Animals were genotyped by PCR analysis on genomic DNA extracted from ear notch biopsies using ChargeSwitch gDNA tissue kit (Invitrogen, Carlsbad, CA). A common upstream primer (5′- gagcgagaagccgtgcagaag-3′) and primers specific for the wild-type allele (5′- aggatgatggtccacactagcc-3′) and the PGK-neo cassette in the mutant allele (5′-ctaccggtggatgtggaatg-3′) were used for amplification. The 710-bp (from wild-type allele) and 562-bp (from targeted allele) PCR products were resolved by electrophoresis on a 1% agarose gel. β-III spectrin deficient mice were fully viable and generally born at a ratio consistent with Mendelian inheritance (1:2:1). Litters comprised of 41 WT, 79 β-III^+/− & 44 β-III^−/− animals.

Whole cerebella were homogenized in 400 μl of ice-cold homogenization buffer [20 mM HEPES, pH7.4, 1 mM EDTA, 1 mM phenylmethylsulfonyl fluoride and protease inhibitor cocktail (Calbiochem, San Diego, CA)] with a Teflon-glass homogenizer. Protein concentrations were determined using Coomassie-Plus Reagent and bovine serum albumin as standard (Pierce, Rockford, IL). Protein samples were resolved by denaturing SDS-polyacrylamide gel electrophoresis and transferred to nitrocellulose membranes (Amersham Pharmacia). The membranes were blocked for 1 h at room temperature with 5% wt/vol non-fat dry milk in Tris-buffered saline/Tween 20 [TBS/T, 20 mM Tris, 17 mM NaCl (pH 7.6) with 0.1% vol/vol Tween-20]. Blots were incubated overnight at 4°C with either rabbit anti-βIII spectrin, -EAAT4, -GLAST (1:200), -GLT1 (1:4,000), -GluR1 (1:1000; AbCam, Cambridge, MA), or mouse anti-actin, -calbindin (1:1,600; Sigma, St. Louis, MO) in blocking buffer. After washing with TBS/T the blots were incubated for 1 h at room temperature with either HRP-conjugated donkey anti-rabbit IgG, or HRP-conjugated sheep anti-mouse IgG (1:4,000; Amersham Pharmacia). Immunoreactive proteins were visualized with enhanced chemiluminescence (ECL, Santa Cruz Insight Biotechnology, Wembley, UK).

Brains were removed and immersion-fixed with 4% paraformaldehyde in 0.1 M sodium phosphate buffer, pH 7.4 overnight at 4°C and cryoprotected by immersion in 0.1 M sodium phosphate buffer (pH 7.4) containing 30% sucrose. Tissue was quick-frozen on dry-ice, then 15 μm-thick cerebellar sections cut and fixed onto microscope slides coated with poly-L-lysine. After air drying for 30 min sections were either stained for Nissl with cresyl violet (0.25%) or immunostained. All quantification was carried out in a blinded manner on Nissl stained sections by counting the number of Purkinje cells along a 1 mm linear length in folia II, III, IV, VI and VIII (these folia were found to be the most consistent in shape between animals) and the counts averaged for each animal. The thickness of molecular layer was measured in each animal at the same three points within each of the five chosen folia and the fifteen measurements averaged. For deep cerebellar nuclei (DCN) counts four areas (150 μm² each) were measured and the counts averaged for each animal. For immunostaining sections were incubated for 1 h with blocking solution [5% normal goat serum with 0.4% Triton X-100 in PBS]. Rabbit anti-β-III spectrin, anti-EAAT4 or mouse anti-calbindin primary antibody (1:50) [2% normal goat serum/0.1% Triton X-100 in PBS] was applied for 1 h at room temperature. Sections were washed three times in PBS before applying goat anti-rabbit fluorescein isothiocyanate (FITC)-conjugated secondary antibody (Cappel, Aurora, OH) or goat anti-mouse Cy-3-conjugated secondary antibody (Jackson ImmunoResearch, West Grove, PA) for 1 h at room temperature followed by three rinses in PBS and coverslipping with Vectashield (Vector Laboratories, Burlingame, CA). Mouse anti-calbindin primary antibody (1:50) and anti-Neu-N antibody (1:100; Chemicon, Temecula, CA) was applied overnight at 4°C and biotinylated universal horse anti-mouse/rabbit IgG, diluted at 1:200 in PBS, was applied for 1 h at room temperature. Detection of biotinylated secondary antibody was carried out using avidin-biotin complex (ABC) method with diaminobenzidine tetrahydrochloride (DAB) and H₂O₂ used as peroxidase substrate (Vector Laboratories). Images were captured with an Olympus IX70 fluorescence microscope using Openlab software (Improvision) or a Zeiss inverted LSM510 confocal laser scanning microscope.

