Spinocerebellar ataxia type 6 (SCA6) is a dominantly inherited degenerative disorder of the cerebellum characterized by nearly selective and progressive death of Purkinje cells. The underlying mutation in SCA6 consists of an expansion of a trinucleotide CAG repeat in the 3′ region of the gene, CACNA1A, encoding the α1Asubunit of the neuronal P/Q-type voltage-gated calcium channel. Although it is known that this mutation results in an expanded tract of glutamine residues in some α1A splice forms, the distribution of these splice forms and the role of this mutation in the highly selective Purkinje cell degeneration seen in SCA6 have yet to be elucidated. Using specific antisera we demonstrate that the pathological expansion in SCA6 can potentially be expressed in multiple isoforms of the α1A subunit, and that these isoforms are abundantly expressed in the cerebellum, particularly in the Purkinje cell bodies and dendrites. Using α1A subunit chimeras expressing SCA6 mutations, we show that the SCA6 polyglutamine expansion shifts the voltage dependence of channel activation and rate of inactivation only when expressed with β4subunits and impairs normal G-protein regulation of P/Q channels. These findings suggest the possibility that SCA6 is a channelopathy, and that the underlying mutation in SCA6 causes Purkinje cell degeneration through excessive entry of calcium ions.
- calcium channel
- cerebellar ataxia
- polyglutamine tract
- Purkinje cell
- β subunit
- neurodegenerative disease
Spinocerebellar ataxia type 6 (SCA6) is a dominantly inherited neurodegenerative disorder characterized by progressive gait ataxia, dysarthria, and incoordination caused by progressive cerebellar atrophy (Zhuchenko et al., 1997). Neuronal loss in SCA6 is limited almost exclusively to Purkinje cells (Gomez et al., 1997a; Sasaki et al., 1998; Takahashi et al., 1998; Ishikawa et al., 1999a). The underlying mutation in SCA6 consists of an expansion of a trinucleotide CAG repeat (CAG21–33) in exon 47 of the gene, CACNA1A, encoding the α1A subunit of the neuronal P/Q-type voltage-gated calcium channel (VGCC) (Zhuchenko et al., 1997;Yabe et al., 1998). The CAG repeat is in frame only in some splice forms of α1A where it codes for a polyglutamine tract in the C terminus (Zhuchenko et al., 1997). Nevertheless, from a genetic standpoint SCA6 belongs to the class of neurodegenerative disorders in which the pathological mutation encodes a protein possessing an expanded polyglutamine tract (Anonymous, 1993; Orr et al., 1993; Kawaguchi et al., 1994; David et al., 1997). These disorders are thought to result from the intranuclear accumulation of ubiquitinated aggregates of the expanded polyglutamine proteins and to have little to do with the function of the native protein (Paulson et al., 1997; Skinner et al., 1997; Becher et al., 1998; Cummings et al., 1998; Li and Li, 1998). Recently nonubiquitinated, intracytoplasmic aggregations of α1A subunits have been detected in brains of SCA6 patients using antisera to the α1A subunit, suggesting a role for aggregates in SCA6 that may differ from that in other disorders (Ishikawa et al., 1999b).
Other episodic or progressive neurological or muscle disorders arise as a direct result of the pathological change in ion channel function produced by mutations in the gene encoding the channel protein (Barchi, 1998; Boonyapisit et al., 1999; Cooper and Jan, 1999). Missense mutations or mutations that predict a truncated protein have been found in the CACNA1A gene in families with episodic or progressive ataxia and hemiplegic migraine (Ophoff et al., 1996; Yue et al., 1997; Denier et al., 1999; Jen, 1999; Tournier, 1999) and in several mouse neurological mutants (Fletcher et al., 1996; Dove et al., 1998; Wakamori et al., 1998; Mori et al., 1999; Wakamori et al., 1999; Zwingman et al., 1999). Thus, rather than relating to the ability of the mutant proteins to form aggregates, the pathogenesis of SCA6 may relate to the effect of the mutation on the function of the P/Q-type calcium channel, making SCA6 a member of the class of disorders called “channelopathies.” This distinction may be artificial, however, because the processes of aggregate formation and ion channel disturbance may be causally related.
Using antisera specific for the alternatively spliced exon 47, along with chimeric α1A subunits expressing the SCA6 mutation, we investigated whether there is restricted expression of exon 47-encoded P/Q-type channels and the functional consequences of the SCA6-associated mutation. We show here that (1) the pathological expansion can be potentially expressed in multiple isoforms of the α1A subunit; (2) the exon 47-encoded polyQ tract is abundantly expressed in Purkinje cells; and (3) heterologous expression of chimeric α1A cDNA harboring wild-type or SCA6-associated polyglutamine tract in exon 47, together with putative cerebellar auxiliary subunits (α2-δ and β2, β3, or β4), reveals a β subunit-specific alteration in the electrophysiological properties of SCA6-mutant P/Q channels.
