The mutant Moonwalker TRPC3 channel links calcium signaling to lipid metabolism in the developing cerebellum

The Moonwalker (Mwk) mouse is a model of dominantly inherited cerebellar ataxia caused by a gain-of-function mutation in the transient receptor potential (TRP) channel TRPC3. Here, we report impairments in dendritic growth and synapse formation early on during Purkinje cell development in the Mwk cerebellum that are accompanied by alterations in calcium signaling. To elucidate the molecular effector pathways that regulate Purkinje cell dendritic arborization downstream of mutant TRPC3, we employed transcriptomic analysis of developing Purkinje cells isolated by laser-capture microdissection. We identified significant gene and protein expression changes in molecules involved in lipid metabolism. Consistently, lipid homeostasis in the Mwk cerebellum was found to be disturbed, and treatment of organotypic cerebellar slices with ceramide significantly improved dendritic outgrowth of Mwk Purkinje cells. These findings provide the first mechanistic insights into the TRPC3-dependent mechanisms, by which activated calcium signaling is coupled to lipid metabolism and the regulation of Purkinje cell development in the Mwk cerebellum.


Introduction
Cerebellar ataxia defines a heterogeneous group of neurological disorders that affect the cerebellum and its afferent and efferent connections and manifest primarily as progressive gait and limb incoordination. The inherited forms of cerebellar ataxia encompass a growing list of genetically diverse disorders with more than 35 genetic loci known to cause the autosomal dominant ataxias alone, which are also referred to as spinocerebellar ataxias (SCAs) (1,2). Despite this genetic heterogeneity, common pathological mechanisms that may be shared among the different forms of cerebellar ataxia are beginning to emerge (2,3). Notably, disruption of intracellular calcium homeostasis and signaling in Purkinje cells has been proposed as a key mechanism in the pathogenesis of SCAs (1,4,5).
The transient receptor potential (TRP) channel TRPC3 is a non-selective cation channel that is highly expressed in the Purkinje cells of the cerebellum (6,7). TRPC3 is required for metabotropic glutamate receptor subtype 1 (mGluR1)-dependent synaptic transmission in Purkinje cells (6). Both genetic loss of Trpc3 and the dominant Moonwalker (Mwk) gain-of-function point mutation in Trpc3 result in cerebellar ataxia in the mouse (6,8), highlighting the importance of Purkinje cell calcium homeostasis for proper cerebellar function. Recently, we have identified the first functionally pathogenic variant (R672H) in the human TRPC3 gene in a patient with adult-onset cerebellar ataxia (9). The human R672H mutation likely acts through a toxic gain-offunction mechanism similar to the Mwk mutation. Moreover, TRPC3 signaling has been linked to several other genetic forms of cerebellar ataxia in human and mouse including SCA1, SCA14, SCA15 and mutations in the GluD2 receptor (1) (reviewed in (7)). Thus, the disruption of the TRPC3 signaling pathway might be a common pathological mechanism underlying cerebellar ataxia in mouse and human.
Interestingly, besides adult-onset Purkinje cell loss, the Mwk mutation in TRPC3 causes impairments in Purkinje cell function and dendritic arborization during cerebellar development (8,10). In contrast, no dendritic phenotype and a milder form of ataxia have been observed in the Trpc3 knockout mice (6,11). Hence, it is the sustained activation of TRPC3 that provides an important regulatory effect on dendritic growth in cerebellar Purkinje cells. This model is consistent with the observation that overexpression of SCA14-associated PKCγ mutants that fail to inhibit TRPC3 (12) causes impairments in the dendritic development of Purkinje cells (13). Similarly, related studies have shown the inhibition of Purkinje cell growth upon chronic activation of mGluR1 or PKC activation (14,15). However, the effector pathways that mediate the inhibition of dendritic growth downstream of activated TRPC3 have remained enigmatic.
Here, we report impairments in dendritic growth and synapse formation early on during Purkinje cell development in the Mwk cerebellum that are accompanied by alterations in key calcium signaling pathways. We also identify Purkinje cell-specific gene expression changes that result in abnormal lipid homeostasis in the mutant mice. Together, our study provides mechanistic insights into the TRPC3-dependent processes that link activated calcium signaling to lipid metabolism and the regulation of Purkinje cell development in the Mwk cerebellum.

