|Home | About | Journals | Submit | Contact Us | Français|
Spinocerebellar ataxia 2 (SCA2) is a neurodegenerative disorder characterized by progressive ataxia. SCA2 results from a polyglutamine (polyQ) expansion in the cytosolic protein ataxin-2 (Atx2). Cerebellar Purkine cells (PC) are primarily affected in SCA2, but the cause of PC dysfunction and death in SCA2 is poorly understood. In previous studies, we reported that mutant but not wild type Atx2 specifically binds the inositol 1,4,5-trisphosphate receptor (InsP3R) and increases its sensitivity to activation by InsP3. We further proposed that the resulting supranormal calcium (Ca2+) release from the PC endoplasmic reticulum (ER) plays a key role in the development of SCA2 pathology. To test this hypothesis, we achieved a chronic suppression of InsP3R-mediated Ca2+ signaling by adeno-associated virus (AAV)-mediated expression of the inositol 1,4,5-phosphatase (Inpp5a) enzyme (5PP) in PCs of a SCA2 transgenic mouse model. We determined that recombinant 5PP overexpression alleviated age-dependent dysfunction in the firing pattern of SCA2 PCs. We further discovered that chronic 5PP overexpression also rescued age-dependent motor incoordination and PC death in SCA2 mice. Our findings further support the important role of supranormal Ca2+ signaling in SCA2 pathogenesis and suggest that partial inhibition of InsP3-mediated Ca2+ signaling could provide therapeutic benefit for the patients afflicted with SCA2 and possibly other SCAs.
Spinocerebellar ataxia type 2 (SCA2) is an autosomal dominant neurodegenerative disorder characterized in patients by a progressive incoordination of movement known as ataxia. SCA2 patients also present with other secondary symptoms including dysarthria, dysphagia, delayed saccadic eye movements and peripheral neuropathy (Lastres-Becker et al., 2008). Autopsy studies of SCA2 patients consistently show olivopontocereballar atrophy with marked reduction of Purkinje cells, degeneration of the inferior olives, pontine nuclei, and pontocerebellar fibers (Geschwind et al., 1997; Lastres-Becker et al., 2008). Atrophy of the cerebellum and brainstem in SCA2 patients can also be visualized in vivo by MRI (Burk et al., 1996). SCA2 is one of the most prevalent dominantly inherited SCAs in a heterogeneous group of 30 spinocerebellar ataxias. Each SCA is linked to a mutation, deletion or polyglutamine expansion at a different gene locus. SCA2 has been genetically linked to an expansion and translation of an unstable CAG repeat track in the gene that codes for ataxin-2 (ATXN2) (Imbert et al., 1996; Pulst et al., 1996; Sanpei et al., 1996). In SCA2 patients, the repeat track is expanded from the normal 22 glutamine repeats (Atx2-22Q) to more than 58 (Atx2-58Q) glutamine repeats. The cause of PC degeneration in SCA2 is currently unknown (Bezprozvanny and Klockgether, 2010).
We previously reported that Atx2-58Q, but not Atx2-22Q, specifically interacts with and increases the sensitivity of inositol 1,4,5-triphosphate receptors (InsP3Rs) to activation by IP3 (Liu et al., 2009). This was supported by our experiments in primary PC cultures from SCA2-58Q transgenic mice that express Atx2-58Q under the control of the PC-specific promoter (Huynh et al., 2000). When compared to PC cultures from their wildtype littermates, SCA2-58Q PC cultures treated with an mGluR agonist (RS)-3,5-dihydroxyphenylglycine (DHPG) exhibited significantly greater InsP3-induced calcium (Ca2+) release (IICR) from the endoplasmic reticulum (ER) and higher intracellular Ca2+ concentrations (Liu et al., 2009). Moreover, we demonstrated that long-term feeding of SCA2-58Q mice with a Ca2+ stabilizer (dantrolene) alleviated the motor phenotype of these mice and reduced PC loss (Liu et al., 2009). Based on these findings, we proposed that the association of polyglutamine-expanded ataxin-2 with InsP3Rs results in excessive Ca2+ release from the PC ER, initiates cytoplasmic Ca2+ dysregulation and cellular dysfunction of PCs which eventually result in ataxia (Kasumu and Bezprozvanny, 2010; Liu et al., 2009). In support of our hypothesis, genetic evidence has been used by others to suggest that InsP3R is the “eye of the storm” of pathogenesis for many SCAs (Schorge et al., 2010).
To further test our hypothesis in the present study we used a specific molecular intervention to normalize InsP3R-mediated Ca2+ signals in PCs of SCA2-58Q mice. Consistent with our hypothesis, we discovered that partial suppression of IICR in SCA2-58Q mice prevented the onset of PC dysfunction, alleviated motor incoordination and reduced age-dependent PC degeneration in SCA2-58Q mice. These results indicate that partial suppression of IICR is a viable therapeutic strategy for treatment of SCA2 and possibly other SCAs.
SCA2-58Q mice on C57/B6 background (Huynh et al., 2000) were kindly provided to our laboratory by Dr Stefan Pulst (Univ of Utah) and used in our previous studies (Liu et al., 2009). In these mice the expression of human Atx2-58Q transgene is driven by the PC-specific L7/pcp2 promoter (Huynh et al., 2000). The mice were back-crossed to FVB/N background for at least 6 generations in our laboratory and used in the previous study (Kasumu and Bezprozvanny, 2010). In this study, the SCA2-58Q (FVB) male hemizygous mice were bred to wildtype (WT) FVB/N females to generate mixed litters. The pups were genotyped by PCR for the presence of human Atx2 transgene and parallel experiments were performed with transgenic and wild type littermates. All mice were housed in a temperature-controlled room at 22–24°C with a 12hr light/dark cycle. Mice had access to standard chow and water ad libitum. All procedures were approved by the Institutional Animal Care and Use Committee (IACUC) of the UT Southwestern Medical Center at Dallas in accordance with the National Institutes of Health guidelines for the Care and Use of Experimental Animals.