MATERIALS AND METHODS
Preparation of antipeptide antisera. Rabbits were immunized with the bovine thyroglobulin-conjugated peptides CT1 (STSGTSTPRRGRRQLPA), corresponding to a sequence 24 amino acids N-terminal to the polyQ tract in the long isoform of the α1A subunit (Zhuchenko et al., 1997), and 2L2 (SELQQREHAPPREHA), which is the mouse homolog of CNA3 (Sakurai et al., 1996) (differing from rat in two amino acids). Peptides were conjugated to bovine thyroglobulin (Sigma, St. Louis, MO) using glutaraldehyde. The conjugated peptide (1 mg/ml) was emulsified with an equal volume of Freund's complete adjuvant (Difco, Detroit, MI) and injected into four New Zealand White rabbits. Subsequent injections were performed every 2 weeks using an equal volume of conjugated peptide and Freund's incomplete adjuvant. Rabbit sera were screened for antibody activity 1 week after the fourth and each subsequent injection.
Western blotting. pQE31-mCT (a gift from Colin Fletcher, Mammalian Genetics Laboratory) consists of a 510 bp Mspfragment beginning from codon 15 of exon 47 of the mouse α1A gene to 79 bases past the second stop codon in exon 47 inserted into the 6xHis vector pQE31 (Qiagen, Hilden, Germany). To perform immunoblots on bacterial proteins, bacterial cell lysates from Escherichia coli transformed with pQE31-mCT were eluted from nickel-nitrilotriacetic acid (NTA) columns according to the manufacturer's instructions (Qiagen). Eluted proteins were separated by SDS-PAGE and electrotransferred to nitrocellulose membranes.
To prepare crude membrane extracts, 0.1–0.2 gm of mouse cerebellum or forebrain was homogenized for 5 sec on ice in 2 ml of extraction buffer [Tris-buffered saline (TBS), pH 7.4, 1% Triton X-100, 10 mm EDTA, 10 mm EGTA, 50 μg/ml antipain, 2 μg/ml aprotinin 2, 50 μg/ml chymotrypsin, 0.5 μg/ml leupeptin, 0.7 μg/ml pepstatin (Roche Molecular Biochemicals, Mannheim, Germany), 0.1 mg/ml calpain inhibitor I, and 0.05 mg/ml calpain inhibitor II ] at a speed of 7 in a tissue homogenizer (Polytron; Brinkman, Westbury, NY) and centrifuged 150,000 ×g for 1 hr at 4°C in an SW28 rotor (Beckman Instruments, Palo Alto, CA). Ten micrograms of affinity-purified antipeptide antibody were added to 1 ml of supernatant, followed by 50 μl of washed staphylococcal protein A-agarose (Life Technologies, Gaithersburg, MD). After shaking gently overnight at 4°C on a tilting mixer, the complexes were centrifuged at 1300 × g and washed three times with 600 μl of TBS and 1% Triton X-100 and once with 200 μl of 10 mm Tris, pH 7.4, and 1% Triton X-100. The proteins were denatured and extracted from the pellet by incubation at 70°C for 10 min in 20 μl of load buffer (190 mm Tris, pH 6.8, 30 mm DTT, 7.5% SDS, 15% sucrose, and 0.01% bromphenol blue), subjected to SDS-PAGE electrophoresis in 6% gels, and electrotransferred at 24 V to nitrocellulose membranes (0.45 μm) for 3 hr in prechilled transfer buffer. Membranes were blocked overnight at 4°C in block solution (TBS, 0.5% Tween 20, 5% dry milk, and 0.5% normal goat serum).
For immunoblotting membranes were reacted with primary antibody at 0.01 mg/ml from 2 hr to overnight at 4°C and washed three times for 10 min each in block solution. Membranes were reacted with horseradish peroxidase-conjugated goat anti-rabbit Ig (Amersham Pharmacia Biotech, Arlington Heights, IL) diluted 1:6000 in block solution at room temperature for 2 hr, washed three times for 10 min each in block solution (minus proteins), and washed twice for 30 min each in TBS. To generate chemiluminescent signal, membranes were incubated 4 min in chemiluminescent substrate (Supersignal; Pierce, Rockford, IL), blotted dry, and exposed to film. To strip membranes for reprobing, blots were incubated in a solution of 100 mm β-mercaptoethanol, 60 mm Tris, pH 8, and 2% SDS at 55°C with gentle rocking and then rinsed in TBS.
Immunohistochemical localization. Male adult Sprague Dawley rats (100–150 gm; Harlan, Madison, WI) were anesthetized with an overdose of a sodium pentobarbital (60 mg/kg) and fixed by transcardial perfusion with calcium-free Tyrode's solution followed by fixative (4% paraformaldehyde and 0.2% picric acid in 0.1 mphosphate buffer, pH 6.9) and finally with 10% sucrose in PBS, pH 7.2. The brain was dissected and stored overnight in 10% sucrose at 4°C. Tissue sections were cut on a cryostat at 14 μm thickness and mounted onto gel-subbed slides. Slides were incubated for 10 min at 92°C in antigen retrieval solution, pH 7.0 (provided by Alex Kalyuzhny, R & D Systems, Minneapolis, MN), rinsed in PBS, and immediately prepared for immunostaining. Sections were preincubated in a humidified chamber in blocking buffer (0.3% Triton X-100, 1% BSA, and 1% normal donkey serum in PBS) for 30 min at room temperature and then incubated in primary antibody [anti-CT1, 1:500; anti-2L1 (CNA1), 1:500 (Alomone Labs, Jerusalem, Israel); anti-calbindin, 1:500 (Sigma); and anti-NF200, 1:1500 (Sigma)] overnight at 4°C. After thorough rinsing in PBS, sections were incubated in species-specific secondary antibody (Cy3-conjugated, 1:200; Jackson ImmunoResearch, West Grove, PA) for 1 hr at room temperature. Absorption controls were performed by preincubating the primary antisera with the peptide used in raising the antiserum at a concentraion of 10 μm. Images were obtained using the BioRad MRC-1000 confocal imaging system (BioRad Microscience Division, Cambridge, MA) and printed using a Fuji (Tokyo, Japan) Pictography 3000 printer.
