal degeneration.     

In vivo toxicity of prion protein in murine scrapie: ultrastructural and immunogold studies

Prion protein (PrP) is a cell surface, host coded, sialoglycoprotein which accumulates in excess in scrapie, Creutzfeldt‐Jakob disease, bovine spongiform encephalopathy and other transmissible spongiform encephalopathies. Infection of mice with the 87 V or ME7 scrapie strains results in distinctive and very different light microscopical patterns of vacuolation and disease specific PrP accumulation. In both of these scrapie strains immunogold electron microscopy was used to locate PrP to the plasmalemma of neurons from where it was released into the neuropil. Initial PrP accumulation around neurons and in early plaques lacking amyloid fibrils was generally not associated with morphological changes either of the neuron or dendrite releasing the PrP or in the adjacent neuropil in which excess PrP accumulated. However, accumulation of pre‐amyloid PrP in some brain areas was associated with specific degeneration of dendritic spines and axon terminals. Initial PrP aggregation into fibrils was also associated with tissue damage with both ME7 and 87 V plaques and diffuse accumulations. Tissue damage associated with fibrillogenesis was localized and would not be expected to have clinical significance. We conclude that pre‐amyloid PrP release and accumulation is not invariably toxic, either to the neuron releasing PrP or to the neuropil into which it is released. However, axon terminal degeneration and dendritic spine loss in some neuroanatomical areas may be indicative of specific PrP toxicity and may be the main cause of neurological dysfunction in murine scrapie.     

An Automated Pipeline for Dendrite Spine Detection and Tracking of 3D Optical Microscopy Neuron Images of In Vivo Mouse Models

The variations in dendritic branch morphology and spine density provide insightful information about the brain function and possible treatment to neurodegenerative disease, for example investigating structural plasticity during the course of Alzheimer’s disease. Most automated image processing methods aiming at analyzing these problems are developed for in vitro data. However, in vivo neuron images provide real time information and direct observation of the dynamics of a disease process in a live animal model. This paper presents an automated approach for detecting spines and tracking spine evolution over time with in vivo image data in an animal model of Alzheimer’s disease. We propose an automated pipeline starting with curvilinear structure detection to determine the medial axis of the dendritic backbone and spines connected to the backbone. We, then, propose the adaptive local binary fitting (aLBF) energy level set model to accurately locate the boundary of dendritic structures using the central line of curvilinear structure as initialization. To track the growth or loss of spines, we present a maximum likelihood based technique to find the graph homomorphism between two image graph structures at different time points. We employ dynamic programming to search for the optimum solution. The pipeline enables us to extract dynamically changing information from real time in vivo data. We validate our proposed approach by comparing with manual results generated by neurologists. In addition, we discuss the performance of 3D based segmentation and conclude that our method is more accurate in identifying weak spines. Experiments show that our approach can quickly and accurately detect and quantify spines of in vivo neuron images and is able to identify spine elimination and formation.     

Dendritic sprouting in Alzheimer's presenile dementia.

The use of Golgi techniques on brain tissue from aging and senescent human individuals has shown a series of progressive deteriorative changes in neurons at a number of sites in cortical structures. These include loss of dendritic spines, irregular swelling of cell body and dendrites, and progressive loss of the dendritic domain, culminating in cell death. These changes which characterize the senile dementias have now been seen also in two cases of Alzheimer presenile dementia. This material is additionally characterized by the presence of clusters of new dendritic growth, developing at one or more sites along the dendritic or somal surface. The dendritic branchlets are densely spine-covered at a time when the original dendritic systems of the neuron may be partially or totally devoid of their spine complements. Because these dendrite clusters appear haphazard in placement and orientation and do not coincide with any known presynaptic terminal fields, we provisionally refer to them as “lawless.” The mechanisms which trigger their development in presenile, but not senile, dementing disease are unknown at present.     