Cerebellar slices were prepared from WT and SCA2 mice at 6, 12, 24, 36, and 48 weeks old. Mice were anesthetized with a ketamine/xylazine cocktail and transcardially perfused with ice-cold aCSF containing (in mM) 85 NaCl, 24 NaHCO3, 25 glucose, 2.5 KCl, 0.5 CaCl2, 4 MgCl2, 1 NaH2PO4, 75 Sucrose. Solutions were equilibrated with 95% O2/5% CO2. Subsequently, the cerebellum was dissected and 300 µm thick sagittal slices were made with a VT1200S vibratome (Leica). Slices were allowed to recover in aCSF at 35°C for 30 minutes and then transferred to room temperature before recordings were made. All recordings were made within 5 hours after dissection. Recordings were made in a chamber (Warner instruments) heated to 34–35°C. Loose-patch recordings were made (Hausser and Clark, 1997; Smith and Otis, 2003) to evaluate spontaneous activity of PCs from WT and SCA2 mice at different ages. Briefly, the patch pipette was filled with 140 mM NaCl buffered with 10 mM HEPES pH 7.3 held at 0 mV. A loose patch (less than 100MΩ) was generated at the PC soma close to the axon hillock. Extracellular recording solution contained (mM) 119 NaCl, 26 NaHCO3, 11 glucose, 2.5 KCl, 2.5 CaCl2, 1.3 MgCl2, 1 NaH2PO4. In some experiments, recordings were performed in the presence of 100 µM Picrotoxin (Sigma) and 10µM DNQX (Sigma) to verify spontaneous nature of observed events. Spontaneous action potential currents were recorded for at least 5 minutes from each cell. These five-minute recordings were analyzed for tonic or burst firing patterns. A PC was characterized as firing tonically if it was found to fire repetitive tonic spikes with relatively constant frequency during the recording duration. A cell was characterized as bursting if it was identified that it had more than 5% of the interspike intervals that fell outside of 3 standard deviations from the mean of all interspike intervals in that cell. The proportion of tonically firing PCs for each genotype at each time-point was calculated and plotted as mean percentage of total PCs recorded ± SE. All tonically firing PCs were analyzed further for firing frequency and firing variability (Alvina and Khodakhah, 2010a; Alvina and Khodakhah, 2010b). The mean firing frequency in the 5 minute recording duration was calculated for each PC and the average firing frequency for each genotype at each time-point was also calculated and plotted as mean ± SE. The correlation of variation of interspike interval (CV ISI) was calculated to represent the firing variability. We calculated CV ISI as the standard deviation divided by the mean inter spike interval in a 5 minute recording period for each PC. To test the acute effects of blocking IP3-Ca2+ signaling on PC dysfunction in SCA2 mice, we treated spontaneously firing 24-week old 58Q PCs with an mGluR1 antagonist, LY367385 ((+)-2-methyl-4-carboxyphenylglycine, Sigma). These recordings were made in the presence of 100 µM Picrotoxin and 10µM DNQX to block fast excitatory and inhibitory transmission. PCs were recorded in the absence of LY367385 for at least 5 minutes and then treated with 100 µM LY367385 for the remainder of the recording (at least 8 minutes). The average firing rate and average CV ISI was analyzed in the 2 minutes immediately prior to treatment with LY367385 and 3 minutes after the addition of 100 µM LY367385.
The wild type and mutant (R343A and R343A/R350A) mouse inositol 1,4,5-triphosphate 5-phosphatase (Inpp5a, 5PP) expression constructs were kindly provided to us by Masamitsu Iino (University of Tokyo, (Kanemaru et al., 2007)). A hemagglutinin (HA) tag was added to the N-terminus of each 5PP construct by PCR. The HA-5PP inserts were excised and individually cloned into the pFBGR (pGANCMVBGHpA) plasmid behind the CMV promoter. The inserts now including a CMV promoter were individually cloned into an pAAV plasmid. The pAAV plasmid also expresses GFP driven by a different CMV promoter. The AAV-5PP plasmids were sent to the University of Iowa Gene Transfer Vector Core where serotype 1 adeno-associated viruses (AAV) were made using the Sf9 cell-based AAV production system. The titer of purified AAV-5PP viruses provided by the University of Iowa Gene Transfer Vector Core was equal to or greater than 1013 vg/ml. In addition, control AAV1-GFP virus of similar titer was also provided by the University of Iowa Gene Transfer Vector Core.
Presenilin double knockout mouse embryonic fibroblast (PS DKO MEF) cells (Herreman et al., 2000) were used to determine the efficacy of the AAV-5PP constructs to suppress InsP3-induced Ca2+ release. Due to the absence of presenilins 1 and 2, these MEF cells have high ER Ca2+ levels that can be released via InsP3Rs with the application of 300nM Bradykinin (Tu et al., 2006). Briefly, MEF cells were cultured on poly-D-lysine (Sigma) coated 12 mm round glass coverslips and infected with AAV-5PP, AAV-RA, AAV-DM or AAV-GFP. Cytosolic Ca2+ imaging experiments in Ca2+-free media were performed 48 hours after viral infection using a Fura-2 fluorescent indicator as previously described (Tu et al., 2006). Infected MEF cells were identified by GFP fluorescence and the Fura-2 data were collected.