Brains were dissected out and immersion-fixed overnight (4°C) with a mixture of 2% paraformaldehyde and 2% glutaraldehyde in 0.1 M sodium cacodylate buffer, pH 7.3. Specimens were post-fixed in 1% osmium tetroxide in 0.1M sodium cacodylate for 45 minutes, washed three times in 0.1M sodium cacodylate buffer and dehydrated in 50%, 70%, 90% and 100% normal grade acetones for 10 minutes each, followed by two 10-minute changes in analar acetone. Samples were embedded in araldite resin and 1 μm-thick sections cut, stained with toluidine blue and viewed in a light microscope to select suitable areas for investigation. Ultrathin sections, 60 nm-thick were cut from selected areas, stained with uranyl acetate and lead citrate and viewed in a Phillips CM120 transmission electron microscope (FEI UK Ltd). Morphological criteria used were irregular cell body, dark cytoplasm and shrinkage. Images were taken on a Gatan Orius CCD camera (Gatan, UK).

Total RNA was extracted from mouse cerebellum using RNeasy Mini kit (Qiagen, Valencia, CA) and RT-PCRs carried out using One-Step RT-PCR kit (Qiagen) according to manufacturer’s instructions. Primers for RT-PCR reactions located in exon 1, 7, 34 and 36 of β-III spectrin were: F1 5′-atgagcagcactctgtcacccact-3′; R7 5′-gccaattcttttgccttccacagc-3′; F34 5′-ggcccagggaagtgtggcctt-3′; R36 5′cttaaagaagctgaatcgtttttctgc-3′. Amplification of the ubiquitously expressed elongation factor alpha was used to control for RNA levels (Stratagene, La Jolla, CA).

β-III spectrin cDNA lacking exons 2-6 was subcloned into the Not I site of pCDNA3.1 (Invitrogen) and a Myc-tagged pRK5 vector. For microscopic observation, HEK293 and Neuro2A cells were plated onto coverslips coated with poly-L-lysine in 35-mm dishes and transfected with 1 μg of each DNA construct using Fugene reagent (Roche, Indianapolis, IN) in accordance with the manufacturer’s instructions. Anti-c-myc (Ab-1; Calbiochem, San Diego, CA) and TRITC-conjugated goat anti-mouse IgG (SouthernBiotech, Birmingham, AL) were used for immunofluorescence. For cell homogenates and biotinylation assays HEK and HEK-rEAAT4 cells, respectively were plated onto 35-mm dishes and transfected with 4 μg of each DNA using Lipofectamine 2000 (Invitrogen) according to manufacturer’s instructions.

Footprint patterns were analyzed using a runway (80 cm by 10.5 cm wide) with white paper at the bottom. Hind paws of animals were dipped in non-toxic, water-soluble black ink (Indian Ink, Winsor & Newton, Harrow, UK). Three consecutive strides were measured for each animal. Stride length measurements were taken from the base of two consecutive paw prints on the same side and the base width was measured as the distance between the centre of one paw print to the centre of the next print on the opposite side. The elevated beam test was performed using a narrow horizontal beam (2 cm wide, 80 cm long, held at a height of 30 cm from the table). The number of hind paw slips the animal made whilst traversing the beam were counted. For hanging wire test mice were placed on a wire cage lid, which was turned upside down, and the latency to fall measured. A 60-second cut-off time was used. In the rotarod test, the ability of mice to maintain balance on a stationary (maximum 60 s) or rotating (3, 5 & 10 rpm) 3-cm-diameter cylinder was assessed (TSE Rotarod, Bad Homburg, Germany) and the time a mouse remained on the accelerating cylinder recorded (maximum 120 s).