Mwk cerebellum
Previously, we demonstrated abnormal Purkinje cell dendritic morphology in 3-week-old Mwk mice (8). To investigate whether the mutant phenotype results from a failure of the dendritic tree to arborize or whether it is a consequence of atrophy following normal outgrowth, we carried out a time course experiment in organotypic cerebellar slice cultures of Mwk mice compared with their wild-type littermates. The initial development of the Purkinje cell dendritic tree was similar in wild-type and mutant mice based on the extent of the Purkinje cell dendritic area (Fig. 1A). However, starting from postnatal day 15 (P9+DIV6), impairments in arborization were observed in the Mwk Purkinje cells (Fig. 1A). Although wild-type Purkinje cells continued to grow their dendritic arbors, mutant Purkinje cells failed to further expand the area of their dendritic tree ( Fig. 1A and B), suggesting that it is the late phase of dendritic outgrowth that is specifically impaired in Mwk Purkinje cells. P/Q-type calcium channels have been shown to contribute to the inhibition of dendritic growth following activation of mGluR1 (11). Moreover, TRPC3 signaling in cardiac myocytes has been linked to activation of voltage-gated calcium channels (VGCCs) (16). These observations raised the question of whether VGCCs might contribute to the dendritic phenotype of Purkinje cells in the Mwk mouse. To assess whether inhibition of VGCCs could rescue the dendritic phenotype in Mwk Purkinje cells, we used a combination of ω-agatoxin IVA and ω-conotoxin MVIIC to Treatment with P/Q block significantly increased the dendritic area of the Mwk but not WT Purkinje cells (P < 0.0001, ANOVA, n = 218).
inhibit P/Q-type calcium channels (P/Q block) (11). Organotypic slice cultures (P9) were treated for 12 days with P/Q block, starting at DIV2. As seen previously, dendritic arborization of Mwk Purkinje cells was significantly impaired compared with wild-type cells (Fig. 1C). Interestingly, inhibition of P/Q-type calcium channels significantly increased the dendritic area of Mwk Purkinje cells when compared with untreated Mwk cultures by 50% (4457 ± 316 μm 2 versus 2176 ± 135.2 μm 2 ) (Fig. 1C). Treatment of wildtype slices with P/Q block had no significant effect (Fig. 1C) (11). Together, these results suggest that P/Q-type calcium channels are involved in the dendritic growth inhibition of Purkinje cells that is caused by mutant Mwk TRPC3.
We next examined whether the dendritic changes observed in the Mwk mice were accompanied by changes in the placement of excitatory synaptic terminals onto Purkinje cell dendritic trees. The distal part of the Purkinje cell dendritic tree receives excitatory input from the parallel fibers (PFs) of the granule cells, whereas the proximal portion of the dendritic tree is innervated by inputs from climbing fibers (CFs). The innervation areas of PFs and CFs are intimately linked, and stabilization of PF synapses restricts the innervation territory of CFs on the Purkinje cell dendritic tree and vice versa (17,18). First, we investigated the PF innervation pattern of wild-type and Mwk Purkinje cells by their immunoreactivity to the vesicular glutamate transporter VGLUT1. We found no detectable differences in the immunostaining pattern for VGLUT1 in cerebellar sections from Mwk mice compared with their wild-type littermates (density VGLUT1 staining WT = 76.26 ± 7.093; Mwk = 74.54 ± 8.694; P = 0.7992; n = 3) ( Fig. 2A). Both wild-type and mutant cerebellar sections exhibited strong VGLUT1-positive puncta representing PF inputs throughout the molecular layer. To determine the spatial distribution of CF terminals on Purkinje cell dendritic arbors, we carried out immunostainings for VGLUT2. In adult wild-type cerebellar sections, the distribution of VGLUT2 staining representing the extent of the CF territory extended to a mean relative height of 81 ± 0.9862% of the molecular layer ( Fig. 2B and C), consistent with previous reports using wild-type mice (17,19,20). In contrast, in cerebellar sections of adult Mwk mice, the mean relative height was reduced to 63 ± 1.108%. We also assessed the extension of the CF territory in cerebellar sections of younger animals. A significant reduction in CF arborization along the Purkinje cell dendrites was observed in the mutants as early as postnatal day 14 (P14) (Fig. 2C), concurrent with the observed dendritic abnormalities (Fig. 1A). Together, these results suggest that Mwk mice exhibit profound dendritic and synaptic impairments during development as well as in adulthood.

