Generation of glutamate receptor δ2 (Grid2)-Cre-dependent CTCF-deficient mice

The experimental procedures for animals were in accordance with the guide for the care and use of laboratory animals of the science council of Japan and were approved by the animal experiment committee of Osaka or Tokushima University. All mice were maintained under specific pathogen-free conditions. Ctcf-floxed mice, in which loxP sites flank exons 3–12, were described previously by Heath et al. [21]. Mice conditionally lacking CTCF (CTCF-cKO) in PCs were generated by breeding Ctcf-floxed mice with Grid2-Cre mice. Grid2-Cre mice were generated using a knock-in strategy, and Cre recombinase was predominantly expressed in PCs and in molecular layer interneurons at lower levels [50]. Ctcf;Grid2-Cre (+/fl; +/+) and Ctcf;Grid2-Cre (fl/fl; +/+) mice were used as controls.

Histological analysis

Mice were deeply anaesthetized and transcardially perfused with phosphate-buffered saline (PBS) followed by 4% paraformaldehyde (PFA) in 0.1 M phosphate buffer (PB). Then, the brains were removed and postfixed in the same fixative for 2 h at 4 °C and subsequently cryoprotected in 30% sucrose in PB. Frozen sections, either 10 or 50 μm thick, were prepared on a microtome. Sections were washed with PBS and incubated for 1 h at room temperature in blocking buffer: 20% Block Ace (KAC Co., Ltd.), 5% normal goat serum (NGS), 0.1% Triton X-100, 0.1% azide in PBS. Then, the sections were incubated overnight with primary antibody in antibody dilution buffer (5% Block Ace, 5% NGS, 0.1% Triton X-100, 0.1% azide in PBS) at 4 °C. Sections were then washed with 0.1% Triton X-100 in PBS and incubated for 1 h with secondary antibody in antibody dilution buffer at room temperature. The antibodies used were as follows: anti-calbindin (1:500; Sigma-Aldrich), anti-CTCF (1:1000; Cell Signaling Technology), anti-VGluT2 (1:10,000; Millipore), anti-active caspase-3 (1:500; Cell Signaling Technology), anti-calnexin (1:500; Enzo), anti-KDEL (1:2000; MBL), and anti-IP3R (1:500; abcam). For haematoxylin and eosin (HE) staining, sections were stained with Mayer’s haematoxylin and eosin Y (Muto Pure Chemicals, Tokyo, Japan).

In situ hybridization analysis

In situ hybridization was performed as described [24]. In brief, mice were deeply anaesthetized, and their brains were removed and embedded in OCT compound (Sakura Finetek, CA, USA) as quickly as possible and then frozen in isopentane cooled with liquid nitrogen. The frozen tissue was cut into 10-μm sections on a cryostat (CM1850, Leica Microsystems). Digoxigenin (DIG)-labelled RNA probes were synthesized from cDNA clones using DIG RNA Labelling Mix (Roche). In situ hybridization RNA probes were as follows: the fragments corresponding to + 1519 (5′-TTTGCTGATCAGACTGGCGT-3′) to + 2303 (5′-ACGGTTGTTCAGTCCCATCC-3′) of mouse Galc cDNA, + 33 (5′-GCAGGAAGATACGGTGCTGT-3′) to + 752 (5′-TCTCCCGACTGTCTGGATGA-3′) of mouse Smn1 cDNA, + 399 (5′-CCCTGAAACCAACAAAAACA-3′) to + 1050 (5′-ATGATAATGCACCAGAGGTC-3′) of mouse Pcdhα12 cDNA and + 318 (5′-CGTGAGCTTTAACATCTTGA-3′) to + 1291 (5′-GAAGGCCACAGATGGTGGAA-3′) of mouse Pcdh γA7 cDNA.

Behavioural assays

Gait analysis was carried out using a footprint test with a runway measuring 30 cm in length and 6 cm in width with walls that were 50 cm high [2]. Habituation was performed by allowing the mice to walk freely along the narrow runway before their footprints were collected. To collect the footprints, a new sheet of white Japanese calligraphy paper was placed on the floor of the runway. Just before the footprint test, the feet of the individual mouse were painted with nontoxic paint (orange or black). Then, the mice were gently held at the entrance of the runway and released. Footprints were analysed for print separation (distance between corresponding fore- and hindpaw prints), front and hind width (distance between left and right prints of the fore- and hindpaws), front- and hind-stride length (distance between each footprint), front and hind ratio (ratio of width/length of each footprint).

The open-field test was performed using IMAGE OF4 (O’Hara & Co., Ltd.), which consists of a white plastic square chamber (50 × 50 × 40 [H] cm) with a CCD camera on the ceiling. Locomotor activity was automatically measured by IMAGE OF4 software (O’Hara & Co., Ltd.) for 10 min at P50, and we analysed total walking distance.