Each cerebellum was homogenized in 800 μl Tissue Buffer (5 mM Tris/320 mM sucrose pH 7.4) with Complete protease inhibitor cocktail, using Teflon-glass homogenizer. Each sample was split and pellets washed twice in ice-cold Tissue Buffer, before being resuspended in either Na⁺- containing Krebs buffer (120 mM NaCl, 25 mM NaHCO₃, 5 mM KCl, 2 mM CaCl₂, 1 mM KH₂PO₄, 1 mM MgSO₄, 10% glucose) or Na⁺-free Krebs (120 mM choline-Cl and 25 mM Tris-HCL pH 7.4 substituted for NaCl and NaHCO₃, respectively). Samples were then incubated with 5 μM ³H-glutamate for 10 min at 37°C and uptake stopped by placing back on ice. Pellets were washed twice with Wash Buffer (5 mM Tris/160 mM NaCl pH 7.4) and radioactivity measured using a scintillation counter. Na⁺-dependent uptake was determined by subtracting Na⁺-free counts.

Extracellular recordings were obtained from Purkinje cells in the vermis of the posterior lobe (lobules V and VI) in WT and β-III^−/− mice anaesthetised with intra-peritoneal injection of 1.5 g/kg urethane (solution at 12.5%). The head of the animal was immobilized in a stereotaxic frame and the cerebellum exposed by a craniotomy extending from the obex to the lamboidal ridge. The dura was opened along the midline and the brain covered with a mixture of paraffin oil and vaseline to prevent dessication. Glass microelectrodes filled with 0.9% NaCl were placed ±1 mm on either side of the midline, then lowered into the vermis using a hydraulic micromanipulator (Narashighe). Recordings of the spontaneous firing activity were made to a depth of 2 mm. Purkinje cells were identified by their characteristic firing of complex spikes followed by a pause in simple spike firing (Fig. 5). Spike activity was digitized at 10 KHz for a minimum of 3 minutes for each cell of interest, using Spike2 software (CED, Cambridge, UK). The mean firing rates of complex and simple spikes were quantified using analysis functions in the Spike2 software.

Cerebella were dissected out into ice-cold modified artificial cerebrospinal fluid (ACSF) containing (in mM): 60 NaCl, 118 sucrose, 26 NaHCO₃, 2.5 KCl, 11 glucose, 1.3 MgCl₂ and 1 NaH₂PO₄ at pH 7.4 when bubbled with 95% O₂:5% CO₂. The cerebellar vermis was glued to the vibratome cutting platform (Dosaka EM Co Ltd, Kyoto, Japan) with cyanoacrylate adhesive. 200 μm-thick sagital slices were cut and incubated for 30 mins at 30°C in standard ACSF composed of the following (in mM): 119 NaCl, 2.5 CaCl₂, 26 NaHCO₃, 2.5 KCl, 11 glucose, 1.3 MgCl₂ and 1 NaH₂PO₄ at pH 7.4 when bubbled with 95% O₂:5% CO₂. Slices were stored at room temperature until required for recording. Slices were transferred to a submerged recording chamber and superfused with standard ACSF (3-5 ml min⁻¹) at room temperature for voltage-clamp experiments and at 32°C ± 2°C for recording spontaneous action potentials. Purkinje cells were visualized with a 40× immersion objective and Normarski differential interference contrast (DIC) optics. Whole-cell recordings were obtained from Purkinje cells using thick-walled borosilicate glass pipettes pulled to 5-8 MΩ. For recording action potentials the internal solution contained (in mM): 125 KGluconate, 15 KCl, 10 HEPES, 5 EGTA, 2 MgCl₂, 0.4 NaGTP, 2 NaATP and 10 Na-phosphocreatine, adjusted to pH7.4 with KOH. For PF-EPSC measurements the internal solution contained (in mM): 108 CsMethanesulfonate, 9 NaCl, 9 HEPES, 1.8 EGTA, 1.8 MgCl₂, 0.4 NaGTP, 2 MgATP, 63 sucrose and 5 QX-314, adjusted to pH 7.4 with CsOH. Picrotoxin (50 μM) was added to the ACSF. PF-EPSCs were evoked by placing a patch-pipette filled with standard ACSF at the same position in the molecular layer and applying a range of stimuli (1.5-5.0 V, 200 μs duration). Pairs of PF-EPSCs (100 ms apart) were evoked at 0.033 Hz and a minimum of 3 PF-EPSCs were averaged under each condition. Series resistances were <15 MΩ and were compensated for by 40-60%. Membrane currents and voltages were filtered at 5 kHz and sampled at 10 kHz and 200 kHz for voltage-clamp and current clamp experiments, respectively. Data was acquired using pClamp 9 (Axon Instruments, Foster City, CA) and analysed using IGOR Pro (Wavemetrics, Lake Oswego, OR). The amplitudes and decay time-constants of PF-evoked EPSCs were measured using the ChanneLab analysis program (Synaptosoft Inc, Leonia, NJ).

