Abstract
The structural maintenance of neural circuits is critical for higher brain functions in adulthood. Although several molecules have been identified as regulators for spine maintenance in hippocampal and cortical neurons, it is poorly understood how Purkinje cell (PC) spines are maintained in the mature cerebellum. Here we show that the calcium channel type 1 inositol trisphosphate receptor (IP3R1) in PCs plays a crucial role in controlling the maintenance of parallel fiber (PF)–PC synaptic circuits in the mature cerebellum in vivo. Significantly, adult mice lacking IP3R1 specifically in PCs (L7-Cre;Itpr1flox/flox) showed dramatic increase in spine density and spine length of PCs, despite having normal spines during development. In addition, the abnormally rearranged PF–PC synaptic circuits in mature cerebellum caused unexpectedly severe ataxia in adult L7-Cre;Itpr1flox/flox mice. Our findings reveal a specific role for IP3R1 in PCs not only as an intracellular mediator of cerebellar synaptic plasticity induction, but also as a critical regulator of PF–PC synaptic circuit maintenance in the mature cerebellum in vivo; this mechanism may underlie motor coordination and learning in adults.
Introduction
The cerebellum is a brain region that plays an important role in motor coordination and learning. In the cerebellar cortex, the Purkinje cells (PCs) are the sole output; they integrate numerous synaptic inputs from granule cells (GCs) and inferior olive neurons (Ito, 2006). In the distal dendritic region of PCs, a high density of spines is innervated by parallel fiber (PF) axons of GCs, while in the proximal region, a few spines are innervated by climbing fiber (CF) axons of inferior olive neurons (Ito, 2002; Lee et al., 2005). Synapses between PF terminals and PC spines are sites for cerebellar synaptic plasticity—long-term depression (LTD) and long-term potentiation—which are involved in certain types of motor learning and are thought to be the primary site where information needed to execute coordinated movements is stored (Jörntell and Hansel, 2006). Therefore, precise control of spine maintenance on distal dendrites of mature PCs is essential to maintain cerebellar neural circuits and their behavioral functions, but the molecular mechanism underlying this phenomenon is largely unknown.
Inositol trisphosphate receptors (IP3Rs) are calcium (Ca2+) channels that are responsible for Ca2+ mobilization from intracellular sores in the endoplasmic reticulum (ER; Berridge et al., 2000). Three subtypes of this receptor have been identified, namely, IP3R1, IP3R2, and IP3R3 (Furuichi and Mikoshiba, 1995; Patel et al., 1999), with IP3R1 predominantly expressed in the CNS, especially in cerebellar PCs (Furuichi et al., 1993). IP3Rs have been shown to play crucial roles in several neuronal functions (Mikoshiba, 2011). In particular, IP3R1-null mice display seizure-like movements and ataxia (Matsumoto et al., 1996), impaired cerebellar LTD (Inoue et al., 1998), and abnormal PC dendritic morphology (Hisatsune et al., 2006). IP3R1-heterozygous mice show motor discoordination (Ogura et al., 2001), suggesting that IP3R1 plays an important role in cerebellar function in vivo. However, because of the broad expression of IP3R1 in the CNS, in addition to juvenile lethality of IP3R1-null mice (Furuichi et al., 1993; Matsumoto et al., 1996), it remains unclear how IP3R1 in PCs contributes to cerebellar function in adult mice.
To investigate the specific contribution of IP3R1 in PCs for cerebellar function in adult mice, we generated mice that specifically lacked IP3R1 in PCs (L7-Cre;Itpr1flox/flox). We found that the loss of IP3R1 in PCs is sufficient to cause severe ataxia, impaired cerebellar learning, and PF–PC LTD. Furthermore, we found that L7-Cre;Itpr1flox/flox mice displayed abnormalities in spine density and morphology, as well as in dendritic branching of mature PCs, despite showing normal PC morphology during development. Our findings suggest that IP3R1 in PCs is essential for motor coordination and learning by controlling the maintenance of structural and functional circuits in the mature cerebellum in vivo.
Materials and Methods
Mice.
Floxed Itpr1 (Itprflox) mice were generated using homologous recombination, which is a standard gene-targeting technique. The structure of the target vector is available in Figure 1A. The targeting construct was created from DNA cloned from a 129/SV mouse genomic DNA library. A neomycin-resistant (neo) gene flanked by 34 bp loxP sites, which was under the phosphogylcerol kinase promoter, was inserted into the intron between exon 2 and exon 3 of the Itpr1 gene. A 34 bp loxP sequence was also inserted downstream of exon 3, which contained the first methionine of Itpr1 gene. The DNA was then electroporated into 129/SvEv embryonic stem (ES) cells. G418-resistant ES cell clones were selected and screened by Southern blot analysis for homologous recombination with a ∼1.0 kb probe. The neo gene of the neo-positive ES clones was removed by electroporation of a Cre recombinase-expressing vector, and neo-deleted ES clones were screened by Southern blotting. Four clones were used for generating chimeras by microinjection into C57BL/6J blastocysts and implantation into pseudopregnant foster mothers. Chimeric male mice were mated to C57BL/6J females, and offspring carrying the floxed Itpr1 were selected by PCR using the following primer pairs: primer A, 5′-CTTCTACCTAATCCCAGCCAGGGAATC-3′; primer B, 5′-CTGGGTTAAGGAATCAAAGCAACAAG-3′; primer H, 5′-ATCAGTTTTGCCTTCTCTAGA-3′ for the 5′ loxP insertion; and primer E, 5′-TACATAGATCCAAAGAAGTGCCTCTG-3′, and primer F, 5′-AGGTTGAGTGATGACTGATTGGAGG-3′ for the 3′ loxP insertion in the gene. The mice were crossed with C57BL/6J at least 10 times. To generate L7-Cre;Itpr1flox/flox mice, homozygous Itpr1flox mice were crossed with mice heterozygous for Itpr1flox and for an L7-Cre transgene (Barski et al., 2000). To generate L7-Cre;Itpr1flox/flox;Pcp2-GFP mice, we used Pcp2-GFP mice obtained from Jackson Laboratory (Tomomura et al., 2001). Mice were bred in a pathogen-free environment, and all experiments were performed in accordance with the guidelines approved by the Animal Experiments Committee of RIKEN Brain Science Institute. Both female and male mice were included in the analysis.
Antibodies.
Rat monoclonal anti-IP3R1 (4C11, 18A10) antibodies were described previously (Maeda et al., 1990; Nakade et al., 1994). Mouse monoclonal and rabbit polyclonal anti-calbindin (Swant), mouse monoclonal anti-β-actin (clone AC-15; Sigma), mouse monoclonal anti-GFP (clone 1E4; MBL), rabbit polyclonal anti-VGluT1 (Synaptic Systems), and mouse monoclonal anti-VGluT2 (Millipore) antibodies were purchased.
Immunostaining and Nissl staining.
