Development of Purkinje cell degeneration in a knockin mouse model reveals lysosomal involvement in the pathogenesis of SCA6

Abstract

Spinocerebellar ataxia type 6 (SCA6) is a neurodegenerative disease caused by the expansion of a polyglutamine tract in the Ca_v2.1 voltage-gated calcium channel. To elucidate how the expanded polyglutamine tract in this plasma membrane protein causes the disease, we created a unique knockin mouse model that modestly overexpressed the mutant transcripts under the control of an endogenous promoter (MPI-118Q). MPI-118Q mice faithfully recapitulated many features of SCA6, including selective Purkinje cell degeneration. Surprisingly, analysis of inclusion formation in the mutant Purkinje cells indicated the lysosomal localization of accumulated mutant Ca_v2.1 channels in the absence of autophagic response. The lack of cathepsin B, a major lysosomal cysteine proteinase, exacerbated the loss of Purkinje cells and was accompanied by an acceleration of inclusion formation in this model. Thus, the pathogenic mechanism of SCA6 involves the endolysosomal degradation pathway, and unique pathological features of this model further illustrate the pivotal role of protein context in the pathogenesis of polyglutamine diseases.

Spinocerebellar ataxia type 6 (SCA6) is a late-onset autosomal dominant neurodegenerative disease characterized by a loss of motor coordination and balance. An important feature of SCA6 pathology is the selective degeneration of Purkinje cells (PCs) (1, 2). Besides SCA6, the polyglutamine (polyQ) diseases include spinobulbar muscular atrophy, dentatorubral-pallidoluysian atrophy, Huntington disease (HD), and SCAs 1, 2, 3, 7, and 17 (3). The pathogenesis of polyQ diseases is complex, but it has been suggested that the expanded polyQ protein causes disease through a toxic gain-of-function mechanism. More recently, protein context and normal protein activity have been shown to be critical in the pathogenesis of these diseases (4, 5). In SCA6, the CAG repeat encodes a polyQ tract in the cytoplasmic tail domain of the Ca_v2.1 channel, a pore-forming subunit of the P/Q type voltage-gated calcium channels (6). Alternative splicing occurring at the beginning of the final exon (exon 47) of CACNA1A produces two major Ca_v2.1 isoforms, MPI and MPc (7). These two isoforms differ in their splice acceptor sites, such that MPI translates a polyQ tract from the CAG coding repeat, whereas MPc splices to an immediate stop codon, resulting in a shorter cytoplasmic tail that lacks the polyQ tract.

Several features of SCA6 set it apart from the other polyQ diseases. Pathologically, the formation of ubiquitinated intranuclear neuronal inclusions (NIs) is a common feature in the majority of polyQ diseases (8). However, mutant Ca_v2.1 channels form inclusions in the cytoplasm of SCA6 PCs that typically lack ubiquitin immunoreactivity (9).

Recently, we generated a knockin (KI) mouse model (Sca6^84Q) carrying an 84 CAG repeat tract in the mouse Cacna1a locus (10). Sca6^84Q mice recapitulate several features of SCA6, including the NI formation in PCs. Nonetheless, the Sca6^84Q mice did not acquire clear neurodegenerative phenotypes, even at later ages. For elucidation of the molecular pathways involved in neurodegeneration, it would be helpful to engender animals that develop PC degeneration that mirrors the degeneration seen in SCA6. In an effort to accelerate the disease progression in mice by enhancing the expression of mutant channels, we created two unique lines of KI mice (MPI-11Q and MPI-118Q) that carried a splice-site mutation so that only the MPI type of splicing occurred at the exon 47 locus.

Results

Generation of Sca6 MPI KI Mice.

