Expanded polyglutamines impair synaptic transmission and ubiquitin–proteasome system in Caenorhabditis elegans


Address correspondence and reprint requests to Nobuyuki Nukina, Laboratory for Structural Neuropathology, RIKEN Brain Science Institute, 2-1 Hirosawa, Wako-shi, Saitama 351-0198, Japan.
E-mail: nukina@brain.riken.jp


Polyglutamine (polyQ) expansion in many proteins, including huntingtin and ataxin-3, is pathogenic and responsible for neuronal dysfunction and degeneration. Although at least nine neurodegenerative diseases are caused by expanded polyQ, the pathogenesis of these diseases is still not well understood. In the present study, we used Caenorhabditis elegans to study the molecular mechanism of polyQ-mediated toxicity. We expressed full-length and truncated ataxin-3 with different lengths of polyQ in the nervous system of C. elegans. We show that expanded polyQ interrupts synaptic transmission, and induces swelling and aberrant branching of neuronal processes. Using an ubiquitinated fluorescence reporter construct, we also showed that polyQ aggregates impair the ubiquitin–proteasome system in C. elegans. These results may provide information for further understanding the pathogenesis of polyQ diseases.

Abbreviations used

anterior lateral microtubule


neuron associated with ALM


bovine serum albumin


canal associated neuron


cyan fluorescent protein


cyto megalo virus






enhanced green fluorescent protein


egg laying defective


Huntington's disease


human embryonic kidney


human homologues of the yeast DNA repair protein RAD23


horseradish peroxidase


Machado-Joseph disease


phosphate-buffered saline




red fluorescent protein


spinocerebellar ataxia


sodium dodecyl sulfate–polyacrylamide gel electrophoresis


ubiquitin-discosoma red fluorescent protein


ubiquitin interacting motif




ubiquitin–proteasome system


wild type

A growing number of neurodegenerative diseases are caused by the expansion of polyglutamine (polyQ) repeats in different proteins. To date, the list includes Huntington's disease (HD), spinal and bulbar muscular atrophy, dentatorubral–pallidoluysian atrophy, and spinocerebellar ataxia (SCA) types 1–3, 6, 7 and 17 (Ross 2002; Gatchel and Zoghbi 2005). A common characteristic of all polyQ diseases is the formation of nuclear or cytoplasmic and axonal or dendritic inclusions of the disease protein (Paulson 1999; Ross 2002). Although these studies support aggregate-induced toxicity, others suggest that intranuclear inclusions may reflect a protective mechanism (Arrasate et al. 2004).

Although the role of polyQ-containing aggregates in the pathogenesis of these diseases is still a matter of debate, several mechanisms have been postulated which support the hypothesis that aggregates trigger toxicity. One of these is conformational toxicity, as it has been proposed for other neurodegenerative diseases with intraneuronal aberrant aggregates of amyloidogenic proteins, such as tau in Alzheimer's disease and α-synuclein in Parkinson's disease (Bucciantini et al. 2002). Another proposed pathogenic mechanism is the sequestration of proteins such as CREB (cAMP-response-element-binding protein) binding protein (Nucifora et al. 2001). Alternatively, because intraneuronal aggregates are stained with anti-ubiquitin and anti-proteasome antibodies, an impairment of the ubiquitin–proteasome system (UPS) by the aggregates has been proposed as the pathogenic mechanism. We and others have confirmed proteasome dysfunction induced by expanded polyQs in cellular model systems (Bence et al. 2001; Jana et al. 2001; Zemskov and Nukina 2003).

The nematode Caenorhabditis elegans provides an excellent model system to address the mechanisms of neurodegenerative diseases genetically (White et al. 1986; Voisine and Hart 2004). As a model of the polyQ repeat diseases we have chosen to study SCA type 3, also known as Machado-Joseph disease (SCA-3/MJD) (Rosenberg 1992). MJD is now regarded as the most common of spinocerebellar degenerative disorders (Maciel et al. 1995). The disease gene MJD1 encodes an intracellular protein ataxin-3. MJD is caused by the expansion of polyQ in ataxin-3; in normal individuals the number of repeats ranges from 13 to 36, whereas in affected individuals this range is increased to 62–82 (Kawaguchi et al. 1994). We reported previously that ataxin-3 interacts with DNA repair protein Human homologues of the yeast DNA repair protein RAD23 (HHR)23A and HHR23B through its N-terminal ubiquitin-like domain (Wang et al. 2000), and it was shown recently that ataxin-3 binds polyubiquitylated proteins and has ubiquitin protease activity (Burnett et al. 2003). We expressed MJD1 in all neurons under the control of a pan neuronal promoter unc-119 to investigate the effects of expanded polyQ in different classes of neurons (e.g. motor neurons, sensory neurons, interneurons).

Here we report that expanded polyQ interrupts synaptic transmission, and induces aberrant branching and swelling of neuronal processes. Using an ubiquitinated fluorescent marker we showed that polyQ aggregates impair the UPS in C. elegans. These abnormalities might reflect early pathological signs of polyQ diseases before cell death.

