Spinocerebellar ataxia type 1 (SCA1) is an autosomal dominant neurodegenerative disease caused by the expansion of a polyglutamine tract within the SCA1 product, ataxin-1. Previously, using transgenic mice, it was demonstrated that in order for a mutant allele of ataxin-1 to cause disease it must be transported to the nucleus of the neuron. Using an in vitro RNA-binding assay, we demonstrate that ataxin-1 does bind RNA and that this binding diminishes as the length of its polyglutamine tract increases. These observations suggest that ataxin-1 plays a role in RNA metabolism and that the expansion of the polyglutamine tract may alter this function.
Received 7 July 2000; Revised and Accepted 25 October 2000.
Spinocerebellar ataxia type 1 (SCA1) is an autosomal dominant neurodegenerative disease caused by the expansion of a polyglutamine tract within the SCA1 product, ataxin-1 (1). SCA1 is characterized by the loss of cerebellar Purkinje cells and neurons within the brainstem. Patients develop ataxia, progressive motor deterioration and problems with breathing and the cough reflex. Onset is usually mid-life, with a 10–15 year progression to death. The polyglutamine tract of ataxin-1 is encoded by CAG trinucleotide repeats. Wild-type alleles contain 6–44 repeats and expanded affected alleles contain 39–82 uninterrupted repeats (1,2).
The mechanism by which expansion of the ataxin-1 polyglutamine tract leads to neuronal degeneration and disease remains unclear. The fact that mice with a targeted deletion in the Sca1 gene fail to show symptoms of SCA1 (3) indicates that the disease is not caused by the loss of ataxin-1 function. Rather, the development of Purkinje cell degeneration in transgenic mice overexpressing a mutant SCA1 allele with 82 glutamines supports a dominant, toxic gain-of-function model for pathogenesis (4). In the brain, ataxin-1 has a nuclear localization (5). A pathological feature of SCA1 and several other polyglutamine diseases is the presence of nuclear aggregates containing the polyglutamine protein in patient neurons (6–10) and in transgenic mouse models (11–13). Mutant ataxin-1 interacts more strongly than its wild-type counterpart with the cerebellar leucine-rich nuclear protein LANP (14) and alters nuclear matrix-associated components (13). The nuclear localization signal (NLS) in ataxin-1 has been identified and transgenic mice expressing ataxin-1 with 82 glutamines with a mutant NLS failed to develop disease (15).
Recently, several genes were identified whose expression is specifically decreased in Purkinje cells at early time points in SCA1 transgenic mice (16). These changes in gene expression correlated with the presence of mutant ataxin-1 in the nucleus and occurred only in mice with disease. A number of these genes were found to encode proteins involved in signal transduction and calcium homeostasis. Thus, there is increasing evidence that pathogenesis is dependent on nuclear localization of expanded ataxin-1 and that altered gene expression is an early effect.
To examine further the role of ataxin-1 in the nucleus, we assessed its ability to bind nucleic acids in vitro. We demonstrate that ataxin-1 does bind to RNA in vitro and that this binding diminishes as the length of its polyglutamine tract increases. These observations suggest that ataxin-1 plays a role in RNA metabolism and that the expansion of the polyglutamine tract may alter this function.