Assembly of α1A chimeras. Rabbit-human chimeric α1A subunits were assembled using an overlap PCR strategy with human and rabbit cDNA and human genomic DNA as template. All amplifications were performed using a thermostable polymerase mixture that reduces errors by a proofreading function (Advantage PCR; Clontech, Cambridge, UK). Each amplified component was subcloned into the plasmid vector pGEMT (Promega, Madison, WI), and its sequence was verified by nucleotide sequence analysis. PCR was used to obtain cDNA correctly spliced for translation of exon 47 from the 3′ region of the human wild-type α1Asubunit. A 150 bp product was amplified from pooled cDNA constructed from a human cerebellar RNA (Marathon Ready cDNA; Clontech) using forward primer A (AGCGCTGGTCCCGCTCGCCCAGCG), corresponding to a conserved sequence near the 3′ end of exon 46, and reverse primer B (GGGGGTCTGGGGGAGCTGGC), corresponding to a position 66 bp upstream of the CAG repeat. A clone, pTail-U, was selected that contained the added five-nucleotide GGCAG sequence. The 680 bp lower half of the cDNA tail was amplified using forward primer C (GCCAGCTCCCCCAGACCCCC), complementary to primer B, and reverse primer D (CGATGATTGGTGCTAAGCCCGGGCGAGG), corresponding to the end of the human α1A subunit cDNA, and subcloned (pTail-Ln). This region contains the polymorphic CAG repeat, which was identified as a CAG13 allele in the human cDNA by dideoxy sequence analysis (pTail-L13). Distinct pathological CAG repeat alleles of 22, 26, and 30 repeats were amplified from genomic DNA of SCA6 patients using forward primer C (GCCAGCTCCCCCAGACCCCC) and reverse primer S-5-R1 (TGGGTACCTCCGAGGGCCGCTGGTG) (Zhuchenko et al., 1997) and inserted in place of the CAG13 allele in pTail-L13 by a unique KpnI restriction site. Full-length cDNA tails, ∼800 bp in length and containing each allele, were assembled by overlap PCR using primers A and D and templates pTail-U and pTail-Ln and subcloned to generate pCT(CAG)n. Except for the different length of the CAG repeat stretch, the 3′ cDNA, pCT(CAG)n, sequence is 100% identical to the published region of the human α1A cDNA clone and consists of 29 bp of exon 46 and the entire exon 47 (Ophoff et al., 1996;Zhuchenko et al., 1997).
The pCT(CAG)n tails were attached to the functional rabbit α1A subunit cDNA, BI-1 (psp72 plasmid backbone, a gift from Y. Mori; Mori et al., 1991), by means of overlap PCR, followed by restriction digestion and ligation. The full-length CT(CAG)n tails were amplified using primers A and D and each pCT(CAG)n as template. A 685 bp 3′ region (BI-1-CT), corresponding to the 3′ end of the rabbit BI-1 cDNA, was amplified using forward primer E (CCATCCTGGGTGACCCAGCG) and reverse primer F (complementary to primer A). The 1470 bp overlap PCR product consisting of the product, BI-1-CT, conjoined to CT(CAG)n was then produced by PCR using primers E and D and BI-1-CT and CT(CAG)n as templates. This chimeric rabbit human 3′ cDNA PCR product was then substituted for the rabbit BI-1 3′ region by means of the common restriction site BstEII. The proper assembly of the final chimeric products was confirmed by dideoxy nucleotide sequence analysis. RNA was transcribed in vitrofrom the BamHI-linearized α1Achimeras.
