Cerebellar amyloid-β plaques: disturbed cortical circuitry in AβPP/PS1 transgenic mice as a model of familial Alzheimer's disease.

Cerebellar amyloid-β (Aβ) deposition in the form of neuritic plaques and Purkinje cell loss are common in certain pedigrees of familial Alzheimer's disease (FAD) mainly linked to PS1 mutations. AβPP/PS1 transgenic mice, here used as a model of FAD, show a few Aβ plaques in the molecular layer of the cerebellum at 6 months, and which increase in number with age. Motor impairment is apparent in transgenic mice aged 12 months. Combined methods have shown degenerated parallel fibers as the main component of dystrophic neurites of Aβ plaques, loss of synaptic contacts between parallel fibers and dendritic spines of Purkinje cells, and degeneration of granule cells starting at 12 months and increasing in mice 18/20 months old. In addition, abnormal mitochondria and focal loss of Purkinje and basket cells, together with occasional axonal torpedoes and increased collaterals of Purkinje cells in mice aged 18/20 months, is suggested to be a concomitant defect presumably related to soluble extracellular or intracellular Aβ. These observations demonstrate serious deterioration of the neuronal circuitry in the cerebellum of AβPP/PS1 transgenic mice, and they provide support for the interpretation of similar alterations occurring in certain pedigrees with FAD.


INTRODUCTION
Alzheimer's disease (AD) is the most common cause of cognitive impairment and dementia in old age. The neuropathological features of AD comprise neuronal loss mainly in the cerebral cortex and hippocampus accompanied by the presence of amyloid-␤ (A␤) plaques and neurofibrillary tangles, the two main histopathological hallmarks of the disease [1,2]. The majority of cases are sporadic (sAD), but a percentage of AD is triggered by dominant genetic factors as in Down syndrome (chromosome 21 trisomy) and in familial cases with autosomal dominant inheritance (familial AD: FAD) caused by mutations in the amyloid-␤ protein precursor (AβPP) gene on chromosome 21, the presenilin-1 (PS1) gene on chromosome 14, and the presenilin-2 gene on chromosome 1 [3].
The cerebellar cortex may be involved at advanced stages of sAD showing diffuse A␤ deposits, but with the absence of neurofibrillary tangles and loss of Purkinje cells [4][5][6][7]. Cerebellar alterations appear earlier and are more severe in FAD, being characterized, in addition to Purkinje cell loss, by the presence of neuritic plaques [8][9][10][11][12]. However, little is known about the consequences on cortical cerebellar circuitry resulting from cerebellar A␤ deposition, and their possible implications in cerebellar dysfunction in AD and, particularly, in FAD.
Double-transgenic AβPP/PS1 mice express a chimeric mouse/human A␤PP (Mo/HuA␤PP695swe: A␤PP Swedish mutation) and a mutant human presenilin 1 (PS1-dE9), both directed to central nervous system neurons under the mouse PRNP, prion protein promoter [13]. The included Swedish mutation (K595N/M596L) elevates the amount of A␤ and the mutant PS1 allele accelerates the A␤ deposition rate. Transgenic mice show impaired memory and learning performance accompanied by histopathological alterations such as A␤ plaques in the neocortex and cerebellum [13][14][15][16]. For these reasons, AβPP/PS1 double-transgenic mice are considered suitable models for FAD.
The present paper is aimed at identifying alterations in the circuitry of the cerebellar cortex that would permit better understanding of the alterations resulting from cerebellar extracellular A␤ deposition in the impaired cerebellar function in AβPP/PS1 transgenic mice, with implications in FAD.

Animals
The experiments were carried out on AβPP/PS1 mice and wild-type (WT) littermates. The generation of mice expressing the human mutated forms A␤PPswe and PS1dE9 has already been described [17]. Identification of transgenic mice was carried out by PCR of genomic DNA isolated from 1 cm tail clips as proposed by the Jackson Laboratories. Animals were maintained under standard animal housing conditions in a 12 h dark-light cycle with free access to food and water. Experiments were carried out according to ethical guidelines (European Communities Council Directive 86/609/EEC) and approved by the local ethics committee (UB-IDIBELL).