Stereotaxic surgery was performed according to (Dodge et al., 2005). Briefly, 4µl of AAV (1013 titer) was bilaterally injected into the deep cerebellar nucleus (DCN) at the coordinates Bregma −5.75; Lateral +1.8; D/V −2.6 mm. PCs project axons into the DCN and retrogradely transport the virus back to the soma where it integrates and is expressed for over 12 months (Dodge et al., 2005; Kaemmerer et al., 2000). AAV GFP expression in PCs was confirmed by scanning cerebellar slices with a fluorescence plate reader or confocal imaging. HA-tag expression was confirmed by western blotting of cerebellar lysates at least 2 weeks after surgery. Infection efficiency was calculated by comparing transduced PCs to PCs from L7-GFP transgenic mice, that express GFP in all PCs driven by the L7/pcp2 PC-specific promoter.
Cell lysates were prepared from infected MEF cell cultures and then from cerebellum of mice post-DCN injection. The expression of AAV-5PP, AAV-RA, AAV-DM was confirmed by probing with monoclonal antibodies against 5PP (clone 3D8, Sigma Aldrich) and HA-tag (clone HA-7, Sigma Aldrich). Monoclonal antibodies against actin were used as a loading control (clone AC-15, Sigma Aldrich).
Rotarod and beamwalk assays were used to assess motor coordination. These were performed as previously described (Liu et al., 2009). For analysis of motor impairment, female mice from each litter were genotyped, weight-matched and divided into 3 WT and 3 SCA2-58Q groups, with each group containing 13-19 mice each. Briefly, SCA2-58Q and WT mice were trained on the accelerating rotarod apparatus (Columbus Instruments). Mice were screened on the rotarod running at a constant speed of 5rpm. Mice that fell off the rod in less than 5 minutes were dismissed. At each time following baseline testing, mice were trained on the rotarod accelerating at 0.2rpm for 4 consecutive days with 3 trials per day. The mean latency to fall off the accelerating rod on day 4 was recorded and analyzed for every animal in all 6 groups. Average group latency was calculated and plotted for all time-points. Three days after rotarod testing, mice were tested on the beam-walk assay using a homemade apparatus. Mice were trained on 3 consecutive days, with 3 consecutive trials on 3 separate beams of differing diameters. A round plastic 17mm beam, a round plastic 11mm beam and a wooden square 5mm beam were used. The mean latencies to traverse the middle 80cm length of the 11mm and 5mm beams on the 3rd day were recorded and analyzed for every animal in all 6 groups. The average group latency and average number of footslips were calculated and plotted for all time-points as mean ± SE. These motor coordination tasks were performed at 4, 6, 8, 10 and 12 months of age. Mice that were found at the 4 month time-point to be severely impaired on both behavioral tasks due to surgery-induced irreversible motor deficits were excluded from the entire study. These included 2 mice in the WT-DM group, 3 from the WT-RA group, 3 mice from the 58Q-DM group and 3 mice from the 58Q-RA group.
Analysis of Purkinje cell dark cell degeneration (DCD) was performed as previously described (Kasumu and Bezprozvanny, 2010). Briefly, 6 mice per group were sacrificed after the analysis of the 12-month motor tasks. Mice were euthanized and transcardially perfused (according to (Custer et al., 2006)) with PBS followed by 2% paraformaldehyde/2% glutaraldehyde in 0.1M cacodylate buffer. The cerebellum was left in fixative overnight. The next day the cerebellum was cut into 1mm3 sagittal sections and post-fixed in 1% Osmium Tetroxide. The specimen were subsequently stained en bloc with aqueous 1% uranyl acetate and lead citrate, dehydrated through a graded ethanol series, and embedded in EMbed 812 resin. Each cerebellum was cut into thinner sections (about 70–90 nanometers in thickness) and placed on copper grids. The grids were stained with aqueous 2% uranyl acetate and lead citrate. Two grids from each animal were examined on a FEI Tecnai G2 Spirit Biotwin transmission electron microscope operated at 120 kV. Digital images were captured with a SIS Morada 11 megapixel side mount CCD camera. At least 6 mice were analyzed per group with 2 grids produced from different cerebellum areas of the same mouse. PCs were judged to be in 1 of 3 stages- normal, moderate or severe. Normal PCs are spherical in shape and have regular alignment in the PC layer. Moderately degenerated PCs have slight shrinkage and moderately electron-dense cytosol that is not as dark as the nucleus. Severely degenerated PCs have markedly shrunken and electron-dense cytosol with similarly darkened nucleus. These PCs are also not regularly aligned in the PC layer. The processing of cerebellar sections for DCD was performed by an independent investigator in the Electron Microscopy core at the University of Texas Southwestern Medical Center at Dallas, who was blinded to mouse genotype and treatment. DCD quantification was performed by an investigator that was blinded to mouse genotype and treatment. The percentage of normal, moderately and severely degenerated PCs in each mouse was calculated. The average percentage of each group was plotted as mean ± SE.
Differences between specific groups were judged by a two-tailed Student’s unpaired t test using a significance level of P < 0.05.
PCs are the sole output of the cerebellar cortex, and the disruption of PC activity impairs cerebellar function in SCAs (Alvina and Khodakhah, 2010a; Alvina and Khodakhah, 2010b; Kasumu and Bezprozvanny, 2010; Mark et al., 2011; Shakkottai et al., 2011). Recent studies demonstrated significantly reduced firing frequency of PC cells in aging SCA2-127Q mice (Hansen et al., 2012). To evaluate the functional state of PCs in SCA2-58Q (58Q) mice, we used well-established protocols (Mark et al., 2011; Shakkottai et al., 2011; Smith and Otis, 2003; Walter et al., 2006) to record spontaneous PC firing pattern in cerebellar slices obtained from the 6, 12, 24, 36 and 48 weeks old 58Q mice and from the age-matched WT littermates. In each experiment the PCs were classified as firing in a tonic or bursting pattern during a 5 minute recording period. The firing pattern of a PC was classified as tonic if it consisted of rarely halting tonic spikes with relatively constant frequency (Fig. 1A). A cell was characterized as bursting if it was identified that it had more than 5% of the event intervals that fell outside of 3 standard deviations from the mean of all intervals in that cell (Fig 1B). From analysis of the data we discovered that >80% of PC cells fire in tonic firing pattern in slices obtained from 6 and 12 weeks old mice for both 58Q and WT groups (Fig. 1C). To analyze these data further, we determined the mean firing rates and the mean correlation of variation of the interspike intervals (CV ISI) for all tonically firing PCs in both groups of mice. We discovered that both the firing frequency and the variability of interspike intervals were similar for 58Q and WT mice at 6 and 12 weeks of age (Fig. 1D–E). From these results, we concluded that PCs function properly in 6 and 12 weeks 58Q mice, consistent with the lack of overt phenotype of 58Q mice at these ages in motor coordination assays (Liu et al., 2009).