To create the Sca6 KI mice that develop PC degeneration, we decided to generate two KI lines that showed modestly enhanced expression of the polyQ-containing Ca_v2.1 channels (MPI KI mice). In addition to the CAG repeat tract of either normal (11 pure CAG repeats for MPI-11Q) or expanded (114 CAG repeats multiply interrupted by 4 CAA units for MPI-118Q) size and sequences immediately flanking the repeat from CACNA1A, we introduced a splice-site mutation (Fig. 1A) that precludes the production of both the MPc and MPII isoforms into Cacna1a by homologous recombination in embryonic stem cells. Germ-line transmission of the targeted alleles in the offspring was confirmed by Southern blot analysis (Fig. S1A). In wild-type (WT) cerebellum, alternative 3′ acceptor use in the exon 47 locus results in the production of the MPI, MPc, and MPII isoforms, with each occupying roughly 60%, 40%, and less than 1% of the whole Ca_v2.1 transcript, respectively. Thus, both the MPI-11Q and MPI-118Q isoforms were expected to be produced from their respective mutant alleles at an ∼67% greater rate than the WT MPI isoform from the normal allele.

Fig. 1.

Generation of Sca6 MPI KI mice. (A) Targeting scheme representing the targeting construct showing the splice-site mutation and polyQ tract, the WT endogenous allele, and the predicted structure of the mutant SCA6-MPI allele generated by homologous recombination and a Cre-mediated excision event. (B) Western blot analysis performed on the cytoplasmic fraction of cerebellar lysates from 6-wk-old WT, MPI^11Q/11Q, and MPI^118Q/118Q mice, all blotted against anti-Ca_v2.1 (Top), A6RPT-polyQ (Middle, Top), 1C2 (Middle, Bottom) and β-tubulin (Bottom, loading control). (C) Analysis of the Sca6 MPI-118Q KI mice on the accelerating Rotarod apparatus. Mice were trained for four trials per day for 4 d (D1–D4). Error bars indicate SEM. (D) Analysis of the Sca6 MPI-11Q KI mice. Error bars indicate SEM. (E) Current density–voltage (I–V) relationships (Left) and inactivation curves (Right) for WT, MPI^118Q/118Q, and MPI^11Q/11Q PCs dissociated from 35- to 42-d-old mice. I–V curves are fitted as in Fig. S2. Inactivation curves are fitted to the Boltzmann’s equation. Reversal potential (E_rev), half-activation or half-inactivation voltage (E_0.5), slope factor (k), and number of cells (n) were summarized as Insets. Error bars show SEM. All recordings were done in PCs dissociated from anterior lobe.

To compare the expression levels of the entire Ca_v2.1 transcripts in the cerebellum, we performed quantitative RT-PCR (qRT-PCR) by using primers designed to amplify exon 3 and exon 4. Similar levels of expression were detected among the WT, homozygous MPI-11Q (MPI^11Q/11Q), and homozygous MPI-118Q (MPI^118Q/118Q) mice, suggesting that the mutations did not affect the transcription of Cacna1a.

We next verified the absence of both the MPc and the MPII transcripts in the homozygous MPI KI cerebella. cDNAs derived from the cerebellar extracts were amplified and subcloned into plasmids. The splice variants were then identified by sequencing each individual clone. In this manner, we confirmed the absence of both the MPc and MPII splice isoforms in the homozygous MPI KI cerebella, whereas 44% of the sequenced clones were found to be MPc in the WT cerebella. These data suggest that the expression levels of the mutant MPI transcripts in the homozygous MPI KI cerebella were nearly 80% greater than those of WT MPI transcripts.

The expression of Ca_v2.1 channels was assessed by immunoblotting with anti-Ca_v2.1 antibodies (Abs) raised against the intracellular loop between domains II and III (Alamone Labs). The cerebellar extracts from the WT mice gave a main, broad Ca_v2.1 immunoreactivity (IR) at ∼200 kDa, whereas the mutant Ca_v2.1 showed less mobility in the lanes of extracts from mutant animals, as expected from the size of the polyQ stretch and its immediate human-specific sequence (Fig. 1B). We also examined the expression of humanized Ca_v2.1 by using an anti–A6RPT-polyQ Ab (11) raised against the C-terminal of human Ca_v2.1 and for the expression of the polyQ-expanded Ca_v2.1 by using Abs specific for expanded polyQ (1C2) (Fig. 1B and Fig. S1B). Cerebellar extracts from the MPI^118Q/118Q mice gave an intense 1C2 IR signal, showing an apparent molecular mass of ∼250 kDa, suggesting that the polyQ-expanded Ca_v2.1 channel was expressed in the cerebellum of the MPI^118Q/118Q mice. Recently, it was reported that the expanded polyQ-containing carboxyl-terminal fragment, which showed molecular masses ranging from 75 to 110 kDa depending on the CAG repeat size, was mislocalized to the nuclear fraction in the SCA6 cerebellum (11). However, we could not detect such a fragment in the cerebellum of the MPI^118Q/118Q mice.