Electrophysiological recordings. Xenopus oocyte preparation and injection (20–40 nl of α1 plus α2-δ plus β cRNA at 0.3 ng/nl) were performed as described elsewhere (Cens et al., 1999). For the G-protein study a mixture of αG0 and μ-opioid receptor cRNA, both at 1 μg/μl, was injected with three volumes of the Ca2+ channel cRNA. The μ-opioid receptor cDNA in the plasmid pBluescript (Stratagene, La Jolla, CA) was linearized with XbaI; the αG0 cDNA (gifts from C. Labarea and L. Yue, Caltech) in pGEM was linearized with NheI; and both were transcribed with T7 polymerase. Oocytes were then incubated for 2–7 d at 19°C under gentle agitation before recording. Whole-cell Ba2+ currents were recorded under two-electrode voltage clamp using the GeneClamp 500 amplifier (Axon Instruments, Burlingame, CA). Current and voltage electrodes (resistance <1 MΩ) were filled with electrode solution (2.8m CsCl and 10 mm BAPTA, adjusted to pH 7.2 with CsOH). Ba2+current recordings were performed after injection of BAPTA solution [one or two 40–70 msec injections at 1 bar, 100 mm BAPTA-free acid (Sigma), 10 mm CsOH, and 10 mm HEPES, pH 7.2]. The composition of the recording solution was 10 mm BaOH, 20 mm TEAOH, 50 mm N-methyl-d-glucamine, 2 mm CsOH, and 10 mm HEPES, adjusted to pH 7.2 with methanesulfonic acid. Only oocytes expressing Ba2+ currents with amplitudes in the range of 0.5–3 μA were analyzed to ensure sufficient resolution and avoid voltage-clamp problems. Currents were filtered and digitized using a DMA-Tecmar labmaster and subsequently stored on a personal computer by using version 6.02 of the pClamp software (Axon Instruments). Ba2+ currents, recorded during a typical test pulse from −80 to +10 mV of 2.5 sec duration, were well fitted using a biexponential function: i(t) =(A 1 * exp(−(t −K)/τ1)+ A2* exp(−(t −K)/τ2)) + C, wheret is the time, K is the zero time,A 1 , A 2, τ1, and τ2 are the amplitudes and time constants, respectively, of the two exponential components, and C is the fraction of noninactivating current. Current amplitudes and inactivation time constants were measured using Clampfit (pClamp version 6.02; Axon Instruments). Pseudo-steady-state inactivation curves (2.5 sec of conditioning depolarization) were fitted using the equationI/I max = R + (1 − R)/(1 + exp((V −V 0.5)/k)), andI–V curves were fitted using the equationI/I max = g * (V − E rev)/[1 + exp((V −V 0.5)/k)], whereg is a normalized conductance,E rev is the extrapolated reversal potential for Ba2+, k is the slope factor, V 0.5 is the potential for half-inactivation or activation, V is the conditioning depolarization for inactivation curves or the membrane potential used to record current for I–V curves, andR is the proportion of noninactivating current. All values are presented as the mean ± SD. The significance of the difference between two means was tested using the Student'st test (p < 0.05). All chemicals were from Sigma. d-Ala2, N-Me-Phe4, Gly-ol5 (DAMGO) was purchased from Research Biochemicals (Natick, MA) and prepared daily at the desired concentration (10 μm).
The long exon 47 α1A sequence is highly expressed in multiple α1A isoforms in cerebellum
To investigate the distribution of exon 47-encoded α1A polypeptides and to determine whether their restricted expression could explain the cell selectivity of the neurodegeneration in SCA6, we generated antisera specific for this domain and characterized their reactivity to proteins immunoprecipitated with other anti-α1A antisera defined in Figure 1 A. The C-terminal antiserum reacted with a highly conserved epitope, CT1, 24 residues N-terminal to the predicted polyglutamine tract in the human α1A and present in the mouse and rabbit C terminus (Zhuchenko et al., 1997). Figure 1 Bdemonstrates reactivity of the anti-CT1 antibody to a 6xHis-tagged fusion protein constructed from the mouse cDNA encoding the C terminus inserted into the 6xHis vector pQE31. The anti-CT1 antiserum reacted with a single species present in transfected bacterial proteins eluted from a Ni-NTA-agarose column (lanes 2, 4). The protein detected corresponds well to the predicted molecular weight of 12 kDa and was also detected by antiserum to the RGS-6xHis tag (lane 1). Binding to the CT1 epitope was blocked by incubation in CT1 peptide (lane 3).
To identify brain α1A polypeptides, we analyzed proteins immunoprecipitated from detergent extracts (Fig.1 C) of forebrain (lane 1) or cerebellum (lanes 2–9) by SDS-PAGE and immunoblotting using antipeptide antibodies. In lanes 1–4, α1A polypeptides were immunoprecipitated using antibody to 2L1, 2L2, and CT1 and reacted with anti-CT1 antibody. 2L1, equivalent to CNA1 (Westenbroek et al., 1995), and 2L2, the mouse homolog of CNA3 (Sakurai et al., 1995, 1996), correspond to sequences found in the LII–III cytoplasmic loop of α1A. Although much less abundant in immunoprecipitates from other regions, CT1 antiserum appeared to detect the same pattern of isoform sizes in forebrain (lane 1) and spinal cord (data not shown). Lanes 5–7 correspond tolanes 2–4 after the blot was stripped and reacted with antibody to 2L1 (CNA1). As has been noted previously for lectin-purified α1A polypeptides, anti-2L1 reacts with polypeptides ranging in size from <160 to >200 kDa (lane 5). The larger sizes isolated than seen previously may relate to the simple one-step extraction procedure. Surprisingly, rather than reacting with a single α1A isoform, the anti-CT1 antiserum binds to a range of polypeptides precipitated by anti-2L1 (lanes 1, 2) that is similar to the pattern recognized by 2L1 (lane 5).
Anti-2L2 (equivalent to CNA3), which reacts with a subset of CNA1-expressing α1A isoforms (Sakurai et al., 1995, 1996), immunoprecipitates several polypeptides that also bear the CT1 epitope (Fig. 1 C, lane 3). The anti-CT1 antisera was less efficient at precipitating the α1Apolypeptides. Moreover, both this antibody and the anti-2L2 (CNA3) precipitate 140 kDa isoforms apparently lacking the 2L1 (CNA1) epitope (lanes 3, 4). These results are consistent with those of Northern blot analysis (Zhuchenko et al., 1997) of cerebellar α1A transcripts and studies of other α1 subunits (Sakurai et al., 1995; Lin et al., 1997) and further indicate that alternative splicing leads to a great diversity of α1A isoforms. Thus, the extended exon 47 sequence, which in humans encodes the polyglutamine tract, is present on multiple α1A isoforms.