For the walking initiation test, each mouse was placed in the middle of a square outlined by white plastic tape (21 × 21 cm) on the smooth black surface of a large tabletop. We measured the time it took each mouse to leave the square (to place all four paws outside of the tape). When the mouse did not leave the square within 60 s, we stopped the test.

For the beam test, we used a beam apparatus with a flat surface (width, 28 mm; length, 700 mm) resting 20 cm above the tabletop on two poles. A box with nesting material from the home cage was placed at one end of the beam to lure the mouse, which was placed at the opposite end of the beam, to the finish point. Before the test, each mouse was allowed to cross the beam two times. During the test, we measured the time it took each mouse to go from the start to the finish point. Two trials were averaged for each mouse. When the mouse did not reach the finish point within 60 s or fell from the beam, the time was set to 60 s.

The platform test was performed as described [14], with some modifications. In brief, each mouse was timed for how long it remained on an elevated, square platform (3 × 3 cm, 1 cm thick) with rounded edges. When the mouse remained on the platform for the entire test trial or fell from the platform, the time was calculated as a maximum score of 60 s.

For the pole test, a mouse was placed with its head upwards on top of a vertical metal rod (diameter, 8 mm; height, 70 cm), the surface of which had been covered by medical tape. We then measured the time the mouse took to descend to the floor, with a maximum duration of 120 s. If a mouse fell from the pole before reaching the floor, it was given the maximum score of 120 s for that trial. Two trial scores were averaged for each mouse.

The screen test was performed as described [14], with some modifications. In brief, each mouse was placed on top of an elevated (30 cm above the floor) wire mesh grid. Then, the screen was inverted by 180°, and we measured how long the mouse was able to remain upside down on the screen. If a mouse did not fall from the screen for the entire duration, it was given the maximum score of 90 s.

The rotarod test was performed as described [14], with some modifications. In brief, the test involved two conditions: a rotating rod with a constant speed (5 rpm for 60 s) and a rod that had an accelerating rotational speed (0–3 rpm for 2 min, with the rod accelerating during the first 60 s) using RRAC-3002 (O’Hara & Co., Ltd.). Mice at P56–62 and P175–182 were used. We conducted three sessions, every 3 days. Each session included two trials at a constant speed and two trials at an accelerating rotational speed with 10 min of rest between each trial.

In vitro whole-cell patch-clamp recordings

Mice at P47–59 were deeply anesthetized by CO2 or isoflurane inhalation and were decapitated. The brains were quickly removed, and parasagittal slices with a thickness of 250 μm were prepared from the cerebellar vermis with a Vibratome slicer (VT1200S, Leica) in chilled normal artificial cerebrospinal fluid (ACSF) containing 125 mM NaCl, 2.5 mM KCl, 2 mM CaCl2, 1 mM MgSO4, 1.25 mM NaH2PO4, 26 mM NaHCO3, and 20 mM glucose, bubbled with 95% O2 and 5% CO2. Cerebellar slices were kept at 25 °C in normal ACSF. Whole-cell recordings were made from visually identified PCs using an upright microscope (BX50WI, Olympus). Normal ACSF supplemented with bicuculline (10 μM, Tocris) or picrotoxin (100 μM, Tocris) was used as a bath solution during recordings to block GABAergic transmission. The pipette solution consisted of 60 mM CsCl, 10 mM Cs D-gluconate, 20 mM TEA-Cl, 20 mM BAPTA, 4 mM MgCl2, 4 mM ATP, 0.4 mM GTP, and 30 mM HEPES (pH 7.3, adjusted with CsOH). The pipette solution also contained 0.5% neurobiotin (Vector labs) for post hoc visualization of the recorded cells. A glass pipette for stimulating climbing fibres was filled with normal ACSF. Its position was moved systematically around each PC soma. The number of climbing fibres innervating the recorded cell was estimated as the number of discrete EPSC steps during a gradually increasing stimulus intensity (0–90 V) at each stimulus location [35, 36]. To calculate the disparity ratio, the amplitudes of each CF-EPSCs in a given multiply innervated PC were measured and numbered in amplitudes (A1, A2, …, AN, N ≧ 2, AN is the largest one). Then, the disparity ratio was obtained from the following formula: Disparity ratio = (A1/AN + A2/AN + ⋯ + AN−1/AN)/(N − 1). If all CFs in a given PC evoke EPSCs with the same amplitude, the disparity ratio will be 1. Conversely, if the differences between the largest (AN) and other smaller CF-EPSC amplitudes are large, the disparity ratio will be small [18]. All experiments were examined and analysed by investigators who were blind to the mouse genotypes. All data were recorded at 32 °C with an EPC10 patch-clamp amplifier with Patch Master software (HEKA Elektronik). Offline data analyses were performed using Fit Master software (HEKA Elektronik), Excel (Microsoft), and IGOR Pro (Wave Metrics). Statistical analyses were conducted with SigmaPlot 12.1 or 12.5 (Systat Software), and differences between the two samples were considered statistically significant if the p-value was < 0.05.