Mwk cerebellum
Having identified these dendritic and synaptic abnormalities, we next investigated the mechanism underlying Mwk TRPC3mediated impairments in Purkinje cell development. As the Mwk phenotype results from a gain-of-function mutation in the calcium-permeable TRPC3 channel (8) and we have also shown that VGCCs are involved (Fig. 1C), we asked which calcium-sensitive downstream signaling pathways might be affected in the developing Mwk cerebellum. We prepared cerebellar extracts from Mwk mice and their wild-type littermates at several time points during postnatal development (P15-P46), followed by immunoblotting for key proteins of neuronal calcium signaling. We found that several calcium-sensitive signaling molecules were altered in the mutant cerebellum compared with wild-type controls. Changes were observed as early as 2 weeks postnatally. The phosphorylation status and thus the activity levels of both the extracellular-regulated kinases 1/2 (ERK1/2) and the transcription factor cAMP response element-binding protein (CREB) were increased in Mwk cerebellum consistent with activated calcium signaling (Fig. 3A). Levels of phosphorylated calcium/calmodulindependent protein kinase type II (CaMKII) were unchanged in Mwk cerebellum at all ages studied (Fig. 3B). In contrast, phosphorylation of calcium/calmodulin-dependent protein kinase type IV (CaMKIV) was reduced in the Mwk cerebellum (Fig. 3C). Noteworthy, phosphorylated CaMKIV is predominantly present in Purkinje cells of the cerebellum (Supplementary Material, Fig. S1). Hence, the changes in phosphorylated CaMKIV that were observed in total cerebellar extracts reflect specific changes in the Purkinje cells. Together, these findings demonstrate alterations in important calcium-sensitive signaling kinases and downstream transcription factors in the developing Mwk cerebellum.

Mwk Purkinje cells display significant gene expression changes
Little is known about the specific targets regulated by calciummediated signaling pathways that control Purkinje cell development. On the basis of our findings, the Mwk mouse represents an ideal system to investigate the intrinsic factors that link altered calcium signaling and Purkinje cell development. To identify gene expression changes specifically in Mwk Purkinje cells, we carried out laser capture microdissection (LCM) of Purkinje cells, followed by RNA isolation and microarray analysis using the Affymetrix Mouse Gene 1.0 ST array. This was done in 18-day-old Mwk mice and their wild-type littermates, thus around the critical period and the onset of molecular and behavioral changes in the mutant mice. Samples were obtained from four sets of Mwk and wild-type littermates, and 1000 Purkinje cells were collected per cerebellum. Applying a stringent detection threshold (1.5-fold change) and cut-off P-value of ≤0.5, we identified statistically significant changes in the expression levels of 634 genes between wild-type and Mwk Purkinje cells. Fifty-one percent (323) of genes were upregulated and 49% (311) of genes were downregulated ( Fig. 4A and Supplementary Material, Table S1). The results were reproducible among the four biological replicates (Fig. 4A).
To verify the microarray results, we selected several genes that showed significant expression changes between Mwk and wild-type Purkinje cells for independent validation by quantitative reverse transcription polymerase chain reaction (qRT-PCR) using LCM-isolated Purkinje cell RNA from an independent set of four Mwk mice and their wild-type littermates at P18. We detected similar changes in the expression levels of these genes by qRT-PCR as by microarray (Supplementary Material, Fig. S2 and Table S2). We also investigated the expression levels of these genes at earlier timepoints to determine how early we could detect changes in the mutant cerebellum. We found three of the selected genes (Ipo5, Opn3, Sv2c) upregulated in Mwk Purkinje cells as early as P11. Others were downregulated in the mutant Purkinje cells from P14 (Stk17b, Car2) (Supplementary Material, Table S2). Consistent with the expression changes observed by microarray and qRT-PCR, differences in mRNA expression levels between Mwk Purkinje cells and wild-type littermates were also detected by in situ hybridization (ISH) (Supplementary Material, Fig. S3). Together, these data show that significant gene transcription changes occur in Mwk mice early on in Purkinje cell development, concurrent with the observed dendritic and synaptic abnormalities.

