Purkinje Cell Pathology and Loss in Multiple Sclerosis Cerebellum

Abstract Cerebellar ataxia commonly occurs in multiple sclerosis, particularly in chronic progressive disease. Previous reports have highlighted both white matter and grey matter pathological changes within the cerebellum; and demyelination and inflammatory cell infiltrates appear commonly. As Purkinje cell axons are the sole output of the cerebellar cortex, understanding pathologic processes within these cells is crucial to develop strategies to prevent their loss and thus reduce ataxia. We studied pathologic changes occurring within Purkinje cells of the cerebellum. Using immunohistochemic techniques, we found changes in neurofilament phosphorylation states within Purkinje cells, including loss of dephosphorylated neurofilament and increased phosphorylated and hyperphosphorylated neurofilament. We also found Purkinje axonal spheroids and Purkinje cell loss, both of which occurred predominantly within areas of leucocortical demyelination within the cerebellar cortex. These changes have important implications for the study of cerebellar involvement in multiple sclerosis and may help design therapies to reduce the burden of ataxia in the condition.


INTRODUCTION
The cerebellum and its efferent and afferent pathways are commonly affected in multiple sclerosis (MS). In patients with established MS, ataxia occurs in about 80% with symptoms and is particularly prevalent in those with progressive disease (24). Both cerebellar tremor and dysarthria may be found commonly in advanced disease.
Cerebellar white matter lesions are commonly found and are often apparent in magnetic resonance imaging (MRI) scans of patients with MS. Recent observations concerning grey matter demyelination in cerebral cortex have led to studies evaluating grey matter disease in the cerebellum (14,17). Indeed, the cerebellar cortex appears a major site for demyelination with one study reporting 38.7% of the cerebellar cortex being affected in a cohort of primary progressive multiple sclerosis (PPMS) and secondary progressive multiple sclerosis (SPMS) patients (9). The same study also showed neuronal pathology with some reductions in Purkinje cell density in lesions (compared with control). No significant reductions in Purkinje cell densities were seen in non-lesional cerebellar grey matter. Other changes in Purkinje cell phenotype have been documented in MS, notably changes in ion channel expression and receptor profiles. The Nav1.8 sensory neuron-specific sodium channel is normally expressed at very low levels in Purkinje cells, but its expression is markedly up-regulated in MS together with annexin light chain (p11), which facilitates the functional expression of this sodium channel (1,2).
Purkinje cells represent the sole output neuron of the cerebellar cortex and thus changes in their function have significant impact on the function of the cerebellum as a whole. The aims of this study were to further characterize Purkinje cell pathology in MS cerebellum particularly with respect to neurofilament phosphorylation states, in light of descriptions of neurofilament abnormalities within white and grey matter of the cerebral hemispheres in MS (5,26). We show increases in neurofilament hyperphosphorylation, loss of dephosphorylated neurofilaments, axonal spheroids and Purkinje cell loss, all of which are linked to lesion formation in the cerebellar cortex.

Cerebellar tissue
Post-mortem cerebellar samples from five control cases and six patients with MS were obtained from the UK Multiple Sclerosis Tissue Bank at the Imperial College London, UK as previously described (6). The clinical background (age, sex, duration of disease, classification of MS, cause of death) of control and MS cohort are present in Table 1. All patients had been clinically diagnosed as having MS and this diagnosis had been confirmed during neuropathologic autopsy examination. Control cerebellum samples were derived from patients who had died from causes not linked to neurologic diseases. Brains were removed, fixed in formalin and embedded in paraffin. Sections of 10 μm in thickness were cut from cerebellar tissue and mounted onto glass slides.

