Neurobiology of Disease 37 (2010) 284–293

Contents lists available at ScienceDirect

Neurobiology of Disease

j ourna l homepage: www.e lsev ie r.com/ locate /ynbd i

A transgenic mouse model of spinocerebellar ataxia type 3 resembling late diseaseonset and gender-specific instability of CAG repeats

Jana Boy a,1, Thorsten Schmidt a,⁎,1, Ulrike Schumann a, Ute Grasshoff a, Samy Unser a, Carsten Holzmann b,Ina Schmitt c, Tim Karl d,2, Franco Laccone e, Hartwig Wolburg f, Saleh Ibrahim g, Olaf Riess a

a Medical Genetics, University of Tuebingen, Calwerstrasse 7, 72076 Tuebingen, Germanyb Medical Genetics, University of Rostock, Rostock, Germanyc Neurology, University of Bonn, Bonn, Germanyd Functional and Applied Anatomy, Medical School of Hannover, Germanye Department for Medical Genetics, Medical University Vienna, Vienna, Austriaf Institute for Pathology, University of Tuebingen, Tuebingen, Germanyg Immunogenetics, University of Rostock, Rostock, Germany

⁎ Corresponding author. Fax: +49 7071 29 5228.E-mail address: Thorsten.Schmidt@med.uni-tuebing

1 These authors contributed equally to this project.2 Present address: Prince of Wales Medical Research

Available online on ScienceDirect (www.scienced

0969-9961/$ – see front matter © 2009 Elsevier Inc. Adoi:10.1016/j.nbd.2009.08.002

a b s t r a c t

a r t i c l e i n f o

Article history:Received 10 March 2009Revised 31 July 2009Accepted 10 August 2009Available online 20 August 2009

Keywords:Spinocerebellar ataxia type 3Machado–Joseph diseaseSCA3MJDPolyglutamineIntranuclear inclusion bodiesTransgenic mouse modelCAG repeat instabilityLate onset

Spinocerebellar ataxia type 3 (SCA3), or Machado–Joseph disease (MJD), is caused by the expansion of apolyglutamine repeat in the ataxin-3 protein. We generated a mouse model of SCA3 expressing ataxin-3with 148 CAG repeats under the control of the huntingtin promoter, resulting in ubiquitous expressionthroughout the whole brain. The model resembles many features of the disease in humans, including a lateonset of symptoms and CAG repeat instability in transmission to offspring. We observed a biphasicprogression of the disease, with hyperactivity during the first months and decline of motor coordination afterabout 1 year of age; however, intranuclear aggregates were not visible at this age. Few and small intranuclearaggregates appeared first at the age of 18 months, further supporting the claim that neuronal dysfunctionprecedes the formation of intranuclear aggregates.

© 2009 Elsevier Inc. All rights reserved.

Introduction

Spinocerebellar ataxia type 3 (SCA3), or Machado–Joseph disease(MJD), is an autosomal-dominantly inherited neurodegenerativedisorder caused by the expansion of a CAG repeat in the MJD1 gene(Kawaguchi et al., 1994). While, in unaffected individuals, the numberof CAG repeats is usually less than 45 (Padiath et al., 2005), it increasesup to 86 repeats in SCA3 patients (Riess et al., 2008). The expandedCAG repeat results in an expanded polyglutamine stretch in theencoded ataxin-3 protein. SCA3 therefore belongs to the group ofpolyglutamine diseases, which includes other types of spinocerebellarataxias as well as SBMA, DRPLA, and Huntington's disease (reviewedin Gatchel and Zoghbi, 2005). The protein harboring the expandedpolyglutamine tract has a strong tendency for aggregation (Scherzin-

en.de (T. Schmidt).

Institute, Randwick, Australia.irect.com).

ll rights reserved.

ger et al., 1999), which usually manifests in the nucleus of neuronalcells in SCA3. Interestingly, while ataxin-3 is ubiquitously expressedthroughout the whole brain, the formation of intranuclear inclusionbodies appears in specific brain regions (Paulson et al., 1997b;Schmidt et al., 1998). There is an ongoing discussion whether theseintranuclear inclusions are toxic or perhaps even protective forneuronal cells (Michalik and Van Broeckhoven, 2003; Sisodia, 1998).