The long exon 47 α1A sequence is expressed heavily in cerebellar Purkinje cells and in neurons throughout the nervous system
To investigate the sites of expression of the long exon 47 α1A isoforms, we studied the pattern of anti-CT1 binding by immunohistochemistry (Fig.2) and compared this with the pattern seen with anti-2L1(CNA1). In the cerebellum anti-CT1 stained the Purkinje cell soma and dendrites intensely (Fig.2 A,B,I,J). Staining was also seen in a smaller population of cells in the molecular layer and more faintly in the granule cell layer. This pattern was similar to that seen in human cerebellar cortex using RNA probes consisting of the exon 46–47 sequence (Ishikawa et al., 1999b). The prominent immunoreactivity within Purkinje cell bodies and proximal dendrites is highly reminiscent of those patterns seen with other antisera to α1A subunits, including anti-CNA1 (Westenbroek et al., 1995; Sakurai et al., 1996). In our hands anti-CNA1 (2L1 in the current nomenclature) strongly stained the Purkinje cells, but the dendritic staining was less prominent, presumably because of differences in fixation, antigen retrieval method, and antiserum lot (Fig. 2 D). Immunostaining by both anti-CT1 (Fig.2 C) and anti-2L1 (data not shown) were blocked by preincubation in the specific peptide.
The pattern of immunostaining with anti-CT1 differs from that of anti-2L1 (CNA1) regionally. Some areas, such as regions of the neocortex, were labeled by both antisera with patterns similar to those seem in cerebellum (Figure 2 E,F). In contrast, numerous other areas, such as the thalamus, the preoptic area, and the piriform cortex, exhibited no immunostaining by anti-CT1 (Fig.2 G), yet neurons of these areas were clearly labeled by anti-2L1 (Fig. 2 H).
In co-localization studies there was strong overlap of the CT1 staining with that of calbindin in Purkinje cells, including in their primary dendrites and terminals (Fig. 2 I). When double labeling was performed using anti-NF200, which labels basket cell fibers intensely, staining with anti-CT1 appeared to be confined to Purkinje cells and some interneurons but was not present in terminals that project to the Purkinje cell layer (Fig. 2 J). These findings indicate that the long exon 47-encoded sequence (which encodes the SCA6-associated polyQ tract) is expressed abundantly in Purkinje cell bodies and dendrites and in some interneurons of the cerebellar cortex. Although these findings cannot account for the absolute selectivity of Purkinje cell death in SCA6, the prominent expression of the expanded polyQ tract in these cells may be a contributing factor to their selective degeneration (Gomez et al., 1997a; Sasaki et al., 1998; Ishikawa et al., 1999a).
Polyglutamine expansions alter activation and inactivation kinetics of P/Q channels
To investigate the effect of the polyglutamine expansion on VGCC function, we expressed chimeric αlA cDNA with five different C-terminal configurations in Xenopus oocytes. The chimeric α1A cDNAs were assembled from the rabbit BI-1 cDNA (Mori et al., 1991) containing part of the sequence of the predicted rabbit exon 46 and the human cDNA encoding the last 29 amino acids of exon 46 and all of exon 47. Separate clones were derived from PCR-amplified genomic DNA containing the wild-type allele with 13 glutamine residues (Q-13) or the SCA6-associated alleles with 22, (Q-22), 26 (Q-26), or 30 (Q-30) glutamine residues. RNA transcribed from the BI-1 plasmid or from plasmids bearing chimeric α1A subunits was co-injected with in vitro-transcribed RNA encoding α2-δ and either β2, β3, or β4 subunits. All five α1A subunits yielded qualitatively normal voltage-activated currents.
We studied the effect of the wild-type and different polyQ SCA6-associated mutant alleles on P/Q channel activation and inactivation using a paired pulse protocol. Results are summarized in Table 1 and Figure3. Oocytes expressing the Q-30 α1A chimera and β4subunits had a significant shift of the point of half-maximal activation, EA0.5, (EA0.5Q-30/β4 = −7 ± 4; n = 12; vs EA0.5Q-13/β4 = 0 ± 5;n = 7; p = 0.005; Fig. 3 A). The oocytes expressing Q-26 and β4 subunits also appeared to have a hyperpolarizing shift, although the change was not statistically significant. In contrast, there was no shift in EA0.5 when the same experiment was performed for the α1A chimeras co-expressed with β2 subunits (EA0.5Q-30/β2 = −11 ± 5;n = 13; vs EA0.5Q-13/β2= −9 ± 4; n = 12; Fig. 3 B) or β3 subunits (EA0.5Q-30/β3 = −4 ± 2;n = 10; vs EA0.5Q-13/β3= −4 ± 4; n = 14; data not shown).
We studied the time course of inactivation of P/Q channels during 2.5 sec voltage steps of 10 mV. Figure4 A shows normalized traces of P/Q channels expressing the BI-1 isoform or the long α1A chimeras expressing Q-13, Q-22, Q-26, and Q-30 alleles in the presence of β2 or β4 subunits. Currents recorded in oocytes expressing the Q-30 allele α1A channel in the presence of the β4 subunit were visibly prolonged, whereas channels expressing the Q-30/β2 combination were no different from the other α1A alleles. To compare the inactivation of each subunit combination quantitatively, we determined the percent of noninactivating current as the ratio of the residual current at the end of the pulse to the peak current at the beginning of the pulse (Fig. 4 B). Q-30/β4 channels inactivate significantly more slowly than the other channels.