DAB staining on paraffin sections
DAB (3,3'-Diaminobenzidine) staining for myelin basic protein (MBP) (1:3200, Serotec, Oxford, UK) and the macrophage/ microglial markers (DP, DQ and DR subregions of MHC class II) (1:800, Dako, Cambridgeshire, UK) were performed in cerebellar sections of MS and control tissue, according to previously published reports from our laboratory (21). MBP staining was used to identify non-demyelinated areas (MBP positive) from areas with cerebellar demyelination (MBP negative); while HLA-DP, -DQ and -DR were used to detect inflammatory cell infiltrates. Images were acquired using an Olympus IX70 microscope coupled with (Media Cybernetics, Rockville, MD, USA).

Quantification of demyelination and Purkinje cells in MS and control cerebellum
The total number of Purkinje cells was counted in MS and control section, scanning the entire length of the Purkinje cell layer. Purkinje cells were identified based on Calbindin positivity and their specific location. To account for the difference in section sizes and plane of cutting, the length of the Purkinje cell layer was traced and measured in each section using Image J software to obtain the percentage of Purkinje cells per standardized unit length. For these sections, the length of Purkinje cell layer was measured and the number of Purkinje cells in each area counted. Additionally, MS sections were subdivided into non-demyelinated ( Figure 1D blue), leucocortical lesion ( Figure 1D red) and intracortical lesion ( Figure 1D green) based on the MBP staining ( Figure 1C) and dependent on whether the region of demyelination was adjacent to the Purkinje cell layer. For each MS section, the total length of the Purkinje cell layer was measured as earlier. The length of demyelinated regions (leucocortical and intracortical) were also measured to obtain the percentage of demyelinated regions in comparison with the total section.

SMI and spheroids quantification in MS and control cerebellum
The total number of spheroids, Purkinje cells bodies positive for SMI-31, Purkinje cells bodies positive for SMI-34 and Purkinje cells bodies negative for SMI-32 were counted in MS and control sections, scanning the entire Purkinje cell layer. As above, MS sections were subdivided into non-demyelinated, leucocortical lesion, intracortical lesion regions and the frequency of spheroids and abnormal Purkinje cell body staining (SMI-31/SMI-34/SMI-32) were counted and divided by the total number of Purkinje cells in the specific region. Furthermore, regions with leucocortical lesions were subdivided based on presence/absence of inflammatory cell infiltrates using the markers HLA-DP, -DQ and -DR ( Figure 2F,G).

Statistic analysis
Statistic analysis was performed using (GraphPad Software, La Jolla, CA, USA). Data were analyzed using one-way analysis of variance (ANOVA) with post hoc Bonferroni's test between groups where appropriate and unpaired t-test or Mann-Whitney test on parametric or non-parametric data, respectively. Values are expressed as the mean ± standard error and differences at P < 0.05 represent statistic significance.

Distribution of SMI-31, SMI-34 and SMI-32 in cerebellar cortex of MS and control patients
To investigate whether abnormal neurofilament phosphorylation was present in Purkinje cell bodies of MS patients, cerebellar sections from control and MS cases were double immunostained. Antibodies used were the Purkinje cell marker Calbindin with SMI-31 (anti-phosphorylated neurofilament), SMI-34 (antihyperphosphorylated neurofilament) or SMI-32 (anti-nonphosphorylated neurofilament). In control sections, Purkinje cell bodies were negative for SMI-31 ( Figure 3B) and SMI-34 ( Figure 3J). Staining for both SMI-31 and SMI-34 was seen in basket cell axons around Purkinje cell bodies and in parallel fibers (axons of granule cells) in the molecular layer ( Figure 3D,L). In sections from MS patients, a number of Purkinje cell bodies were positive for both SMI-31 ( Figure 3F) and SMI-34 ( Figure 3N). An overall decrease of SMI-31 and SMI-34 labeling of surrounding cell processes was also observed in MS sections ( Figure 3H,P).
Immunostaining using SMI-32 (directed against nonphosphorylated neurofilaments) in control cerebellum revealed that almost all Purkinje cells bodies were positive for this marker ( Figure 3R); while in MS sections, there was a loss of expression in a number of Purkinje cell bodies ( Figure 3V).