MPI-118Q Mice Exhibit Age- and Dosage-Dependent Motor Incoordination.

Both the homozygous and heterozygous MPI-118Q and MPI-11Q mice were fertile and had normal lifespans. At 6 wk of age, the MPI^118Q/118Q KI mice began to show ataxic gait, which progressively deteriorated as they aged (Movie S1). Although the MPI^118Q/+ mice did not show an obvious phenotype at a young age, they also began to show similar gait abnormalities at ∼1 y of age, indicating a gene dosage effect. In contrast, both the MPI^11Q/11Q and MPI^11Q/+ mice were indistinguishable from their WT littermates, even at 18 mo of age. We evaluated the gait patterns of 15-wk-old animals by using a CatWalk automated system (12). The footprints of the MPI^118Q/118Q mice revealed wobbling and a short-stepped gait (Fig. S1C).

To assess motor impairment in more detail, the MPI KI mice were tested on an accelerating Rotarod at several time points. At 6 wk of age, the MPI^118Q/118Q mice performed worse than the WT mice, spending significantly less time (Fig. 1C; P < 0.01 by one-way ANOVA with repeated measures) on the Rotarod. The MPI^118Q/+ mice performed similarly to their WT littermates at 6, 12, and 52 wk of age. At 18 mo of age, however, the performance of the MPI^118Q/+ mice became significantly worse than that of the WT mice (Fig. 1C). In contrast, both the MPI^11Q/11Q and MPI^11Q/+ mice performed as well as their WT littermates up to 52 wk of age (Fig. 1D). These results suggest that the mice bearing the MPI-118Q allele develop impairments in motor coordination in an age- and gene dosage-dependent manner.

Electrophysiology of PCs from Young MPI KI Mice.

To test the properties of the mutant MPI channels in vivo, we recorded whole-cell Ca²⁺ channel currents in acutely dissociated PCs by using 2 mM Ba²⁺ as the charge carrier (SI Methods). Fig. 1E and Fig. S2 show the results obtained from PCs isolated from postnatal (P)35–42 and P28–32 animals, respectively. Both of these two independent studies failed to reveal any significant differences in the averaged current–voltage (I-V) relationships among the WT, MP^11Q/11Q, and MPI^118Q/118Q mice. (Fig. 1E and Fig. S2 B and C). For all three groups of P28–32 animals, the majority of inward currents were blocked by 100 nM ω-Aga IVA, a specific peptide blocker of P/Q-type calcium channels (Fig. S2 D and E). These data suggest that the polyQ expansion did not affect the basic properties of the Ca_v2.1 channel in the PCs of MPI KI mice.

Next, we examined the influence of the polyQ expansion on the firing pattern of PCs in brain slices prepared from WT and MPI^118Q/118Q mice (P38–41) (SI Methods). The PCs of both groups showed Na⁺ bursts in response to an injection of a small depolarizing current (Fig. S3A). In 4 of 18 WT PCs, the injection of large depolarizing current evoked Ca²⁺ spikes, which often led to oscillating behavior in the Na⁺ spike bursts (Fig. S3B) (13). Cd²⁺ terminated the bursting activity (Fig. S3C), which resembled the firing pattern of the MPI^118Q/118Q PCs (Fig. S3D). Zero of the 22 MPI^118Q/118Q PCs we recorded exhibited the oscillatory pattern. The change of firing pattern likely reflects the dendritic degeneration seen in MPI^118Q/118Q PCs (see below).

Progressive PC Degeneration in MPI-118Q Cerebella.