Clinically, SCA3 presents with a highly heterogenous phenotype,leading to differentiation into clinical subtypes (reviewed in Riess etal., 2008). In addition, the disease is characterized by a late onset,usually appearing in the late third decade of life (Dürr et al., 1996;Schöls et al., 1997), and by slow progression. In contrast to the diseasecourse in human patients, most of the previously generated mousemodels of SCA3 are characterized by an early onset and rapidprogression (Bichelmeier et al., 2007; Cemal et al., 2002; Chou et al.,2008; Goti et al., 2004; Ikeda et al., 1996) with rather distinct regionalmanifestation. Thus, not all the pathogenic aspects of the disease inhumans are reproduced by these models.

Here, we present a mouse model for SCA3 with late onset ofsymptoms that resembles major genetic and pathogenetic character-istics in humans and demonstrates that motor symptoms precede the

285J. Boy et al. / Neurobiology of Disease 37 (2010) 284–293

occurrence of intranuclear inclusion bodies. For this new transgenicmouse model, we utilized the well-characterized rat huntingtinpromoter (Holzmann et al., 1998, 2001) to express a full-lengthataxin-3 construct containing 148 CAG repeats ubiquitously through-out the brain.

In transgenic mice, the instability of the CAG repeat expansion wasobvious, and clearly correlated with the sex of the transmittingparent. We additionally observed a biphasic course of the disease anddetected hyperactivity in HDPromMJD148 mice long before the onsetof motor deficits, but the formation of intranuclear inclusion bodieswas no prerequisite for the onset of symptoms.

The novel mouse model of SCA3 presented in this study is avaluable addition to previous published mouse models, since itsslowly progressing phenotype allows us to study aspects of thedisease that cannot be analyzed in other models of this disease.

Materials and methods

Transgenic mice

To generate a novel transgenic mouse model for SCA3, a 764 bpXbaI restriction fragment of the rat huntingtin promotercorresponding to nucleotide positions −777 to −14 was cloned tothe 5′ end of the full-length ataxin-3 cDNA (isoform ataxin-3c) (Gotoet al., 1997; Schmitt et al., 1997) in the pBluescript SK vector. Thetranscriptional activity of this promoter fragment has been demon-strated before (Holzmann et al., 1998). The SV40 early mRNApolyadenylation signal was amplified from the vector pEGFP-C(Clontech, Mountain View, CA) using the primers pEGFP-C-PolyAKpn-For (5′-GTT GGT ACC AAC TTG TTT ATT GCA GCT TA-3′) andpEGFP-C-PolyA Kpn-Rev (5′-TTT GGT ACC TAA GAT ACA TTG ATG AGTTT-3′). The primers attach KpnI restriction sites to both sides of thePCR product which were used to insert the polyadenylation signal atthe 3′ end of the ataxin-3 cDNA. 148 CAG repeats were cloned into theconstruct using the BglII and XhoI restriction sites. Extension of theCAG repeat to 148 CAG repeats was performed as described earlier(Laccone, 2002). The transgene construct, comprising the rat hun-tingtin promoter fragment, the full-length ataxin-3 cDNA with 148CAG repeats and the polyadenylation signal, were linearized andseparated from the vector backbone using the restriction enzymesNotI and PvuI. Seven injections into fertilized murine oozytes of theC57BL/6N mouse strain were performed, and the resulting 25offspring were genotyped. The transgene was detected in fourindividuals. After crossing these founders with wildtype mice, onestable mouse line (line 3746, designated HDPromMJD148) wasgenerated that expresses the ataxin-3 transgene.