Materials and methods

Maintenance of C. elegans strains

C. elegans were cultured using standard methods (Brenner 1974). The standard N2 Bristol strain was used as the wild type (WT) in this study. All strains were grown at 20°C, unless otherwise indicated, on a lawn of Esherichia coli.

Scoring of dead embryos

Healthy L4 larvae were seeded on individual plates and allowed to lay about 100 eggs per plate. Hatched F1 larvae were scored and removed from each plate; only the eggs were left for hatching. For transgenic animals green fluorescent protein (GFP)-positive eggs were observed for hatching over 7 days.


PolyQ transgenic animals were first fixed for 16 h at 4°C in phosphate-buffered saline (PBS) containing 4% paraformaldehyde and 1% glutaraldehyde, and then washed three times with PBS. Fixed worms were incubated for 48 h at 37°C with 5%β-mercaptoethanol, 1% Triton X-100 and 125 mm Tris (pH 7.4), washed again with PBS, and incubated for 1–2 h with vigorous shaking in 2 mg/mL collagenase prepared in buffer containing 100 mm Tris, (pH 7.5) and 1 mm CaCl2 until a few animals were broken. Samples were washed three times in PBS, and incubated with primary antiserum (α-synaptotagmin, 1 : 50; kindly provided by Dr Nonet, Washington University, St Louis, MO, USA) in incubation buffer [1% bovine serum albumin (BSA), 0.5% Triton X-100, 0.05% sodium azide, 1 mm EDTA in PBS] for 24 h at room temperature (25°C). The samples were washed four times (2.5 h) in wash buffer (0.1% BSA, 0.5% Triton X-100, 0.05% sodium azide, 1 mm EDTA in PBS). They were then incubated with Alexa fluor-conjugated goat anti-rabbit IgG (Molecular Probes, Eugene, OR, USA) for 3 h at room temperature, washed four times in wash buffer (Nonet et al. 1993) and mounted on Vectashield [4’,6-diamidino-2-phenylindole (DAPI) included; Vector Laboratories, Burlingame, CA, USA] or 3% agar pad for microscopic observation.

Cell culture transfection and treatment

Human embryonic kidney (HEK)293 cells were grown in Dulbecco's modified Eagle's medium (Sigma, St Louis, MO, USA) supplemented with 10% fetal bovine serum and penicillin–streptomycin (Invitrogen, Carlsbad, CA, USA) at 37°C in an atmosphere containing 5% CO2. Cells were transfected with ubiquitin (Ub)-Discosoma Red fluorescent protein (dsRed)2/N1 plasmid and plasmids encoding the truncated N-terminal of human huntingtin with either 17 or 150 glutamine repeats fused to enhanced GFP (tNhtt17Q-EGFP/N1 and tNhtt150Q-EGFP/N1) (Wang et al. 1999) using Lipofectamine 2000 (Invitrogen). Some 6 h later, cells were treated with 5 or 10 µm MG132 (carbobenzoxyl-leucinyl-leucinyl-leucinal), a proteasome inhibitor, for 12 and 24 h. The amount of transfected plasmid DNA used in western blot experiments was 4 µg Ub-dsRed2/N1, and 4 µg tNhtt17Q-EGFP/N1 or tNhtt150Q-EGFP/N1, per 6-cm dish. The cells cultured for immunocytochemistry in four-well chamber slides were transfected with 0.25 µg DNA for single and 0.5 µg DNA for double trasnsfection. For quantification experiments, cells were grown in 12-well plates and transfected with 0.5 µg of each plasmid.

Western blotting

Cells were sonicated briefly in PBS containing 0.5% Triton X-100 and supplemented with protease inhibitors. The cell lysate was clarified by centrifugation at 15 000 g for 10 min at 4°C, and the resulting supernatant was used for the assays. After determination of total protein (Bradford method), the samples were adjusted with PBS. Samples were boiled for 5 min in the presence of sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS–PAGE) loading buffer and then loaded on to the gel. Proteins were separated on precast 20–5% gradient gels (Atto Corporation, Tokyo, Japan) and detected by western blotting. Western blots were performed using primary antibody (mouse anti-dsRed antibody; Clontech, Mountain View, CA, USA) at a dilution of 1 : 2500 and secondary antibody [goat anti-mouse horseradish peroxidase (HRP)-conjugated antibody; Amersham, Piscataway, NJ, USA] at 1 : 4000. As a loading control, rat tubulin antibody (Chemicon, Temecula, CA, USA) at 1 : 5000 dilution and goat anti-rat HRP-conjugated antibody were used.

Fluorescence microscopy

Transgenic animals (C. elegans) were placed on a 2% agar pad and anesthetized with 1% sodium azide. The animals were observed with either an Olympus BX50 and/or a Leica DM RXA2 confocal laser scanning microscope. Images were acquired by Fluoview (for Olympus Corporation, Tokyo, Japan) and Leica Confocal Software (LCS) (for Leica Microsystems, Wetzlar, Germany). C. elegans neurons were identified based on previous descriptions (White et al. 1986; Chalfie and White 1988).

Transfected HEK293 cells were grown in four-well chamber slides and then treated with MG132 for 12 and 48 h. Treated cells were washed twice in cold PBS and then fixed in PBS containing 4% paraformaldehyde for 30 min, washed with PBS and analyzed by confocal microscopy (Leica DM RXA2).