For immunohistochemistry of cerebellar sections, mice were deeply anesthetized with pentobarbital and transcardially perfused with saline and 0.1 m phosphate buffer (PB) containing 4.0% paraformaldehyde (PFA). The brains were dissected and postfixed in PFA at 4°C for 3 h, and then immersed in 30% sucrose in PB overnight at 4°C. Except for GFP and type 2 vesicular glutamate transporter (VGluT2) immunostaining, the brains were sectioned sagittally at 8.0 μm thickness using a cryostat (HM550; MICROM). The sections were blocked with 1.0% skim milk and 1.0% normal goat serum (Vector Laboratories) in PBS, and then incubated with the primary antibodies overnight at 4°C. After being washed with PBS, sections were incubated with Alexa Fluor 488-conjugated and Alexa Fluor 594-conjugated secondary antibodies (Invitrogen). Stained sections were mounted with Vectashield (Vector Laboratories). For GFP and VGluT2 immunostaining, the brains were sectioned sagittally at 30 μm thickness using a Vibratome tissue slicer (VT1000S; Leica Microsystems). PBS containing 0.3% Triton X-100 was used in GFP immunostaining. Other staining steps are described above. The primary antibodies used were 4C11 (1.0 μg/ml), monoclonal anti-calbindin (1:1000), polyclonal anti-calbindin (1:2000), monoclonal anti-GFP (1:500), polyclonal anti-VGluT1 (1:2000), and monoclonal anti-VGluT2 (1:2000) antibodies. Distal dendrites of PCs located in lobule VI where IP3R1-positive PCs could be detected even in adult L7-Cre;Itpr1flox/flox mice (Fig. 2C) were examined to assess the relationship between IP3R1 expression and spine density in L7-Cre;Itpr1flox/flox mice PCs (Fig. 3C,D).
For Nissl staining, sections were incubated with warmed cresyl violet solution [0.1% cresyl violet acetate (Sigma)/0.07% acetic acid]. Nissl-stained sections were differentiated in 80% ethanol, dehydrated in 99.5% ethanol, cleared in xylene, and then mounted with Permount (Fisher Scientific). All images were obtained using a Bio-zero BZ-8000 microscope (Keyence), with 4× [PlanApo, numerical aperture (NA) 0.2; Nikon] and 20× (PlanApo, NA 0.75; Nikon) objectives. Brain size was measured using ImageJ software (http://rsbweb.nih.gov/ij/).
Golgi staining.
Three- or 10-week-old brains were stained using modified Golgi–Cox impregnation of neurons following the manufacturer's protocol (FD NeuroTechnologies). The brains were sectioned sagittally at 100 μm thickness using a Vibratome tissue slicer. Golgi-impregnated sections were counterstained with Nissl staining. All images were obtained by using an ECLIPSE-80i phase-contrast microscope (Nikon) with a 40× objective (Plan Fluor, NA 0.75; Nikon) or 100× oil-immersion objective (Plan Fluor, NA 1.3; Nikon). The branch points of PCs were manually counted and were used as a measure for PC dendritic branching. Because the deletion of IP3R1in L7-Cre;Itpr1flox/flox mice was significant particularly in the anterior lobule (Fig. 2C), we assayed PCs that were located in the anterior lobule for analysis in this study (unless otherwise noted).
Immunoblotting.
Crude microsome fractions of mouse cerebella were obtained as described previously (Yoshikawa et al., 1999; Mizutani et al., 2008). Fractions were loaded on 6.5% SDS-PAGE and transferred onto a polyvinylidene difluoride membrane (Millipore). The membrane was blocked with 5.0% skim milk in PBST (PBS containing 0.05% Tween 20) and then incubated with the primary antibodies for 1 h at room temperature or overnight at 4°C. The primary antibodies used were 18A10 (1.0 μg/ml) and monoclonal anti-β-tubulin (1:5000) antibodies. After washing with PBST, the membrane was incubated with horseradish peroxidase-conjugated secondary antibody (1:5000; GE Healthcare) for 1 h at room temperature. The blot was developed using chemiluminescence reagents (Immobilon Western Chemiluminescent HRP Substrate; Millipore) and detected using an image analyzer (LAS-4000 mini; FUJIFILM).
Footprint analysis.
Mouse hindpaws were dipped in nontoxic water-based black paints. Mice were then allowed to walk down an enclosed runway lined with white paper. The footprint patterns were analyzed for the following two parameters: stride length and base width. Stride length was measured as the average distance between two consecutive paw prints on the same side. Three consecutive strides were measured for each paw per animal. Base width was measured as the average distance between the center of the left and right hind footprints. Four base widths were measured for each animal.
Fluorescent image acquisition and image analysis.
Fluorescent images were obtained using a fluorescent microscope (Biozero BZ-8000) with a 4× objective (PlanApo, NA 0.2; Nikon) and a confocal laser microscope (FV-1000; Olympus) with 20× (UPlanApo, NA 0.7; Olympus), 40× (UPlanApo, NA 1.0; Olympus), or 60× (PlanApoN, NA 1.42; Olympus) oil-immersion objective. For quantification of PC density and soma area, a series of images of sections at 1.0 μm intervals were obtained. Calbindin-positive cells were counted for PC density and used for the measurement of PC soma area. For spine analysis, a series of images of sections at 0.1–0.5 μm intervals were obtained. Spine density and morphology were quantified. For synapse analysis, the percentage of spines contact with PFs was determined by colocalization analysis. Each spine of GFP-positive PCs was classified as positive or negative depending on its colocalization with the presynaptic terminal signal of VGluT1. To assess the thickness of the molecular layer and the height of the CF extending along the PC dendrite, a series of images of sections at 1.0 μm intervals was obtained. Molecular layer thickness was measured as the distance from the top of the PC soma to the end of the dendrite near the pial surface. The VGluT2 signal was used to determine CF terminal deposition. For determining the extent of CF innervation along PC dendrites, the maximal CF extent from the top of the PC soma was calculated relative to the molecular layer thickness. All images were analyzed by MetaMorph software (Universal Imaging) and ImageJ software.
Electron microscopy.
Anesthetized mice were intracardially fixed with 2.0% paraformaldehyde and 2.5% glutaraldehyde in 0.1 m phosphate buffer, pH 7.4. Brain slices (1.0 mm) prepared with a microslicer were postfixed in 1.0% osmium tetroxide, dehydrated, and then embedded in epoxy resin. Ultrathin sections were cut with an ultramicrotome, stained with uranyl acetate and lead citrate, and observed under a transmission electron microscope (1200EX; JEOL). For quantitative analysis, images were analyzed using Photoshop 7.0 (Adobe Systems) and ImageJ software.
Electrophysiology.