Mwk Purkinje cells exhibit gene expression changes linked to lipid metabolism
To gain insights into the potential molecular pathways affected in Mwk Purkinje cells, we next analyzed the differentially expressed genes using Ingenuity® Pathway Analysis (IPA). This analysis revealed several biological pathways and functions that were significantly enriched in dysregulated genes of the Mwk cerebellum including cellular assembly and organization  Table S5).  Tables S6 and S7). We identified 27 differentially expressed genes that are associated with lipid metabolism, of which 10 were upregulated in Mwk Purkinje cells and 17 were downregulated (Table 1). We classified these genes further into sub-categories of lipid metabolism using a combination of pathway analysis tools. Consistent with the results described above, we found that a large number of affected genes encode proteins that are key enzymes in sphingolipid metabolism, particularly the synthesis of ceramide (Fig. 4C). Notably, we observed reciprocal expression changes in the enzymes sphingosine kinase 2 (SPHK2) and sphingosine-1phosphate phosphatase 1 (SGPP1) as well as changes in the expression of acid ceramidase (ASAH1), sphingomyelin synthase 1 (SGMS1) and sphingomyelin phosphodiesterase 1 (SMPD1), suggesting a coordinated deregulation of ceramide biosynthesis (Fig. 5A). To validate these gene expression changes at the protein level, we prepared cerebellar extracts from three-week-old Mwk mice and their wild-type littermates and subjected these to immunoblotting. Consistent with the observed microarray changes, we found the protein levels of ASAH1 and SPHK2 upregulated in the Mwk cerebellum (Fig. 5B). Conversely, SGPP1, SMPD1 and SGMS1 levels were reduced in the mutant cerebellum (Fig. 5B). Together, these enzyme expression changes point towards a reduction in the pool of available ceramide within the Mwk Purkinje cells. In addition, we observed expression changes in UDP-glucose ceramide glycosyltransferase (UCGC) and G M2 ganglioside activator protein (GM2A), two enzymes that are involved in the synthesis and degradation of complex glycosphingolipids that are linked to the ceramide pathway (Fig. 5A).
Having identified that the expression levels of enzymes involved in lipid metabolism are significantly changed in the Mwk mutant, we next investigated whether levels of lipids were changed as a consequence in the Mwk cerebellum. We carried out lipid analysis by mass spectrometry in the cerebella of three Mwk mice and their wild-type littermates at P18. Consistent with the expression changes observed in Mwk Purkinje cells, we identified alterations of lipid levels in the mutant cerebellum including a reduction in distinct ceramides (Fig. 6A). We also observed changes in other lipids in the Mwk cerebellum, including fatty acyls and sterol lipids (data not shown). Interestingly, changes in lipid metabolism were not only evident in our analysis of the Mwk cerebellum, but serum levels of free fatty acids, triglycerides and cholesterol have also been found to be significantly changed in the Mwk mice in the EuroPhenome mouse phenotyping project (Supplementary Material, Fig. S4) (21).
We next tested whether treatment of Mwk organotypic slice cultures with ceramide might alleviate the dendritic phenotype observed in the mutant Purkinje cells. Sphingolipids have previously been shown to be important for Purkinje cell dendritic differentiation (22,23). Moreover, treatment of organotypic cerebellar cultures with sphingomyelinase was demonstrated to enhance dendritic branching complexity of Purkinje cells (23). In mice sphingomyelinase is encoded by the gene Smpd1, the expression of which is downregulated in Mwk cerebellum ( Fig. 5A and B, Table 1). We prepared organotypic cerebellar slice cultures from Mwk mice and their wild-type littermates at P9 and treated these for 12 days in vitro with C6-ceramide or its inactive form dihydroceramide C6 as control. We found that treatment of the Mwk cerebellar slices with C6-ceramide significantly improved Purkinje cell dendritic outgrowth compared with control ( Fig. 6B and C). Together, these findings suggest that the Mwk gain-of-function mutation in TRPC3 results in aberrant calcium signaling and gene expression changes in Purkinje cells that lead to alterations in lipid homeostasis and impairments in dendritic growth.