A kinetic analysis of the inactivation phases has shown that the time course of inactivation can be fit to two exponents, τfast and τslow. Only the fast component is affected by the expanded polyQ (τfastQ-30/β4 = 166.4 ± 27.1;n = 7; vs τfastQ-13/β4= 135.5 ± 13.5; n = 7; p = 0.005; Fig. 5 A). The fast component of the Q-30 channels also appears to be prolonged in the β2-expressing channels (τfastQ-30/β2 = 205.8 ± 114.9;n = 10; vs τfastBI-1/β2 = 125.8 ± 114.9;n = 4; p = 0.005). In the β2 channels the slow component represents ∼99% of the current decay. This accounts for the lack of effect of the polyQ expansion on inactivation in β2-expressing oocytes. In β4-expressing channels, however, the slow component represents ∼30% of the current decay. The increase in τfast in Q-30/β4 channels leads to the slowed inactivation.
Although the effect on inactivation was statistically significant, it was important to estimate the potential impact of such a change in channel kinetics on the intracellular ionic milieu. Figure5 B compares the proportion of Ca2+ entering the cell with a given activation of P/Q channels of each type, as determined by the integral of the current traces over time. With a single event a significantly greater amount of Ca2+ enters the cell through Q-30/β4 channels than through wild-type or β2 channels.
During normal cellular activity the effect of a reduced threshold for P/Q channel activation in SCA6 Purkinje cells might potentially be blunted if the mutation also reduced channel reactivation. Therefore, we studied the effect of the elongated α1A C terminus with normal and expanded polyQ tract on recovery from inactivation. To assess reactivation, the currents evoked by test pulses delivered at varying intervals after a conditioning pulse (2.5 sec at 10 mV) were plotted as a function of pulse interval. Figure6 A shows typical current traces for oocytes expressing BI-1 plus α2-δ plus β4. The recovery time course was not significantly affected by α1A chimeras Q-13, Q-22, Q-26, and Q-30. Furthermore, because calcium channel activation occurs repetitively, we investigated the effect of the modified P/Q channel C termini on the response to repetitive stimulation (Fig. 6 B). Ba2+ currents were recorded during stimulation (10 mV, 1 Hz), and the peak amplitude was plotted as a function of stimulus number. Current decay was assessed by determining the ratio of the amplitude of the 1st response to that of the 10th response (I 1/I 10). No statistical differences inI 1/I 10(at the p = 0.05 level) were noted between current decay in BI-1 and normal (Q-13) or pathological (Q-22, Q-26, and Q-30) chimeric α1A subunits expressed with α2-δ and either β4 or β2.
G-protein regulation of the long exon 47 α1A subunit is abolished by the expanded polyQ tract
There are at least two recognized G-protein interaction domains on α1A subunits, one in the cytoplasmic loop between repeat domains I and II (LI–II), adjacent to the primary β subunit interaction domain (De et al., 1997), and a second in the C terminus N-terminal to the putative polyQ tract (Qin et al., 1997; Furukawa et al., 1998; Simen and Miller, 1998). Because the β4 subunit also interacts with the α1A subunit C terminus (Walker et al., 1998), we sought to determine whether G-protein regulation of the long exon 47 α1A isoform differed from that of the BI-1 isoform or whether this regulation was perturbed by the presence of an expanded polyglutamine tract.
To investigate G-protein regulation, we used the activity of the G protein-coupled μ-opioid receptor. We co-injected in vitro-transcribed cRNAs for the μ-opioid receptor and Gαo cDNA together with combinations of Ca2+ channel subunits, including normal or mutated α1Asubunit and α2-δ and β4 subunits. We tested the effect of perfusion of the μ-opiod agonist DAMGO on Ba2+currents evoked by consecutive pulses (50 msec at 0 mV) every 3 sec. Figure 7 A shows typical current traces and time course of inhibition and recovery for a single subunit combination. To quantitate the extent of regulation, we calculated the degree of inhibition at the steady state of DAMGO (Fig.7 B). We found that addition of the sequence encoded by the wild-type extended 47 (Q-13 clone) conferred a greater degree of G-protein regulation than seen with the BI-1 isoform without the exon 47 extension. Similar findings were obtained with Q22 chimera. In contrast, the G protein regulation conferred by the long exon 47 sequence appeared to be abolished in the Q26 and Q30 chimeric α1A subunits. Thus, the wild-type long exon 47 α1A subunit is sensitive to G-protein regulation, which is in turn impaired by the expanded polyQ tract in SCA6.