For electrophysiological experiments, 9- to 10-week-old Itpr1flox/flox and L7-Cre;Itpr1flox/flox mice were used. Cerebellar slices were prepared and whole-cell patch-clamp recordings from PCs were made as described previously (Le et al., 2010). Under ether anesthesia, mice were decapitated and the cerebellum was excised. Sagittal cerebellar slices (300 μm thickness) were prepared from the vermis using a Vibratome tissue slicer (DTK-1000; Dosaka EM) in 1 μm tetrodotoxin (Wako Pure Chemical Industries Ltd.) that contained saline and were kept at room temperature (23–25°C) for at least 1 h in normal saline. The recording chamber was perfused with standard Krebs–Henseleit Ringer's solution containing (in mm) 118 NaCl, 4.7 KCl, 2.5 CaCl2, 25.0 NaHCO3, 1.18 KH2PO4, 1.19 MgSO4, and 11.0 glucose, which was added with 100 μm picrotoxin, equilibrated with 95% O2 and 5% CO2 gas, and warmed to maintain the temperature of the perfusion chamber at 30–31°C. Under an upright microscope, a whole-cell clamp pipette (resistance, 3–4 MΩ) was attached to the soma of a PC located in anterior lobule. The pipette contained (in mm) 134 K-gluconate, 6 KCl, 4 NaCl, 10 HEPES, 0.2 EGTA, 4 MgATP, 0.3 Tris-GTP, and 14 phosphocreatine, pH 7.25. A Multiclamp700A amplifier and pClamp 9 software (Molecular Devices) were used. In experiments on the input–output relationship of PF-EPSPs, the stimulation pipette was placed on the molecular layer at a distance of 150 μm from PC soma and 5 μm in depth from the slice surface. The stimulus intensity threshold (T) was determined when the peak of EPSP was positive (after subtracting the base noise). Then, an input–output curve was recorded by gradually increasing the stimulus intensity (1–5 T). The paired-pulse facilitation (PPF) ratio was determined as the rising slope of the second PF-EPSP divided by that of the first PF-EPSP at interpulse intervals of 50 ms. In the LTD sessions, we used two different LTD induction protocols. The first conjunctive stimulation consisting of single CF stimulation (0.1 ms in duration) and double-pulse PF stimulation (each 0.1 ms in duration) paired at 50 ms intervals timed in such a way that the second pulse fell at the same time as the CF stimulation. The second conjunctive stimulation consisted of a current pulse-induced membrane depolarization (200 ms, within 2 nA) and a double-shock PF stimulation timed in such a way that the first pulse fell 30 ms after the onset of each depolarizing pulse. The conjunctive stimulation was repeated at 1 Hz for 5 min (300 pulses) in each trial of LTD induction.
Vestibulo-ocular reflex and optokinetic response eye movements.
Eye movement was measured by the infra-red TV method as described previously (Katoh et al., 1998; Shutoh et al., 2006). Under isofluorane (Escain; Mylan) anesthesia and aseptic conditions, a platform for head fixation was made on the mouse cranial bone by synthetic resin (Super-Bond C&B; Sun Medical) and one 15-mm-long stainless bolt. Two days after surgery, a mouse was mounted on the turntable surrounded by a checked-pattern (check-size, 4°) screen (diameter, 60 cm; height, 60 cm), with the head fixed and the body loosely strained in a plastic cylinder. Two types of reflex eye movements, i.e., the horizontal vestibulo-ocular reflex (HVOR) and horizontal optokinetic response (HOKR), were measured. The HOKR was tested with the sinusoidal screen oscillation by 10–20° (peak to peak) at 0.11–0.33 Hz (maximum screen velocity, 3.5–10.5°/s) on the same plane in the light. The HVOR was tested by sinusoidal turntable oscillation (frequency, 0.11–0.5 Hz; peak-to-peak amplitude, 10°) on the plane in parallel with the bilateral horizontal semicircular canals in the dark. More than 10 cycles of the evoked eye movements that were free from artifacts due to blinks and saccades were averaged, and the mean amplitude and phase were calculated by a modified Fourier analysis (Jastreboff, 1979). The gain of the eye movement was defined as the ratio of the peak-to-peak amplitude of eye movements to that of the turntable or screen oscillation. The phase was defined as 0° when the peak of the eye movement was opposite to the peak of turntable oscillation in the HVOR and when the peak of the eye movement matched the screen oscillation in the HOKR. To induce the adaptation of HOKR, the mice were trained to view 600 cycles of 0.16 Hz at 15° screen oscillation continuously for 1 h. The HOKR gains were measured by 50 cycles of oscillation at the start and end of 1 h of training.
Statistical analysis.
All data were shown as the mean ± SEM, and statistical significance was defined as p < 0.05, as determined using the paired Student's t test, the Mann–Whitney U test, or a two-way ANOVA multiple measurements. Correlation analysis was performed using the nonparametric Spearman test.
Results
Severe cerebellar ataxia in adult L7-Cre;Itpr1flox/flox mice
To establish PC-specific IP3R1 knock-out mice (L7-Cre;Itpr1flox/flox), we generated Itpr1flox mice (Fig. 1A–C; see Materials and Methods) and crossed them with transgenic mice expressing Cre recombinase under control of the L7 promoter (Barski et al., 2000). L7-Cre;Itpr1flox/flox mice were born in the expected Mendelian ratios. In contrast to IP3R1-null mice that died around postnatal day 21 (P21; Matsumoto et al., 1996), L7-Cre;Itpr1flox/flox mice could survive to adulthood. They displayed cerebellar ataxia beginning ∼6 weeks after birth (6W), which worsened as they continued to develop; they exhibited severe ataxia, and their footprints were not visibly discernible after 8W (Fig. 1D). Quantitative analysis of footprint patterns for stride length (Fig. 1E) and base width (Fig. 1F) clearly revealed the age-dependent onset and worsening of ataxia in L7-Cre;Itpr1flox/flox mice. Although IP3R1-null mice presented seizure-like movements, such as opisthotonus (Matsumoto et al., 1996), L7-Cre;Itpr1flox/flox mice did not.
Severe cerebellar ataxia in adult L7-Cre;Itpr1flox/flox mice. A, Schematic diagram of the procedure for generating floxed Itpr1 ES clones. Top, Wild-type Itpr1 locus. Open box indicates the exon 3 of Itpr1 in which the initiation ATG codon is localized. Middle, Targeting vector in which a loxP site was inserted at both sides of exon 3; a phosphogylcerol kinase-neo gene cassette and a diphtheria toxin A marker (DT) was inserted at the 5′ end at the exon 3. Bottom, Floxed Itpr1 locus with deleted neo cassette gene. Restriction enzyme abbreviations are as follows: E, EcoRI; S, SalI; B, BamHI; X, XbaI. B, Southern blot of genome digested with BamHI using a probe indicated in A. C, Genotyping of floxed Itpr1 by PCR. Asterisk indicates a hybrid product of PCR product derived from wild-type and floxed alleles. D, Representative footprints of Itpr1flox/flox and L7-Cre;Itpr1flox/flox mice at 4W, 5W, 6W, 7W, 8W, and 9W. The walking patterns were recorded from mice hindpaws. E, F, Quantitative analysis of footprint patterns for stride length (E) and base width (F) from Itpr1flox/flox (white circles) and L7-Cre;Itpr1flox/flox (black circles) mice at 4W, 5W, 6W, 7W, 8W, and 9W (n = 3 mice; **p < 0.01, ***p < 0.0001, t test and Mann–Whitney U test). n.d., Not determined.
We examined IP3R1 protein levels in the brain of L7-Cre;Itpr1flox/flox mice at 10W using immunohistochemistry and found that the IP3R1 signal was abolished in PCs, but not in other brain regions such as the cerebral cortex and hippocampus (Fig. 2A). Immunoblot analyses also revealed that IP3R1 expression levels were decreased in the cerebellum, but not in the cerebral cortex, of L7-Cre;Itpr1flox/flox mice at 8W (Fig. 2B). The low level of IP3R1 protein in the cerebellum of L7-Cre;Itpr1flox/flox mice was detectable after long exposure times, which may be attributable to the residual expression of IP3R1 in a small number of PCs and to the presence of IP3R1 protein in GCs (Hisatsune et al., 2006).