Discussion
In this study, we describe early dendritic and synaptic impairments and altered calcium-activated signaling pathways during Purkinje cell development in the Mwk mouse, an ataxic mouse model in which we previously identified a novel gain-of-function mutation in the non-selective cation channel TRPC3 (8). Combining Purkinje cell-specific transcriptomic and cerebellar lipid analysis, this study provides the first mechanistic insights into the molecular pathways underlying the Mwk phenotype. Together, our findings support the model that activated TRPC3 leads to aberrant calcium signaling in Purkinje cells and subsequent gene Representative blots are shown for 3-week-old animals. Densitometry quantifications are shown (right panels). Statistical significance was determined by one-way ANOVA (n = 3). *P < 0.05; **P < 0.01. expression changes that result in altered lipid homeostasis and consequently abnormal dendritic development.
Purkinje cells form one of the most elaborate dendritic arbors among all neurons in the brain. Although the morphological stages of Purkinje cell dendritic development have extensively been characterized, relatively little is known about the molecular mechanisms that underlie these processes (24)(25)(26). Previous work has demonstrated that chronic activation of Purkinje cell post-synaptic signaling including mGlur1 (14), PKC (15,27) and TRPC3 (8) causes inhibition of dendritic growth. Consistent with this hypothesis, Purkinje cells in the lurcher mouse mutant, which harbors a gain-of-function mutation in the δ2 glutamate receptor (GluD2) resulting in chronic depolarization, also exhibit impaired dendritic growth (28). These signaling events may provide a negative feedback mechanism for limiting the size of the dendritic tree once Purkinje cells have established appropriate synaptic contacts with the parallel fibers in the developing cerebellum. However, the downstream effector pathways mediating this inhibitory effect remain elusive. Calcium signaling is thought to regulate dendritic growth and patterning through distinct mechanisms that are dependent on calcium entry (local/global), timing (acute/chronic), neuronal cell type and  Table 1.
Only categories with two or more genes are listed. Annotations were made using a combination of pathway analysis tools including IPA, Kyoto Encyclopedia of Genes and Genomes (KEGG) and SuperPaths (http://www.genecards.org/info.shtml#pathways_interactions). developmental stage (29,30). Our study points to specific calciumsensitive pathways that may restrict dendritic growth in Purkinje cells upon sustained TRPC3 activation and chronic Purkinje cell depolarization. Interestingly, we found that CaMKIV is hypophosphorylated in the Mwk cerebellum compared with wild-type controls (Fig. 3B). Our findings of ataxia, impaired dendritic Purkinje cell growth and reduced CaMKIV activity in Mwk mutants are consistent with the phenotype observed in CaMKIV-deficient mice. The latter also display cerebellar ataxia and reduced dendritic arborization of Purkinje cells (31). CaMKIV is predominantly nuclear and is thought to regulate gene transcription either directly or through co-factors (32). Notably, one of the candidate transcription factors regulated by CaMKIV is RORα, an orphan member of the nuclear hormone-receptor superfamily (33,34). RORα is abundantly expressed in Purkinje cells (35) and strongly implicated in Purkinje cell development (36) as well as maintenance (37). Moreover, RORα has been linked to gene expression regulation that controls lipid metabolism (36). We have performed an initial bioinformatics-based analysis to identify transcription factor enrichment among the genes that are differentially expressed in Mwk Purkinje cells, which identified RORα among others (Supplementary Material, Table S8). In view of our findings, these observations raise the interesting possibility that RORα might link TRPC3 and CaMKIV signaling to gene expression changes in the Mwk cerebellum. Interestingly, staggerer mice with reduced RORα expression display decreases in triglycerides and cholesterol levels similar to the changes we observed in the Mwk mice (38).
Deficiencies in sphingolipid metabolism have been implicated in a range of neurological diseases, many of which are associated with cerebellar ataxia. Examples include autosomalrecessive cerebellar spasticity, which was recently been found to be caused by mutations in the GBA2 gene (39,40) and many lysosomal storage disorders (41). Notably, mice that are deficient in enzymes that we found to be downregulated in the Mwk cerebellum also display cerebellar dysfunction including Scp2 knockout mice (42), Smpd1-deficient mice, a model of Niemann-Pick disease (43), and Gm2a knockout mice, a model of G M2 -gangliosidosis (44). Moreover, a deficiency of ceramide biosynthesis caused by mutations in the Lass1 gene encoding ceramide synthase 1 (CerS1) (Fig. 5A) has been shown to cause cerebellar ataxia and Purkinje cell degeneration (45). Interestingly, Purkinje cell arborization is also greatly reduced in the Lass1 mutants (45). The overlap in affected proteins and cerebellar phenotypes between the Mwk mouse and models of lysosomal storage diseases might point to converging pathological pathways underlying the cerebellar dysfunction in these models. It will be interesting to determine in future studies whether lipid-related pathways are also affected in other genetic forms of cerebellar ataxia and contribute to pathology more generally. This might be particularly relevant for those cerebellar ataxias that are linked to TRPC3 signaling such as SCA1, SCA14 and SCA15 (1) (reviewed in (7)).  Sphingosine kinase 2 1.5 Crh Corticotropin releasing hormone 1.5 Scp2 Sterol carrier protein 2 −5.7 Ptpla Protein tyrosine phosphatase-like ( proline instead of catalytic arginine), member a −4.9

