A Novel Central Nervous System–Enriched Spinocerebellar Ataxia Type 7 Gene Product FREE

NINE INHERITED neurodegenerative diseases caused by the expansion of CAG repeat-encoded polyglutamine (polyQ) tracts are characterized by loss of specific neuronal populations despite widespread expression of the mutant proteins.^{1 - 2} Several nonmutually exclusive hypotheses have been proposed to explain selective neurotoxicity in polyQ diseases and invoke involvement of neuronal intranuclear inclusion (NII) formation, alterations in gene expression, and disruption/promotion of normal/inappropriate protein-protein interactions involving mutant polyQ-containing gene products.^{3 - 7} However, the possibility that multiple mutant polyQ proteins act in concert to generate the spatially restricted neurodegenerative patterns typical of CAG repeat disorders has received relatively little attention. Multiple distinct proteins are frequently produced from a single gene by alternative splicing, a mechanism of transcriptional regulation that exerts profound effects on normal and pathologic animal physiologic function.⁸ Alternative splicing is particularly prevalent in the central nervous system (CNS), where recently conducted human genome-wide analyses estimate that greater than 60% of all genes are represented as multiple messenger RNA (mRNA) species.⁹ Although alternate splice forms of several polyQ disease-associated genes have been identified, their unique contributions to pathology have not been thoroughly examined.^{10 - 15} Spinocerebellar ataxia type 7 (SCA7) is caused by CAG repeat expansion within the SCA7 gene and involves degeneration of cerebellar Purkinje cells, brainstem neurons, and retinal photoreceptors.^{16 - 17} We hypothesize that multiple products of the SCA7 gene contribute to this degeneration and report the identification of SCA7b and ataxin-7b, a novel SCA7 splice variant and the ataxin-7 isoform it encodes. In contrast with the previously described ataxin-7 protein, ataxin-7b expression occurs predominantly within the CNS and may modulate the spatial distribution of SCA7-mediated neurodegeneration.

Spinocerebellar ataxia type 7 complementary DNA (cDNA) and ataxin-7 amino acid sequences were entered as GenBank queries using the basic local alignment search tool.¹⁸ Significant matches identified in the initial search were used as second-pass queries of GenBank. Significant matches were then used to search the Expressed Sequence Tags database for clones with high sequence similarities.¹⁹

Human genomic DNA was amplified using the Expand Long Template Polymerase Chain Reaction (PCR) system (Roche, Indianapolis, Ind) and standard thermocycling conditions. Primer pairs B (5′-CCACTCACACACTCCTCTAGACAA-3′) and C (5′-CCTGTTCGTAAGCAAGGTGA GGAG-3′) were used to amplify intron 12 and atx7-2F (5′-CTCCTCACCTTGCTTACGAAC AGG-3′) and D (5′-GTAACTTGTCCTCAAAAGTGTCCGC-3′) to amplify the following intron. Agarose-embedded PCR products were purified with Prep-A-Gene (Bio-Rad, Hercules, Calif) and cloned into the pCRII TOPO vector (Invitrogen, Carlsbad, Calif). Recombinant plasmids were isolated with Qiaprep spin columns (Qiagen, Valencia, Calif) and subjected to automated cycle sequencing. Spinocerebellar ataxia type 7 genomic sequence was analyzed with DNAStar software (Madison, Wis).

Total RNA was prepared from whole human fetal brain with TriReagent (Molecular Research Center, Cincinnati, Ohio), and cDNA was generated with the SuperScript Preamplification System (Invitrogen) using an oligo dT primer. Spinocerebellar ataxia type 7b–specific fragments from fetal brain cDNA were amplified with Platinum Pfx DNA polymerase (Invitrogen) using primer pairs B and C, B and D, or A (5′-GGGCGCGAATTCATGTCGGAGCGG GCCGCG-3′) and SCA7b-R (5′-GGGTGATGTTGCTGAAATTCC-3′) using a thermocycling protocol recommended by the supplier. The provided enhancer solution was used at 3× concentration. Products were electrophoresed through 2% Tris/acetate–buffered agarose gels containing ethidium bromide, and UV light was used to visualize bands. Appropriately sized products were excised from the agarose, cloned into pCRII TOPO, and sequenced.