Western blot

Mice were killed by CO2 inhalation, the tissue was freshlyprepared, immediately snap frozen, and stored at−80 °C. For proteinisolation, tissue was homogenized at 30,000 rpm using a tissuehomogenizer (Ultra-Turrax; IKA Werke, Staufen, Germany) in TESbuffer (50 mM Tris, pH 7.5, 2 mM EDTA, and 100 mM NaCl)supplemented with a mixture of protease inhibitors (complete; RocheApplied Science, Mannheim, Germany). After the addition of NonidetNP-40 (Sigma-Aldrich, Seelze, Germany) to a final concentration of1%, and incubation at 4 °C for 15 min, debris was removed bycentrifugation (15 min each, 20,000 relative centrifugal force (rcf),4 °C). The cleared protein extract was supplemented with glycerol(final concentration of 10%; VWR International, Darmstadt, Germany)and stored at −80 °C. The protein concentration was determinedusing a protein assay (Protein Assay Dye Reagent Concentrate; Bio-Rad, Munich, Germany) based on the method described by Bradford(1976) according to the manufacturer's instructions. Protein extracts(30 μg of each) were supplemented with loading buffer (80 mM Tris

pH 6.8, 0.1 MDTT, 2% SDS, 10% glycerol, bromphenol blue), denatured,and analyzed in PAGE buffer (192 mM glycine, 25 mM Tris, 1% SDS)using SDS-PAGE (Blue Vertical 100/C; Serva, Heidelberg, Germany),according to the method described by Laemmli (1970). The separatedproteins were transferred to a Polyvinylidenedifluoride (PVDF)membrane (Immobilon-P Transfer-Membran, Millipore, Schwalbach,Germany) in transfer buffer (0.2 M glycine, 25 mM Tris, 20%methanol). The detection of the protein was performed essentiallyas described previously (Schmidt et al., 1998). Briefly, the membranewas blocked in 5% dry milk in TBST buffer (10 mM Tris, pH 7.5, 0.15 MNaCl, 0.1% Tween 20) for 2 h at room temperature. The primaryantibody was diluted in TBST. The generation of our anti-ataxin-3antibody (diluted 1:1000) has been described previously (Schmidt etal., 1998). The 1H9 antibody against ataxin-3, as well as the 1C2antibody directed against expanded polyglutamines, were purchasedfrom Chemicon (Hofheim, Germany). After incubation for 2 h, themembrane was washed four times with TBST for 15 min. Thesecondary antibody, coupled to horseradish peroxidase (GE Health-care, Freiburg, Germany), was incubated with the membrane for75min. After four washing steps with TBST (15min each), bandswerevisualized using the enhanced chemiluminescence method (ECL; GEHealthcare) and by exposure to Hyperfilm ECL (GE Healthcare).

Immunohistochemistry

Mice were deeply anaesthetized by CO2 inhalation and transcar-dially perfused using 4% paraformaldehyde (PFA) in 0.1 M PBS. Brainswere removed from the skull and postfixed overnight in fixative,embedded in paraffin and cut into 7 μm sagittal sections. Theimmunohistochemical staining of paraffin-embedded tissue wasperformed as described previously (Schmidt et al., 2002). Briefly,after deparaffinization of sections in xylene and rehydration in agraded alcohol series, slides were microwaved for 15 min in 10 mMsodium citrate, pH 6.0. The slides were washed with PBS; endogenousperoxidases were blocked using 1% hydrogen peroxide in 40%methanol for 10 min and blocked using 5% normal goat serum in PBSsupplemented with 0.3% Triton. After washing with PBS (three timesfor 10 min), the primary antibody (diluted in PBS plus 3% goat serum)was added and incubated at 4 °C overnight in a humid chamber. Thesecondary antibody, coupled with biotin (Vector Laboratories, Burlin-game, CA), was diluted the same way in PBS plus 1.5% goat serum andadded after washing the slides with PBS. After incubation for 30min atroom temperature and a brief wash with PBS, an ABC enhancercomplex coupled with peroxidase (Vector Laboratories) was addedand incubated for 30 min at room temperature. After washing withPBS, the substrate (DAB; Sigma-Aldrich) was added, and the reactionwas stopped in distilled water after the desired degree of staining wasreached. Finally, slides were dehydrated again and mounted using CVmount (Leica, Bensheim, Germany).