Spheroid formation and pathologic abnormalities in MS cerebellum
Double immunostaining for Calbindin and SMI-31/SMI-34 revealed other pathologic changes in sections from MS cases, besides the abnormal Purkinje cell body staining. In the granular layer of MS cerebellum (immediately adjacent to the Purkinje cell layer) spheroids were observed ( Figure 4A,B) These spheroids were ovoidal structures, negative for SMI-32, but strongly immunoreactive for Calbindin and both SMI-31 (A) and SMI-34 (B), suggesting that they consist of accumulated phosphorylated and hyperphosphorylated neurofilament. In some cases, a segment of axon was visible in continuity with the spheroid ( Figure 4C) and some axonal collaterals were seen ( Figure 4D) confirming that these structures represent axonal spheroids. Furthermore, Purkinje cells bodies positive for SMI-34 often displayed an abnormal morphology, with a less flask-like shape ( Figure 4E,F).
Spheroids appeared to arise from the proximal portion of the Purkinje cell axon ( Figure 4E). MS sections were therefore viewed using a confocal microscope. From the three-dimensional image, it was possible to visualize the Purkinje cell body, its axon and the axonal spheroid all in continuity ( Figure 5B). The spheroid structure indeed had no nucleus ( Figure 5A) and was positive for SMI-34 ( Figure 5C).

Quantification of demyelination in MS cases
Areas of cerebellar demyelination were present in all MS cases and not seen in controls. Two main types of demyelination were seen: demyelination of cerbellar cortex in continuity with demyelination in the white matter (leucocortical); and demyelination restricted to cerebellar cortex and not expending to white matter (intracortical). MS sections were subdivided into non-demyelinated regions ( Figure 1D

Purkinje cell number in MS and control cerebellum
In order to determine whether there was a decrease in Purkinje cell number in MS cases compared with control, the entire length of Purkinje cell layer was measured for each section and Purkinje cells counted to obtain the percentage of Purkinje cells per standardized unit length. There was no significant difference in Purkinje cell number in MS cerebellum compared with control sections ( Figure 1A). Interestingly, the number of Purkinje cells were similar between control and nondemyelinated areas of MS cases, but significantly reduced when comparing either control cases or non-demyelinated regions with leucocortical lesion regions of MS ( Figure 1B; *P < 0.05). The Purkinje cell number within intracortical lesions was slightly reduced compared with non-demyelinated areas of MS and control cases, but this reduction was not significant ( Figure 1B).

Correlation between number of spheroids and inflammation in demyelinated areas of MS cerebellum
To investigate whether inflammation could have a role in spheroid formation or in the abnormal Purkinje cell body staining, MS sections were further studied. Leucocortical lesions negative for MBP ( Figure 2F) were divided based on presence/absence of a marked inflammatory cell infiltrate, according to the HLA-DP, -DQ and -DR staining ( Figure 2G). Interestingly, spheroids positive for SMI-31 ( Figure 2C; *P < 0.05) and SMI-34 ( Figure 2D; *P < 0.05) were present with a high frequency in leucocortical lesions positive for HLA compared with lesions without inflammation, suggesting that spheroid formation may be driven by inflammation. No significant differences were present in the frequency of Purkinje cell body staining for SMI-31 (Figure 2A), SMI-34 ( Figure 2B) and SMI-32 ( Figure 2E) between demyelinated areas positive or negative for HLA.