Oligonucleotide SCA7b-R was end labeled by a standard procedure, and the β-actin control probe supplied by the manufacturer (Clontech, Palo Alto, Calif) was random hexamer-labeled with Rediprime II (Amersham Pharmacia, Piscataway, NJ).²⁰ Unincorporated radionucleotides were eliminated from the probe preparations using Sephadex G-25 fine Quick Spin columns (Roche). Hybridization of 10⁷ cpm of each probe to Multiple Tissue Northern blots (Clontech) was performed overnight at 45°C for SCA7b-R or at 60°C for the β-actin control, and each blot was washed using a standard protocol.²⁰ Autoradiographic exposures were performed at −80°C between 2 intensifying screens.

Generation and purification of polyclonal antiserum samples was performed as previously described using an oligopeptide corresponding to amino acids 928 through 943 (TFEDKLHLHSAL WTPR) of the predicted unique ataxin-7b C-terminus.²¹

Glutathione S-transferase (GST)–ataxin-7b fusion proteins were expressed and purified as previously described except that the excision of a 1119–base pair (bp) StuI/XbaI fragment from the SCA7b cDNA resulted in the expression of ataxin-7b proteins lacking amino acids 455 through 828.²¹ Tissue and cell extracts were prepared, electrophoresed on 8% polyacrylamide gels, and transferred to nitrocellulose membranes as previously described.²¹ Immunoblots were blocked for 1 hour at room temperature in Tris-buffered saline with 0.04% Tween-20, containing 5% nonfat dried milk and incubated overnight at 4°C with either Ab161 or anti-GST diluted 1:1000 or anti–β-actin diluted 1:5000. Antibody A, preimmune serum, and Ab161 preadsorbed with 100µM oligopeptide antigen were used at 1:1000. After washing in TBS-T, blots were incubated with either horseradish peroxidase–conjugated donkey antirabbit (1:2000) or goat antimouse (1:3200) secondary antibodies for 1 hour at room temperature. Detection was performed with ECL-Plus (Amersham Pharmacia).

Fixation, slide preparation, antigen recovery, and tissue staining were performed by standard methods.²² Ab161 was diluted 1:50 and Monoclonal Antibody 1C2 (Chemicon, Temecula, Calif) was diluted 1:4000. Goat antirabbit and horse antimouse biotin-conjugated secondary antibodies were diluted 1:1000 and 1:300, respectively. Preadsorption of Ab161 was performed with 75µM oligopeptide antigen. Detection was performed with Vectastain Elite ABC and either the 3, 3′-diaminobenzidine or Novared substrate (Vector Laboratories, Burlingame, Calif).

To examine SCA7 gene expression, portions of SCA7 cDNA sequence were systematically searched against the National Center for Biotechnology Information Expressed Sequence Tags database using the basic local alignment search tool.^{18 - 19} A single clone was identified from human testis that contains 549 bp of nonvector DNA corresponding to exons 12 and 13 of SCA7 cDNA but that also harbors a novel intervening sequence of 67 nucleotides between the 2 exons. Alignment of SCA7 genomic DNA with the Expressed Sequence Tags revealed that the 67-bp insertion lies within the 2881-bp SCA7 intron 12 such that 1119 bp lies upstream of the insertion and 1762 bp lies downstream (Figure 1A). Canonical splice acceptor and donor sites flank the insertion, indicating that it likely represents an uncharacterized SCA7 exon.²³ Polymerase chain reaction amplification of human fetal brain cDNA using an insertion-specific primer yielded a product migrating at the expected size of 221 bp, indicating that an SCA7 mRNA containing the 67-bp sequence is expressed (Figure 1B, lane B/C). Primers flanking the insertion amplified 2 products, one containing and one lacking the insertion (Figure 1B, lane B/D). DNA sequence analysis of these amplicons confirmed that at least 2 distinct SCA7 mRNAs are expressed within fetal brain tissue, the previously described SCA7 mRNA (SCA7a) and a novel transcript (SCA7b) containing the additional exon. The entire 2.7-kilobase (kb) SCA7b coding region was amplified and DNA sequence analysis confirmed that, other than the 67-bp exon, SCA7b is identical to SCA7a, including the polymorphic CAG repeat tract near the 5′ end (Figure 1B, lane A/C).