Acat2
Acetyl-Coenzyme A acetyltransferase 2 −1.5 Although long implicated in neuronal differentiation, the precise role of sphingolipids in dendritic development is unclear. Membrane traffic is required to supply growing dendrites with building material and hence, abnormalities in the production of required lipids might impede this process. It is also conceivable that ceramide and other sphingolipids control dendritic differentiation through their role in the formation of lipid rafts and the organization and activation of signaling receptors (46). Finally, ceramides have been demonstrated to act as a second messenger in signal transduction where they participate in growth inhibition, induction of differentiation and programmed cell death (47).
Although this study is focused on altered lipid homeostasis in the Mwk cerebellum, changes in other pathways are likely to contribute to the Mwk phenotype. The observed gene expression changes in several genes associated with vesicle fusion and exocytosis including Sv2c (Supplementary Material, Figs S2 and S3) point towards alterations in the secretory pathway in the Mwk cerebellum, which has a known role in dendritic arborization in many neuronal cell types (48). Of interest are also gene changes that have been observed in other ataxic mouse models such as the downregulation of the Car2 gene (Supplementary Material, Fig. S2), encoding carbonic anhydrase II, which has also been reported in a mouse model of spinocerebellar ataxia type 7 (SCA7) (49). Moreover, protein levels of carbonic anhydrase II are decreased in the ataxic lurcher mouse mutant (50). Carbonic anhydrase II is highly and specifically expressed in developing Purkinje cells and has been postulated to be involved in differentiation and synaptogenesis of Purkinje cells (50). These findings highlight another potential converging pathological mechanism in the Mwk mouse and other models of cerebellar ataxia that will be interesting to explore further.
Our study identifies molecular alterations in the Mwk cerebellum that are common to other mouse models of cerebellar ataxia. Future studies will reveal whether these alterations pinpoint to converging pathways that could be targeted to modulate or monitor the pathogenesis of genetically different cerebellar diseases.

Animals and reagents
The generation of Mwk mice has been described elsewhere (8). The Mwk colony was maintained by backcrossing to C3H/HeH. All animal work was approved by the University of Oxford Ethics Panel and in accordance with UK Home Office regulations. All reagents were purchased from Sigma unless otherwise noted.