Characterization of L7-Cre;Itpr1flox/flox mice. A, Immunohistochemical analysis of IP3R1 proteins in the cerebellum, cerebral cortex, and hippocampus. Sagittal sections prepared from control Itpr1flox/flox and L7-Cre;Itpr1flox/flox mice at 10W were stained with anti-IP3R1 antibody (4C11). Scale bars, 100 μm. B, Immunoblot analysis of IP3R1 proteins in the cerebral cortex and cerebellum. Membrane fractions prepared from cortex or cerebellum of Itpr1flox/flox and L7-Cre;Itpr1flox/flox mice at 8W were probed with the antibodies indicated. C, Immunohistochemical analysis of the expression level of IP3R1 protein in the cerebellar PCs at different ages. Sagittal sections prepared from Itpr1flox/flox and L7-Cre;Itpr1flox/flox mice at P15, P21, 6W, 8W, and 14W were stained with anti-IP3R1 antibody (4C11). Scale bar, 500 μm. D, Nissl staining of sagittal sections of the cerebellar vermis prepared from mice at 10W. The bottom panels show magnified views of the middle panels. Scale bars: top, 1 mm; middle, 500 μm; bottom, 50 μm. E, Immunostaining for calbindin of sagittal sections of the cerebellar vermis prepared from mice at 10W. Scale bar, 50 μm.
It has been reported that the L7-Cre transgenic mice we used in this study started to exhibit the Cre-mediated recombination in a few PCs from P6, and it progressed in most PCs until P21 (Barski et al., 2000). To determine the interaction between the time course of IP3R1 deletion in PCs and the onset of ataxia during development, we examined the expression level of IP3R1 in L7-Cre;Itpr1flox/flox mice at P15, P21, 6W, 8W, and 14W by immunohistochemistry. As shown in Figure 2C, the IP3R1 signal was strongly detected in most PCs at P15 and P21 (∼80%), but the number of IP3R1-positive PCs in L7-Cre;Itpr1flox/flox mice was dramatically decreased at 6W (<25%) and abolished by 8W (<5%; Fig. 2C). The recombination efficiency of Itpr1flox allele by Cre recombinase and/or protein turnover of IP3R1 might be responsible for the time lag of IP3R1 deletion to the expression of Cre recombinase. Since cerebellar ataxia began in L7-Cre;Itpr1flox/flox mice around 6W (Fig. 1D–F), the result suggests that the onset of ataxia is associated with the deletion of IP3R1 protein in PCs.
Abnormal maintenance of PC spine and dendritic morphology in adult L7-Cre;Itpr1flox/flox mice
Although cerebellar degeneration is associated with many forms of hereditary ataxia (Dueñas et al., 2006), Nissl staining revealed that L7-Cre;Itpr1flox/flox mice had normal overall cerebellar morphology, including the organization of the cortical layer at 10W (Fig. 2D). The cerebellar sizes in L7-Cre;Itpr1flox/flox mice were comparable to those of age-matched control Itpr1flox/flox mice at 10W, as assessed by normalization to longitudinal size in the cerebral cortex (cerebellum/cerebrum ratio: Itpr1flox/flox mice, 0.491 ± 0.029; L7-Cre;Itpr1flox/flox mice, 0.458 ± 0.027; n = 5 mice; p = 0.41, t test; Fig. 2D). Immunostaining for calbindin, which is a marker of PC, revealed that the molecular layer thickness of L7-Cre;Itpr1flox/flox mice was thinner than that of Itpr1flox/flox mice at 10W (Itpr1flox/flox, 148.6 ± 1.9 μm; L7-Cre;Itpr1flox/flox, 140.7 ± 1.5 μm; n = 3 mice; p < 0.001, Mann–Whitney U test; Fig. 2E). In addition, the PC soma area of L7-Cre;Itpr1flox/flox mice was smaller than those of Itpr1flox/flox mice (Itpr1flox/flox, 285.4 ± 5.2 μm2, n = 54 cells from three mice; L7-Cre;Itpr1flox/flox, 225.7 ± 4.7 μm2, n = 56 cells from three mice; p < 0.0001, t test; Fig. 2E). However, apparent degeneration of PCs was not observed in L7-Cre;Itpr1flox/flox mice at 10W; PC density was comparable between Itpr1flox/flox and L7-Cre;Itpr1flox/flox mice (Itpr1flox/flox, 3.82 ± 0.15 cells/100 μm; L7-Cre;Itpr1flox/flox, 4.00 ± 0.12 cells/100 μm; n = 3 mice; p = 0.37, t test; Fig. 2E).
To further investigate the precise morphology of individual PCs in adult L7-Cre;Itpr1flox/flox mice, neurons were visualized by Golgi staining. PCs from Itpr1flox/flox mice displayed well developed dendritic arbors, whereas PCs from L7-Cre;Itpr1flox/flox mice had ∼40% fewer dendritic branching points than those from Itpr1flox/flox mice at 10W (Itpr1flox/flox: 232.2 ± 10.8/cell, n = 11 neurons from two mice; L7-Cre;Itpr1flox/flox: 140.8 ± 6.5/cell, n = 13 neurons from two mice; p < 0.0001, t test; Fig. 3A). Interestingly, a closer examination revealed the striking increase in the spine density of PCs in the L7-Cre;Itpr1flox/flox mice (Fig. 3B). This increased spine density was apparent in the distal dendritic region of PCs, but not in the proximal region.
Abnormal PC dendritic and spine morphology from adult L7-Cre;Itpr1flox/flox mice. A, Golgi-stained PCs from Itpr1flox/flox and L7-Cre;Itpr1flox/flox mice at 10W. Low-magnification images (top) and magnified images (bottom) are shown. Scale bars: top, 100 μm; bottom, 50 μm. B, Higher-magnification views of distal dendritic (left) and proximal dendritic (right) regions of Golgi-stained PCs. Scale bars, 2 μm. C, Relationship between abnormal spine density and loss of IP3R1 in PCs of adult L7-Cre;Itpr1flox/flox mice. Cerebellar sagittal sections prepared from L7-Cre;Itpr1flox/flox mice at 10W were coimmunostained with anti-GFP (green) and anti-IP3R1 antibodies (4C11; red). Representative images of distal dendrites of IP3R1-positive (top panels) and IP3R1-negative (bottom panels) PCs located in lobule VI are shown. Scale bar, 2 μm. D, Plot of spine density along distal dendrites versus average fluorescence intensity for IP3R1 in distal dendrites in L7-Cre;Itpr1flox/flox;Pcp2-GFP mice. Each circle represents a single neuron. The black line is plotted by linear regression (n = 53 neurons from two mice; r = −0.743, p < 0.0001, Spearman correlation analysis). A.U., Arbitrary units. E, Spines on distal (left) and proximal (right) dendrites of PCs expressing GFP in Itpr1flox/flox;Pcp2-GFP and L7-Cre;Itpr1flox/flox;Pcp2-GFP mice at 10W. Scale bars, 2 μm.