Quantification of Ub-dsRed fluorescence

For quantification analysis, fixed and washed cells were incubated with Hoechst 33258 dye (Amersham) at a dilution of 1 : 1000 in PBS at 4°C overnight. The next day, cells were washed three times in PBS and analyzed by ArrayScan (Cellomics Inc., Pittsburgh, PV, USA) using Target Activation application software.

Pharmacological assay

Pharmacological tests were carried out as described previously (Kraemer et al. 2003). Nematode growth plates were prepared with 1 mm aldicarb or 200 µm levamisole. The plates were seeded with E. coli OP50 and used within 1 week. For each experiment, 20 animals were placed on the aldicarb and levamisole plates, and incubated at 20°C. The numbers of motile animals were counted after 2 h for levamisole and after 6 h for aldicarb. Animals that could move their bodies or pharynx after prodding the plate were scored as motile.

Locomotion assay

Transgenic and WT L4 stage animals were picked and transferred to a new nematode growth medium plate and allowed to crawl for 1 h at room temperature (25°C). Body bends per minute were counted under a microscope. Twenty animals were analyzed for each experiment, and each experiment was repeated three times.

Quantification of Ub-dsRed in C. elegans neurons

unc-47-19Q-GFP/Ub-dsRed and unc-47-127Q-GFP/Ub-dsRed transgenic animals were scanned by confocal microscopy (Fluoview software) for red fluorescence. Digital images of neurons were taken, subtracting the background fluorescence. We analyzed these digital images usingy Macscope software (Mitani Corporation, Fukui, Japan) to determine the intensity of Ub-dsRed fluorescence in neurons.


Expression and aggregation of expanded polyQ in the C. elegans nervous system

In these studies, we expressed full-length and truncated MJD1 with different numbers of glutamine repeats in the nervous system of C. elegans under the control of a pan neuronal promoter unc-119 (hereafter the full-length MJD1 constructs will be called MJD1-17Q-GFP, MJD1-91Q-GFP and MJD1-130Q-GFP; and the truncated MJD1 constructs 19Q-GFP, 33Q-GFP, 63Q-GFP and 127Q-GFP) (Supplementary Fig. S1 and Tables S1 and S2). The transgenic animals were investigated for the expression pattern of polyQ. We found that GFP(pPD95.77::punc-119), MJD1-17Q-GFP, MJD1-130Q-GFP and 19Q-GFP remained diffuse in the embryo (Fig. 1a, i–iv), 63Q-GFP aggregated in a few neurons in the embryo (Fig. 1a, v), and 127Q-GFP aggregated in most of the neurons in the embryo (Fig. 1a, vi). We found that 80% of embryos of the 127Q-GFP line remained unhatched, whereas most of the embryos from WT and other transgenic lines hatched (Fig. 1b).

Figure 1.

 Effects of expanded polyQ in C. elegans embryos. (a) expression of GFP and polyQ-GFP in C. elegans embryos: (i) GFP, (ii) MJD1-17Q-GFP, (iii) MJD1-130Q-GFP, (iv) 19Q-GFP, (v) 63Q-GFP and (vi) 127Q-GFP. Arrowheads show aggregates. Scale bar 10 µm. (b) Quantification of embryos hatched (n = 300).

In postembryonic stages we found that MJD1-17Q-GFP and MJD1-91Q-GFP remained diffuse throughout the cell body (Fig. 2a, i and ii). MJD1-130Q-GFP remained diffuse in young animals (Fig. 2a, iii) but aggregated in older animals (6 days old) (Fig. 2a, iv). GFP, 19Q-GFP and 33Q-GFP remained diffuse throughout the cell body and processes (Fig. 2b, i–iii). On the other hand, 63Q-GFP and 127Q-GFP aggregated in the cytoplasm (Fig. 2b, iv and v). The polyQ aggregates were localized mainly in the perinuclear region and rarely in the nucleus (Fig. 2b, vi). Aggregation of polyQ was confirmed by immunocytochemistry and western blot with an antibody against polyQ (data not shown). These results indicated that the aggregating properties of polyQ depend on the length of the glutamine repeats and flanking sequence.

Figure 2.

 Expression and aggregation of polyQ in the nervous system of C. elegans. (a) Aggregation of full-length MJD1 in C. elegans neurons. (i) Diffuse expression of MJD1-17Q-GFP throughout the cell body (arrow) (CANL neuron, left CAN neuron), (ii) MJD1–91Q-GFP, diffuse expression all over the cell body (arrow) (ADAL neuron, left anterior deirid ring interneuron), (iii) MJD1-130Q-GFP remained diffuse in young (4 day old) animals (arrow) (CANL neuron), and (d) MJD1-130Q-GFP aggregation in older (6 day old) animals around the perinuclear region (arrowhead) (DB7 neuron, ventral cord motor neuron). Scale bar 10 µm in (i–iv). (b) Aggregation of truncated MJD1 with normal and expanded polyQ in neurons of C. elegans. CAN neurons are shown (i) Diffuse expression of GFP all over the cell body (arrow), (ii) 19Q-GFP, diffuse expression all over the cell body (arrow), (iii) 33Q-GFP, diffuse expression all over the cell body (arrow), (iv) 63Q-GFP, aggregation (punctate GFP, arrowhead) all over the cell body, (v) 127Q-GFP, aggregation around perinuclear region (arrowhead), (vi) 127Q-GFP, aggregation (green) around perinuclear region (arrowhead) and DAPI staining (blue) of the nucleus (arrow) (in this figure the cell boundary is not visible because of the low resolution). Scale bar 10 µm.