Amyotrophic lateral sclerosis is a distal axonopathy: evidence in mice and man

The SOD1 mutant mouse is the most widely used model of human amyotrophic lateral sclerosis (ALS). To determine where and when the pathological changes of motor neuron disease begins, we performed a comprehensive spatiotemporal analysis of disease progression in SOD1(G93A) mice. Quantitative pathological analysis was performed in the same mice at multiple ages at neuromuscular junctions (NMJ), ventral roots, and spinal cord. In addition, a patient with sporadic ALS who died unexpectedly was examined at autopsy. Mice became clinically weak at 80 days and died at 131 +/- 5 days. At 47 days, 40% of end-plates were denervated whereas there was no evidence of ventral root or cell body loss. At 80 days, 60% of ventral root axons were lost but there was no loss of motor neurons. Motor neuron loss was well underway by 100 days. Microglial and astrocytic activation around motor neurons was not identified until after the onset of distal axon degeneration. Autopsy of the ALS patient demonstrated denervation and reinnervation changes in muscle but normal appearing motor neurons. We conclude that in this widely studied animal model of human ALS, and in this single human case, motor neuron pathology begins at the distal axon and proceeds in a "dying back" pattern.     

Impact of dendritic spine preservation in medium spiny neurons on dopamine graft efficacy and the expression of dyskinesias in parkinsonian rats

Dopamine deficiency associated with Parkinson’s disease (PD) results in numerous changes in striatal transmitter function and neuron morphology. Specifically, there is marked atrophy of dendrites and dendritic spines on striatal medium spiny neurons (MSN), primary targets of inputs from nigral dopamine and cortical glutamate neurons, in advanced PD and rodent models of severe dopamine depletion. Dendritic spine loss occurs via dysregulation of intraspine Cav1.3 L‐type Ca²⁺channels and can be prevented, in animal models, by administration of the calcium channel antagonist, nimodipine. The impact of MSN dendritic spine loss in the parkinsonian striatum on dopamine neuron graft therapy remains unexamined. Using unilaterally parkinsonian Sprague–Dawley rats, we tested the hypothesis that MSN dendritic spine preservation through administration of nimodipine would result in improved therapeutic benefit and diminished graft‐induced behavioral abnormalities in rats grafted with embryonic ventral midbrain cells. Analysis of rotational asymmetry and spontaneous forelimb use in the cylinder task found no significant effect of dendritic spine preservation in grafted rats. However, analyses of vibrissae‐induced forelimb use, levodopa‐induced dyskinesias and graft‐induced dyskinesias showed significant improvement in rats with dopamine grafts associated with preserved striatal dendritic spine density. Nimodipine treatment in this model did not impact dopamine graft survival but allowed for increased graft reinnervation of striatum. Taken together, these results demonstrate that even with grafting suboptimal numbers of cells, maintaining normal spine density on target MSNs results in overall superior behavioral efficacy of dopamine grafts.     

Proliferative and degenerative changes in striatal spiny neurons in Huntington's disease: a combined study using the section-Golgi method and calbindin D28k immunocytochemistry.

Dysmorphic alterations of dendritic arbors and spines in spiny striatal neurons were identified in section-Golgi impregnations of moderate and severe grades of Huntington's disease (HD). These alterations could be characterized as either proliferative or degenerative changes. Proliferative changes included prominent recurving of distal dendritic segments, short-segment branching along dendrites, and increased numbers and size of dendritic spines. Degenerative alterations consisted of truncated dendritic arborizations, occasional focal dendritic swellings, and marked spine loss. Proliferative changes were found primarily in moderate grades of HD, while degenerative changes were predominantly found in severe grades. Cytopathologic changes increased with neuropathologic severity. Similar morphologic alterations were observed in calbindin D28k (Calb) stained neurons in HD striatum. The immunoreactive intensity of Calb staining was increased in the distal dendrites of positive neurons in HD striatum. The present findings provide morphologic and quantitative evidence that confirms an early and marked involvement of spiny striatal neurons in HD and suggest that neuronal growth, rather than degeneration, may be the harbinger of cell death in this disorder.     