Ataxin-1 binds RNA in vitro
To test directly the RNA-binding capability of ataxin-1, we cloned the SCA1 cDNA into an expression vector that contains a T7 polymerase promoter for in vitro transcription followed by in vitro translation of the transcription product in the presence of [35S]methionine. The ataxin-1 protein was assayed for its ability to bind RNA using an equivalent amount, 40 µg, of each RNA homopolymer immobilized onto agarose beads. The translation product of SCA1 with 30 CAG repeats, ataxin-1[30Q], had an apparent size of 97 kDa on SDS–polyacrylamide gel electrophoresis (SDS–PAGE), as had been observed previously (5). Ataxin-1[30Q] showed strong binding to poly(rG), weaker but substantial binding to poly(rU) and very little binding to poly(rA) and poly(rC) (Fig. 1A). Binding to poly(rG) was stable in NaCl concentrations up to 0.50 M and binding to poly(rU) was stable in NaCl concentrations up to 0.20 M. Ataxin-1 showed weak binding to single-stranded DNA and no binding to double-stranded DNA (data not shown). A similar binding profile is characteristic of RNA-binding proteins. For example, the RNA-binding protein ICP27 from herpes simplex virus showed strong binding to poly(rG), weaker binding to poly(rU) and no binding to poly(rC) or poly(rA) (17). The binding of ataxin-1[30Q] to poly(rG) was competed with an excess of unbound poly(rG), but not with an excess of either unbound poly(rA), transfer RNA or heparin (Fig. 1B).
Whether ataxin-1 was able to bind directly to poly(rG) homopolymers was assessed using immunopurified ataxin-1. Figure 1C demonstrates that ataxin-1 purified using the 11750 antibody (5,16) coupled to Sepharose beads has the same ability to bind to poly(rG) as non-purified ataxin-1 (Fig. 1A). These results indicate that ataxin-1 binds directly to the RNA homopolymer as opposed to an RNA association mediated via ataxin-1 interacting with another RNA-binding protein produced by the in vitro translation reaction.
To determine whether the RNA-binding activity of ataxin-1 could be localized to a specific segment of the protein, translated products were derived from a series of SCA1 cDNA deletion clones (Fig. 2A). Using this approach, a substantial proportion of the RNA-binding activity of ataxin-1 was localized to its C-terminal half. The fragment of ataxin-1[30Q] containing the N-terminal portion to residue 540 showed very little RNA binding compared with full-length ataxin-1[30Q] (Fig. 2C, lanes 1 and 3). In contrast, the C-terminal fragment spanning residues 277–816 retained RNA binding (Fig. 2C, lane 6). Deletion of the C-terminal residues 767–816 had no effect on RNA binding (Fig. 2C, lane 2). These results localized the RNA-binding activity of ataxin-1[30Q] to a 226 amino acid region between residues 541 and 767.
Ataxin-1 RNA-binding capability is inversely related to length of its polyglutamine tract
To assess whether the length of the polyglutamine tract in ataxin-1 alters its ability to bind RNA in vitro, ataxin-1 with 82 glutamines (ataxin-1[82Q]) was assayed for its capability to bind RNA using RNA homopolymers immobilized onto agarose beads. As seen for ataxin-1[30Q], ataxin-1[82Q] bound most strongly to poly(rG) (Fig. 3A) and more weakly to poly(rU) (Fig. 3B). However, the stability of ataxin-1[82Q] binding to RNA with increasing concentration of NaCl was substantially less than the stability of RNA binding seen with ataxin-1[30Q]. The binding of ataxin-1[82Q] to poly(rG) was stable up to an NaCl concentration of 0.20 M (Fig. 3A) compared with ataxin-1[30Q] RNA binding that was stable to an NaCl concentration of 0.50 M (Fig. 1A). Likewise, the binding of ataxin-1[82Q] to poly(rU) was stable only to 0.05 M NaCl (Fig. 3B) versus 0.20 M NaCl for ataxin-1[30Q] (Fig. 1A). Thus, ataxin-1[82Q] showed weaker binding to ribohomopolymers than ataxin-1[30Q].
The effect of the polyglutamine tract on ataxin-1 binding to RNA homopolymers was examined further by assessing the ability of ataxin-1 lacking a polyglutamine tract to bind RNA. Using a PCR approach, a deletion from amino acid 197 to 226 was generated that included the entire polyglutamine-encoding portion of SCA1. Ataxin-1 lacking a polyglutamine tract had an increased RNA-binding activity compared with full-length ataxin-1[30Q] (Fig. 4). The proportion of input ataxin-1 bound to RNA increased 3-fold, from 4% for full-length ataxin-1[30Q], 30Q, to 12% for ataxin-1 lacking a polyglutamine tract, dCAG. Thus, the absence of a polyglutamine stretch increased the capacity of ataxin-1 to bind to a ribohomopolymer, poly(rG), in vitro.