To quantify the spine abnormalities of PCs in L7-Cre;Itpr1flox/flox mice, we crossed L7-Cre;Itpr1flox/flox mice with Pcp2-GFP mice expressing GFP in the PCs (Tomomura et al., 2001), resulting in L7-Cre;Itpr1flox/flox;Pcp2-GFP mice. Since a few PCs still expressed IP3R1 in the adult L7-Cre;Itpr1flox/flox mice, we immunostained the IP3R1 of PCs and assessed the relationship between IP3R1 expression and PC spine density. As shown in Figure 3C, an abnormal increase of spine density was clearly observed in the IP3R1-negative PCs, but not in the IP3R1-positive PCs of L7-Cre;Itpr1flox/flox;Pcp2-GFP mice at 10W (Fig. 3C), and we found a negative correlation between IP3R1 expression levels and spine densities (Fig. 3D). Spine densities along distal dendrites of the L7-Cre;Itpr1flox/flox;Pcp2-GFP PCs increased to 143% of those of control Itpr1flox/flox;Pcp2-GFP PCs (Itpr1flox/flox;Pcp2-GFP, 2.402 ± 0.096 spines/μm; L7-Cre;Itpr1flox/flox;Pcp2-GFP, 3.436 ± 0.109 spines/μm; n = 17 neurons from two mice; p < 0.0001, t test; Fig. 3E). In addition, the spine length of the L7-Cre;Itpr1flox/flox;Pcp2-GFP PCs appeared to be longer than that of the control PCs (Itpr1flox/flox;Pcp2-GFP, 0.898 ± 0.018 μm, n = 230 spines from two mice; L7-Cre;Itpr1flox/flox;Pcp2-GFP, 1.174 ± 0.027 μm, n = 231 spines from two mice; p < 0.0001, Mann–Whitney U test). In contrast, spine head diameter was not significantly different between L7-Cre;Itpr1flox/flox;Pcp2-GFP and control PCs (Itpr1flox/flox;Pcp2-GFP, 0.542 ± 0.011 μm, n = 230 spines from two mice; L7-Cre;Itpr1flox/flox;Pcp2-GFP, 0.544 ± 0.010 μm, n = 231 spines from two mice; p = 0.9, t test). The spine density was also not changed at the proximal dendrites in the L7-Cre;Itpr1flox/flox;Pcp2-GFP PCs.
PC dendritic outgrowth and formation of synapses with PFs and CFs have been shown to occur extensively during the first 3 postnatal weeks (Sotelo and Dusart, 2009; Tanaka, 2009). To examine whether abnormal dendritic development and spinogenesis during this developmental period was responsible for dendritic and spine abnormalities of adult L7-Cre;Itpr1flox/flox mice, we examined the morphology of PCs in L7-Cre;Itpr1flox/flox mice at 3W. Golgi staining revealed that the apparent PC dendritic morphology was comparable between control and L7-Cre;Itpr1flox/flox mice at 3W (branching point; Itpr1flox/flox, 211.3 ± 9.9/cell; L7-Cre;Itpr1flox/flox, 204.3 ± 13.1/cell; n = 10 neurons from two mice; p = 0.67, t test). We also could not observe any significant changes in spine density, length, and head diameter between L7-Cre;Itpr1flox/flox;Pcp2-GFP mice and age-matched control Itpr1flox/flox;Pcp2-GFP mice at 3W (spine density: Itpr1flox/flox;Pcp2-GFP, 2.488 ± 0.070 spines/μm, n = 20 neurons from two mice; L7-Cre;Itpr1flox/flox;Pcp2-GFP, 2.448 ± 0.087 spines/μm, n = 28 neurons from two mice; p = 0.74, t test; spine length: Itpr1flox/flox;Pcp2-GFP, 0.761 ± 0.012 μm, n = 402 spines from two mice; L7-Cre;Itpr1flox/flox;Pcp2-GFP, 0.733 ± 0.010 μm, n = 541 spines from two mice; p = 0.07, t test; spine head diameter: Itpr1flox/flox;Pcp2-GFP, 0.470 ± 0.008 μm, n = 402 spines from two mice; L7-Cre;Itpr1flox/flox;Pcp2-GFP, 0.485 ± 0.007 μm, n = 541 spines from two mice; p = 0.45, Mann–Whitney U test).
Because most PCs in L7-Cre;Itpr1flox/flox;Pcp2-GFP mice expressed IP3R1 at 3W (Fig. 2C), we further assessed the relationship between IP3R1 expression levels and spine density of PCs in the L7-Cre;Itpr1flox/flox;Pcp2-GFP mice. We could not observe a correlation between IP3R1 expression levels and spine densities (n = 26 neurons from two mice; r = 0.246, p = 0.22, Spearman correlation), indicating that PC IP3R1 was not essential for spine development. Together, these results suggest that IP3R1 in PCs plays a crucial role in the maintenance of spines in the mature cerebellum.
Formation of functional PF–PC synapses in adult L7-Cre;Itpr1flox/flox mice
We next investigated whether abnormal spines observed in adult L7-Cre;Itpr1flox/flox mice form synapses with PF terminals. To this end, we stained presynaptic PF terminals with an anti-VGluT1 antibody and assessed the contact of GFP-labeled spines with the VGluT1-labeled PF terminals in L7-Cre;Itpr1flox/flox;Pcp2-GFP mice at 10W (Fig. 4A). The mean percentage of spines that were in contact with PF terminals was comparable between L7-Cre;Itpr1flox/flox;Pcp2-GFP mice and control Itpr1flox/flox;Pcp2-GFP mice (Itpr1flox/flox;Pcp2-GFP, 86.9 ± 2.1%; L7-Cre;Itpr1flox/flox;Pcp2-GFP, 88.4 ± 2.0%; n = 12 neurons from three mice; p = 0.62, t test), suggesting that abnormally formed PC spines were able to make synapses with PF terminals in adult L7-Cre;Itpr1flox/flox mice.
Formation of PF–PC and CF–PC synapses in adult L7-Cre;Itpr1flox/flox mice. A, Immunohistochemical analysis of PF innervation on PC distal dendritic spines in Itpr1flox/flox;Pcp2-GFP and L7-Cre;Itpr1flox/flox;Pcp2-GFP mice at 10W. Sagittal cerebellar sections were coimmunostained with anti-GFP (green) and anti-VGluT1 antibodies (blue). Representative images of PC distal dendritic region are shown. Scale bar, 5 μm. B, Immunohistochemical analysis of CF innervation on PCs in Itpr1flox/flox and L7-Cre;Itpr1flox/flox mice at 10W. Sagittal cerebellar sections were coimmunostained with anti-calbindin (green) and anti-VGluT2 antibodies (red). Representative images are shown. Dotted lines indicate the pial surface. Scale bar, 20 μm. C, Ultrastructure of the molecular layer of the cerebellum in Itpr1flox/flox and L7-Cre;Itpr1flox/flox mice at 10W. Spines containing PSD (S), PF presynaptic terminals containing active zone and synaptic vesicles (P), and PC dendrites (d) are indicated. Scale bar, 1 μm.
To examine the distribution patterns of CF terminals—another type of excitatory projection to PCs—in the molecular layer of adult L7-Cre;Itpr1flox/flox mice, presynaptic CF terminals were immunostained with an anti-VGluT2 antibody. As shown in Figure 4B, there was no obvious difference in the distribution pattern of VGluT2-immunoreactive puncta between L7-Cre;Itpr1flox/flox and control Itpr1flox/flox mice at 10W (Fig. 4B). The extent of CF innervation along the PC dendrites, quantified as the relative height of VGluT2 puncta to the molecular layer thickness, was not significantly different between the L7-Cre;Itpr1flox/flox and the Itpr1flox/flox mice (Itpr1flox/flox, 0.832 ± 0.006; L7-Cre;Itpr1flox/flox, 0.831 ± 0.004; n = 3 mice; p = 0.60, Mann–Whitney U test). These data indicate that the innervation of PCs by CFs in adult L7-Cre;Itpr1flox/flox mice was comparable to that in Itpr1flox/flox mice.