Morphometric analysis of PC dendrites

Neurobiotin was injected into PCs after electrophysiological analysis, after which the slices were fixed with 4% PFA. After incubation in blocking buffer (5% NGS, 0.5% Triton X-100, 0.1% azide) for 1 h at room temperature, neurobiotin-injected PC dendrites were stained with Alexa 488-conjugated streptavidin (1:1000; Invitrogen) in antibody dilution buffer (5% Block Ace, 5% NGS, 0.1% Triton X-100, 0.1% azide in PBS) overnight at 4 °C. Three-dimensional image stacks were acquired on a confocal microscope (Olympus, FV1000), and PC dendrites were manually traced using Neurolucida-8 software (MicroBrightField). The total dendrite length, total dendrite area, and number of dendritic branches were calculated by Neurolucida Explorer (MicroBrightField). Crossing branch points were calculated from three-dimensional reconstructions as described [10]. Grids with individual squares of 10 × 10 μm were overlaid on top of the z projection of the confocal stacks, and randomly selected areas corresponding to a minimum of 20% of all grid squares covering the entire dendritic area of a PC were analysed.

Analysis of CF synapses in PCs

For the analysis of CF synapses, images were obtained with a confocal microscope (Olympus, FV1000). The molecular layer, which is where PC dendrites reside, was divided into five equal bins, each with a width of 100 μm along the dorsoventral axis, for quantification. The number of VGluT2 puncta in each bin at P60 was counted in sagittal sections. Three control mice and three CTCF-cKO mice were used for quantification experiments. Counts were performed on more than three different regions from each mouse (total 42,019 μm2 for control mice, 29,914 μm2 for CTCF-cKO mice).

Electron microscopy analysis

Mice at P60 were deeply anaesthetized and perfused transcardially with 25 mM PBS, followed by perfusion with 1.6% PFA and 3% glutaraldehyde in 0.1 M PB (pH 7.3–7.4) for 12 min. Coronal slices (50 μm thick) were cut using a vibrating microslicer (VT-1000; Leica) in 0.1 M PB. After being washed in 0.1 M PB several times, sections were treated with 1% OsO4 in 0.1 M PB for 40 min, stained en bloc with 1% uranyl acetate for 40 min, dehydrated with ethanol, and flat-embedded in Durcupan resin (Fluka) [46]. Serial ultrathin sections were prepared at a thickness of 70 nm (Ultracut S; Leica). Images were captured using a transmission electron microscope (TEM1010; JEOL).

SBF-SEM analysis

The mice were deeply anaesthetized and transcardially perfused with 25 mM PBS followed by 4% PFA with 0.5% glutaraldehyde in 0.1 M PB (pH 7.3–7.4). Then, the cerebellum was removed and postfixed in the same fixative at 4 °C. The brains were sectioned at a thickness of 50 µm on a Vibratome slicer. The slices were postfixed in 4% PFA and 1% glutaraldehyde in 0.1 M cacodylate buffer (pH 7.4) at 4 °C overnight. Slices were cut into small pieces and then treated with 2% OsO4 (Nisshin EM, Tokyo, Japan) in 0.1 M cacodylate buffer containing 1.5% K4[Fe(CN)6] (Nacalai Tesque, Kyoto, Japan), washed four times with cacodylate buffer, and incubated with 1% thiocarbohydrazide (Sigma-Aldrich) for 20 min and then with 2% OsO4 for 30 min at room temperature. The slices were then treated with 2% uranyl acetate at 4 °C overnight and stained with Walton's lead aspartate at 60 °C for 30 min. The slices were dehydrated by passing the tissue through an ethanol series (60%, 80%, 90%, 95%, and 100% ethanol) at 4 °C; infiltrated sequentially with acetone dehydrated with a molecular sieve, a 1:1 (v/v) mixture of resin and acetone, and 100% resin; and then embedded in Epon 812 with carbon (Ketjen black) [38]. The specimen-embedding resin was polymerized at 40 °C for 6 h, 50 °C for 12 h, 60 °C for 24 h, and, finally, 70 °C for 2 days. After trimming to the region of interest, the samples were imaged via single electron beams with a Sigma or Merlin field emission-type scanning electron microscope (Carl Zeiss, Munich, Germany) equipped with the 3View system and a backscattered electron detector (Gatan, Pleasanton, CA).

Statistical analysis

Statistical analyses were performed using GraphPad Prism 7 software (GraphPad Software Inc., San Diego, CA). Comparisons of two samples were performed by unpaired Student’s t-test, with Welch’s correction for unequal variances when appropriate. Multiple comparisons were made by two-way analysis of variance (ANOVA) with a Bonferroni post hoc test. The sample size is defined in the figure legends and is also shown as individual dots in the bar graphs.