In contrast to the results obtained with slices from the young WT and 58Q mice, the fraction of tonically firing PCs was significantly lower in 58Q mice than in WT mice at 24 weeks old (Fig. 1C). On average, 91 ± 10% (n = 4 mice) of WT PCs and only 64 ± 9% (n = 7 mice) of 58Q PCs were firing tonically at this age (p < 0.05; Fig. 1C). Moreover, at 24 weeks of age, the tonically firing PCs in slices from the 58Q mice were firing less frequently that the tonically firing PCs in WT mice at the same age (Fig. 1D). On average, tonically active WT cells were firing at 51 ± 3.3 Hz (n = 19 neurons) and tonically active 58Q cells were firing at 37 ± 3.1 Hz (n = 41 neurons) at this age (p < 0.01; Fig. 1D). The variability of interspike intervals was not significantly different between 58Q and WT mice at this age (Fig 1E). These results were replicated in recordings made from 36-week old mice. The fraction of tonically firing PCs was reduced to 87 ± 3% (n = 2 mice) in WT mice and to 62 ± 6% (n = 3 mice) in 58Q mice (p < 0.01; Fig. 1C). Furthermore, the firing frequency of these tonically firing WT cells remained relatively constant at 58 ± 6.8 Hz (n = 14 neurons) and as expected, the firing frequency of 58Q PCs was further reduced to 29 ± 3.9 Hz (n = 22 neurons; p < 0.001; Fig. 1D). We also found that the variability of interspike intervals (CV ISI) for tonically firing 58Q PCs was significantly increased 0.28 ± 0.05 (n = 22 neurons) compared to WT PCs (0.13 ± 0.03; n=14; p< 0.05, Fig. 1D).
The functional differences between 58Q and WT mice became even more dramatic with increased age. When slices from 48 week old mice were used, the fraction of tonically firing PCs was reduced to 76 ± 4% (n = 4 mice) in WT mice and to 56 ± 10% (n = 3 mice) in 58Q mice (p < 0.01; Fig. 1C). For tonically firing cells, the firing frequency of WT cells remained relatively constant at 55 ± 3.7 Hz (n = 24 neurons) but for 58Q cells the frequency of firing was further reduced to 25 ± 6.3 Hz (n = 13 neurons; p < 0.001; Fig. 1D). Moreover, at 48 weeks of age, the variability of interspike intervals (CV ISI) for tonically firing cells remained constant for WT cells at 0.20 ± 0.03 (n= 24) but was significantly increased to 0.49 ± 0.05 (n = 13 neurons) for 58Q cells (p< 0.0001, Fig. 1D). Thus, we concluded that aging 58Q PCs increasingly fire in a bursting pattern and even tonically firing cells fire less frequently and with reduced precision when compared to age-matched WT cells. Interestingly, the age of onset of 58Q PCs electrophysiological abnormalities at 24 weeks of age (Fig. 1C–D) closely mirrors the age of onset of behavioral symptoms in SCA2-58Q mice in motor coordination assays observed in our previous studies (Liu et al., 2009). Based on this coincidence, we suggest that the reduced precision in PC firing in SCA2-58Q mice is causing the impaired performance of these mice in motor coordination tasks starting at 24 weeks of age. The continuous breakdown of the PC firing pattern at older ages (Fig. 1C–E) is likely to cause a progressive worsening of the motor incoordination phenotype of aging SCA2-58Q mice in motor coordination assays (Liu et al., 2009).
To determine the importance of excessive InsP3R-mediated Ca2+ release from the ER for SCA2 pathogenesis, we chose to chronically suppress InsP3-induced Ca2+ release (IICR) in PCs from 58Q mice. To specifically achieve this, we used serotype 1 adeno-associated viruses (AAV) to stably express mouse inositol 1,4,5-triphosphate 5-phosphatase (Inpp5, 5PP) in PCs of 58Q mice. InsP3-mediated Ca2+ release from the ER is triggered in response to generation of a cytosolic second messenger InsP3. The 5PP enzyme converts 1,4,5-InsP3 to an inactive form 1,4-InsP2 and terminates the InsP3-induced Ca2+ signals. It has been previously demonstrated that stable overexpression of recombinant 5PP can be used to chronically suppress InsP3-mediated Ca2+ signals in astrocytes (Kanemaru et al., 2007). We adapted the same approach for chronic suppression of InsP3-mediated Ca2+ signals in PCs of 58Q mice. Two arginine residues (R343 and R350 in mouse Inpp5a sequence) have been previously demonstrated to be critical for catalytic activity of 5PP (Communi et al., 1996; Kanemaru et al., 2007). Replacing Arg343 with Ala (R343A) decreases the ability of 5PP to bind InsP3, thus causing a 10-fold increase in the Km for InsP3 and consequently allowing partial suppression of IICR in cells when RA mutant is overexpressed (Communi et al., 1996; Kanemaru et al., 2007). A double R343A/R350A mutation (DM) creates a null mutant of 5PP that is unable to suppress IICR and can be used as a negative control (Communi et al., 1996; Kanemaru et al., 2007).