The discovery of the genetic basis of SCA6 was accompanied by the recognition of a novel splice form of the α1Asubunit expressing a previously unrecognized reading frame of exon 47 of the gene CACNA1A (Zhuchenko et al., 1997). In the human gene this exon encodes the polymorphic polyglutamine tract that is expanded in SCA6 (Zhuchenko et al., 1997). Because the distribution and properties of α1A subunits expressing this novel sequence were unknown, and because of its importance in the disease, we studied the pattern of α1A polypeptide isoforms and tissue distribution of the exon 47-encoded sequence. Our immunoblot and immunohistochemical analysis indicates that exon 47 is frequently translated, is present on multiple α1A isoforms, and is expressed heavily in Purkinje cells and in lesser amounts throughout the CNS. Therefore, although restricted expression of the extended sequence of exon 47 cannot completely account for the selectivity of the Purkinje cell death in SCA6, its abundance in these cells may play an important role.
Expansion of the polyQ tract in the C terminus of the α1A subunit has three discrete effects when P/Q channels are expressed in oocytes: (1) the voltage dependence of channel activation evoked by 10 mV steps is shifted in the hyperpolarizing direction; (2) the rate of inactivation during a 10 mV pulse is slowed; and (3) regulation by G-proteins conferred by the C terminus is abolished. The first two of these effects exhibit a β subunit dependence and are not seen when the chimeric subunits are co-expressed with β2 or β3 subunits. These studies support the model of SCA6 as a channelopathy.
The findings in this study are distinct from those of other groups.Matsuyama et al. (1999) have found that expression of recombinant α1A subunits encoding a Q-30 or Q-40 (exaggerated) polyglutamine tract caused reduced current density and an 8 mV shift in the voltage dependence of inactivation in the hyperpolarizing direction. The net effect of this change would be to reduce the number of channels available for activation. Recent studies using an α1A subunit with a tagged epitope in a similar system showed an increased current density and increased level of expression (Piedras-Renteria et al., 1999). In both studies there was a small shift in the voltage dependence of activation in the hyperpolarizing direction that did not reach statistical significance for either spontaneous (Q-27 or Q-30) or exaggerated (not found in humans, Q-40 or Q-72) alleles.
At least three explanations may account for these different findings. (1) In the present study α1A subunit alleles were co-expressed with β2, β3, or β4 subunits, whereas the β1a (the muscle-specific) and β1b isoforms were used in the other studies (Matsuyama et al., 1999; Piedras-Renteria et al., 1999). The clear dependence of the SCA6 effect on β subunit subtype in our study further indicates the importance of an auxiliary subunit subtype in shaping P/Q channel activity and may account for the different findings. (2) In this study, the use of the Xenopus oocyte expression system rather than mammalian cells may have led to distinct or more robust responses. It will be important to determine whether the recordings in oocytes represent an exaggeration of the effect in vivo or whether the recordings in mammalian cells with distinct β subunit subtypes exhibit the same effect. (3) In the present study the 3′ portion of the correctly spliced human 46–47 cDNA (with the polyQ allele substituted) was added to the cDNA encoding the rabbit BI-1 isoform of the α1A subunit rather than inserting the polyglutamine tract directly into the rabbit in frame 3′ untranslated region (UTR) (Matsuyama et al., 1999). Thus, the chimeric α1A subunits in the present study possess 253 amino acids encoded by human exons 46 and 47. Although the rabbit BI-1 3′ UTR encodes a sequence 90% identical to the human extended α1A isoform, subtle differences may be important (Zhuchenko et al., 1997).
The C terminus of the α1A subunit in P/Q channels participates in β subunit interactions and Ca2+-calmodulin-mediated regulation and contains an interaction domain for G-protein regulation. (Qin et al., 1997; Furukawa et al., 1998; Simen and Miller, 1998). The sequence encoded by the entire exon 47 nearly doubles the length of the C terminus of the α1A subunit by the addition of 246 amino acids (Zhuchenko et al., 1997). Although this exon encodes the polymorphic polyglutamine tract and SCA6-associated expanded polyQ alleles, the wild-type Q-13 and mildly expanded Q-22 alleles appear to have no effect on P/Q channel activation or inactivation. On the other hand, chimeric Q-13 and Q-22 α1A subunits are more susceptible to inhibition by G-protein than is BI-1, as demonstrated by the inhibitory effect of the opioid agonist DAMGO when channels were co-expressed with the μ-opioid receptor. This may account for some differences in disease severity and disease phenotype associated with different CAG repeat sizes (Kaseda et al., 1999). Curiously, the larger expanded alleles, Q-26 and Q-30, appear to interfere with the regulatory role of the C terminus. This provides a second aspect of the molecular phenotype of the mutant P/Q channels in SCA6 and may explain the differences found in the activation properties of the different alleles.
The dependence of β subunit subtype in determining the phenotype of the SCA6 mutation is a plausible finding. β subunits have a prominent effect on gating of VGCC (De and Campbell, 1995; Cens et al., 1999). β subunit subtypes diverge greatly in primary structure in their C termini, and the β4 subunit has a unique C terminus interaction with the α1A subunit C terminus (Walker et al., 1998). The expanded polyglutamine tract in the SCA6-associated α1A subunit may alter channel gating by interfering with this β4-specific interaction with the α1A C terminus.