Human cases
The cerebellum of three cases bearing mutations in PS1 with neuritic plaques [17] and four advanced sAD cases with diffuse cerebellar plaques were examined for immunohistochemistry in paraffin sections.

Motor function
Motor function was evaluated in mice aged 6 and 12 months (11 AβPP/PS1 mice from different litters and 8 WT littermates per age) with two different behavioral paradigms: the wire hanging test and the rotarod. For the wire-hanging test, the apparatus consisted of a coat hanger (50 cm long; 2 mm in diameter) suspended 32 cm above a sawdust-covered surface. Latency-tofall was measured from the time animals were placed hanging by their forepaws on the wire until they fell, with a maximum time of 1 min. Three days after the exposure to the wire hanging test, animals were placed in the rotarod (Rotamex 4/8; Columbus Instruments, OH, US). The apparatus consisted of a plastic roller (3 cm in diameter) with small grooves running along its turning axis. Mice received four trials per day for two consecutive days. During each test session, animals were placed on the rod rotating at a constant speed (4 rpm), and then the rod started to accelerate continuously from 4 to 40 rpm over 600 s. The latency-to-fall off the rotarod was recorded. The improvement in the motor coordination throughout the different trials was calculated as the difference between the latency-to-fall in each trial with respect to the first one. A criterion of a minimum latency-to-fall of 10 s during the last trial was established in order to classify animals depending on their ability to exhibit motor function in the rotarod.
Data were analyzed with Student's t-test. χ 2 analysis was performed to compare the percentage of mice that acquired motor coordination criteria in the rotarod between the different experimental groups. In all the experiments, the significance level was set at p < 0.05.

Histological procedures
Animals aged 6, 12, and 18/20 months were examined for the chronological appearance of cerebellar A␤ plaques (3 AβPP/PS1 mice and 3 WT per time-point). With this aim the number of plaques per section was counted in 5 sections per animal.
However, further histological studies were centered in animals aged 12 months (n = 4) and 18/20 months (n = 3), and corresponding age-matched control littermates (n = 3 per time-point). AβPP/PS1 mice and control animals were deeply anesthetized by intraperitoneal injection (0.2 mL/10 g body weight) with a mixture of ketamine (100 mg/kg) and xylazine (20 mg/kg) prior to intra-cardiac perfusion with 4% paraformaldehyde in PBS delivered at a pressure of 19 mL/min. Brains were removed and post-fixed in the same fixative for 4 h. The cerebella were dissected and embedded in paraffin. Ten-m-thick mid-sagittal sections were collected on slides and used for the morphometric and morphological analyses.

Morphometric analysis
The morphometric analysis was performed on digital images of sections of cerebellar vermis immunostained for calbindin (3 AβPP/PS1 mice and 3 WT littermates and 5 sections per animal), captured using an Olympus camera mounted on an Olympus BX51 microscope at 4X magnification. All images were analyzed using the NIH ImageJ software.
The average area of cerebellar cortex in mm 2 was obtained by outlining the contour of sections. The data were analyzed with Student's t-test. The significance level was set at p < 0.05.
For Purkinje cell counts, the total number of Purkinje cells per section was counted and the depth of the Purkinje cell layer measured. The Purkinje cell linear density was obtained by dividing the number of Purkinje cells by the depth of the layer. Statistical evaluation of differences was performed by means of the two-level nested ANOVA on ranks test.