To examine whether the manifestation of gait ataxia in the young MPI^118Q/118Q mice might be associated with any pathological changes in the PCs, an inspection of cerebellar sections with anticalbindin immunofluorescence was performed. In the 6-wk-old MPI^118Q/118Q cerebellum, we could not detect significant cell loss, but the size of the PC cell bodies appeared smaller than those in the WT cerebellum (Fig. 2A). We also observed reduced dendritic arborization and numerous axonal swellings of the PCs in the mutant cerebellum, suggesting that neuritic degeneration of the PCs took place before PC loss in the MPI^118Q/118Q cerebellum. We next evaluated temporal changes in the number of PCs in midsagittal cerebellar sections stained with anticalbindin Abs. The number of PCs in the MPI^118Q/118Q cerebellum steeply declined after 6 wk of age (Fig. 2B), indicating an age-dependent progressive loss of PCs. Although no significant PC loss was detected in the MPI^118Q/+ cerebellum for up to 30 wk, there were significantly (P < 0.01) fewer PCs in the 100-wk-old MPI^118Q/+ cerebellum compared with the WT cerebellum. Both a TUNEL assay and cleaved caspase-3 immunohistochemistry revealed scattered positively stained PCs in the MPI^118Q/118Q cerebellum (Fig. S4A), indicating apoptotic features of PC death. In contrast, we could not detect any significant neuritic changes or PC loss in the cerebellum of the MPI^11Q/11Q or MPI^11Q/+ mice for up to 14 mo of age (Fig. S4B). Moreover, we were unable to detect any significant loss of other types of neurons on HE-stained brain sections of aged MPI^118Q/118Q mice.

Fig. 2.

Neuropathology of Sca6 MPI KI mice. (A) Confocal microscopy of anticalbindin-labeled PCs of a 7-wk-old MPI^118Q/118Q mouse showed reduced dendritic arborization and swollen axons (arrowheads). (B) Quantitative analysis of remaining PCs in midsagittal cerebellar sections. Error bars indicate SEM. (C) Immunohistochemical analysis using A6RPT-polyQ Ab. Cytoplasmic inclusion formation in PCs was detected in the cerebellum of 7-wk-old (Center) and 25-wk-old (Right) MPI^118Q/118Q mice. (D) Double immunofluorescence analysis using anticalbindin (red) and A6RPT-polyQ (green) Abs. In the PCs of 9-wk-old MPI^118Q/118Q mice, but not in those of 9-wk-old MPI^11Q/11Q mice, mutant channels formed both cytoplasmic and dendritic inclusions. (Scale bars: A, 200 μm; C, 100 μm; D, 50 μm.)

NIs Colocalize with Lysosome Markers in MPI-118Q PCs.

We analyzed the formation of NIs in the mutant PCs by immunohistochemistry using an A6RPT-polyQ Ab (10, 11). This Ab reacts specifically with humanized Ca_v2.1 and revealed NI formation in the cytoplasm of the Sca6^84Q/84Q Purkinje neurons (10). Although no staining was observed in the 4-wk-old MPI^118Q/118Q cerebellum, many PCs showed the formation of cytoplasmic inclusions in the 7-wk-old MPI^118Q/118Q cerebellum (Fig. 2C). Scattered staining, which corresponded to an accumulation of mutant Ca_v2.1 in the dendrites of PCs, was also observed in the molecular layer of the 7-wk-old MPI^118Q/118Q cerebellum (Fig. 2D). These changes became more evident as the animals aged, and similar age-dependent changes were also observed in the MPI^118Q/+ cerebellum after 15 wk of age, but not in the MPI^11Q/11Q cerebellum (Fig. S4 C–E).