Expanded PolyQs cause neuronal dysfunction in C. elegans

Full-length MJD1-130Q remained diffuse in young animals but aggregated in mid-aged animals(> 6 days old). We found that young transgenic animals were like WT in locomotion but the aged transgenic animals were uncoordinated (Unc) in locomotion (data not shown). On the other hand, the 127Q-GFP transgenic animals showed pleiotropic behavior. For example, the animals displayed Unc, Egl (egg laying defective) or Exp (expulsion step of defecation abnormal) phenotypes. We characterized the 19Q-GFP and 127Q-GFP transgenic animals for the Unc and Egl phenotypes. 19Q-GFP and 127Q-GFP animals were placed on an E. coli lawn and allowed to crawl overnight. 19Q-GFP animals explored the entire E. coli lawn during the overnight incubation (Fig. 3a, i) whereas the 127Q-GFP animals explored only a part of the E. coli lawn (Fig. 3a, ii). We quantified the Unc phenotype of WT, GFP, 19Q-GFP and 127Q-GFP animals by counting body bends in 1 min. The locomotion of GFP and 19Q-GFP animals was similar to that of WT but 127Q-GFP was uncoordinated (Fig. 3a, iii). In the egg laying assay we found that 19Q-GFP animals retained few eggs in the uterus (Fig. S2a), but the 127Q-GFP animals retained a huge number of eggs in the uterus (Fig. S2b). We quantified the Egl phenotype of WT and transgenic animals. WT, GFP and 19Q-GFP animals retained about 10 eggs whereas 127Q-GFP animals retained more than 40 eggs in the uterus (Fig. S2c).

Figure 3.

 Expanded polyQ causes behavioral defects and neuronal dysfunction. (a) Locomotion of 19Q-GFP and 127Q-GFP transgenic animals. (i) The 19Q-GFP transgenic animal (arrowhead) made a large number of tracks (arrow) and explored the entire E. coli lawn during an overnight incubation. (ii) The 127Q-GFP transgenic animal (arrowhead) produced few tracks (arrow) and explored only part of the lawn during the overnight incubation. (iii) Quantification of locomotion (body bends per minute) of WT, GFP, 19Q-GFP and 127Q-GFP animals. Data represent the mean ± SD of three trials. *p < 0.0001 (Student's t-test). (b) Expanded polyQs impair synaptic transmission. (i) 19Q-GFP animals were sensitive to aldicarb, an acetylcholine esterase inhibitor (left bar), whereas 127Q-GFP animals were resistant (right bar). *p < 0.001 (Student's t-test). (ii) Both 19Q-GFP and 127Q-GFP animals were sensitive to levamisole, an acetylcholine receptor agonist. Both 19Q-GFP and 127Q-GFP animals were 100% motile in normal nematode growth medium plates (data not shown).

Expanded polyQs interrupt synaptic transmission

To further understand the cause of uncoordinated locomotion we carried out a pharmacological study to determine whether synaptic transmission was affected by polyQ aggregation. We used aldicarb, an acetylcholine esterase inhibitor, and levamisole, a nicotinic acetylcholine receptor agonist, to study synaptic transmission. Both of these drugs produce muscle hypercontraction and paralysis in C. elegans (Nonet et al. 1993). We found that the 127Q-GFP transgenic animals were resistant to 1 mm aldicarb but sensitive to 200 µm levamisole, whereas the control animals (19Q-GFP) were sensitive to both aldicarb and levamisole (Fig. 3b, i and ii). These data indicated that polyQ aggregates disrupt acetylcholine release, causing 127Q-GFD animals to become resistant to the acetylcholine esterase inhibitor. On the other hand, because the animals were sensitive to the acetylcholine receptor agonist, the postsynaptic acetylcholine receptors were intact. Acetylcholine release could be disrupted owing to damage of the neuronal processes, or the synaptic vesicles might be trapped in the aggregates.

We stained WT, unc-104 (e1265) (a kinesin knockout mutant), 19Q-GFP and 127Q-GFP animals using synaptotagmin antibody (Nonet et al. 1993) to determine whether synaptic vesicles interact with polyQ aggregates. As shown in Fig. 4(a), synaptotagmin antibody predominantly stained the ventral cord, where many neuromuscular synapses are found, in WT animals. In contrast to WT, anti-synaptotagmin staining was restricted to the cell body of unc-104 (e1265) mutants (Fig. 4b), because the vesicles cannot be transported to the synapses in the absence of kinesin motor protein (also see Nonet et al. 1993). In 19Q-GFP animals the synaptotagmin antibody predominantly stained the ventral cord but not the cell bodies, similar to WT (Figs 4c–e). However, in 127Q-GFP animals the synaptotagmin antibody strongly stained the aggregates localized mainly in the cell body although the ventral cord was stained weakly (Figs 4f–h). These data indicated that synaptic vesicles may interact with the aggregates and their transport to the synapses may be interrupted.