Dendritic Spine Pathology: Cause or Consequence of Neurological Disorders?

Altered dendritic spines are characteristic of traumatized or diseased brain. Two general categories of spine pathology can be distinguished: pathologies of distribution and pathologies of ultrastructure. Pathologies of spine distribution affect many spines along the dendrites of a neuron and include altered spine numbers, distorted spine shapes, and abnormal loci of spine origin on the neuron. Pathologies of spine ultrastructure involve distortion of subcellular organelles within dendritic spines. Spine distributions are altered on mature neurons following traumatic lesions, and in progressive neurodegeneration involving substantial neuronal loss such as in Alzheimer’s disease and in Creutzfeldt–Jakob disease. Similarly, spine distributions are altered in the developing brain following malnutrition, alcohol or toxin exposure, infection, and in a large number of genetic disorders that result in mental retardation, such as Down’s and fragile-X syndromes. An important question is whether altered dendritic spines are the intrinsic cause of the accompanying neurological disturbances. The data suggest that many categories of spine pathology may result not from intrinsic pathologies of the spiny neurons, but from a compensatory response of these neurons to the loss of excitatory input to dendritic spines. More detailed studies are needed to determine the cause of spine pathology in most disorders and relationship between spine pathology and cognitive deficits.     

A comparison of the effects of climbing fiber deafferentation in adult and weanling rats

The climbing fiber input to the cerebellar cortex was destroyed using both electrolytic and chemical (3-acetylpyridine) lesions. The long-term effects of climbing fiber deafferentation on the ansiform lobule of weanling and adult rats were examined at both the light and electron microscopic levels. Image analysis of Golgi-impregnated Purkinje cells indicated a significantly lower number of smooth branches and spiny branchlets following climbing fiber deafferentation of both adult and weanling rats. The results suggest that the lower number of smooth branches and spiny branchlets following climbing fiber deafferentation of the weanling rat is the result of a loss of postnatal growth rather than transneuronal degeneration. Ultrastructural evidence is provided in confirmation of these quantitative findings. Formation of ectopic dendritic spines was found following climbing fiber deafferentation of the weanling rat, but not the adult. It is shown that ectopic spines and the denervated dendritic thorns of these animals were synaptically innervated by the parallel fiber system and basket axons. The formation of ectopic spines on climbing fiber deafferentated Purkinje cells may represent a form of dendritic plasticity. Ultrastructurally, the dendritic arborizations of weanling deafferentated Purkinje cells showed no signs of transneuronal degeneration. However, the primary response to climbing fiber deafferentation in the adult rat was marked transneuronal degeneration of the Purkinje cell dendrites. It is suggested that the inability of the adult Purkinje cell to form ectopic spines and to replace the excitatory postsynaptic potential of the climbing fiber varicosity is directly related to the Purkinje cell's subsequent transneuronal degeneration.     

Degeneration of the human Betz cell due to amyotrophic lateral sclerosis.

The giant pyramidal cell of Betz is known to be partially affected in cases of amyotrophic lateral sclerosis (ALS). Though biochemical, physiologic, and histologic properties of the diseased Betz soma have been investigated, the morphologic status of the largest portion of cell membrane, that of its vast dendritic array, has not. Precentral cortex from six patients, ages 51 to 64 years, who had succumbed to the sequelae of ALS was examined using variants of the Golgi techniques. ALS is shown to be a degenerative disorder which causes a decline in the integrity of the Betz dendritic arbor as well as of the soma of origin. Dendritic fragmentation occurs and numerous irregularities appear, while the number of dendritic spines declines. Concomitantly, a reactive gliosis encroaches upon the soma and may extend onto initial dendritic segments. These observations are remarkably similar to those of the Betz cell in normal aging. The correlation of qualitative histologic data in these conditions is meaningful in light of suggestions that aging and ALS may be related processes. This could provide some clue as to the etiology of Betz cell degeneration and of motor neuron disease.     