Morphological analysis
Light microscopy immunohistochemistry: For immunohistochemistry the following antibodies were used: mouse anti-A␤ IgG (1 : 50, Dako, DK), rabbit anti-GFAP serum (1 : 500, Dako, DK), voltage dependent anion channel (VDAC) (1 : 100, Calbiochem, Spain), Iba1 rabbit polyclonal antibody for microglia Two slides per animal were used for each immunoreaction. After de-waxing, sections were boiled with citrate buffer for 20 min to enhance antigenicity. Sections for A␤ immunocytochemistry were pre-treated with 98% formic acid for 3 min. The endogenous peroxidases were blocked by incubation in 10% methanol-1% H 2 O 2 solution for 5 min. To block unspecific sites, sections were incubated with 3% normal horse serum for 1 h. Incubation with the primary antibodies was carried out at 4 • C overnight. The immunoreaction was revealed with biotinylated secondary antibodies (Dako, DK), followed by EnVision+ system peroxidase (Dako, DK) and 3 ,5 -diaminobenzidine tetrahydrochloride. Some sections were incubated only in buffer instead of primary antibodies; no immunostaining was detected in these sections.
In situ end-labeling of nuclear DNA fragmentation (TUNEL) was carried out as previously reported [18] using cerebellar sections of x-irradiated newborn rats (2 Gy) killed 6 h after irradiation as positive controls. The same controls were used for immunocytochemistry to active caspase 3.
Human cases were processed in the same way excepting double-labeling immunohistochemistry to A␤ and VDAC, in which the peroxidase reaction of A␤ was visualized with diaminobenzidine, NH 4 NiSO 4 and H 2 O 2 . The immunoreaction results in a blue-grey precipitate.
To block lipofuscin fluorescence, de-waxed sections were stained with a saturated solution of Sudan black B (Merck, DE) for 30 min. After treatment with citrate buffer to enhance antigenicity, the sections were incubated at 4 • C overnight with combinations of primary antibodies. Sections for A␤ immunocytochemistry were pre-treated with 98% formic acid for 3 min. After washing, the sections were incubated with the appropriate secondary antibodies conjugated with Alexa488, Alexa546, or Alexa633 (1 : 400, Molecular Probes, CA, US). After counterstaining of nuclei with DRAQ5 ™ (1 : 2,000, Biostatus, UK), the sections were mounted in Immuno-Fluore mounting medium (ICN Biomedicals, CA, US), sealed, and dried overnight. Sections were examined with a Leica TCS-SL confocal microscope.
The quantitative estimation of co-localized antigens was performed calculating the 'co-localization coefficients' on plotted cytofluorograms [19]. The data were analyzed with Student's t-test. The significance level was set at p < 0.05.

Transmission electron microscopy
AβPP/PS1 and WT mice aged 12 and 18/20 months (n = 3 per age and group) were deeply anaesthetized and perfused with 4% paraformaldehyde and 0.1% glutaraldehyde in PBS. Sagittal sections of cerebellar vermis (1 mm thick) were post-fixed in 1% OsO 4 for 2 h and embedded in Epon 812. Semithin sections were stained with toluidine blue and ultrathin sections with uranyl acetate and lead with citrate. Samples were examined with a Jeol Jem 1011 electron microscope.

Motor function
AβPP/PS1 mice exhibited preserved motor function at the age of 6 months but decreased motor function in both the wire hanging and the rotarod tests at the age of 12 months. For this reason, no further tests were carried out on animals older than one year. In the wirehanging test, the fall latency of AβPP/PS1 mice was significantly lower (−55%) than in WT mice (p < 0.05) (Fig. 1a). In addition, AβPP/PS1 mice exhibited less improvement in the rotarod test throughout the consecutive trials. However, differences between AβPP/PS1 and WT mice were not statistically significant in any test trial (Fig. 1b). The percentage of animals that reached the motor coordination criterion in the rotarod after 8 trials was 75% for WT mice and 27% in the AβPP/PS1 group, revealing a significant ability reduction in transgenic animals (p < 0.05) (Fig. 1c).