In this report we demonstrate that ataxin-1, the product of the SCA1 gene, can bind RNA homopolymers in vitro in a manner similar to that of other RNA-binding proteins. The RNA-binding activity of ataxin-1 was localized to a 200 amino acid segment, C-terminal to the polyglutamine tract. RNA-binding proteins bind RNA via several well characterized RNA-binding domains. These include the ribonucleoprotein/RNA recognition motif (RNP/RRM), the KH motif, the RGG box, an arginine-rich motif and the double-stranded RNA-binding motif (18). A scan of the RNA-binding region of ataxin-1 failed to reveal any elements with convincing sequence similarity to one of these previously characterized RNA-binding motifs.
RNA-binding proteins are known to function in a wide variety of nuclear and cytoplasmic cellular processes, including regulation of RNA splicing, mRNA stability, transport of RNA from the nucleus to the cytoplasm and mRNA translation. Several findings indicate a possible function of ataxin-1 in a nuclear aspect of RNA metabolism. First, at least in neurons including cerebellar Purkinje cells, a prominent site of pathology in SCA1, ataxin-1 localized predominantly to the nucleus (5). This was also true when ataxin-1 was transfected into COS cells (13); a functional NLS has been identified in ataxin-1 (15). Moreover, ataxin-1 fractionated with the nuclear matrix in extracts prepared from transfected COS cells and SCA1 transgenic mouse Purkinje cells. Evidence is accumulating that the nuclear matrix may have an important role in RNA metabolism. For example, pre-mRNA (19) and protein components of the splicing complex (20) have been shown to be part of the nuclear matrix. Interestingly, in transfected COS cells, ataxin-1 with an expanded polyglutamine tract altered the distribution of the nuclear matrix-associated promyelocytic leukemia protein disrupting structures known as nuclear bodies of promyelocytic oncogenic domains (13). Thus, mutant ataxin-1 may disrupt the function of nuclear matrix-associated complexes involved in nuclear RNA metabolism.
Ataxin-1 has also been shown to interact with two other proteins found in the nucleus. One is the leucine-rich acidic nuclear protein LANP (14). LANP is a 28.6 kDa nuclear protein that is predominantly expressed in cerebellar Purkinje cells. The interaction between ataxin-1 and LANP was found to be stronger when ataxin-1 contained an expanded polyglutamine tract. In addition, when LANP was co-transfected into COS cells with ataxin-1 it was recruited into structures associated with the nuclear matrix. Although the function of LANP remains unclear, it consists of two distinct structural domains: an N-terminal leucine-rich repeat (LRR) and a C‐terminal cluster of acidic amino acids. The leucine-rich domain belongs to a superfamily of LRR proteins that function in protein–protein interactions (21). On the basis of this sequence similarity to other LRR-containing proteins, Matsuoka et al. (22) suggested that LANP may have a function in nuclear signal transduction. However, LANP is also homologous to the U2A small nuclear ribonucleoprotein particle, snRNP U2A, that is involved in alternative splicing of RNA. This latter observation is consistent with the ataxin-1–LANP complex having a role in RNA processing.
The second ataxin-1-interacting protein to be identified has been designated A1Up (23). A1Up localized to the nucleus and cytoplasm of transfected COS-1 cells. In the nucleus, A1Up co-localized with mutant ataxin-1 aggregates, further demonstrating that A1Up interacts with ataxin-1. Sequence analysis of A1Up revealed that it contains an N-terminal ubiquitin-like (UbL) region, placing it within a large family of similar proteins. In addition, A1Up has substantial homology to human Chap1/Dsk2, a protein which binds the ATPase domain of the HSP70-like Stch protein (24). Thus, A1Up may link ataxin-1 with the chaperone and ubiquitin/proteasome pathways. In addition, these data support the concept that ataxin-1 may function in the formation and regulation of multimeric protein complexes within the nucleus. The RNA-binding capability of ataxin-1 indicates that ataxin-1 may also recruit nuclear RNAs to a multimeric complex.