Development of neurons in the visual cortex (area 17) of the monkey (Macaca nemestrina): a Golgi study from fetal day 127 to postnatal maturity.

The morphological maturation of several varieties of neurons of cortical area 17 have been followed in Golgi Rapid preparations from Macaque monkeys ranging in age from fetal day 127 to maturity. A developmental sequence common to all varieties of neuron is described. Maturation occurs at the same rate at all cortical depths and appears to relate to the size of the neuron rather than to factors such as generation time, arrival at a final laminar position or cell type. The characteristic laminar patterns of cell type distribution and the specific axonal and dendritic arborisations seen in the adult are generated in the earliest stages of growth and do not occur as the result of elimination from a wider, less precise, distribution. During the period from birth to postnatal week 8 a marked increase in the numbers of dendritic spines is seen in all varieties of neuron including those which will be spine-free in the adult. Following this period an equally marked reduction in spine numbers occurs, initially rapid but continuing at a slower rate even nine months postnatally. Possible relationships between these postnatal dendritic spine changes and the extreme sensitivity of the system to visual input during the early postnatal weeks are discussed.     

Purkinje cell spinogenesis during architectural rewiring in the mature cerebellum

Spines can grow and retract within hours of activity perturbation. We investigated the time course of spine formation in a model of plasticity involving changes in brain architecture where spines of a dendritic domain become innervated by a different neuronal population. Following a lesion of rat olivocerebellar axons, by severing the inferior cerebellar peduncle, new spines grow on the deafferented proximal dendrite of the Purkinje cells (PCs) and these new spines become innervated by parallel fibres (PFs) that normally contact only the distal dendrites. The varicosities of climbing fibre (CF) terminal arbors disappear within 3 days of the lesion. Spine density in the proximal dendritic domain begins to rise within 3 days and continues to increase towards a plateau at 6-8 days. In 'slow Wallerian degeneration' mice, in which axonal degeneration is delayed, climbing fibre varicosities virtually disappear at 14 rather than 3 days. Spine density in the proximal dendritic domain is similar to control Purkinje cells up to 14 days and increases significantly 18 days postlesion. The delayed spinogenesis in the latter mutant is the result of a persistence of the climbing fibre presynaptic structure in the absence of activity. Therefore, climbing fibre activity itself is not directly responsible for the suppression of spine formation, but suppression mechanisms tend to become weaker as long as the structural dismantling of the presynaptic varicosities proceeds. Thus, spinogenesis is guided by two different mechanisms; a rapid one related to changes in homotypic remodeling and a slower one, which requires the removal of a competitive afferent.     

Regulation of hippocampal synapse remodeling by epileptiform activity

We examined the regulation of dendritic spines and synapses by epileptiform activity (EA) in rat hippocampal slice cultures. EA, which was induced by a GABA(A) receptor inhibitor, gabazine, reduced pyramidal neuron spine density by approximately 50% after 48 h and also caused an increase in the average length of remaining spines. To directly determine the effects of EA on synapses, we used fluorescent protein-tagged PSD95, which marks postsynaptic densities. EA induced a net loss of synapses on spines but not shafts; conversely, activity blockade (TTX) induced a loss of shaft synapses. Time-lapse confocal imaging in live tissue slices revealed that EA (1) shifts the balance of synapse gain and loss in dendrites leading to a net loss of spine synapses and (2) induces the formation of new filopodia-like dendritic structures having abnormally slow motility. These results identify EA-induced changes in the density and distribution of synaptic structures on dendrites.     