Gross changes in the cerebellum
Measurement of the mid-sagittal section area revealed that the cerebellar vermis of AβPP/PS1 mice Fig. 1. Behavioral testing of the motor function of AβPP/PS1 mice in comparison with age-matched WT at 12 months. a) In the wire hanging test, transgenic mice exhibit a significant decrease in the latency-to-fall with respect to WT animals. b) Deterioration of motor function in transgenic animals is manifested by reduced skills in rotarod performance throughout eight consecutive trials. c) Latencyto-fall of each experimental animal during the last test trial in the rotarod. Only three of the eleven AβPP/PS1 mice reached the motor function criterion in the rotarod (minimum latency-to-fall of 10 s, dotted line), revealing impairment in the motor function of transgenic mice compared to WT littermates (p < 0.05). Data are expressed as mean ± SD. *p < 0.05, compared to WT mice (Student's t-test).
was smaller than the cerebellum of their WT littermates. A 20% reduction was found in transgenic mice when compared with age-matched controls at the age of 18/20 months (WT versus AβPP/PS1: 9.96 ± 0.16 mm 2 versus 8.35 ± 0.33 mm 2 , p < 0.01).

General microscopical findings
In the cerebellum of 6-month-old AβPP/PS1 mice, very few A␤ deposits were observed, and these were exclusively found in the molecular layer (ML). A␤ deposits were frequent in the ML in 12-month-old mice (Fig. 2). A few A␤ deposits were also present in the internal granular layer immediately beneath the Purkinje cell layer. The number of deposits increased with disease progression in mice aged 18/20 months, the last time-point examined ( Fig. 3A-C). A␤ deposits were not observed in WT mice. At early stages, A␤ extracellular deposits were mainly diffuse; a few neuritic plaques were observed at 12 months, whereas such neuritic plaques were predominant at the age of 18/20 months. Neuritic plaques were characterized by a central core of A␤ surrounded by microglia (Fig. 3D) and astrocytes, and by dystrophic neurites clearly visualized with antibodies to mitochondrial porin VDAC (Fig. 3E), synaptophysin (Fig. 3F), among other proteins.
Cotton-wool plaques as seen in FAD cases due to PS1 mutations were not seen in AβPP/PS1 mice. Yet A␤ angiopathy was common, involving the meningeal and cerebellar blood vessels (data not shown).

Extracellular Aβ deposits and glial cells
Extracellular A␤ deposits in the cerebellum were focal (Fig. 4a). However, astrocytes and Bergmann glial cells throughout the ML were more reactive to GFAP immunohistochemistry in AβPP/PS1 mice aged 18/20 months when compared with age-matched WT animals (Fig. 4b, c). In addition, astrocyte processes, as identified with double-labeling immunofluorescence to A␤ and GFAP, and confocal microscopy, surrounded A␤ deposits in the ML (Fig. 4d). Furthermore, with electron microscopy, microglial cells with dense nucleus and cytoplasm filled with lysosomes and residual bodies were often encountered adjacent to dystrophic neurites surrounding A␤ plaques (Fig. 4e).

Purkinje cells
Calbindin immunohistochemistry revealed various alterations of Purkinje cells, including the presence of enlarged axons (Fig. 5a), axonal spheroids (Fig. 5b), and hypertrophic infra-ganglionic plexuses (Fig. 5c). In addition, some convolutions were devoid of Purkinje   (Fig. 5d). Cell counting revealed that the linear cell density of Purkinje cells was reduced by 18% and 25% in transgenic mice in comparison with WT animals at the age of 12 and 18/20 months, respectively (p < 0.05 and p < 0.001, at the ages of 12 months and 18/20 months).
Electron microscopic examination of the molecular layer revealed the presence of large mitochondria in dendrites and small branches of Purkinje cells, accompanied by occasional dark mitochondria at the ages of 12 and 18/20 months (Fig. 5e). Dendritic arbors of Purkinje cells were distorted around the amyloid deposits, as shown by doublelabeling immunofluorescence to calbindin and A␤, and confocal microscopy ( Fig. 5f, g). No calbindinimmunoreactive dystrophic processes were evidenced with double-labeling immunofluorescence and confocal microscopy.
Electron microscopy of the Purkinje cell layer disclosed occasional Purkinje cells with shrunken cytoplasm and dark nuclei, and increased lipofuscin granules at the ages of 12 and 18/20 months (Fig. 5h).