DISCUSSION
Symptoms of cerebellar dysfunction-ataxia, dysarthria and tremor-are common manifestations of MS. In patients with established progressive disease, cerebellar symptoms are difficult to treat and generally worsen as time progresses. Understanding key pathologic changes in the cerebellum may help design therapies for progressive ataxia in the disease. In this report, we show loss of Purkinje cells within demyelinated areas of cerebellar cortex. In addition, we show a number of pathologic changes including loss of dephosphorylated neurofilament labeling in Purkinje cells, increases in phosphorylated and hyperphosphorylated neurofilament staining in Purkinje cells and Purkinje axonal spheroids.
The cerebellar cortex appears a major site for demyelination in MS with one study reporting 38.7% of the cerebellar cortex being affected in a cohort of PPMS and SPMS patients (9). Interestingly, in this study, many grey matter lesions appeared independent of white matter lesions, with some tissue blocks showing very extensive grey matter demyelination in the near absence of underlying white matter disease. The same study also showed reductions in Purkinje cell density in lesional grey matter (compared with control), but no significant reductions in Purkinje cell densities were seen in non-lesional cerebellar grey matter. The current study has confirmed this finding.
In addition, we have shown pathologic changes within Purkinje cells that appear most marked within demyelinated areas. Neurofilament changes in MS have been documented in the cerebral cortex and white matter tracts. Neuronal and axonal cytoskeletons are composed of variably phosphorylated   neurofilaments of low, medium and high molecular weight (MW) that perform many diverse cellular functions, but are predominantly involved in providing structural stability (10,21). Neurofilament (NF) subunits are the most extensively phosphorylated proteins in neurons and this phosphorylation is highly regulated, with an intense phosphorylation in axons and little or no phosphorylation in cell bodies and dendrites (16). Antibodies directed against different neurofilament phosphoisoforms have long been used to study changes in neurofilament phosphorylation as a marker of axonal pathology in several neurologic diseases. In MS, reductions in neurofilament phosphorylation within white matter axons are found, most markedly within lesions (26). In addition, neurons containing hyperphosphorylated (SMI-34 staining) are found in MS cerebral cortex (5). Furthermore, the correlation of clinical disability with elevated neurofilament phosphoisoforms in cerebrospinal fluid (CSF) of MS patients highlights its importance as a biomarker for axonal and neuronal injury (18). Abnormal phosphorylation and accumulation of perikaryal neurofilaments has been found to occur also in Alzheimer's disease (AD) and amyotrophic lateral sclerosis (ALS), highlighting the critical role of neurofilament alteration in the settings of neurodegeneration (11). Accumulation of phosphorylated neurofilament in motor neurons are found in ALS (23); as well as a sevenfold elevation of phosphorylated NF heavy subunit (pNFH levels) in patient's CSF, supporting the potential use of pNFH as a new diagnostic biomarker (4). Neurofilament levels are altered also in AD (28) and neurofilaments found in neurofibrillary tangles are extensively phosphorylated (15). Furthermore, β-amyloid plaques from patients in a preclinical stage of the disease are associated with abnormal accumulation of NF in the absence of tau abnormalities, suggesting that neurofilament changes may represent the earliest cytoskeleton alteration associated with dystrophic neurite formation (3,27). In addition, loss of non-phosphorylated neurofilament in neuronal cell bodies in the spinal cord from ALS (20) and temporal cortical areas in AD (25) has been documented. The current study has revealed neurofilament phosphorylation changes in Purkinje cells, namely reduced numbers of cells staining for SMI-32 (non-phosphorylated NF) and increased numbers of cells staining for SMI-31 and SMI-34 (phosphorylated and hyper-phosphorylated NF). Reduced staining for SMI-32 in ventral spinal neurons has previously been documented in MS samples (20). In keeping with our previous study of MS cerebral cortex, we have found that neurofilament phosphorylation changes are most marked in Purkinje cells within demyelinated areas (5). Furthermore, we subdivided demyelinated areas into leucocortical lesions (cortical demyelination with demyelination in the adjacent white matter) and purely intracortical lesions. Abnormal phosphorylation was present in both lesions, but only significant for leucocortical lesions. It is not clear why Purkinje cells within leucocortical lesions are more susceptible to neurofilament changes. White matter plaques in continuity with demyelinated