In transgenic mice, it has been demonstrated that in order for ataxin-1 with an expanded polyglutamine tract to cause disease it must localize to the nucleus of Purkinje cells (15). Thus, the nucleus appears to be the subcellular site of SCA1 pathogenesis. Interestingly, a genetic screen in Drosophila to identify modifiers of ataxin-1-induced neurodegeneration identified a nuclear pore protein and five proteins containing RNA-binding domains (25). These results further indicate that in vivo the nuclear localization of ataxin-1 and an alteration in RNA processing are important for pathogenesis.
The ability of an expanded allele of ataxin-1 to alter the expression of neuronal genes was recently examined using the SCA1 transgenic mice and a PCR-based cDNA subtractive hybridization strategy (16). Several genes, all expressed by Purkinje cells, were found to be down-regulated at an early stage of disease, prior to any detectable pathological or neurological alteration. Interestingly, a number of the genes found to be down-regulated encoded proteins involved in neuronal calcium signaling. These included IP3R1, SERCA2, TRP3 and inositol polyphosphate 5-phosphatase type 1. Intriguingly, all of the genes whose expression was found to be altered early on in the SCA1 transgenic mice were down-regulated. Although the mechanism of this down-regulation could involve either transcription and/or post-transcription events, the finding that ataxin-1 is an RNA-binding protein in vitro raises the possibility that the decreases in gene expression may be due to altered RNA metabolism.
What might be the biological relevance of the observation that the RNA-binding activity of ataxin-1 decreases as the length of the polyglutamine tract increases? Clearly, loss of RNA-binding capability with increasing polyglutamine length does not fit simply into a gain-of-function pathogenic model. Perhaps, if ataxin-1 is involved in the recruitment of several entities (RNA, LANP, A1Up etc.) into a nuclear complex, a decrease in its interaction with one component, for example RNA, might affect its interactions with the remaining components leading to enhancement in some cases which would be manifested as a gain-of-function alteration. It also should be considered that the RNA-binding activity of ataxin-1 is more relevant to its normal function than to SCA1 pathogenesis. If this is the case, then one might argue that the decrease in RNA-binding with increased length of the polyglutamine tract is involved in the selective pressure that drives the high polymorphism that is typical of ataxin-1 wild-type alleles.
Finally, it is worth noting that recently a protein with RNA-binding motifs has been shown to interact with the SCA2 gene product ataxin-2 (26). Thus, alterations in RNA metabolism may have a role in pathogenesis for other polyglutamine disorders.
MATERIALS AND METHODS
Ataxin-1 expression plasmids
The coding region of SCA1 cDNA (27) containing 30 CAG repeats (pCITE-SCA1[30Q]) was cloned into the EcoRI site of pCITE2b (Novagen). The pCITE-SCA1[82Q] construct was constructed by replacing the SfiI fragment containing the 30 CAG repeats with a fragment containing 82 CAG repeats. pCITE-27 (17), containing the wild-type herpes simplex virus ICP27 gene, was obtained from S. Rice (University of Minnesota). The control luciferase plasmid was supplied by Promega.