Figure 4.

 Synaptic vesicle protein synaptotagmin co-localizes with polyQ aggregates. (a) Synaptotagmin immunoreactivity was predominantly restricted to the ventral cord (arrowheads) where many neuromuscular synapses are found. (b) Anti-synaptotagmin serum mainly stained the cell bodies (arrow) in unc-104 (e1265) mutant animals. (c) Expression of 19Q-GFP (green) in the ventral cord motor neurons (arrow) and in the ventral cord axons (arrowheads). (d) Anti-synaptotagmin serum stained (red) mainly the ventral cord axons (arrowheads) but not the cell bodies. (e) Merge of (c) and (d). (f) 127Q-GFP aggregated in the ventral cord motor neurons (arrow). (g) Anti-synaptotagmin serum stained (red) ventral cord axons (arrowheads) and also the cell bodies (arrow). (h) Merge, indicates co-localization of polyQ aggregates and synaptotagmin (arrows). Scale bar 10 µm.

Expanded PolyQ causes abnormal collateral branching and swelling of neuronal processes

To further investigate the nature of polyQ-mediated neuronal dysfunction, we examined the morphologies of the neuronal cell bodies and processes in expanded polyQ animals. To visualize neuronal processes clearly, we co-expressed GFP with polyQ-cyan fluorescent protein or polyQ-red fluorescent protein. We found abnormal collateral branching in neurites of the 127Q-CFP/GFP animals (Figs 5a, ii and iii) and 127Q-RFP/GFP animals (Fig. S3a, iii and iv). In the control animals (19Q-CFP/GFP and 19Q-RFP/GFP) we did not see any collateral branching (Fig. 5a, i; Fig. S3a, i and ii). We observed abnormal collateral branching of neuronal processes in 70% of 127Q-CFP/GFP animals (Table S3). Aberrant branching was observed only when the animals had passed the midpoint of their lives (≥ 7 days old). We did not observe aberrant branching of processes in the 127Q-CFP/GFP animals up to 6 days old (Table S4). In some 127Q-RFP/GFP animals, an excess of processes was observed extending from the cell body (Fig. S3b, ii). In the control 19Q-RFP/GFP animals we did not see this phenotype (Fig. S3b, i).

Figure 5.

 Expanded polyQ causes aberrant branching and beading of neuronal processes. (a) Aberrant branching in the processes of left CAN (CANL) neurons shown by co-expression of 127Q-CFP and GFP. (i) Visualization of the cell body (arrow) and processes (arrowheads) of a CAN neuron by expression of 19Q-CFP and GFP, revealing two unbranched processes (arrowheads). (ii) Co-expression of 127Q-CFP and GFP, showing branching of the anterior process (arrowheads); an arrow indicates the soma. (iii) CFP fluorescence of the CAN neuron shown in (ii), illustrating aggregation of 127Q-CFP in the soma (arrow). Scale bar 10 µm. (b) Expanded polyQs cause swelling of neuronal processes. (i) Co-expression of 19Q-RFP and GFP in the right ALM neuronal cell body (arrow) and process (arrowheads). (ii) Co-expression of 127Q-RFP and GFP in the right ALM neuron. Red dots indicate aggregation of 127Q-RFP in the cell body (arrow) and the arrowheads indicate swelling of the process. (iii) Co-expression of 127Q-RFP and GFP in the left ALN neuron. Yellow dots indicate aggregation of 127Q-RFP in the cell body (arrow). (iv) Swelling (arrowheads) in the anterior process of the same ALN neuron. Scale bar 20 µm.

In addition to abnormal branching, we observed swelling or beading in the neuronal processes in 127Q-RFP/GFP animals (Fig. 5b, ii and iv). Swelling was observed in the neurons of young animals (4 days old). Swellings were mostly observed in the long processes. There was swelling in the anterior lateral microtubule (ALM) (Fig. 5b, ii) and neuron associated with ALM (ALN) neurons (Fig. 5b, iv). Fig. 5(b, iii) shows the cell body of the same ALN neuron. We did not see this type of swelling in 19Q-RFP/GFP animals (Fig. 5b, i).

Expanded PolyQs impair the UPS system

In eukaryotic cells the UPS degrades unfolded or misfolded proteins. We and others have reported that polyQ and other protein aggregation impairs the UPS in cellular systems (Bence et al. 2001; Jana et al. 2001; Zemskov and Nukina 2003). To determine whether polyQ aggregates affect the UPS in vivo, we used an ubiquitin-conjugated reporter (Ub-dsRed). First we tested the reporter gene in HEK cells. Because dsRed is fused to ubiquitin, the recombinant protein is targeted for degradation by the proteasome system. We expressed Ub-dsRed in HEK cells with and without polyQ and treated these cells with MG132, a proteasome inhibitor (Jana et al. 2001). We found a much larger number of red cells in the MG132-treated cells and in the 150Q-transfected cells (either MG132 treated or untreated) but not in the 17Q cells without MG132 treatment (Figs 6a and b). We also confirmed by western blotting that Ub-dsRed was efficiently degraded in 17Q cells but not in 150Q and MG132-treated cells (Fig. 7a). Polyubiquitinated GFPu was previously shown as a high molecular weight band in western blot analysis using an anti-ubiquitin antibody (Bence et al. 2001). We did not see a band of this size, but longer exposure or blotting with anti-ubiquitin antibody might show such species on a western blot.