areas in the cortex may have a combined deleterious effect on Purkinje cells. Transected Purkinje cell axons in the white matter may lead to abnormal axonal transport to and from the cell body which may ultimately result in an abnormal accumulation of perikaryal neurofilament. We have shown that Purkinje cell bodies positive for SMI-34 displayed often an abnormal morphology, with a less flask-like shape. Accumulation of hyperphosphorylated neurofilaments in Purkinje cells of the cerebellar cortex in MS may represent an important neurodegenerative mechanism. Although the precise relationship between hyperphosphorylated neurofilament accumulation and neuronal damage remains unknown, aberrant perikaryal accumulation may contribute to neuronal dysfunction and eventually to cell death.
Axonal spheroids (torpedoes) are found extensively within MS white matter lesions (26). Spheroids represent transected axons and are a common finding in a number of neuro-inflammatory and neurodegenerative disorders. Spheroids arising from Purkinje cell axons have been noted in a number of ataxic conditions and in essential tremor (12,13). Spheroids typically occur in the proximal portion of the Purkinje cell axon and may be a common response to cerebellar injury. The relationship between spheroid formation and Purkinje cell loss, however, appears complex, and it is not clear whether spheroid formation represent an early reversible process that precedes Purkinje cell loss (13). In this study, we have demonstrated Purkinje axonal spheroids, which stain for SMI31 and SMI34, with a clear increase in spheroid formation in demyelinated regions of the cerebellum, suggesting a relationship between demyelination, inflammation and axonal pathology.
Regions with leucocortical lesions were subdivided based on presence/absence of inflammatory cell infiltrates using the markers HLA-DP, -DQ and -DR. This subdivision was not made for intracortical lesions as it was harder to define the inflammation within this lesion. This study has shown that spheroid formation in the cerebellar cortex appear most marked in leucocortical lesions with high levels of inflammatory cell infiltration. This suggests a possible link between axonal transection/spheroid formation and inflammation. In MS white matter, acute axonal injury (measured by amyloid precursor protein accumulation) is elevated in acute lesions and correlates to the degree of inflammation (8). In addition, levels of axonal transection are most marked in acute white matter lesions (26). Our data suggest injury to the proximal Purkinje cell axon with spheroid formation may also be linked to inflammatory cell activity. Our study, however, has shown that the presence of an inflammatory infiltrate was not associated with higher levels of abnormal Purkinje cell body neurofilament phosphorylation changes. There is likely to be a dynamic relationship between axonal spheroid formation, alteration in structural proteins (eg, phosphorylated neurofilaments) and Purkinje cell (F) and SMI-34 (G) were higher in leucocortical lesions compared with non-demyelinated areas of MS. Purkinje cell bodies staining for SMI31/34 and the frequency of spheroids in areas with intracortical lesions were increased comparing with non-demyelinated areas, although not significant. Results are expressed as the mean ± standard error, *P < 0.05; **P < 0.01.
loss, which the current study, by nature of its neuropathologic material, cannot determine.
Other changes in Purkinje cell phenotype have been documented in MS, notably changes in ion channel expression and receptor profiles. The Nav1.8 sensory neuron-specific sodium channel is normally expressed at very low levels in Purkinje cells, but its expression is markedly up-regulated in MS (1). In addition, annexin light chain (p11), which facilitates the functional expression of this ion channel, is also up-regulated in Purkinje cells (2). In experimental models, aberrant expression of Nav1.8 in Purkinje cells causes significant abnormalities of firing patterns of these cells (19,22).
Optimal Purkinje cell function is vital to the overall function of the cerebellum, as Purkinje cell axons represent the sole output of the cerebellar cortex. Loss of Purkinje cells is a common finding in a range of neurologic conditions characterized by ataxia, and in all likelihood, is irreversible. Thus, designing therapies to protect Purkinje cells in MS and related conditions is vital to preserve cerebellar function. Recent reports of bi-nucleate Purkinje cells and heterokaryon formation in experimental models and in human disease have suggested a potential intrinsic repair mechanism of Purkinje cells, whereby nuclear material is donated to an injured cell (6,7). Understanding the precise pathologic changes in Purkinje cells will aid the search for reparative and regenerative therapeutic approaches.