Organotypic slice cultures
Organotypic slice cultures were prepared as previously described (8). The medium was changed every 2-3 days. For experiments using P/Q block, 100 nM ω-agatoxin IVA (Bachem) and 1 μM ω-conotoxin MVIIC (Bachem) were added from day 2 onwards (11). For ceramide treatment, 50 μM of C6-ceramide (N-hexanoyl-sphingosine) or its inactive form dihydroceramide C6 were added from day 2 onwards. At days-in vitro (DIV) 5-12, slice cultures were fixed in 4% paraformaldehyde followed by brief methanol permeabilization. Slices were blocked in PBS containing 0.3% Triton-X and 3% normal goat serum for 1 h at room temperature and incubated with rabbit anti-calbindin D-28K (Swant; 1:5000) for two nights at 4°C, followed by incubation with Alexa Fluor 594 goat anti-rabbit antibody (Invitrogen; 1:2000). For the quantification of Purkinje cell dendritic arbors, the size of the dendritic area was measured by tracing the outline of the dendritic tree using Axiovision 4.3 software (Zeiss) in a blinded manner. Cells were acquired from three independent experiments. Data were analyzed using GraphPad Prism. All data are represented as mean ± SEM. Statistical significance was assessed by one-way ANOVA followed by Bonferroni's multiple comparison test (*P < 0.05; **P < 0.01; ***P < 0.001).

Biochemical analysis
Three-week-old mice were sacrificed and cerebella were dissected out and kept on ice. Protein extracts were prepared by 10-s sonication in an RIPA buffer (50 m Tris pH 8, 150 m sodium chloride, 1% Nonidet P40, 0.5% sodium deoxycholate, 0.1% sodium dodecyl sulfate (SDS), 100 μM sodium vanadate, 10 m sodium fluoride, 1× Complete Protease Inhibitors The intensity of the bands was quantified using ImageJ software (NIH). Data were normalized to actin levels and wild-type bands and analyzed using GraphPad Prism. All data are represented as mean ± SEM. Statistical significance was assessed by one-way ANOVA (*P < 0.05; **P < 0.01; ***P < 0.001).

LCM and RNA preparation
Freshly dissected cerebella from Mwk mice and wild-type littermates were embedded in OCT compound (TissueTek) and stored at −80°C until sectioning. 10-μm sagittal midline sections were cut and mounted onto PEN membrane-coated slides (Carl Zeiss), before fixation in a graded series of ice-cold ethanol solutions (95, 75 and 50%) (containing ProtectRNATM RNase inhibitor (Sigma)) for 30 s each, and staining for 1 min with 1% cresyl violet acetate. Sections were rinsed in 50% ethanol solution and dehydrated in 75, 95 and 100% ethanol solutions for 30 s each, and dehydrated twice in xylene for 30 s. Slides were left to air-dry briefly before proceeding with the LCM. Purkinje cells were microdissected from all cerebellar lobes of the sections using the P.A.L.M.
Micro Beam system in combination with Robo software (Zeiss) and directly catapulted into a tube cap containing RLT lysis buffer with β-mercaptoethanol (RNeasy Micro kit (Qiagen)). One thousand Purkinje cells were collected for microarray analysis and 300-500 cells were collected for qRT-PCR. Total RNA was extracted using the RNeasy Micro Kit (Qiagen) and RNA quality was assessed on a 2100 BioAnalyzer using the RNA 6000 Pico Assay (Agilent Technologies). All samples used for microarray and qRT-PCR analysis had an RNA Integrity Number (RIN) of ≥6.

Microarray analysis
Fragmented and labeled cDNA was generated from 1 ng of total RNA using the Ovation Pico WTA System, the WT-Ovation Exon Module and the Encore Biotin Module (all NuGEN) according to manufacturer's instructions and hybridized to the Affymetrix Mouse Gene 1.0 ST Array. Chips were processed on an Affymetrix GeneChip Fluidics Station 450 and Scanner 3000.
Microarray data were normalized by Probe Logarithmic Intensity Error (PLIER) using GeneSpring GX11.0 (Agilent), and differentially expressed genes (fold change difference ≥1.5) were identified using a moderated t-test with a Benjamini and Hochberg multiple testing correction of 0.05. Differentially expressed genes were hierarchically clustered. Gene interaction networks and canonical pathways were analyzed using IPA (www. ingenuity.com). Further functional annotation was carried out using a combination of IPA, Kyoto Encyclopedia of Genes and Genomes (KEGG) and SuperPaths (http://www.genecards.org/ info.shtml#pathways_interactions). Microarray data are available in the ArrayExpress database (www.ebi.ac.uk/arrayexpress) under accession number E-MTAB-3517.

Supplementary Material
Supplementary Material is available at HMG online.