Figure 6.

 Expanded polyQ impairs the UPS in HEK293 cells. (a) Inhibition of UPS by MG132 in control cells caused accumulation of Ub-dsRed. HEK293 cells were transfected with Ub-dsRed plasmid and grown for 24 h without treatment (upper panel) or treated with 5 µm MG132 (lower panel). Scale bar 16 µm. (b) Cells were transfected with Ub-dsRed and 17Q-GFP or 150Q-GFP plasmids. After 6 h, MG132 was added to the medium and cells treated for 24 h (+ MG132). Scale bar 40 µm.

Figure 7.

  (a) Western blot analysis. Lysates were prepared from cells expressing Ub-dsRed alone (control) or in combination with 17Q or 150Q in the presence or absence of MG132, and analyzed by SDS–PAGE and western blotting using dsRed antibody. Lane 4 shows that Ub-dsRed accumulated in 150Q-expressing cells. (b, c) Quantification of dsRed fluorescence in control (ctrl), polyQ and MG132-treated cells. (b) To determine the percentage of cells that showed fluorescence, about 6000 cells were scanned for each sample and the experiment was repeated three times. Data represent mean ± SD. *p < 0.05 (t-test). (c) Mean total intensity of ds-Red fluorescence per cell. Values are mean ± SD (n = 6000, repeated three times). *p < 0.05 (t-test).

We quantified the proportion of Ub-dsRed-positive cells in control (without any treatment), MG132-treated, and 150Q- and 17Q co-transfected (MG132 treated and untreated) cells by ArrayScan. We found a higher proportion of Ub-dsRed-positive cells in 150Q-transfected and MG132-treated plates (Fig. 7b). We also measured the mean total intensity of Ub-dsRed/cell in control (without any treatment), MG132-treated, 150Q- and 17Q-transfected (MG132 treated and untreated) cells by ArrayScan. There was no Ub-dsRed signal in the control and 17Q cells, but we detected a high Ub-dsRed intensity in 150Q (MG132 treated and untreated) cells (Fig. 7c). These data suggested that polyQ aggregates inhibit proteasomal function in HEK cells and confirm the feasibility of this reporter gene for monitoring the UPS system in animals.

To assay the effects of polyQ aggregates on the UPS in vivo we co-expressed the Ub-dsRed with 19Q-GFP or 127Q-GFP in GABA neurons of C. elegans under the control of the unc-47 promoter. We used unc-47 promoter to express Ub-dsRed and polyQ in a small number of neurons, 26 GABergic neurons, to make it easier to follow the fluorescence intensity. We analyzed L1 (larval stage 1) larvae from both unc-47-19Q-GFP/Ub-dsRed and unc-47-127Q-GFP/Ub-dsRed transgenic lines. In the unc-47-19Q-GFP/Ub-dsRed control animals no red fluorescence was detected (Fig. 8a, upper panel). In the unc-47-127Q-GFP/Ub-dsRed animals, cells possessing polyQ aggregates had bright red fluorescence whereas cells without aggregates lacked red fluorescence (Fig. 8a, middle panel). At higher magnification (Fig. 8a, bottom panel), it could be seen that Ub-dsRed accumulated throughout the cell body. We quantified the number of neurons with accumulated Ub-dsRed. We found that 94% of the neurons that had aggregated polyQ showed Ub-dsRed fluorescence, whereas only 3.5% of the neurons that had diffuse polyQ showed Ub-dsRed fluorescence (Table S5). It may be that neurons with a diffuse pattern of polyQ-GFP did not accumulate the UPS reporter because they did not express the reporter transgene. We could rule this out because the neurons with diffuse polyQ-GFP in early-stage (L1) larval animals did not show Ub-dsRed, but when the animals reached adulthood the polyQ-GFP aggregated in the same neurons and Ub-dsRed became visible (Table 1). We also found Ub-dsRed accumulation in unc-47-19Q-GFP animals at later stages (larval stage 3 to adult) but the intensity of fluorescence was about 40% lower than that of 127Q-GFP animals (Fig. 8b). To ensure that the observed increase in Ub-dsRed fluorescence was not due to increased transcription of Ub-dsRed mRNA, we also determined the mRNA level of Ub-dsRed by quantitative RT–PCR. We did not find any difference in mRNA levels between control and expanded polyQ animals (Fig. S4). These data indicated that Ub-dsRed accumulates as a result of an impaired UPS and not as a consequence of increased transcription of the reporter gene.

Figure 8.