Morphological development of dendritic spines on rat cerebellar Purkinje cells

The posterior cerebellum is strongly involved in motor coordination and its maturation parallels the development of motor control. Climbing and mossy fibers from the spinal cord and inferior olivary complex, respectively, provide excitatory afferents to cerebellar Purkinje neurons. From post-natal day 19 climbing fibers form synapses with thorn-like spines located on the lower primary and secondary dendrites of Purkinje cells. By contrast, mossy fibers transmit synaptic information to Purkinje cells trans-synaptically through granule cells. This communication occurs via excitatory synapses between the parallel fibers of granule cells and spines on the upper dendritic branchlets of Purkinje neurons that are first evident at post-natal day 21. Dendritic spines influence the transmission of synaptic information through plastic changes in their distribution, density and geometric shape, which may be related to cerebellar maturation. Thus, spine density and shape was studied in the upper dendritic branchlets of rat Purkinje cells, at post-natal days 21, 30 and 90. At 90 days the number of thin, mushroom and thorn-like spines was greater than at 21 and 30 days, while the filopodia, stubby and wide spines diminished. Thin and mushroom spines are associated with increased synaptic strength, suggesting more efficient transmission of synaptic impulses than stubby or wide spines. Hence, the changes found suggest that the development of motor control may be closely linked to the distinct developmental patterns of dendritic spines on Purkinje cells, which has important implications for future studies of cerebellar dysfunctions.     

Effects of intraocular tetrodotoxin on dendritic spines in the developing rat visual cortex: a Golgi analysis

The effect of tetrodotoxin (TTX)-induced monocular impulse blockade on the growth of dendritic spines in the developing rat primary visual cortex was analysed by quantitative Golgi techniques. Between 5 and 21 days postnatal (dpn), rats were injected with TTX every 2 days into the right eye to chronically eliminate optic impulses. Effectiveness of TTX was monitored by loss of the pupillary light reflex. At 21 dpn, the number of spines located on the portion of the apical dendrite within layers III, IV and the superficial region of layer V was reduced by approximately 26%. These decreases were found on the apical dendrites of both large and medium sized pyramidal cells. TTX also reduced the number of spines on the proximal portion of oblique dendrites in layer IV by 16%, yet did not change the number of spines on basilar dendrites. No evidence of transneuronal degeneration was seen following long-term TTX treatment. These data indicate that dendritic spine development in the visual cortex is sensitive to the loss of optic impulses and that the decrease in spine population is principally due to a reduction in spine growth.     

Synapse replacement in the striatum of the adult rat following unilateral cortex ablation

Defining the selective pattern of synapse replacement that occurs in different areas of the damaged brain is essential for predicting the limits of functional compensation that can be achieved after various types of brain injury. Here we describe the time course of dendritic reorganization, spine loss and recovery, and synapse replacement in the striatum following a unilateral cortex ablation. We found that the time course for the transient loss and recovery of dendritic spines on medium spiny I (MSI) neurons, the primary postsynaptic target for corticostriatal axons, paralleled the time course for the removal of degenerating axon terminals from the neuropil and the formation of new synapses on MSI neurons. Reinnervation of the deafferented striatum occurred chiefly by axon terminals that formed asymmetric synapses with dendritic spines of MSI neurons, and the mean density of asymmetric synapses recovered to 86% of the sham-operated rat value by 30 days postlesion. In addition, the synaptic circuitry of the reconstructed striatum was characterized by an increase in the number of multiple synaptic boutons (MSBs), i.e., presynaptic axon terminals that make contact with more than one dendritic spine. Whether the postsynaptic contacts of MSBs are formed with the dendritic spines of the same or a different parent dendrite in the striatum is unknown. Nevertheless, these data suggest that the formation of MSBs is an essential part of the compensatory response to the loss of input from the ipsilateral cortex following the aspiration lesion and may serve to modulate activity-dependent adaptive changes in the reconstructed striatum that can lead to functional recovery.     