We generated serotype 1 AAV viruses encoding HA-tagged versions of wild type 5PP, RA and DM mutants (Fig. 2A). In addition to 5PP, these viruses also encoded EGFP protein to allow easy identification of infected cells by GFP fluorescence. In validation experiments, mouse embryonic fibroblast (MEF) cell cultures were infected with AAV-5PP viruses. Expression of recombinant 5PP in infected cells was confirmed by Western blotting of MEF cell lysates with anti-HA and anti-5PP antibodies (Fig. 2B). To validate the functional effects of 5PP-overexpression, we performed a series of Fura-2 Ca2+ imaging experiments with AAV-5PP infected MEF cells. The AAV-GFP virus was used as an additional control in these experiments. InsP3-coupled hormone Bradykinin (BK) was used to trigger IICR in these experiments. We found that the application of 300 nM BK resulted in strong Ca2+ responses in MEF cells infected with AAV-GFP or AAV-DM viruses (Fig. 2C,D). Consistent with the previous report (Kanemaru et al., 2007), IICR was significantly suppressed in cells infected with AAV-RA viruses and was completely abolished in cells infected with AAV-5PP viruses (Fig. 2C,D).
InsP3-mediated signaling is important for Purkinje cell function and the genetic deletion of InsP3R1 results in severe epileptic phenotype and early death in mice (Matsumoto et al., 1996). To avoid such problems, we restricted our in vivo experiments to the less potent 5PP-RA mutant. The inactive 5PP-DM mutant was used as a negative control in these experiments. The AAV-RA and AAV-DM viruses were delivered by bilateral stereotaxic injection into the deep cerebellar nuclei (DCN) region of 7-week old WT and 58Q mice (Fig. 3A). PCs project axonal processes to the DCN and transport the virus load retrogradely to the cell body in the molecular layer of the cerebellar cortex. As previously described, this protocol effectively transduces PCs with recombinant AAV in vivo (Dodge et al., 2005; Kaemmerer et al., 2000). Indeed, in our experiments we discovered that 2 weeks after injection AAV-encoded GFP signal that can be detected in an average of 5.8 ± 0.4 cerebellar lobes and 71.1 ± 4.6 % PCs (n = 17 mice) (Fig. 3C). The cerebellar expression of HA-tagged recombinant RA and DM constructs was further confirmed by immunohistochemistry using anti-HA antibodies (data not shown). Injection of AAV into the DCN at these coordinates leads to highest expression in the cerebellar lobules, with some expression in the DCN, pons and medulla (Dodge et al., 2005). The continuous expression of RA and DM constructs in injected mice was confirmed by preparing cerebellar lysates from injected mice and probing with antibodies against the HA-tag (Fig. 3B).
To determine the effect of IICR suppression on the progressive dysfunction of PCs in the SCA2 mouse model, we injected the DCN of 7 week old WT and 58Q mice with RA and DM viruses and prepared cerebellar slices from 24- and 48-week old injected mice for the recording of spontaneous PC activity. The transduced cells in these experiments were identified by GFP imaging. At 24 weeks of age, we found that most 58Q PCs overexpressing RA fired spontaneous action potential currents in a tonic fashion with a firing frequency similar to WT PCs. Specifically, we observed that the overexpression of RA prevented the irregularity in firing patterns of transduced 58Q PCs (89% tonically firing PCs; n = 27 neurons; n = 3 mice) when compared to 58Q-DM PCs (65% tonically firing PCs; n = 34 neurons; n = 4 mice; Fig. 4A). The fraction of tonically firing RA-infected 58Q cells was similar to the fraction of tonically firing WT-DM PCs of the same age (83% tonically firing PCs; n = 42 neurons; n = 4 mice; Fig. 4A). We also found that RA overexpression normalized the firing frequency and precision firing of tonically firing 58Q PCs (FF = 53 ± 2.5 Hz, p<0.001, CV ISI = 0.22 ± 0.07; n = 22 neurons) when compared to tonically firing 58Q-DM PCs (FF = 37 ± 4.4 Hz; CV ISI = 0.22 ± 0.06; n = 21 neurons; Fig. 4B,C). The fraction of tonically-firing PC cells, the frequency of tonically firing cells and the variability of interspike interval were not significantly different between RA-transduced 58Q PC cells, RA-transduced WT PCs and non-infected age-matched WT PC cells (Fig. 4A–C).
At 48 weeks, we found that most 58Q PCs overexpressing RA fired spontaneous action potential currents in a tonic fashion with a firing frequency similar to WT PCs. Specifically, we observed that the overexpression of 5PP prevented the irregularity in firing patterns of transduced 58Q PCs (86% tonically firing PCs; n = 31 neurons; n = 3 mice) when compared to DM-transduced 58Q PCs (57% tonically firing PCs; n = 19 neurons; n = 2 mice; Fig. 4E). The fraction of tonically firing RA-infected 58Q cells was similar to the fraction of tonically firing DM-transduced WT PCs of the same age (83% tonically firing PCs; n = 16 neurons; n = 2 mice; Fig. 4E). We also found that RA overexpression restored the firing frequency and precision firing of 58Q PCs (FF = 75 ± 7.4 Hz, p<0.001; CV ISI = 0.19 ± 0.04, p<0.0001; n = 25 neurons) when compared to DM-transduced 58Q PCs (FF = 27 ± 7.2 Hz; CV ISI = 0.59 ± 0.05; n = 11 neurons; Fig. 4D–G). The fraction of tonically-firing PC cells, the frequency of tonically firing cells and the variability of interspike interval were not significantly different between RA-infected 58Q PC cells and non-infected age-matched WT PC cells (Fig. 4D–G). Interestingly, the overexpression of RA slightly reduced the firing frequency and impaired regularity of firing of WT PCs at this age (Fig. 4E–G), but the difference from non-infected WT cells was not statistically significant.