 Expanded polyQ impairs the UPS in C. elegans. (a) The upper panel shows co-expression of Punc-47::Ub-dsRed and Punc47::19Q::GFP in the GABA neurons of L1 larvae. Ub-dsRed was not visible in the GFP-expressing cells (arrows). Anterior is to the right, ventral is to the lower left. The middle panel shows co-expression of Punc-47::Ub-dsRed and Punc-47::127Q::GFP in GABA neurons of L1 larvae. Ub-dsRed accumulated in the polyQ-aggregating cells (arrowheads), but was not visible in the non-aggregating cells (arrow). Anterior is to the right and ventral is to the top. Scale bar 20 µm. Lower panel shows an expanded polyQ-expressing cell at higher magnification, revealing that Ub-dsRed remained diffuse all over the cell body. Scale bar 5 µm. (b) Quantification of Ub-dsRed intensity/cell in adult C. elegans neurons. In each experiment 10 animals were used and the experiment was repeated three times. Data represent mean ± SD. *p < 0.001 versus 19Q-GFP. t-test.

Table 1.   Number of neurons with diffuse and aggregated pattern of polyQ that accumulated Ub-dsRed in L1 larval and adult stages of C. elegans
MarkersL1 larvaeAdult
  1. n = 50 animals of each age.



Although a number of neurodegenerative diseases are now known to be caused by expanded polyQ, the mechanism of pathogenesis is not well understood. Moreover, whether polyQ aggregates are toxic or protective is still being debated (Ross 2002; Arrasate et al. 2004).

We used the simple animal C. elegans to study the polyQ diseases. We expressed full-length and truncated human ataxin-3 with WT and expanded polyQ in the nervous system of C. elegans to investigate neuronal dysfunction. We found that truncated ataxin-3 (without ubiquitin interacting motifs; UIMs) with expanded polyQ forms aggregates at a very early stage (comma stage of egg) whereas the full-length ataxin-3 with expanded polyQ forms aggregates at a later stage (animals ≥ 6 days old). This result suggests that the aggregation property of polyQ peptides depends on protein length and the number of glutamine repeats. Truncated peptides aggregated faster than the full-length protein. A similar observation has been reported in a cellular system (Jana and Nukina 2004).

In the 127Q-GFP transgenic lines, 80% of the transgenic embryos remained unhatched and many of them had severe morphological abnormalities, indicating that expanded polyQ either caused severe neuronal dysfunction, preventing the larvae from hatching, or caused cellular dysfunction in the early stage leading to developmental abnormalities.

The surviving 127Q-GFP transgenic animals were uncoordinated in locomotion and defective in egg laying, whereas GFP and 19Q-GFP animals were similar to WT in locomotion and egg laying behavior. These phenotypes suggest that the expanded polyQ causes neuronal dysfunction in C. elegans. We examined acetylcholine release in the transgenic animals. Our pharmacological assay showed that acetylcholine release was impaired by the aggregates but that postsynaptic acetylcholine receptor function was unaffected. Immunostaining of the polyQ animals using synaptotagmin antibody showed that synaptotagmin co-localized with polyQ aggregates. These immunostaining data indicate that a proportion of the synaptic vesicles may be trapped in polyQ aggregates and so cannot reach the synapses. This may lead to impaired acetylcholine release and uncoordinated locomotion of expanded polyQ animals. Previous reports showed that striatal amino acid neurotransmitter release was altered in R6/1 Huntington transgenic mice (Nicniocaill et al. 2001) and the expression of pathogenic polyQ in neurons of Drosophila causes a defect in axonal transport (Gunawardena et al. 2003). Similarly, expression of a huntingtin fragment containing expanded polyQ inhibits fast axonal transport in isolated squid axoplasm (Szebenyi et al. 2003). Our results also indicated that expanded polyQ disturbed axonal transport and synaptic transmission.

We further analyzed the transgenic animals for morphological abnormalities in their neuronal cell bodies or processes. We observed aberrant branching of the neuronal processes in mid-aged animals (≥ 7 days old). Branching was seen in mechanosensory neurons, interneurons and canal associated neuron (CAN) neurons, usually in the long processes. Age-dependent aberrant branching (Tables S3 and S4) suggests that this is a progressive abnormality. Alterations in dendritic branching have also been reported in post-mortem brain tissues of patients with HD (Ferrante et al. 1991).

A similar type of aberrant branching has been reported in the neurons of C. elegans ced-10;mig-2 double mutants. Both CED-10 and MIG-2 are members of the Rac small GTPase family and are involved in actin polymerization (Lundquist et al. 2001). It is possible that polyQ aggregates sequester actin and/or Rac proteins causing disorganization of the cytoskeleton leading to aberrant axonal branching.

We also found beading or swelling in the neuronal processes in 127Q-CFP/GFP animals. Swelling in expanded polyQ-expressing neurons was mainly found in long processes. These swellings were progressive, growing as the animals aged. PolyQ aggregates were present in most of the swollen areas analyzed. In another C. elegans HD model, neuronal process swelling was observed when polyQ (exon 1 of huntingtin) was expressed in mechanosensory neurons (Parker et al. 2001). We found that expression of polyQ from the MJD1 gene in all neurons causes swelling in the mechanosensory neurons, interneurons and, occasionally, in the CAN neurons. Collectively, these results suggest that polyQ aggregates, whether in the context of huntingtin or ataxin-3, cause swelling of neuronal processes.