Simultaneous analysis of dendritic spine density, morphology and excitatory glutamate receptors during neuron maturation in vitro by quantitative immunocytochemistry

Alterations in the density and morphology of dendritic spines are characteristic of multiple cognitive disorders. Elucidating the molecular mechanisms underlying spine alterations are facilitated by the use of experimental and analytical methods that permit concurrent evaluation of changes in spine density, morphology and composition. Here, an automated and quantitative immunocytochemical method for the simultaneous analysis of changes in the density and morphology of spines and excitatory glutamate receptors was established to analyze neuron maturation, in vitro. In neurons of long-term neuron-glia co-cultures, spine density as measured by drebrin cluster fluorescence, increased from DIV (days in vitro)10 to DIV18 (formation phase), remained stable from DIV18 to DIV21 (maintenance phase), and decreased from DIV21 to DIV26 (loss phase). The densities of spine-localized NMDAR and AMPAR clusters followed a similar trend. Spine head sizes as measured by the fluorescence intensities of drebrin clusters increased from DIV10 to DIV21 and decreased from DIV21 to DIV26. Changes in the densities of NR1-only, GluR2-only, and NR1+GluR2 spines were measured by the colocalizations of NR1 and GluR2 clusters with drebrin clusters. The densities of NR1-only spines remained stable from the maintenance to the loss phases, while GluR2-only and NR1+GluR2 spines decreased during the loss phase, thus suggesting GluR2 loss as a proximal molecular event that may underlie spine alterations during neuron maturation. This study demonstrates a sensitive and quantitative immunocytochemical method for the concurrent analysis of changes in spine density, morphology and composition, a valuable tool for determining molecular events involved in dendritic spine alterations.     

Disseminated corticolimbic neuronal degeneration induced in rat brain by MK-801: potential relevance to Alzheimer's disease

Blockade of N-methyl-D-aspartate (NMDA) glutamate receptors by MK-801 induces neuronal degeneration in the posterior cingulate/retrosplenial cortex and other corticolimbic regions although damage in the latter has not been adequately characterized. This disseminated corticolimbic damage is of interest since NMDA hypofunction, the mechanism that triggers this neurodegenerative syndrome, has been postulated to play a role in the pathophysiology of Alzheimer's disease (AD). Several histological methods, including electron microscopy, were used to evaluate the neurotoxic changes in various corticolimbic regions of rat brain following MK-801 or a combination of MK-801 plus pilocarpine. We found that MK-801 triggers neuronal degeneration in a widespread pattern similar to that induced by phencyclidine and that females showed more damage than males. The neurotoxic reaction involved additional brain regions when muscarinic receptors were hyperactivated by administering pilocarpine with MK-801. Ultrastructural evaluation revealed that a major feature of the neurotoxic action involves degeneration of dendritic spines which entails loss of synaptic complexes. The ultrastructural appearance of degenerating neurons was generally inconsistent with an apoptotic mechanism, although evidence equivocally consistent with apoptosis was observed in some instances. The cell death process evolved relatively slowly and was still ongoing 7 days posttreatment. Relevance of these results to AD is discussed.     

Long-term effect of mossy fiber degeneration in the rat

Mossy fiber-deafferentated rats (20) were permitted to survive from 34 to to 120 days and subsequently examined following Golgi-Cox preparation or after processing for electron microscopy. The primary response to mossy fiber deafferentation was transneuronal degeneration of the granule cell system. Morphological evidence is provided that suggests that the mossy fiber varicosity plays an important role in the fragmentation and removal of the granule cell digitiform dendrite. Computer-assisted image analysis of Golgi-impregnated Purkinje cells indicated significant losses in both smooth branch and spiny branchlet numbers following loss of the mossy fiber input. Ultrastructural examination revealed that a secondary transneuronal degeneration occurred within the dendritic arborization of both Purkinje cells and molecular layer interneurons. Although an overall reduction in the number of dendritic spines occurred along the terminal branchlets following mossy fiber deafferentation, several of the existing spines underwent marked changes in length, with some elongating to more than twice their size. By increasing the length of their spines, denervated Purkinje cells may acquire new synaptic contacts.