Parallel fibers and granule cells
Calretinin immunohistochemistry, used as a marker of parallel fibers, revealed dystrophic terminals around A␤ deposits in the ML in mice aged 18/20 months (Fig. 6a). This was further demonstrated by doublelabeling immunofluorescence to calretinin and A␤ and confocal microscopy (Fig. 6b, c). Electron microscopy revealed the dystrophic nature of these processes, showing huge processes filled with altered mitochondria, and large numbers of membrane-bound pleomorphic bodies and other debris (Fig. 6d). Since pentraxin 1 (NP1) is part of the program of apoptotic death in cerebellar granule neurons [20], antibodies to NP1 were used to detect degeneration in cerebellar dystrophic neurites. Strong NP1 immunoreactivity was observed in dystrophic neurites of the cerebellum in AβPP/PS1 transgenic mice (Fig. 6e).
Electron microscopic examination of the granular layer disclosed occasional granule cells with degenerative features such as compacted peripheral chromatin and electron-dense cytoplasm (Fig. 6k, l).
Caspase 3 immunoreaction failed to show positive cells in the AβPP/PS1 mice cerebellum. Also, after TUNEL reaction, no cells were labeled at any time-point of the study. In contrast, consistent numbers of caspase-3-immunoreactive cells and of neurons stained with TUNEL were present in the cerebella of newborn x-ray irradiated rats, killed 6 hours later, which were used as positive controls (data not shown).

Basket neurons
Parvalbumin immunoreactivity was used to detect Purkinje cells and interneurons of the molecular layer. Dystrophic neurites were not stained with antiparvalbumin antibodies. Not only Purkinje cells (as already noted) but also local-circuit neurons were lost or unstained in AβPP/PS1 transgenic mice aged 18/20 months (Fig. 8b) when compared with WT (Fig. 8a). The consequences were best illustrated by doublelabeling immunofluorescence to calbindin and RT97, Fig. 9. Double-labeling immunofluorescence and confocal microscope to A␤ (a, d, g, green) and VDAC (b, red), calretinin (e, red) and VGluT1 (h, blue) in the molecular layer of the cerebellum of mice aged 12 months. Dystrophic neurites are seen surrounding A␤ deposits.
which decorates the axons of basket cells, showing reduced baskets around apparently preserved Purkinje cells in AβPP/PS1 transgenic mice (Fig. 8d) when compared to controls (Fig. 8c).

Dystrophic neurites with disease progression in AβPP/PS1 mice
Similar studies were carried out in AβPP/PS1 mice aged 12 months. Most A␤ deposits were devoid of dystrophic neurites, and, therefore, their morphology was reminiscent of diffuse plaques. However, some A␤ deposits were surrounded by dystrophic neurites. Double-labeling immunofluorescence and confocal microscopy disclosed altered neurites filled with VDAC (Fig. 9a) indicating accumulation of mitochondria, and increased expression of calretinin-immunoreactive granules (Fig. 9b) and weak enhancement in the expression of VGluT1 (Fig. 9c), thus revealing their origin in parallel fibers.

Neuritic plaques in the cerebellum of FAD
For comparative purposes, paraffin sections of the cerebellar vermis of sAD and FAD cases were examined for immunohistochemistry. Unfortunately, calretinin and VGluT1 antibodies did not work properly in human cases. However single-and doublelabeling immunohistochemistry to synaptophysin, VDAC and A␤ disclosed enhanced synaptophysin and VDAC immunoreactivity around A␤ plaques in FAD (Fig. 10). No similar alterations were found in relation with diffuse A␤ plaques in sAD (data not shown).