We also performed a transmission electron microscopy analysis to visualize synaptic ultrastructure (Fig. 4C). Consistent with the immunohistochemical analysis result, free spines—spines without any attached PF terminals (Guastavino et al., 1990; Kurihara et al., 1997)—were rarely seen in L7-Cre;Itpr1flox/flox mice at 10W, similar to controls (percentage of spines contacted with PF terminals: Itpr1flox/flox, 99.4 ± 0.4%; L7-Cre;Itpr1flox/flox, 98.8 ± 0.9%; n = 12 sections from two mice; p = 0.86, Mann–Whitney U test). In addition, almost all spines formed synapses with PF terminals in a one-to-one fashion, and spines contacting two or more synaptic terminals were not observed in either Itpr1flox/flox or L7-Cre;Itpr1flox/flox PCs (percentage of multiple synapse: Itpr1flox/flox, 0.030 ± 0.011%, n = 396 synapses from two mice; L7-Cre;Itpr1flox/flox, 0.041 ± 0.009%, n = 249 synapses from two mice; p = 0.44, t test). Although abnormal enlargement of PF presynaptic terminals and vesicle accumulation have been reported in IP3R1-null mice (Hisatsune et al., 2006), the L7-Cre;Itpr1flox/flox mice did not show these abnormalities in PF–PC synapses (PF presynaptic area: Itpr1flox/flox, 0.281 ± 0.012 μm2, 167 synapses from two mice; L7-Cre;Itpr1flox/flox, 0.284 ± 0.013 μm2, n = 150 synapses from two mice; p = 0.90, t test; PF presynaptic vesicle density: Itpr1flox/flox, 102.8 ± 2.8 vesicles/μm2, 72 synapses from two mice; L7-Cre;Itpr1flox/flox, 101.1 ± 3.9 vesicles/μm2, n = 61 synapses from two mice; p = 0.71, t test).
To further examine whether the PF–PC synapses of adult L7-Cre;Itpr1flox/flox mice were functional, we compared the input–output relationship of PF-EPSP amplitudes of PCs in acute cerebellar slices by using the whole-cell patch-clamp method. We found that PCs of L7-Cre;Itpr1flox/flox mice at 10W displayed significantly larger PF-EPSPs than age-matched Itpr1flox/flox mice (Fig. 5A). We also assessed the probability of presynaptic transmitter release by measuring the PPF of the PF-EPSPs, and found that the PPF ratio at interpulse intervals of 50 ms was comparable between L7-Cre;Itpr1flox/flox and control Itpr1flox/flox mice at 10W (Itpr1flox/flox, 1.353 ± 0.028, n = 15 cells from 10 mice; L7-Cre;Itpr1flox/flox, 1.363 ± 0.019, n = 22 cells from 11 mice; p = 0.77, t test; Fig. 5B), indicating that the large PF-EPSPs of the input–output relationship in the L7-Cre;Itpr1flox/flox mice could not be dependent on PF presynaptic effect. These results suggest that abnormally increased synapses formed on the distal dendrites of L7-Cre;Itpr1flox/flox PCs are functional, and that functional synapse density in the stimulated region increases in L7-Cre;Itpr1flox/flox mice at 10W. Together, these findings suggest that abnormal spines of PCs form functional synapses with PF in adult L7-Cre;Itpr1flox/flox mice.
Formation of functional PF–PC synapse in adult L7-Cre;Itpr1flox/flox mice. A, Averaged input–output relationship of PF-EPSPs in acute cerebellar slices from 9- to 10-week-old Itpr1flox/flox and L7-Cre;Itpr1flox/flox mice. Plot of EPSP amplitude versus the T from Itpr1flox/flox (white circles) and L7-Cre;Itpr1flox/flox (black circles) mice are shown (Itpr1flox/flox, n = 6 neurons from three mice; L7-Cre;Itpr1flox/flox, n = 8 neurons from three mice; *p < 0.05, **p < 0.01, t test). Insets show representative EPSP traces evoked by PF stimuli of different intensities. B, The PPF ratio of the PF-EPSPs from acute cerebellar slices from 9- to 10-week-old Itpr1flox/flox or L7-Cre;Itpr1flox/flox mice. Representative EPSP traces evoked by PF stimuli at interpulse intervals of 50 ms are shown.
Impaired cerebellar LTD and motor learning in adult L7-Cre;Itpr1flox/flox mice
PF–PC synapses are sites where cerebellar LTD is induced by a simultaneous activation of PFs and CFs, and these are known to be involved in certain types of motor learning (Ito, 1989). We previously showed that IP3R1 plays an essential role in the induction of cerebellar LTD by using IP3R1-null mice (Inoue et al., 1998). Therefore, we examined whether IP3R1 in PCs was necessary for the induction of LTD by using 9- to 10-week-old L7-Cre;Itpr1flox/flox mice. As shown in Figure 6A, conjunctive stimulation of PF and CF induced LTD of PF-EPSPs in acute cerebellar slices obtained from the Itpr1flox/flox mice. In contrast, in the L7-Cre;Itpr1flox/flox mice, induction of LTD was blocked (Fig. 6A). We also used another LTD-induction protocol that replaced CF stimulation with direct depolarization of PCs to exclude any effect from CF properties. This protocol also induced LTD in the Itpr1flox/flox mice, whereas it failed to induce LTD in the L7-Cre;Itpr1flox/flox mice (Fig. 6B).
Lack of cerebellar LTD in L7-Cre;Itpr1flox/flox mice. A, Averaged time course of LTD induced by conjunctive stimulation of double-shock PF stimulation (2PF) and CF stimulation (CF) from Itpr1flox/flox (white circles) and L7-Cre;Itpr1flox/flox (black circles) mice at 10W. Rising slopes of PF-EPSPs relative to average baseline values during the 5 min preconjunction period are shown. The obliquely shaded band indicates period of conjunction. LTD magnitude, the average decrease in PF-EPSP slopes at 41–50 min, was used for statistical analysis (Itpr1flox/flox, n = 6 neurons from four mice; L7-Cre;Itpr1flox/flox, n = 6 neurons from six mice; p < 0.05, t test). Inset shows superimposed PF-EPSP traces recorded before (gray line) and after (black line) conjunctive stimulation. B, Averaged time course of LTD induced by conjunctive stimulation of 2PF and membrane depolarization (DEPO) from Itpr1flox/flox (white circles) and L7-Cre;Itpr1flox/flox (black circles) mice at 10W (Itpr1flox/flox, n = 7 neurons from six mice; L7-Cre;Itpr1flox/flox, n = 6 neurons from six mice; p < 0.05, t test). Inset shows superimposed PF-EPSP traces recorded before (gray line) and after (black line) conjunctive stimulation.
Finally, to evaluate the motor learning ability of adult L7-Cre;Itpr1flox/flox mice, we tested the adaptation of the HOKR, which is a simple form of cerebellar motor learning (Katoh et al., 1998, 2000). Since the cerebellar flocculus is an essential site for HOKR adaptation (Katoh et al., 1998), we first assessed the expression level of IP3R1 in floccular PCs and found that most L7-Cre;Itpr1flox/flox PCs lacked IP3R1 expression at 10W (Fig. 7A). In addition, the L7-Cre;Itpr1flox/flox mice displayed no detectable loss of PCs or other cells in the flocculus and had normal overall flocculus morphologies (Fig. 7A,B).