Calcium is an intracellular second messenger involved in a multitude of pathways downstream of IP3Rs (Kasumu and Bezprozvanny, 2010). A disruption in these pathways decreases cell health and stimulates cell death cascades. We believe that the PC dysfunction is a representation of declining cell health caused by excess calcium release. Thus, to study the efficacy of acute normalization of intracellular calcium in restoring normal PC function, we tested the acute effect of blocking IP3-induced Ca2+ release in SCA2-58Q PCs. We treated 24-week old slices from 58Q mice with an inhibitor of metabotropic glutamate receptors, (+)-2-methyl-4-carboxyphenylglycine (LY367385). We found that 100uM of LY367385 increased the firing frequency of 58Q PCs from 31.2Hz ± 4 Hz to 40.1Hz ± 4 Hz (n = 14 neurons; data not shown). However, this was not statistically significant. Consequently, the acute treatment of 58Q PCs with LY367385 decreased the irregularity in firing of 58Q PCs from 0.10 ± 0.02 to 0.06 ± 0.01 (p<0.05; n = 14 neurons; data not shown).
To determine if preventing the PC dysfunction also reverses the motor incoordination in 58Q mice, WT and 58Q mice were divided into 6 groups (Table 1). Female WT and 58Q were age-matched and were used for this series of experiments. One group of WT mice and another group of 58Q mice underwent surgery for the DCN injection of AAV-RA. Two control groups were included that were injected with AAV-DM and 2 naïve groups of mice were also included that never experienced any surgery (non-injected, NI). All 6 groups of mice were then tested in motor coordination assays over a 10-month period using the beamwalk (11 mm round and 5 mm square beam) and accelerating rotarod behavioral tasks. Both tasks were performed starting at 2 months after surgery, (i.e. with 4, 6, 8, 10 and 12 months old mice). Previously, we reported that the onset of motor incoordination in 58Q mice (C57/B6 strain) is between 6 and 8 months of age (Liu et al., 2009). Consistent with this earlier finding, NI 58Q mice (FVB strain) used in the present study also demonstrated impaired beamwalk performance starting at 6 months of age when compared to NI WT mice (Fig 5). We further found that at 6 months of age, 58Q mice injected with RA made significantly less footslips on 5 mm square beam than NI 58Q and 58Q mice overexpressing DM (Fig. 5F). Starting at 8 months of age the 58Q mice expressing RA made fewer errors while traversing the entire length of the 11 mm or 5 mm beam (Figs. 5B, 5F) and had a shorter latency to traverse the 11 mm or 5 mm beam (Figs. 5D and Fig 5H). There was no significant difference between the performance of 58Q-RA, WT-NI and WT-DM mice (Fig. 5). SCA2-58Q mice expressing RA continued to perform significantly better on the 11 mm and 5mm beam than 58Q mice expressing DM at the 8, 10 and 12 month time-points. Overexpression of RA in PCs of WT mice did not affect its performance on 11 mm beam (Figs 5A and 5C). However, it increased the number of errors made on the 5mm beam at 10-month and 12-month time-points (Fig. 5E) and increased the latency to traverse the entire 5mm beam length at 8 months (Fig. 5E,G).
The same cohort of mice was also tested on the accelerating rotarod task. At 6 months of age, 58Q mice injected with RA performed significantly better on the rotarod task than 58Q-DM and 58Q-NI mice (Fig. 6B). There was no significant difference between the performance of 58Q mice expressing RA and WT mice expressing DM (Fig. 6A–B). 58Q mice expressing RA continued to perform significantly better on the rotarod than 58Q mice expressing DM at the 8, 10 and 12 month time-points. WT mice expressing RA had a trend to perform worse on the rotarod than WT-NI mice and WT-DM mice (Fig. 6A). However, this difference was not statistically significant. We monitored the body weight of all the mice in this study and did not find any significant differences between the 6 groups at all ages (data not shown). From these experiments, we concluded that partial and chronic suppression of IICR in PCs of 58Q mice resulted in significantly improved beamwalk and rotarod performance and that the same manipulation in PCs of WT mice resulted in somewhat impaired beamwalk and rotarod performance.
To assess if the benefit of chronic 5PP-RA overexpression also extended to preventing SCA2-58Q pathology, this cohort of mice was sacrificed at 12 months old for pathological analysis. We previously reported that calbindin-staining and stereological counting of PCs only detected a 15% loss of PCs in 12-month old 58Q mice (Liu et al., 2009). An alternative method used to analyze PC health status is by the quantification of dark cell degeneration (DCD). DCD is a form of cell death induced by excitotoxicity, which has been used previously to assess the health state of PCs in mouse models of SCA7 and SCA28 (Barenberg et al., 2001; Custer et al., 2006; Maltecca et al., 2009; Strahlendorf et al., 2003). Using this method in previous studies we demonstrated that DCD quantification is a more sensitive way to analyze PC death in SCA2 mice (Kasumu and Bezprozvanny, 2010). We previously reported that at 12 months of age, less than 20% of PCs in 58Q mice appear normal, compared to 70% of PCs in age-matched nontransgenic mice (Kasumu and Bezprozvanny, 2010).