The formation of cytoplasmic inclusions is a key feature of SCA6 and contrasts with the NIs seen in most of other polyQ diseases in the sense that it appears to lack ubiquitin IR (9). Consistent with this observation, antiubiquitin staining failed to detect any inclusions in the MPI^118Q/118Q PCs (Fig. S5A). To further analyze the subcellular localization of NIs in the MPI^118Q/118Q PCs, we performed double immunofluorescence studies by using 1C2 and Abs to several organelle marker proteins. Interestingly, we found that the 1C2-positive showed coimmunofluorescence (co-IF) with the lysosomal marker cathepsin D (CatD) (85 ± 0.83%, n = 3; Fig. 3A). These inclusions also revealed co-IF with other lysosomal markers such as cathepsin B (CatB) and LIMP2 (Fig. 3A), but not with KDEL, a marker of endoplasmic reticulum, or Tom20, a mitochondrial marker (Fig. S5 B and C). We also examined the expression of CHOP, a well-known marker of ER stress, in the MPI^118Q/118Q cerebellum (Fig. S6), but failed to detect significant differences in the patterns of CHOP expression between the WT and the MPI^118Q/118Q mice. These results strongly suggest that the mutant Ca_v2.1 accumulated in the lysosome of MPI-118Q PCs without undergoing polyubiquitination or giving rise to ER stress. The inclusions in the molecular layer were also costained with lysosomal markers, although lysosomes are normally limited within the soma (Discussion). Ultrastructural analysis of the 15-wk-old MPI^118Q/118Q cerebellum by transmission electron microscopy (TEM) revealed that most of the remaining PCs had irregularly shaped nuclei and contained many structurally abnormal mitochondria, such as those showing broken cristae (Fig. 3B). We also observed many electron-dense deposits in the cytoplasm of the MPI^118Q/118Q PCs (Fig. 3B) and in the molecular layer of the cerebellum (Fig. 3C). Immunoelectron microscopy revealed the localization of Lamp1, a major lysosomal membrane protein, in the large electron-dense structures (Fig. 3D and SI Methods). Similar structures were also observed in the PCs of 1-y-old Sca6^84Q/84Q mice (Fig. S7A). We also conducted a double-immunofluorescence analysis on the postmortem cerebellar tissues of an SCA6 patient who had 22 CAG repeats in her mutant allele (11) and found that many of the 1C2-positive inclusions showed co-IF with LIMP2 (38 ± 6.4%, n = 5; Fig. 3E).

Fig. 3.

Lysosomal markers colocalize with NIs in MPI^118Q/118Q PCs. (A) Double immunofluorescence microscopy for polyQ (1C2, red) and CatD, CatB, and LIMP2 (green) demonstrates that the polyQ-containing inclusions colocalize with lysosomal marker proteins. DAPI was used for nuclear staining. (B) Representative EM images of PCs. Large electron dense structures (arrowhead) were detected in the PC cytoplasm of 15-wk-old MPI^118Q/118Q mice. An image of a PC of a 15-wk-old WT mouse is shown as a control (Left). Scale bars for the left and the middle panels indicate. A magnified image of the electron dense structure is shown in the right. (C) Representative EM images of the molecular layer (D) A representative EM image of ultrathin cryosections of a PC stained with an anti-Lamp1 Ab (10-nm colloidal gold particles). In the PC cytoplasm of a 10-wk-old MPI^118Q/118Q mouse, Lamp1 is specifically localized to the lysosomes containing electron-dense structures (asterisks). (Scale bars: A, 50 μm; B Left and Center, 5 μm; B Right, C, and D, 500 nm.) (E) The 1C2-postitive NIs are shown co-IF for LIMP2 in the postmortem tissue of an SCA6 patient.

Autophagic Response in MPI-118Q PCs.

The accumulation of autophagosomes has been observed in several neurodegenerative diseases, such as HD (14, 15), and in multiple sulfatase deficiency, a lysosomal storage disease in which an impairment of the autophagic flux has been demonstrated (16, 17). We could, however, hardly detect autophagic vacuoles in the cytoplasm of the MPI^118Q/118Q PCs by TEM. Both the LC3 immunoblot analysis and LC3 immunohistochemistry of the cerebellum consistently failed to reveal significant differences between the 7-wk-old MPI^118Q/118Q mice and their wild-type littermates (Fig. S8 A and D). Moreover, the expression of phospho-S6, a protein whose abundance is inversely related to the level of mTOR-dependent cellular autophagy, was not significantly altered in the 7-wk-old MPI^118Q/118Q cerebellum (Fig. S8 B and C). Together, these data indicated that the autophagic response was not activated in degenerating MPI^118Q/118Q PCs.