Characteristics of eye movements in L7-Cre;Itpr1flox/flox mice. A, Immunohistochemical analysis of IP3R1 protein in the cerebellar flocculus (FL). Coronal sections prepared from Itpr1flox/flox and L7-Cre;Itpr1flox/flox mice at 10W were stained with anti-IP3R1 (4C11, red) and anti-calbindin (green) antibodies and DAPI (blue). Scale bar, 500 μm. B, Nissl staining of coronal sections of cerebellar flocculus prepared from Itpr1flox/flox and L7-Cre;Itpr1flox/flox mice at 10W. Bottom panels show a magnified view of the top panels. Scale bars: top, 500 μm; bottom, 100 μm. C, HOKR gain of Itpr1flox/flox (white bars) and L7-Cre;Itpr1flox/flox (black bars) mice at 10W. HOKR was measured by sinusoidal screen oscillation at 5–20° (peak-to-peak) and 0.11–0.17 Hz (maximum screen velocity, 2.6–10.5°/s) in the light (Itpr1flox/flox, n = 13 mice; L7-Cre;Itpr1flox/flox, n = 5 mice; two-way repeated-measures ANOVA). p value is shown in the panel.
We next investigated the basic properties of the HOKR and HVOR, which are eye movements that compensate for the movements of the visual field and head, respectively, to assess the dynamic characteristics of eye movements in L7-Cre;Itpr1flox/flox mice at 10W. We characterized the amplitude and temporal properties of eye movements by the gain and phase of the eye movement response. Gain is the ratio of the amplitude of eye movement to the amplitude of visual (HOKR) or head (HVOR) movement. Phase is a parameter described as the temporal relationship between eye movement and visual or head movement. During eye movement testing, spontaneous nystagmic movements were rarely seen in the L7-Cre;Itpr1flox/flox mice in both light and dark conditions. The HOKR gains (amplitude) of the L7-Cre;Itpr1flox/flox mice were smaller than those of the control Itpr1flox/flox mice at all screen velocities examined (Fig. 7C), whereas the HOKR phases (timing) were similar to those of the Itpr1flox/flox mice (Itpr1flox/flox, n = 13 mice; L7-Cre;Itpr1flox/flox, n = 5 mice; p = 0.10, two-way repeated-measures ANOVA). No significant changes were observed in the HVOR gains or phases between the Itpr1flox/flox and the L7-Cre;Itpr1flox/flox mice (HVOR gain: Itpr1flox/flox, n = 7 mice; L7-Cre;Itpr1flox/flox, n = 5 mice; p = 0.97, two-way repeated-measures ANOVA; HVOR phase: Itpr1flox/flox, n = 7 mice; L7-Cre;Itpr1flox/flox, n = 5 mice; p = 0.07, two-way repeated-measures ANOVA).
Sustained exposure of the eyes to screen oscillations in the presence of sufficient retinal slips for 1 h induces adaptive changes in the HOKR gain in mice (Ito, 1989; Katoh et al., 1998). Ten-week-old Itpr1flox/flox mice demonstrated an adaptive increase in HOKR gain (delta gain) by 0.05 on average during 1 h. In contrast, no adaptive increase in HOKR gain was observed in the L7-Cre;Itpr1flox/flox mice (Fig. 8A,B). These results indicate that IP3R1 in PCs is essential for the induction of PF–PC LTD and HOKR adaptation; this supports the hypothesis that PF–PC LTD underlies cerebellar motor learning (Ito, 1989; Yuzaki, 2012).
Lack of HOKR adaptation in L7-Cre;Itpr1flox/flox mice. A, Examples of eye movement traces at 0 h (dashed lines) and after 1 h (solid lines) of sustained screen oscillation (15° at 0.17 Hz) in Itpr1flox/flox (top) and L7-Cre;Itpr1flox/flox (middle) mice at 10W. Averaged traces of eye position obtained from >10 cycles of screen oscillation are shown. The trace of screen position is shown in the bottom panel. B, The mean HOKR gain changes during 1 h of sustained screen oscillation of Itpr1flox/flox (white bar) and L7-Cre;Itpr1flox/flox (black bar) at 10W. HOKR gain change was defined as the HOKR gain at 1 h − HOKR gain at 0 h (n = 7 mice; Mann–Whitney U test). p value is shown in the panel.
Discussion
In this study, we present two major findings. First, the loss of IP3R1 in PCs causes cerebellar dysfunction, motor discoordination, and impaired cerebellar learning and PF–PC LTD in adult L7-Cre;Itpr1flox/flox mice. Second, PC IP3R1 is critical for the maintenance of cerebellar synaptic circuits, especially the spatial arrangement of PF–PC synapses in the mature cerebellar cortex.
Ca2+ influx through ligand-gated or voltage-gated Ca2+ channels and Ca2+ release from the ER are two major Ca2+ sources in spines; ryanodine receptors (RyRs) or IP3Rs are responsible for the latter. Interestingly, RyRs and IP3Rs show distinct expression patterns in the spines of hippocampal neurons and PCs (Sharp et al., 1993): RyRs, but not IP3Rs, are expressed in hippocampal neuron spines, whereas IP3Rs, but not RyRs, are localized in the PC spines. Although it was previously reported that Ca2+ release from the ER through RyRs induced changes in spine morphology and the formation of new spines in cultured hippocampal neurons (Korkotian and Segal, 1999), the role of IP3R1 in controlling spine morphology was unclear. In this study, we provide the first evidence that loss of IP3R1 function produces an abnormal increase in spine density and length in mature PCs. Thus, our results suggest that the IP3R has a critical role in the regulation of spine distribution and morphology of adult PCs in vivo.
Although PCs in L7-Cre;Itpr1flox/flox mice have quite normal morphology at 3W, they display a dramatic increase in spine density and length along distal dendrites, and a decrease in dendritic branch points in adults. We previously reported that IP3R1-null mice had normal PC spine density at P20 but showed abnormal PC dendrites, which is caused by the absence of IP3R1 in GCs but not in PCs (Hisatsune et al., 2006). Therefore, the abnormal spine and dendritic morphology of the mature PCs in L7-Cre;Itpr1flox/flox mice is likely caused by defects in spine and dendritic morphology maintenance, rather than defects during neuronal development. Concerning normal spine morphogenesis in IP3R1-null mice, an alteration in the presynaptic function of the PF–PC synapses was also observed in IP3R1-null mice (Matsumoto et al., 1996; Hisatsune et al., 2006); consequently, there is a possibility that normal spine morphogenesis in PCs of IP3R1-null mice is due to alterations in PF presynaptic function. However, several studies using mice lacking GCs provide evidence that spinogenesis in the distal dendrites of PCs is intrinsic and not dependent on presynaptic axons (Mariani et al., 1975; Hirano et al., 1977; Sotelo, 1977). Thus, if IP3R1 in PCs is required for normal spinogenesis in developing PCs, spine abnormality is expected to occur in IP3R1-null mice. Together, we conclude that IP3R1 plays a critical role in the maintenance of spine and dendritic morphology in mature PCs.