To analyze DCD, cerebellar sections from each of 6 experimental groups of mice (Table 1) were processed for transmission electron microscopy (TEM) and the number of normal, moderately and severely degenerated PCs was quantified. According to (Custer et al., 2006; Kasumu and Bezprozvanny, 2010), PCs spherical in shape and with regular alignment in the PC layer were classified as “normal” (Fig. 7A). PCs with slight shrinkage compared to surrounding PCs and with moderately electron-dense cytosol that is not as dark as nucleus were classified as “moderate” (Fig. 7A). PCs with markedly shrunken and electron-dense cytosol with similarly darkened nucleus were classified as “severe” (Fig. 7A). In our analysis, we discovered that the expression of RA in 58Q mice significantly increased the percentage of normal PCs when compared to DM-injected 58Q mice (Fig. 7B). On average, in samples from RA-injected 58Q mice 54% of PCs were normal, 29% were moderately degenerated and 17% were severely degenerated (n = 218 PCs; Fig. 7B, Table 1). In contrast, in samples from DM-injected 58Q mice 12% of PCs were normal, 47% were moderately degenerated and 41% were severely degenerated (n = 185 PCs; Fig. 7B; Table 1). When compared to 58Q-DM mice, the increase in the fraction of “normal” cells and the reduction in the fraction of “severely degenerated cells in 58Q-RA mice were statistically significant (p < 0.05; Fig. 7B, Table 1). Interestingly, and in sharp contrast to 58Q mice, RA expression in WT PCs resulted in obvious worsening of DCD phenotype (Fig. 7B; Table 1).
It is generally assumed that neuronal dysfunction prior to symptom presentation is a common occurrence in cerebellar ataxias (Alvina and Khodakhah, 2010a; Alvina and Khodakhah, 2010b; Kasumu and Bezprozvanny, 2010; Mark et al., 2011; Shakkottai et al., 2011; Walter et al., 2006). However, only a few functional studies of PCs in ataxia mouse models have been performed. In a recent study, PCs in pre-symptomatic SCA3 mice were reported to fire less tonically than PCs in WT mice (Shakkottai et al., 2011). PCs in mouse models of episodic ataxia fire less regularly than PCs in age-matched wildtype mice (Alvina and Khodakhah, 2010a; Alvina and Khodakhah, 2010b; Walter et al., 2006). Recent studies also demonstrated significantly reduced firing frequency of PCs in aging SCA2-127Q mice (Hansen at al, submitted). These findings agree with results that we obtained for PC spontaneous firing in the SCA2-58Q mouse model (Fig 1). We discovered that the fraction of tonically firing PCs is lower in SCA2 mice when compared with age-matched wild type mice (Fig 1C). We further discovered that for the aging tonically firing SCA2 PC cells the frequency of firing is reduced (Fig 1D) and the correlation of variability of interspike intervals is increased (Fig 1E) when compared to age-matched wild type cells. Interestingly, the age of onset of electrophysiological abnormalities in 58Q PCs at 24 weeks of age (Fig 1C–D) closely mirrors the age of onset of behavioral symptoms in SCA2-58Q mice in previous (Liu et al., 2009) and present (Fig 5 and Fig 6) studies. The findings obtained in our study (Fig 1) and in previous analyses of ataxic mouse models (Alvina and Khodakhah, 2010a; Alvina and Khodakhah, 2010b; Shakkottai et al., 2011) support the hypothesis that the burst firing of PCs reflects the dysfunctional state of these cells and is directly linked with ataxic symptoms. This conclusion is in agreement with the well-established importance of PC firing for maintaining cerebellar timing and function (Walter et al., 2006).
What is the cause of neuronal dysfunction and death in cerebellar ataxias? In previous studies, we suggested that supranormal InsP3R-mediated Ca2+ release from the ER may play an important role in pathogenesis of SCA2 and SCA3 (Chen et al., 2008; Liu et al., 2009). Based on these results we proposed a “calcium hypothesis” of cerebellar ataxias (Kasumu and Bezprozvanny, 2010). Genetic analysis was used by another group to independently suggest an importance of InsP3-mediated Ca2+ signaling in the pathogenesis of many SCAs (Schorge et al., 2010). We also demonstrated the neuroprotective effects of dantrolene, a Ca2+ stabilizer, in SCA3 and SCA2 mouse models (Chen et al., 2008; Liu et al., 2009). In the present study, we utilized a highly specific molecular tool to further test our “Ca2+ hypothesis” in the SCA2 mouse model. By using an adeno-associated viral approach, we achieved stable expression of inositol 1,4,5-triphosphate 5-phosphatase (5PP) enzyme in PCs of SCA2 transgenic mouse model. The 5PP enzyme converts the active messenger InsP3 to its inactive InsP2 form. As it has been demonstrated previously (Kanemaru et al., 2007), heterologous overexpression of 5PP results in potent inhibition of InsP3R-mediated Ca2+ signaling in cells. In our experiments, we used the partially active R343A mutant of 5PP (RA) with reduced enzymatic activity. We reasoned that chronic expression of RA mutant in SCA2 PCs will normalize supranormal InsP3R-mediated Ca2+ signals in SCA2 PCs without completely suppressing them. The catalytically inactive double mutant 5PP-DM was used in our studies as a negative control. Consistent with our predictions, we found that chronic expression of RA in SCA2 PCs normalized their electrophysiological phenotype (Fig. 4), alleviated motor coordination deficit of SCA2 mice in beamwalk and rotarod assays (Fig. 5 and and6),6), and prevented DCD form of PC death in aging SCA2 mice (Fig 7, Table 1). These effects required the enzymatic activity of 5PP, as the chronic expression of the catalytically dead R343A/R350A mutant 5PP (DM) had no effect in behavioral assays with SCA2 mice (Fig. 5 and and6)6) and did not protect SCA2 PC cells in DCD assay (Fig. 7). These findings provided strong support for the “calcium hypothesis of SCA2” and suggested that partial inhibition of InsP3-mediated Ca2+ signaling could provide therapeutic benefit for the patients afflicted with SCA2 and possibly other SCAs
Autophagy plays an important role in neurodegenerative disorders (Ravikumar et al., 2010). Previous studies have shown that decreasing total InsP3 levels with agents such as Lithium (inhibits inositol recycling, (Shakkottai et al., 2001)), L-690330 (inhibits inositol recycling, (Simon et al., 2010)), valproic acid (inhibits inositol biosynthesis, (Kuznetsov et al., 2006)) and cytosolicInsP3 kinase A (diminishes total InsP3, (Simon et al., 2010)) induces autophagy. Studies have also shown that InsP3R inhibition with Xestospongin B (Ito, 2002) or InsP3R knockdown by siRNA (Fouquet et al., 2011) strongly stimulated autophagy. Thus, it is possible that chronic overexpression of 5PP in our experiments acted by stimulating autophagy in SCA2 PCs by reducing InsP3 levels. The 5PP intervention functions by hydrolyzing InsP3 after it is produced (Kanemaru et al., 2007). Thus, 5PP overexpression only decreases the lifetime of cytosolic InsP3 without affecting InsP3 production or changing levels of phosphoinositides present in the membrane. There is no evidence in the literature for the direct involvement of InsP3 in up-regulating autophagy, and the inhibition of autophagy appears to be another downstream mechanism of excessive Ca2+ release (Brady et al., 2007; Decuypere et al., 2011a; Decuypere et al., 2011b; Hosy et al., 2011; Laver and Lamb, 1998). Therefore, beneficial effects of our 5PP intervention in SCA2-58Q mice arose primarily from suppressing excessive Ca2+ release from the ER in PCs cells, with possible downstream effects of Ca2+ on the autophagy along with effects of Ca2+ on mitochondrial dysfunction, calpain activation and other Ca2+-dependent cellular processes.