Lack of CatB Exacerbates Motor Impairment in MPI-118Q KI Mice.

The cellular mechanisms implicated in the formation of mutant Ca_v2.1 inclusion and its role in SCA6 pathogenesis are not well understood. Our morphological analysis strongly suggests lysosomal involvement in the formation of inclusions in the MPI-118Q PCs. To evaluate the role of the lysosomal degradation pathway in SCA6 and to determine how general impairment of lysosomal function affects disease pathogenesis, we cross-bred the MPI-118Q mice with well-characterized mutant animals lacking CatB (ctsb^−/−). CatB is one of the major endosomal/lysosomal cysteine proteases in the cerebellum, but ctsb^−/− mice do not have an obvious phenotype (Fig. S9) (18). Matings between male and female MPI^118Q/+/ctsb^+/− mice produced litters with the expected ratios of each genotype. Both the MPI^118Q/118Q; ctsb^−/− and the MPI^118Q/118Q; ctsb^+/− mice developed normally and were indistinguishable from littermates of the other genotypes with respect to cage behavior until P21.

To examine the effects of reduced expression of CatB on motor coordination, we tested the animals on the Rotarod at 9 wk of age (Fig. 4A). As expected, ctsb^−/− mice performed similarly to WT littermates, whereas the motor impairments of the MPI^118Q/118Q; ctsb^+/+ mice were reproduced in these litters. Interestingly, the performance of both the MPI^118Q/118Q; ctsb^−/− and the MPI^118Q/118Q; ctsb^+/− mice was significantly poorer than the MPI^118Q/118Q; ctsb^+/+ mice (Fig. 4A; *P < 0.05, **P < 0.01, by one-way ANOVA with repeated measures followed by a Tukey–Kramer post hoc test). These results indicated that the reduced expression levels of CatB deteriorated motor coordination in the MPI^118Q/118Q mice.

Fig. 4.

The lack of CatB exacerbated PC death and motor impairment in MPI^118Q/118Q mice. (A) Rotarod analysis of the double mutant mice. Mice were trained for five trials per day for 5 d (D1–D5). Error bars indicate SEM. n = 7–15. (B) Anticalbindin immunohistochemistry on cerebellar sections of 9-wk-old WT, MPI^118Q/118Q; ctsb^+/+, and MPI^118Q/118Q; ctsb^−/− mice. (Lower) Quantitative analysis shown that a decrease in the number of PCs was more pronounced in the MPI^118Q/118Q; ctsb^−/− mice compared with the MPI^118Q/118Q; ctsb^+/+ mice at this age (P < 0.05 by Student’s t test). Error bars indicate SEM. (C) Immunohistochemical analysis using A6RPT-polyQ Ab on cerebellar sections of 5-wk-old MPI^118Q/118Q; ctsb^+/+ and MPI^118Q/118Q; ctsb^−/− mice. (Scale bars: B, 1 mm; C, 20 μm.) (D) The percentage of inclusion-harboring PCs is shown for 5-wk-old mice. Four animals were counted for both groups. Error bars indicate SEM. *P < 0.01.

Inclusion Formation and Neurodegeneration in MPI^118Q/118Q Mice Lacking CatB.

To evaluate the effects of CatB deficiency on PC loss in the MPI^118Q/118Q mice, we counted the number of remaining PCs in midsagittal cerebellar sections stained with anticalbindin Abs. Although there was no significant neurodegeneration at 4 wk of age (Fig. S10), the MPI^118Q/118Q; ctsb^−/− mice showed a marked loss of PCs, which was most prominent in the anterior lobe at 9 wk of age. In fact, the number of remaining PCs was significantly reduced in the MPI^118Q/118Q; ctsb^−/− cerebellum compared with those in the MPI^118Q/118Q; ctsb^+/+ cerebellum (Fig. 4B), suggesting that the lack of CatB exacerbated PC loss in the MPI-118Q KI mice. Consistent with these results, significant PC loss was observed in the cerebellum of 30-wk-old Sca6^84Q/84Q; ctsb^−/− mice (Fig. S7 B and C) but not in age-matched Sca6^84Q/84Q; ctsb^+/+ mice.