Staining was visualized using an Axioplan 2 imaging microscope(Carl Zeiss Microimaging, Oberkochen, Germany) equipped with anAxio-Cam MR color digital camera (Carl Zeiss Microimaging) using a40× Plan Neofluar and a 63× Plan/Apochromat objective and theAxioVision 4.6 software package (Carl Zeiss Microimaging).

Genotyping and fragment analysis

For the genotyping of transgenic mice, DNA isolated from earbiopsy tissue and the following primer combination was used:MJDPromA-Int-F (5′-CTT TGG TTC CGC TTC GGT CT -3′), MJDPromA-Int-R (5′-CTT CTC CTC CTC ATC CAG CT -3′).

In order to determine the CAG repeat number, the primershMJD1c-CAG-Frag-F (5′-Cy5-GCT AAG TAT GCA AGG TAG TTC C-3′)and hMJD1c-postGAG-R (5′-CAA GTG CTC CTG AAC TGG TG -3′) wereused to specifically amplify the CAG repeats. For detection, theforward primer was marked with Cy5, and for size determination, an

Fig. 1. Generation of HDPromMJD148 transgenic mouse model of SCA3. (A) Constructused for the generation of transgenic mice. A full-length ataxin-3 cDNA with 148 CAGrepeats is expressed under the control of a 764-bp fragment of the rat huntingtinpromoter. To ensure stability of the encoded mRNA, the construct also contains apolyadenylation (polyA) signal. (B) Expression of transgenic ataxin-3 in differenttissues. The expression of the ataxin-3 transgene was analyzed in different tissues. Intissue culture (“construct”, transgene constructs overexpressed in SK-N-AS neuroblas-toma cells) and in the brain of our transgenic mouse line, the ataxin-3 transgene isdetected using an anti-ataxin-3 antibody (1H9, Chemicon) while no expression can bedetected in the negative control (brain of a wildtype mouse), testis, muscle, lung, liver,and kidney. The band detected in the heart is a background band since it is not detectedusing the 1C2 antibody directed against expanded polyglutamine repeats (polyQ). TBP,which is co-detected by the 1C2 antibody (see below), is not shown on this blot since itwas allowed to run out of the gel for optimal separation. (C) Comparison of transgeneexpression in whole brain lysates of mice transgenic for ataxin-3 with 148 CAG repeatsunder the control of a prion protein promoter (PrpProm, line 148.19) (Bichelmeier et al.,2007) and of the huntingtin promoter (HDProm, line 3746). The Prp promoter results inabout 2.5-fold higher expression than the huntingtin promoter. The transgene band (tg)is not present in the wildtype mouse (wt). (D) The huntingtin promoter results in aneven distribution of transgene expression throughout the whole brain. Whencomparing different brain regions, no difference in transgene expression is observedbetween the whole brain (brain), the telencephalon (tel), the cerebellum (ceb), and thebrainstem (bs). The western blot in B (bottom), C, and D were processed using the 1C2antibody directed against expanded polyglutamines. TBP marks the TATA bindingprotein which is usually co-detected when the antibody 1C2 is used, since this antibodywas originally generated against TBP (Trottier et al., 1995).

286 J. Boy et al. / Neurobiology of Disease 37 (2010) 284–293

internal standard (DNA-Size Standard Kit-600, Beckman Coulter,Krefeld, Germany) was used. Fragment detection was performedusing a CEQ8000 Genetic Analysis System (Beckman Coulter). Foradditional calibration, we used ataxin-3 constructs for which repeatlengths were confirmed by sequencing (15, 77, or 148 CAG repeats).

Home cage activity

To analyze the spontaneous home cage activity of the mice, theLabMaster system (TSE Systems, Bad Homburg, Germany) was used.The mouse cages were placed into sensor frames, the number of beambrakes was counted, and the total activity, ambulatory and finemovements, as well as rearings, were quantified. In addition, drinkingand feeding behavior was analyzed. The analysis was performed for23 h, starting with 5 h of light phase, continuing with 12 h of darknessand concluding with an additional 6 h of light. For analysis, beambreaks were counted in 15 min intervals. In each run, four mice wereanalyzed in parallel using four separate test systems. In order toexclude any bias, to serve as internal control, and for reproduction,both transgenic and control mice were routinely analyzed in paralleland were randomly distributed between the test systems.