cDNA fragments encoding the C-terminal truncated forms of ataxin-1 were generated by subcloning restriction fragments cleaved from pCITE-SCA1[30Q] using restriction enzymes that cleaved once within the ataxin-1 reading frame. cDNAs encoding the C-terminal truncated forms of ataxin-1[30Q] (1–767, 1–540, 1–489 and 1–425) were produced by cleavage with SacII, BstEII, AlwNI and PinAI, respectively. The N-terminus deletion construct (277C) and the internal CAG deletion construct (dCAG) were generated by the ExSite PCR-Based Site-Directed Mutagenesis system (Stratagene) using the double stranded pCITE-SCA1[30Q] vector. Primers (Gibco BRL) flanking the regions to be deleted were designed and the reactions conducted according to the manufacturer’s directions. Mutants were confirmed by DNA sequencing. The primers used to generate the CAG deletion mutant were 5′CAG (5′-CACCTCAGCAGGGCTCCGGGGCTC-3′) and 3′CAG (5′-CTCAGCCTTGTGTCCCGGCGTCTG-3′), and the primers used to generate the N-terminal deletion mutant, 277C, were 5′277C (5′-ATGATCCCACACACGCTCACCCTG-3′) and 3′277C (5′-CACTGTCTGGATGGCTCTGATTTT-3′). All truncation mutations were confirmed by DNA sequencing.
In vitro transcription and translation
Plasmids were linearized using the appropriate restriction enzyme to generate templates for in vitro RNA synthesis by the T7 RNA polymerase. In vitro translation of the resultant transcript was in rabbit reticulocyte lysate (Boehringer Mannheim) in the presence of [35S]methionine (Amersham) according to the supplier’s directions.
RNA binding assay
In vitro translated proteins were subjected to binding with ribonucleotide homopolymers, double-stranded DNA and single-stranded DNA (28,29). Briefly, an equivalent amount, 40 µg, of ribohomopolymer attached to agarose beads (Sigma) was added to binding buffer (10 mM Tris pH 7.6, 2.5 mM MgCl2, 0.5% Triton X-100 and NaCl at varying concentrations), followed by the addition of 145 000 c.p.m. of trichloroacetic acid-precipitatable protein from the translation reaction to a final volume of 0.5 ml. The mixture was incubated for 1 h at 4°C, the beads pelleted by centrifugation and washed five times in binding buffer. Bound protein was eluted from the beads by boiling in SDS–PAGE sample buffer for 5 min and run on a 12% SDS–PAGE gel. Resulting bands were visualized by autoradiography and quantitated by densitometry (BioRad). Competitive binding assays were performed in the same manner with the addition of increasing concentrations of unbound ribopolymer or tRNA.
To assess the RNA-binding ability of purified ataxin-1, 75 µl of in vitro translated protein in 120 µl Tris 0.25 M (pH 7.5) plus 5 µl of protease inhibitor (Boehringer Mannheim) was immunopurified using 0.2 µl of the 11750 ataxin-1 antibody (5.16). The mixture was incubated overnight at 4°C. Ten microliters of Protein G–Sepharose beads (Amersham Pharmacia) was added to the 200 µl protein volume and incubated at 4°C for 1 h. The beads were centrifuged at 500 g for 3 min and washed three times with 500 µl of RIPA buffer (150 mM NaCl, 1.0% NP-40, 0.5% DOC, 0.1% SDS, 50 mM Tris pH 8.0) and resuspended in RNA binding buffer. Immunopurified protein (145 000 c.p.m.) was added to three different 1.5 ml tubes containing 40 µg of ribohomopolymer attached to agarose beads (polyguanilic acid; Sigma) and RNA binding buffer (10 mM Tris pH 7.6, 2.5 mM MgCl, 0.5% Triton X-100 and NaCl at varying concentrations) to a final volume of 500 µl. The mixture was incubated for 1 h at 4°C, centrifuged and washed five times in RNA binding buffer. Bound protein was eluted from the beads by boiling in SDS–PAGE sample buffer for 5 min, run on a 12% SDS–PAGE gel and the resulting bands were visualized by autoradiography.
This work was supported by grant NS22920 from the NINDS/NIH.
To whom correspondence should be addressed. Tel: +1 612 625 3647; Fax: +1 612 626 2600; Email: email@example.com