Purkinje cells were isolated from P16–P20 mice using dissociation techniques modified from Raman and Bean (1997). Isolated Purkinje cells were visually identified by their large, pear-shaped soma (due to the stump of the apical dendrite). The control extracellular recording solution contained (in mM) 110 TEA-Cl; 25 NaCl; 2 BaCl₂; 0.3 CdCl₂; 10 HEPES and 10 glucose buffered to pH 7.4 with NaOH. Recordings were made at room temperature with borosilicate pipettes (3-5 MΩ) containing (in mM): 117 CsCl; 9 EGTA; 9 HEPES; 1.8 MgCl₂; 14 Na-phosphocreatine; 4 MgATP and 0.3 NaGTP, adjusted to pH 7.4 with CsOH. To isolate the TTX-sensitive Na⁺ current voltage protocols were repeated in the presence of 300 nM TTX and substracted from the control recordings.

Purkinje neurons provide the sole output from the cerebellar cortex, in the form of inhibitory inputs to the DCN, and so it is their firing rate that encodes the timing signals essential for motor planning, execution and coordination. They are one of a few neuronal populations that fire regularly in the absence of synaptic input, but integration of excitatory and inhibitory inputs modulates their output (Hausser and Clark, 1997). We therefore recorded the in vivo firing pattern of Purkinje cells in 12-week and 8-month old β-III^−/− and WT mice (Fig. 5). We found that the simple spike firing rate was much less at 8-months than 12-weeks in β-III^−/− mice compared to WT animals, indicating an age-related decrease in output from surviving Purkinje cells. There was no difference in the firing rate of complex spikes but some Purkinje cells from 8-month old animals appeared to fire only complex spikes with the number of such cells being greater in β-III^−/− mice (WT 1 out of 9; β-III^−/− 5 out of 15 cells). These cells were not included in the calculation of mean simple spike firing rate.

We found that in older β-III^−/− mice, when motor deficits are more pronounced, there is, in addition to the loss of EAAT4 seen in young β-III^−/− mice, a loss of the astroglial transporter GLAST and corresponding reductions in glutamate uptake. These results and the evidence of glutamate-mediated excitotoxicity in Purkinje cells of β-III^−/− mice from 6 months onwards suggest that it may be the delayed and progressive loss of GLAST activity that leads to Purkinje cell degeneration and the worsening ataxic phenotype. Therefore a non-cell-autonomous mechanism involving GLAST dysfunction might be a pathogenic mechanism common to several SCAs. A complete loss-of-function mutation in EAAT1 (GLAST), with a dominant negative effect on wild type protein, was recently identified in a child with episodic and progressive ataxia (Jen et al., 2005), and this is in accordance with the finding that a GLAST knockout mouse possesses motor deficits (Watase et al., 1998). Furthermore, Purkinje cell degeneration is also seen when expanded ataxin-7 is expressed solely in Bergmann glia, resulting in impaired glutamate uptake and loss of GLAST (Custer et al., 2006). Therefore, glutamate-mediated excitotoxicity, involving GLAST dysfunction, appears to play a critical role in Purkinje cell degeneration. Exactly how loss of β-III spectrin gives rise to this is not yet determined as β-III spectrin does not appear to be expressed in Bergmann glia (unpublished data). One possibility is that β-III spectrin, by orchestrating specialized microdomains in Purkinje cells, assembles a multi-protein transmembrane complex that maintains cell-to-cell contact with Bergmann glia and retains GLAST at the glial membrane. Alternatively given the role of spectrin in vesicular transport it may be that β-III spectrin is involved in the trafficking and secretion of trophic factors from Purkinje cells which modulate GLAST protein levels.