Northern blot analysis with an SCA7b-specific probe revealed a 4.5-kb transcript migrating distinctly from previously reported SCA7 mRNAs and occurring in greater abundance in the brain than in non-CNS tissues, such as those of the heart, placenta, lung, liver, skeletal muscle, kidney, and pancreas (Figure 2). The 4.5-kb transcript is expressed at similar levels within all CNS regions examined, including the cerebellum, cerebral cortex, medulla, spinal cord, occipital pole, frontal lobe, temporal lobe, and putamen (Figure 2). Reverse transcription PCR analysis of SCA7b distribution also demonstrated greater product abundance in the CNS than in peripheral tissues (data not shown). Taken together, these data indicate that SCA7b is a novel 4.5-kb transcript expressed throughout the CNS, where it occurs at significantly higher levels than in peripheral tissues.

Spinocerebellar ataxia type 7b is predicted to encode ataxin-7b, a 101-kd protein comprised of 945 amino acids. The N-terminal 887 amino acids of this protein, including the polymorphic glutamine repeat, nuclear localization sequence, and arrestin-like phosphoprotein–binding domain are identical to the previously described ataxin-7A protein, which is 892 amino acids in length (Figure 3A).^{17 ,24} The 67-bp insertion and resultant frameshift within SCA7b mRNA encode 58 amino acids comprising the novel ataxin-7b C-terminus, which bears no significant homologic characteristic to described protein domains. To examine ataxin-7b expression, we raised an antibody (Ab161) against the unique C-terminus and confirmed specificity by demonstrating that Ab161 detects recombinant ataxin-7b proteins on immunoblots (Figure 3B). Examination of ataxin-7b tissue distribution with Ab161 revealed expression of a single 101-kd protein species, the signal intensity of which was consistently higher in the cerebellum, medulla, occipital pole, retina, heart, and pons compared with that in the liver, kidney, or pancreas (Figure 3C). Signal intensity was also relatively strong in the frontal lobe, skeletal muscle, and spinal cord but markedly reduced in the lung, lymphoblast, and intestine (data not shown). The 101-kd protein was neither detected by Ab161 preadsorbed with its oligopeptide antigen nor by rabbit serum collected prior to immunization, further demonstrating the specificity of Ab161 (data not shown). These data indicate that Ab161 specifically detects the SCA7b gene product, ataxin-7b, a 101-kd protein expressed predominantly within tissues of the CNS.

Ab161 also allowed immunohistochemical analysis of ataxin-7b expression and revealed intense cytoplasmic staining predominantly within cerebellar Purkinje cells and basis pontis and inferior olivary neurons of the brainstem (Figure 4A-C). Neurons of the locus coeruleus and anterior horn also exhibited significant cytoplasmic staining (data not shown). In addition, omission of the hematoxylin counterstain revealed Ab161 immunoreactivity in cerebellar granule and molecular layer neurons (data not shown). Retinal staining with Ab161 revealed strong immunoreactivity within the photosensitive inner segments of photoreceptor cells as well as within large cell bodies residing within the ganglion cell layer, likely corresponding to displaced amacrine cells (Figure 4D). Significant staining was also detected within the cell bodies of the inner nuclear layer. In contrast, little to no Ab161 immunoreactivity occurred within photoreceptor outer segments or their cell bodies comprising the outer nuclear layer. Immunohistochemical specificity of Ab161 was demonstrated by a reduction of Purkinje cell staining intensity to background levels when cerebellar sections were probed with Ab161 preadsorbed with oligopeptide antigen (Figure 4E). These data indicate that ataxin-7b is a cytoplasmic component of neurons comprising major cerebellar and brainstem tracts and that this novel protein likely functions within the photosensitive machinery of the retina.