Among several allelic mouse neurological mutants affecting CACNA1A, the recessively inherited tg la(leaner) mouse bears the greatest similarity to SCA6. Homozygous tg la mice develop dystonia, severe ataxia, and absence seizures after weaning (Fletcher et al., 1996). They develop Purkinje cell degeneration, but it is confined to parasagittal striped domains overlapping the pattern of zebrin expression in the cerebellar cortex (Fletcher et al., 1996). Thetg la mutation disrupts the C terminus of the α1A subunit and presumably interactions with the β4 subunit, G-protein βγ, and Ca2+-calmodulin kinase (Furukawa et al., 1998; Walker et al., 1998; Lee et al., 1999). Cultured tg la Purkinje cells have reduced Ca2+ current density, a molecular phenotype that is replicated in recombinant expression studies, as well as an increase in the noninactivating component of calcium currents (Wakamori et al., 1998).
In SCA6, Purkinje cell degeneration is not confined to parasagittal zones (Gomez et al., 1997a; Sasaki et al., 1998; Ishikawa et al., 1999a). Moreover, we found that immunostaining with anti-CT1 did not reflect any compartmentalization of exon 47 expression into parasagittal zones in rat cerebella (data not shown). Thus, although the two types of Ca2+ channel mutations appear to have some opposing effects on P/Q-type Ca2+ channel activity, differences in the inheritance pattern, age of onset, and tissue selectivity between the two resulting diseases suggest a model in which Purkinje cell development and viability depend on tight control of Ca2+ channel function and the avoidance of either abnormally increased or decreased Ca2+ current. The congenital myasthenic syndromes represent a similar case in which weakness and impaired neuromuscular transmission can arise from either homozygous inactivating mutations or heterozygous “gain-of-function” mutations of the acetylcholine receptor subunits (Engel et al., 1996a,b; Gomez et al., 1996).
The changes in P/Q channel properties resulting from the SCA6 mutations will have several effects on the electrophysiological and cellular properties of the Purkinje cells. In isolation, a delay in inactivation of the magnitude measured in this study will result in entry of 1.3-fold greater Ca2+ per activation. This could significantly increase the amount of Ca2+ entering the Purkinje cell during climbing fiber activity or other influences that activate 1 Hz complex spike activity (Kitazawa et al., 1998; Lang et al., 1999).
The effect of the shift in activation of P/Q channels toward the hyperpolarizing direction could have a more dramatic effect on both Purkinje cell firing and intracellular ion content. In the absence of any compensatory response, this change would potentially lead to opening of a greater number of VGCC and entry of more Ca2+. At the physiological level, excessive Ca2+ ions entering through the SCA6-mutant channels might be compensated for by downregulation of gene expression of the α1A subunit. This is apparently not the case, because immunohistochemical staining of α1A subunits in SCA6 cerebellar Purkinje cells shows normal levels (Ishikawa et al., 1999b). If sufficient to exceed the intracellular buffering capacity, Ca2+overload could result in activation of Ca2+-sensitive enzymes, such as caspases, calpain, and phospholipases, leading to cell death, as has been suggested for other conditions characterized by excitotoxity or Ca2+ overload (Mattson, 1992; Choi, 1994;Dugan and Choi, 1994; Gomez et al., 1997b).
Neurodegeneration in SCA6 is highly selective for Purkinje cells (Gomez et al., 1997a; Sasaki et al., 1998; Ishikawa et al., 1999a). On the basis of their prominent involvement in other multisystem degeneration (Durr et al., 1995; Durr et al., 1996; Koeppen, 1998), Purkinje cells appear to be a vulnerable cell type. However, expression of the SCA6-associated exon 47 is more prominent in the Purkinje cells than in many other regions. Furthermore, expression of the β4 subunit is more abundant in Purkinje cells than in other areas (Burgess et al., 1999). There are conflicting reports on the relative level of expression of β3 (Tanaka et al., 1995; Ludwig et al., 1997;Volsen et al., 1997; Burgess et al., 1999). The strikingly selective degeneration of Purkinje cells may relate in part to the combined effect of a high level of Purkinje cell expression of the pathogenic splice form and its co-localization with the β4subunit. Regional co-localization studies at the subcellular level combined with co-immunoprecipitation studies may clarify this issue.
Another study has demonstrated the presence of cytoplasmic aggregates in Purkinje cells in SCA6 (Ishikawa et al., 1999b). These aggregates may arise because of self-association properties of the polyglutamine tract. Although it is difficult to envision how cytoplasmic aggregates could contribute to altered channel function, the formation of these associations while inserted in the membrane of the Purkinje cell soma or dendrite could have a prominent effect on channel gating.
These studies show that the SCA6 polyglutamine expansion causes a change in the function of an endogenous ion channel molecule, suggesting that SCA6 differs qualitatively from the other characterized forms of SCA. Although abnormal protein folding may yet underlie this perturbed channel function in SCA6, recognition that SCA6 is a channelopathy may lead to development of rational drug therapies. Further advances using co-localization studies and neuronal expression studies in vitro and in vivo may reveal more details of the pathogenesis of SCA6.
This work was supported by National Institutes of Health Grants RO1 NS37211 and RO1 NS36809 (C.M.G.), NATO Grant CRG 972059, Association Francaise contre les Myopathies, Association pour la Recherche contre le Cancer, Ligue Nationale contre le Cancer (P.C.), and Groupement de Reflexion sur la Recherche Cardio-Vasculaire (S.R.).
Correspondence should be addressed to Dr. Christopher M. Gomez, Box 295, Center for Clinical and Molecular Neurobiology, Departments of Neuroscience and Neurology, University of Minnesota, 420 Delaware Street Southeast, Minneapolis, MN 55455. E-mail:.