DISCUSSION
Motor function was evaluated in AβPP/PS1 and WT mice by using two common tests, the wire hanging test and the rotarod, which demonstrate altered motor performance. It may be argued that the observations made using these methods are indicative of poor motor function that may result from causes other than cerebellar impairment, such as reduced motor strength and altered basal ganglia motor coordination. More specific methods could have been employed but morphological examination of the cerebellum promptly served to focus the study on the objective visualization of altered neurons and damaged intrinsic cerebellar pathways.
The present observations have shown increased A␤ deposition with disease progression in the cerebellar cortex of AβPP/PS1 mice. A␤ plaques without dystrophic neurites were predominant in mice aged 12 months although a small percentage of A␤ deposits showed incipient dystrophic neurites at this age. Cerebellar A␤ deposits at advanced stages are reminiscent of A␤ neuritic plaques in the cerebral cortex and other brain regions in AβPP/PS1 mice [14], and they are similar to those seen in the cerebellum in FAD [12]. Dystrophic neurites of A␤ plaques in the cerebellum of AβPP/PS1 mice contain phospho-tau and several proteins including ␣-synuclein, synaptophysin and ubiquitin, and they are filled with abnormal mitochondria and autophagic inclusions, as are plaques of the cerebral cortex and other regions [14,16]. This is an important point, as dystrophic neurites are a fair manifestation of altered, non-functional neuronal processes associated with A␤ deposits.
Cell composition and connectivity of the cerebellar cortex is well established [22][23][24][25]. Briefly, major afferents are the mossy fibers and climbing fibers. Mossy fibers establish excitatory synapses with dendrites of granule cells in the glomeruli; climbing fibers promote excitatory impulses in the proximal dendrites of Purkinje cells, type II Golgi cells, stellate cells, and basket neurons. Granule cells throughout the ascending axons, divided into parallel fibers in the molecular layer, form excitatory contacts with spines of dendritic arbors of Purkinje cells. The system is modulated by distinct types of local neurons covering type II Golgi cells in the granular layer which make inhibitory synapses at granule cell dendrite, and mossy fibers at the glomeruli; stellate cells in the molecular layer which are inhibitory of Purkinje cells at the dendritic arbors; and basket cells, just above the Purkinje cell layer, whose axons embrace the cell body of Purkinje cells and provide robust inhibition of these cells. Purkinje cells are inhibitory and are the only efferents of the cerebellar cortex to the deep cerebellar nuclei and vestibular nuclei.
The present observations have shown degeneration of parallel fibers associated with A␤ plaques in the molecular layer as revealed by calretinin staining and further supported by the expression of the vesicular glutamate transporter 1 (VGluT1), as a marker of parallel fibers [26]. This is associated with reduced focal synaptic contacts of parallel fibers to Purkinje cells, as shown by combined cytofluorograms. The toxic effect of A␤ on parallel fibers is further demonstrated by the observation of local over-expression of pentraxin 1 (NP1) which is associated with neuronal degeneration in AD [20], and eventual cell death under particular conditions [15]. Whether loss of parallel fibers results in the retrograde degeneration of individual granule cells, as evidenced with electron microscopy, remains to be determined, but appears likely.
In addition, other cells are altered without apparent direct contact with A␤ deposition. This was seen by using antibodies against calbindin and parvalbumin as markers of Purkinje cells and other local cortical cerebellar neurons [27,28]. Quantitative studies have shown a 25% decrease of Purkinje cells in AβPP/PS1 mice aged 18/20 months. In addition, abnormalities of Purkinje cells include changes of dendritic arbors in the molecular layer, axonal torpedoes, increased axon collaterals, and, importantly, the presence of large mitochondria mainly in the dendritic arbors, even at the thin terminal branchlets. Similar mitochondrial changes are not seen in normal Purkinje cells [24]. Focal loss of basket cells is shown by focal loss of parvalbumin-immunoreactive cells just above the Purkinje cell layer and by the reduced number of basket terminals around the soma of groups of Purkinje cells.