The formation of NIIs containing the expanded polyQ tract is a hallmark of CAG repeat disorders and has been reported in tissues of patients with SCA7, using antibodies directed at either the ataxin-7a N-terminus or the pathogenic glutamine repeat.^{21 ,25 - 26} However, Ab161 failed to detect NIIs within inferior olivary neurons, neocortical pyramidal cells, and retinal photoreceptors of patients with SCA7 with either 66 (Figure 5A-C) or 41 CAG repeats (data not shown). Instead, the neuronal distribution of Ab161 immunoreactivity in each patient with SCA7 was predominantly cytoplasmic and resembled staining within neurons of control tissues. Staining with an N-terminal anti–ataxin-7 antibody, however, revealed numerous NIIs within sections of the same tissues from the patient with 66 CAG repeats (Figure 5D-F). Since Ab161 recognizes the ataxin-7b C-terminus, these data suggest that proteolytic cleavage of ataxin-7 proteins is involved in NII formation and/or their subsequent processing.

Since the defective gene products associated with polyQ expansion diseases are widely expressed, recognition of the molecular mechanisms underlying the restricted patterns of cell death characteristic of these disorders is not straightforward.² Several explanations have been proposed for this paradox but no conclusive evidence has been presented. The formation of polyQ-containing NIIs in specific neuronal populations of patients was initially thought to mark those cells for dysfunction and death. However, more recent studies have demonstrated that NII distribution in patients does not correlate well with the neurodegenerative patterns and further suggested that nuclear aggregation of expanded polyQ proteins may be benign or even beneficial.²⁷ In addition, numerous interactions between polyQ-containing proteins and tissue-specific partners have been identified, the disruption of which by polyQ expansion has been proposed to cause spatially restricted neuronal cell death.^{4 ,6 ,28} However, most of such partnerships are not modulated by polyQ tract length, and the consequences of disrupting those that are polyQ-length–dependent are yet unclear. Finally, abnormal gene expression may be an important component of polyQ disease and may contribute to tissue-specific degeneration owing to the tendency of polyQ tracts to self-associate and promote inappropriate association of mutant proteins with polyQ-containing transcription factors, some of which may be neuron-specific.^{5 ,29 - 30}

This report raises the additional possibility that cell-specific toxicity in polyQ diseases involves expression of multiple gene products with distinct tissue or cellular distributions. Our expression analyses indicate that CNS tissues contain significant levels of ataxin-7a and 7b isoforms, whereas cells within peripheral tissues predominantly express only ataxin-7a, potentially leading to cell-specific differences in toxic polyQ loads and increased degeneration of those neuronal populations that express multiple ataxin-7 isoforms. It is possible that threshold polyQ levels are required to initiate cell dysfunction and death, thus requiring only minor differences in expanded polyQ expression to produce profound cell-specific susceptibilities to polyQ toxicity. The fact that significant ataxin-7b expression is detected within cardiac and skeletal muscle, and that patients with SCA7 have been reported to suffer from defects within these tissues, is consistent with this model and further suggests that ataxin-7b expression may contribute to non-CNS degeneration in SCA7 (Ptacek, unpublished data, 1994; Huda Zoghbi, MD, unpublished data, 1998).^{31 - 32}