Expanded PolyQs also cause swelling of neuronal processes in cultured motor neurons. These swellings are rich in mitochondria and occasionally contain accumulations of kinesin (Piccioni et al. 2002). Roediger and Armati (2003) reported that axonal swelling and beading is consistent with a disruption of microtubules by oxidative stress and the subsequent arrest of axonal transport. Interestingly, a similar type of neurite swelling has been described in other neurodegenerative disorders, for example amyotropic lateral sclerosis (Harry 1992).

We have shown previously that expanded polyQ reduces proteasomal activity in a cellular system (Jana et al. 2001; Zemskov and Nukina 2003; Nagaoka et al. 2004). Expanded polyQ-containing molecules are also degraded inefficiently by the proteasome (Venkatraman et al. 2004; Holmberg et al. 2004). In this study we asked whether polyQ aggregates affect the UPS in vivo by directly visualizing its substrate (Ub-dsRed) in C. elegans. We expressed the Ub-dsRed in HEK cells under the control of the cytomegalovirus (CMV) promoter and in C. elegans under the control of the GABA transporter promoter (unc-47). We found that expanded polyQ impaired UPS both in HEK cells and C. elegans neurons. Our quantitative RT–PCR data showed that Ub-dsRed reporter mRNA levels in control and expanded polyQ cells and animals were almost the same. Based on these data, we conclude that the accumulation of Ub-dsRed in HEK cells and in C. elegans was due to impairment of the UPS. Bowman et al. (2005) also found accumulation of Ub-GFP reporter in expanded polyQ-expressing mouse retina. They observed that an increase in Ub-GFP mRNA levels accounted for the increase in protein levels. It was suggested that increased accumulation of Ub-GFP was not due to impairment of the UPS. Transcription of the Ub-dsRed reporter may depend on promoter type, cell type and/or on the system used in the study. We used CMV promoter in HEK cells and unc-47 promoter in C. elegans, and did not find enhancement of Ub-dsRed transcription. Bowman et al. used CMV enhancer-chick β-actin promoter to express the reporter in their experiments. It is possible that under their experimental conditions the expanded polyQ either directly or indirectly enhanced promoter activity because different types of promoters require different transcription enhancers or suppressors. In our systems we overexpressed polyQ, whereas Bowman et al. (2005) used a knockin model mouse. These differences might explain the observed discrepancy. In particular, overexpression of polyQ may lead to UPS dysfunction in our cellular and C. elegans systems. Because knockin mouse models better resemble SCA7 disease, it might be reasonable to use other promoters to express Ub-GFP, to test the effect of expanded polyQ on UPS function in vivo.

Our previous study showed that HHR23A and HHR23B proteins interact with the N-terminal fragment of both normal and mutant ataxin-3 (Wang et al. 2000). Another group has shown that the N-terminal part of ataxin-3 has ubiquitin protease activity, and that the UIM domain binds polyubiquitylated proteins (Burnett et al. 2003). Our results suggest, even without those UPS-related domains, that truncated expanded polyQ domain could inhibit UPS in C. elegans.

As mentioned above, we have observed altered proteasomal functions in cells expressing expanded polyQ. Others have also reported that expanded polyQ and other aggregate-prone proteins caused UPS dysfunction (Bence et al. 2001; Burnett et al. 2003; Bennett et al. 2005). Aggregate-prone proteins have also been reported to cause UPS dysfunction in other neurodegenerative diseases, for example mutant α-synuclein in Parkinson's disease (Tanaka et al. 2001), suggesting that dysfunction of the ubiquitin–proteasome pathway might be involved in the mechanism of such diseases.

In summary, expression of expanded polyQ in the nervous system of C. elegans caused behavioral abnormalities. We found defective synaptic transmission, morphological abnormalities in neuronal processes and impaired UPS function in these animals. These cellular abnormalities may account for the observed behavioral abnormalities. Our results indicate that impaired synaptic transmission and UPS dysfunction contribute to the early pathogenesis of neurodegenerative diseases. Several different model systems are now available and used for the study of polyQ diseases. Some abnormalities may be system specific. Findings in each system should be carefully examined to understand the real pathological process of human disorders.


We would like to thank to Professor Roger Y. Tsien (University of California, San Diego, CA, USA) for mRFP; Dr Nonet (Washington University, St Louis, MO, USA), for synaptotagmin antiserum; Dr T. Ishihara (NIG, Mishima, Japan) for Fire Laboratory vectors; the Caenorhabditis elegans Genetics Center (which is supported by National Institutes of Health) for nematode strains; Dr J. Doumanis for reading the manuscript; and our laboratory members for their generous support. This project was supported partly by the Grant-in-Aid for Scientific Research on Priority Areas from MEXT, Japan (17025044), Japan Society for the Promotion of Science (JSPS), and Grant-in-Aid by the Ministry of Health, Labour and Welfare. LAK was a JSPS postdoctoral fellow.