Since loss of Purkinje cells and axonal torpedoes, and loss of basket cells, is not physically related to A␤ plaques, it can be suggested that they are the consequence of other factors rather than of the presence of plaques. In this line, several studies have shown that soluble A␤ oligomers are causative of neurodegeneration [29][30][31][32][33][34], and that maintenance of A␤ oligomers does not impair disease progression in spite of A␤ plaque depletion [35].
Increased soluble A␤ has been demonstrated in the cerebral cortex of AβPP/PS1 mice [14,36], and the level of A␤ 42 is higher than the level of A␤ 40 in the frontal cortex, hippocampus, and cerebellum in these mice [36].
Furthermore, intracellular A␤PP and A␤ are observed in AD under appropriate staining conditions [37]. Microdissected human neurons analyzed by a sensitive ELISA have shown increased intracellular Fig. 11. Diagram showing the changes in cerebellar cortex circuitry in AβPP/PS1 mice compared with wild type animals (WT). 1) Degeneration of parallel fibers (PF) in association with A␤ plaques (BA) in the molecular layer; 2) Focal loss of excitatory synaptic contacts following damage to parallel fibers on dendritic spines of Purkinje cells (Pc), and probably also on Golgi neurons (G), stellate (S) and basket (B); 3) Loss of granule cells presumably secondary to damage of parallel fibers; 4) Focal loss of basket cells and loss of basket terminals around Purkinje cells may account for decreased Purkinje cell inhibition; 5) Focal degeneration of Purkinje cells, altered dendritic arbors, axonal swellings (torpedoes) and increased collateral plexus, probably resulting in reduced cortical inputs to the deep cerebellar nuclei. levels of A␤ 42 in pyramidal neurons and Purkinje cells in AD when compared with corresponding neurons in age-matched controls [38]. Therefore, although not specifically examined in the present study, it can be assumed that both extracellular soluble and intracellular A␤ are putatively harmful to cerebellar neurons in A␤PP/PS1 mice. Whether a link exists between putative intracellular A␤ and abnormal mitochondria in Purkinje cells in A␤PP/PS1 mice should be examined by using suitable experimental paradigms.
As shown in Fig. 11, morphological cerebellar abnormalities in AβPP/PS1 transgenic mice can be enumerated as follows: i) loss of parallel fibers in association with A␤ plaques; ii) loss of synaptic contacts of granule cells to Purkinje cells as a result of focal degeneration of parallel fibers; iii) focal loss of granule cells probably resulting from retrograde degeneration of parallel fibers; iv) focal loss of baskets probably leading to focal impairment of Purkinje cell inhibition; and v) axonal torpedoes, increased axon collaterals, and focal loss of Purkinje cells probably resulting in reduced cortical cerebellar outputs. Motor impairment is not uncommon at advanced stages of FAD due to PS1 mutations. Spastic paraparesis and parkinsonism have been observed in several families [39][40][41][42][43]. Cerebellar symptoms occur in some pedigrees [43] and are marked in others [10,12]. Although the majority of FAD cases with cerebellar involvement are associated with PS1 mutations, rare mutations in AβPP are manifested with generalized cerebellar A␤ deposition [9]. On the other hand, transgenic mice with PS1 mutations alone have not shown clear AD pathology so far. Therefore, it is difficult at this moment to attribute to A␤PP or to PS1 mutations their impact on the cerebellar damage in AβPP/PS1 mice.
The present findings in AβPP/PS1 mice, together with previous observations showing axonal degeneration in the spinal cord [44], may help to increase our understanding of motor and cerebellar impairment in certain advanced FAD cases manifested with motor and cerebellar symptoms with Purkinje cell loss and abundant A␤ neuritic plaques.