Our results indicate that excessive Ca2+ release from InsP3R-sensitive Ca2+ stores is likely to play a key role in dysfunction and eventual death of PCs in SCA2. This conclusion is consistent with early-onset PC degeneration and the ataxic phenotype in Inpp5a (5PP) knockout mice (Andy W Yang; Andrew J Sachs; Emily M Strunk; Arne M Nystuen, SfN-2011 abstract). The 5PP enzyme is highly expressed in cerebellar PCs; thus, it is likely that knockout of 5PP causes delayed termination of InsP3 signals and supranormal Ca2+ release. Remarkably, insufficient InsP3R-mediated Ca2+ signaling also leads to the ataxic phenotype and PC degeneration, such as observed in SCA15/16 patients haploinsufficient for the InsP3R1 gene (Hara et al., 2008; Iwaki et al., 2008; van de Leemput et al., 2007). An ataxic phenotype is also observed in opt mice with reduced levels of InsP3R1 protein (Street et al., 1997) and a severe ataxia is observed in InsP3R1 knockout mice (Matsumoto et al., 1996). These findings indicate that the reduced Ca2+ release via InsP3R1 also leads to PC dysfunction and ataxic phenotype. Some of the data in the present manuscript support this conclusion. Despite using the RA version of 5PP with reduced activity in our experiments, we observed that chronic expression of RA in wildtype PC cells resulted in mild impairment in the precision of PC firing (Fig 4G), impaired beam walk and rotarod performance of wild type mice in behavioral studies (Fig 5 and and6)6) and a reduced fraction of normal cells in DCD analysis of wild type mice (Fig 7B, Table 1). Although relatively mild, these effects were in sharp contrast with the beneficial effects of RA expression in SCA2-58Q mice (Fig. 4–7, Table 1).
We conclude from these findings that there is a relatively narrow range of optimal InsP3-mediated Ca2+ signaling that is compatible with proper function and long-term survival of PCs. Deviation from this optimal range in either direction of InsP3-mediated Ca2+ signaling results in PC dysfunction and an ataxic phenotype. According to this model (Fig 8), supranormal InsP3-mediated Ca2+ signals are responsible for PC dysfunction and ataxic phenotype in SCA2 patients, as well as in Inpp5a knockout mice (Fig 8). The subnormal InsP3-mediated Ca2+ signals is likely responsible for PC dysfunction and the ataxic phenotype in SCA15/16 patients and InsP3R1 knockout mice (Fig 8). Chronic expression of 5PP-RA enzyme in our studies shifted Ca2+ signals in SCA-58Q PCs to a normal range and rescued the SCA2 phenotype (Fig 8). However, chronic expression of the same enzyme in WT PC cells in our studies shifted Ca2+ signals to subnormal range, resulting is detrimental effects (Fig 8). This hypothesis agrees well with independent genetic evidence that placed InsP3R in the “eye of the storm” of pathogenesis for many SCAs (Schorge et al., 2010).
We thank Ying Li for help with electrophysiological experiments and Leah Taylor for administrative assistance. We thank Stefan Pulst (Univ of Utah) for generously providing us with SCA2 mouse model and Dr. Masamitsu Iino (University of Tokyo) for generously providing 5PP expression constructs. We thank Dr. Beverly Davidson, Maria L. Scheel and the staff of the University of Iowa Gene Transfer Vector Core for help with AAV production. We thank Dr Melanie Mark (Ruhr University Bochum) and Drs Tom Otis and Meera Pratap (UCLA) for advice on electrophysiological recordings. We thank Stefan Pulst and Tom Otis for comments on the manuscript. AWK is a Howard Hughes Medical Institute Med into Grad scholar. IB is a holder of the Carl J. and Hortense M. Thomsen Chair in Alzheimer’s Disease Research. These studies were supported by the R01NS056224, R01NS38082 and R01NS074376 NIH grants (IB) and by the contract with the Russian Ministry of Science 14.740.11.0924 (IB).
AUTHOR CONTRIBUTIONSA.W.K. and I.B. designed research; A.W.K, X.L., P.E., and D.V. performed research; A.W.K. and P.E. analyzed the data. The manuscript was written by A.W.K with assistance from I.B.
COMPETING FINANCIAL INTERESTS
The authors declare no competing financial interests.