Spatially restricted cell death in SCA7 may also involve the differences in subcellular localization of the ataxin-7 isoforms. Although previous studies have demonstrated that nuclear localization of mutant polyQ proteins is necessary for pathologic neuronal function, it is possible that simultaneous expression of expanded polyQ tracts in the cytoplasm exacerbates cytotoxicity.³³ In support of this hypothesis, significant levels of cell death were reported in cultured striatal neurons, wherein expression of mutant huntingtin was restricted to the cytoplasm.³⁴ Additionally, the immunohistochemical data presented in this report, together with previous reports of patients with SCA7 and murine models wherein ataxin-7a N-terminal, but not C-terminal, epitopes were detected, suggest that ataxin-7 proteins undergo proteolytic cleavage.^{35 - 36} Although we were unable to detect ataxin-7 fragments in human tissues, likely owing to sample degradation, we did observe cleavage events involving mutant ataxin-7 constructs in preliminary cell culture experiments and observed spatial distinctions between the ultimate cellular destinations of the N- and C-termini of each isoform (D.D.E., Said Bendahou, PhD, L.J.P., Y-H.F., unpublished data, 2001). Taken together, these data suggest that cleavage of ataxin-7 proteins is involved in SCA7 degeneration and raise the possibility that isoform-specific fragments exert distinct toxic influences that could contribute to selective neurodegeneration.

Recent investigation of SCA7 genomic structure delineated intron-exon boundaries and estimated intron sizes but did not identify additional coding DNA.³⁷ This study furthers the current understanding of SCA7 gene expression with the identification of SCA7b, a 4.5-kb CAG repeat–containing transcript with a coding sequence identical to that of the previously described SCA7a mRNA except for the inclusion of a 67-bp exon between exons 12 and 13.^{17 ,38} Therefore, a nomenclature revision of the SCA7 gene structure is proposed whereby the 67-bp sequence in SCA7b is denoted exon 12b and is flanked on the 5′ and 3′ ends by introns 12a and 12b, respectively. In summary, CNS-enriched expression of ataxin-7b fits well with neuropathologic findings in patients with SCA7 and may contribute to selective neurodegeneration in SCA7. Furthermore, the distinct subcellular localization of ataxin-7b raises the possibility that ataxin-7 isoforms contribute differently to the neurodegenerative pattern of SCA7. These findings suggest unique models to explain the restricted pattern of cell death in SCA7 and, since polyQ diseases involve similar genetic mutations and degeneration of specific neuronal populations, in polyQ diseases in general. Finally, these data clearly illustrate the importance of considering the complete expression profiles of CAG repeat-containing disease genes to gain a comprehensive understanding of the molecular mechanisms underlying polyQ-induced cell death.

Corresponding author and reprints: Ying-Hui Fu, PhD, 505 Parnassus Ave, Suite M798, San Francisco, CA 94122 (e-mail: yinghui@itsa.ucsf.edu).

Accepted for publication June 5, 2002.

This work was supported by Public Health Service research grant M01-RR00064 from the National Center for Research Resources, Bethesda, Md (Dr Ptacek), by grant NS41331 from the National Institutes of Health, Bethesda, Md (Dr Fu), and by a memorial fund generously established by a large Utah family with SCA7 (Salt Lake City).

Author contributions: Study concept and design (Drs Einum, Ptacek, and Fu); acquisition of data (Dr Einum); analysis and interpretation of data (Drs Einum, Clark, Townsend, Ptacek, and Fu); drafting of the manuscript (Dr Einum); critical revision of the manuscript for important intellectual content (Drs Einum, Clark, Townsend, Ptacek, and Fu); obtained funding (Drs Ptacek and Fu); administrative, technical, and material support (Drs Ptacek and Fu); study supervision (Drs Ptacek and Fu).

We are indebted to Kelley Murphy, BS, Huntsman Cancer Institute, Salt Lake City, for expert immunohistochemical assistance, and to Launce Gouw, MD, PhD, for bioinformatics support.