Next, we examined the effects of the ctsb mutation on the formation of inclusions in the MPI^118Q/118Q cerebellum. Similar to the MPI^118Q/118Q; ctsb^+/+ PCs, the 1C2-positive inclusions in the cerebellum of MPI^118Q/118Q; ctsb^−/− mice showed co-IF with lysosomal markers. We evaluated the abundance of PC inclusions by immunohistochemical analysis using A6RPT-polyQ Abs and found that the percentage of PCs harboring cytoplasmic inclusions in the MPI^118Q/118Q; ctsb^−/− mice was significantly (P < 0.01, by Student's t test) larger than that in the MPI^118Q/118Q; ctsb^+/+ mice at 5 wk of age, a time at which neither group developed significant PC loss (Fig. 4 C and D). These results suggest that the lack of CatB expression accelerated the formation of mutant Ca_v2.1 inclusions in the MPI^118Q/118Q cerebellum. Overall, our data indicate that the lack of CatB expression accelerates mutant Ca_v2.1 inclusion formation and increases PC loss in the MPI^118Q/118Q cerebellum.

Discussion

The MPI^118Q/118Q mice expressed a polyQ-expanded humanized Ca_v2.1 under the control of its endogenous promoter at modestly enhanced levels and faithfully recapitulated many phenotypic features of SCA6. The animals developed progressive PC degeneration, accompanied by the formation of cytoplasmic inclusions. Motor incoordination was apparent as early as 5 wk of age by using Rotarod testing and later developed into distinct gait ataxia. The MPI^118Q/+ mice developed similar key behavioral and neuropathological phenotypes at later ages than the MPI^118Q/118Q mice, indicating a gene dosage effect.

Before undergoing selective cell loss, the PCs of MPI^118Q/118Q mice developed distinctive degenerative changes that were strikingly similar to those seen in the SCA6 cerebellum. These changes include the formation of cytoplasmic inclusions, reduced dendritic arborization, and the presence of swollen axons (19). More importantly, our neuropathological analysis revealed colocalization of lysosomal markers with the 1C2-positive cytoplasmic inclusions. Together with the presence of Lamp1-positive electron-dense deposits in the PCs of the mutant mice, these data strongly suggest that mutant Ca_v2.1 forms inclusions in the lysosomes of SCA6 PCs. It is of great interest that we also detected co-IF of polyQ and lysosomal markers in deposits in the molecular layer of both the MPI^118Q/118Q and the MPI^118Q/+ cerebellum. Lysosomes are typically restricted to the soma of neurons, but the translocation of lysosomes into dendrites has been reported in neurons in AD brains and in hippocampal neurons treated with colchicine (a microtubule inhibitor) or ZPAD (a selective inhibitor of cathepsins B and L; refs. 20–22). Therefore, we speculate that the mislocalization of lysosomes caused by either cytoskeletal abnormalities or lysosomal dysfunction may be relevant to the dendritic localization of these inclusions.

PolyQ expansion is believed to alter protein conformation, making the host protein harder to degrade and resulting in the formation of insoluble aggregates. The mechanisms that control the degradation of Ca_v2.1 are not well understood. It has been shown that Ca_v1.2 and Ca_v2.2 undergo internalization and translocation to lysosome for degradation in response to activation of either the NMDA receptor or the nociceptin receptor (23, 24). Therefore, although the mechanisms by which Ca_v2.1 is internalized remain elusive, it is likely that Ca_v2.1 is also sequestered to lysosome for degradation. The apparent lysosomal localization of the mutant Ca_v2.1 inclusions most likely reflects the incomplete degradation of mutant Ca_v2.1 aggregation by the endolysosomal degradation pathway. There could be many explanations for how the lack of CatB lead to an enhancement of the formation of cytoplasmic inclusions in the MPI-118Q cerebellum. We speculate that a general impairment of lysosomal function might accelerate the accumulation of the mutant channel. Another plausible explanation is that the mutant Ca_v2.1 is a substrate for CatB and that its degradation in the lysosome is attenuated when CatB expression is down-regulated.