Studies using mice lacking IP3R1 and other upstream molecules of IP3R1 in PCs (Aiba et al., 1994; Matsumoto et al., 1996; Kim et al., 1997; Offermanns et al., 1997; Kano et al., 1998; Ichise et al., 2000; Miyata et al., 2001; Nakao et al., 2007; Frederick et al., 2012) have suggested that type 1 metabotropic glutamate receptor (mGluR1)-heteromeric G-protein α q subunit (Gαq)-phospholipase C (PLC) β4-IP3-IP3R1 signaling is essential for motor coordination. However, ataxia observed in L7-Cre;Itpr1flox/flox mice appeared to be more severe than that in other mutant mice, since mGluR1 or PLCβ4 knock-out mice were reported to present clear footprints (Kim et al., 1997; Ichise et al., 2000; Nakao et al., 2007). mGluR1, Gαq, or PLCβ4 knock-out mice were able to perform the rota-rod test (Offermanns et al., 1997; Kano et al., 1998; Ichise et al., 2000; Miyata et al., 2001; Nakao et al., 2007; Frederick et al., 2012), but L7-Cre;Itpr1flox/flox mice older than 8W could not even retain normal posture for the rota-rod test. Why did such unexpectedly severe ataxia occur in adult L7-Cre;Itpr1flox/flox mice? During locomotion, PCs receive sensory and motor inputs through the mossy fiber–GC–PF pathway and integrate them to provide feedback to cerebellar nuclei (Arshavsky et al., 1983; Ito, 2002). Since cerebellar nuclei have a somatotopic map of the body, the connection of PFs to the cerebellar nuclear map by PCs could influence coordinated movements of multiple muscles and joints (Thach et al., 1992). Therefore, precise topographic organization of the PF–PC synaptic circuit is thought to be important for coordinated movements of body parts in animals. Because L7-Cre;Itpr1flox/flox mice have increased spine densities and decreased dendritic arbors, and these spines formed intact synapses with PF terminals in a one-to-one fashion, the spatial distribution of PF–PC synapses in the cerebellar cortex is presumably abnormal in adult L7-Cre;Itpr1flox/flox mice. Judging from the evidence that mGluR1, Gαq, or PLCβ4 knock-out mice have not shown such apparent morphological abnormalities in PCs, the unexpected severity of ataxia in L7-Cre;Itpr1flox/flox mice might be caused by the abnormal rewiring of PF–PC synaptic circuits in the mature cerebellar cortex. The redundant contributions of AMPA receptor and PLCβ isoforms other than β4 for IP3 production in PCs (Kano et al., 1998; Hirono et al., 2001; Okubo et al., 2001) most likely underlies the differences in PC morphological phenotypes between L7-Cre;Itpr1flox/flox and mGluR1, Gαq, or PLCβ4 knock-out mice. Besides the abnormalities in PF–PC synaptic circuits, mild atrophic PC phenotypes with small soma size and decreased dendritic branching may also, at least in part, contribute to severe ataxic phenotype in L7-Cre;Itpr1flox/flox mice.
Recent studies have shown that the disruption of IP3R1-dependent signaling is linked to several types of human spinocerebellar ataxia (SCA; Schorge et al., 2010; Kasumu and Bezprozvanny, 2012). In SCA2 pathogenesis, polyglutamine-expanded ataxin-2 specifically binds IP3R1 and increases the sensitivity of IP3R1 to activation by IP3, and this supranormal Ca2+ release via IP3R1 is important in cerebellar dysfunction (Liu et al., 2009; Kasumu et al., 2012). Alternatively, SCA15 patients have genetic deletion or mutation of the Itpr1 gene (van de Leemput et al., 2007; Hara et al., 2008; Iwaki et al., 2008; Di Gregorio et al., 2010; Novak et al., 2010). Moreover, an ataxic phenotype is observed in several spontaneous mutant mice with less IP3R1 expression (Street et al., 1997; van de Leemput et al., 2007). However, it is still not understood how disruption of IP3R1-dependent signals leads to cerebellar dysfunction in SCA pathogenesis. In this study, we demonstrated that the loss of IP3R1 function specifically in PCs is sufficient to cause ataxia without apparent PC loss, suggesting that PC loss itself is not essential for causing ataxia in the situation of IP3R1 loss of function. In addition, L7-Cre;Itpr1flox/flox mice showed abnormal PC spine maintenance, which was possibly associated with severe ataxia. Therefore, it is possible that the disruption of PC spine maintenance by IP3R1 deletion might be associated with the severity of ataxia in SCA15 pathology.
Although Kasumu et al. (2012) demonstrated that chronic depletion of IP3R-dependent signaling by expressing the inositol 1,4,5-phosphatase enzyme in intact PCs caused PC degeneration and ataxia at 12 months of age, our L7-Cre;Itpr1flox/flox mice did not show PC loss at 10W (Fig. 2E), and even at 5 months of age (our preliminary observation; n = 1). The low survival rate with less feeding due to severe ataxia in the adult L7-Cre;Itpr1flox/flox mice hinders gathering the data in aging L7-Cre;Itpr1flox/flox mice. Nevertheless, because SCA15 is characterized by late-onset ataxia in humans (Storey et al., 2001), further analysis with more aging L7-Cre;Itpr1flox/flox mice is required for clarifying this issue.
Recent studies have revealed that the structural rearrangement of spines caused by the induction of synaptic plasticity is associated with learning and memory in the cerebral cortex and hippocampus (Kasai et al., 2010). Several studies have also suggested that motor learning is associated with morphological changes in PC spines and PF–PC synapses in the cerebellum in vivo (Black et al., 1990; González-Burgos et al., 2011). However, in contrast to the cerebral cortex and hippocampus, PF–PC LTD induction did not cause any change in the spine density or size of PCs (Sdrulla and Linden, 2007), and many genetically modified mice lacking cerebellar LTD did not show any significant morphological abnormalities of PC spines (Aiba et al., 1994; De Zeeuw et al., 1998; Feil et al., 2003; Hansel et al., 2006; Steinberg et al., 2006; van Woerden et al., 2009), suggesting that PF–PC LTD does not trigger morphological changes in PC spines. Thus, structural PC spine rearrangement and PF–PC LTD are likely to be independently involved in cerebellar motor learning. PC IP3R1 is therefore thought to regulate spine morphology in addition to inducing PF–PC LTD, and the abnormalities in both spine morphology and PF–PC LTD induction may cause a severe impairment of motor learning in L7-Cre;Itpr1flox/flox mice.
In conclusion, we propose a specific role for IP3R1 in PCs not only as an intracellular mediator of LTD induction but also as a critical regulator of PF–PC synaptic circuit maintenance in the cerebellum of adult mice. Our L7-Cre;Itpr1flox/flox mice may provide a valuable experimental model to clarify the relationship between the maintenance and modification of PF–PC synapses and the expression of cerebellar functions in vivo.
Footnotes
This work was supported by grants from the RIKEN Brain Science Institute (BSI) and The Moritani Scholarship Foundation (to C.H.); and by Japan Society for the Promotion of Science Grants-in-Aid for Scientific Research 20500301 (to C.H.) and 20220007 (to K.M.). We thank Dr. K. Kawaai, Dr. H. Ando, Dr. A. Mizutani, and Dr. Y. Kuroda for valuable discussions. We also thank Ms. M. Kudo, Mr. T. Suzuki, and Dr. T. Okamoto for technical assistance. We thank the staffs of the BSI Research Resources Center for animal care. We also thank Dr. C. Yokoyama and Dr. A.V. Terashima for editing of this manuscript.
The authors declare no competing financial interests.
- Correspondence should be addressed to either Chihiro Hisatsune or Katsuhiko Mikoshiba, RIKEN Brain Science Institute, Laboratory for Developmental Neurobiology, 2-1 Hirosawa, Wako, Saitama 351-0198, Japan. chihiro{at}brain.riken.jp or mikosiba{at}brain.riken.jp