Although many reports studying cellular and mouse models of other polyQ diseases report that polyQ-containing inclusions are not toxic to neurons (25–27), the accumulation of mutant Ca_v2.1 in the lysosome seemed to correlate with disease progression in the SCA6 model. The lysosomal accumulation of protein substrates has also been observed in certain types of inherited lysosomal storage diseases (LSDs), such as CLN2 and CLN10. It is of note that the presence of common pathogenic mechanisms among the different LSDs has been recently proposed despite the diversity of the accumulating substrates (28). Thus, we hypothesize that the lysosomal accumulation of mutant Ca_v2.1 may cause SCA6, at least in part, through such mechanisms. The mechanisms may induce the destabilization of the lysosomal membrane and the release of lysosomal contents into the cytosol (29, 30), initiating the apoptotic pathway through lysosomal-mitochondrial cross-talk. The presence of morphologically abnormal mitochondria, as well as the activation of caspase-3 in the MPI^118Q/118Q PCs, appeared to be consistent with this model, although more direct evidence is needed.

Accumulating evidence indicates the importance of protein context and protein–protein interactions in the pathogenesis of the polyQ diseases. MPI and MPc are two major CACNA1A splice isoforms in the cerebellum, but the functional differences between the two variants are not well understood. Our electrophysiological analysis of the homozygous MPI-11Q PCs suggests that these two subtypes are not different in terms of the basic properties of the channel. However, recent studies identified several protein–protein interactions that specifically occurred on the MPI isoform and appeared to play important physiological roles in the cerebellum (31). We speculate that the expanded polyQ tract in Ca_v2.1 could alter such pivotal interactions and play important roles in the pathogenesis of the disease, because some of these interactions were shown to be repeat-length dependent in mammalian cells.

One limitation of the current study was that the length of polyQ tract expressed in the MPI-118Q KI mice was beyond the disease range. Future studies, such as generation of BAC/YAC transgenic models that develop similar phenotypes while expressing full-length Ca_v2.1 with a modest polyQ expansion, would be helpful for validating our results.

In conclusion, the MPI-118Q KI mice faithfully recapitulated the key phenotypic features of SCA6 and revealed a lysosomal contribution to SCA6 pathogenesis. Given that the lysosomal accumulation and storage of mutant Ca_v2.1 within PCs is associated with toxicity, decreasing the mutant Ca_v2.1 load either by activating the autophagic response or by lysosomal exocytosis (32) may be an effective therapeutic strategy for this disease.

Methods

Generation of Sca6 MPI KI Mice.

A 129/SvEv mouse genomic clone used for the generation of Sca6^84Q mice was used to construct a targeting vector. In brief, two splice acceptor AG sequences that were used for the production of MPc and MPII were mutated to TC (“CAG GGC AGT AGT” to “CAG GGC TCT TCT”). This mutation does not lead to any amino acid substitutions in the MPI channel. A SacII fragment that includes part of the mouse Sca6 exon 47 was replaced with a homologous fragment prepared from human Ca_v2.1 cDNAs containing either 11 pure CAG repeats or 118 repeats multiply interrupted by 4 CAA units. The interrupted repeats were used to prevent instability in the tract length. Mating the chimeric mice with mice expressing CAG-Cre (33) allowed the excision of the Neo/Tk selection cassette. The mutant mice were analyzed after a minimum of five backcrosses to C57BL/6 mice. All of the animal procedures were performed in accordance with the protocols approved by the Animal Experiment Committee of the Tokyo Medical and Dental University (0090078).

PC Counting.

Midsagittal brain sections were prepared from three to five animals for each group. After anticalbindin staining, the number of PCs in three to four adjacent sections from each brain were counted under a light microscope and were then averaged.

Statistical Analysis.

A one-way repeated measures ANOVA with a Tukey–Kramer’s post hoc test was used for the Rotarod analysis. A one-way ANOVA was used for the comparison of PC numbers. A Student’s t test was used for the biochemical analyses and for the quantitative analysis of inclusion-harboring PCs. Additional information is published in SI Methods.