Open field analysis

For the assessment of the explorative behavior and emotionality ofthe mice, open field tests were performed. Mice were placed in a50×50 cm arena with 50 cm high walls and their movement activitywas recorded for 15 min using the TSE VideoMot2-Video ActivityTracking System (TSE Systems, Bad Homburg, Germany). The lightintensity was set to be at least 150 lux in the corners, and not higherthan 200 lux in the center of the arena (Brown et al., 2005; Green etal., 2005). To analyze the collected data, the arena was divided intodifferent regions. Region 1 is the border, with a width of 8 cm, region 3the center, which comprises 16% of the overall area, and region 2 thearea between the border and the center.

Rotarod

To measure the motor coordinative abilities and balance of thetransgenic mice, Rotarod analyses were performed. At a maximumillumination of 100 lux, mice were tested on an accelerating Rotarod(TSE Systems, Bad Homburg, Germany) starting at 4 rpm andaccelerating to 40 rpm over a period of 300 s (5 min). Three trialsper test day, in which the latency to fall off the rotating rod wasrecorded, were carried out with 15 min rest between trials (Brown etal., 2005; Green et al., 2005). The tests were repeated approximatelyevery 6 weeks. Statistical analysis of the data thus obtained wasperformed using Excel (Microsoft, Unterschleiβheim, Germany),calculating the mean values (in s) for each group. Both t-tests(Excel) and two-way ANOVA (GraphPad Prism, GraphPad Software,La Jolla, CA) were performed. Data are represented as mean±SEM.

Electron microscopy

Electron microscopic analyses were essentially performed asdescribed previously (Lundkvist et al., 2004). Briefly, after beingdeeply anaesthetized by CO2 inhalation, mice were transcardiallyperfused using 2.5% glutaraldehyde in 0.1 M cacodylate buffer (pH7.4). Brains were removed and postfixed in fixative for 30 min to 1 h.After dehydration in ethanol, brain pieces were incubated overnightin 70% ethanol saturated with uranyl acetate, followed by anadditional dehydration in absolute ethanol and propylene oxide.Finally the samples were embedded in Araldite 502 (Sigma-Aldrich)and sectioned on a Leica FCR Ultracut ultramicrotome (Leica).Ultrathin sections were stained with lead citrate and examined witha Zeiss EM 10 electron microscope.

Animal experiments

All animal experiments were conducted in accordance withinternational standards on animal welfare, and comply with thestandards defined by the European Communities Council Directive of24 November 1986 (86/609/EEC). Animal health was routinelymonitored by local veterinarians and supervised by the regional board.

Results

The huntingtin promoter led to transgene expression throughoutthe whole brain

To control the expression of a full-length construct of ataxin-3containing 148 CAG repeats, a 764-bp fragment of the well-characterized rat huntingtin promoter was used (Fig. 1A). Thispromoter fragment contains numerous conserved putative

287J. Boy et al. / Neurobiology of Disease 37 (2010) 284–293

transcription factor binding sites, which have led to robust expressionin a variety of cell lines derived from different rodents (Holzmann etal., 1998). A comparable fragment of this promoter has beensuccessfully used before to generate a transgenic rat model ofHuntington's disease (von Hörsten et al., 2003). At the 3′-end ofataxin-3, a SV40 early mRNA polyadenylation signal was introduced.The integrity of the construct was confirmed by sequencing, and theexpression of the ataxin-3 transgene was verified in tissue culture(Fig. 1B).

Using this construct, one stable mouse line (line 3746, designatedHDPromMJD148) was generated. In this mouse line, the transgene isexpressed exclusively in the brain, while no expression was observedin peripheral tissue (Fig. 1B). We previously generated a SCA3transgenic mouse line under the control of a Prion protein promoter,which manifests with early onset and a rapid disease progression(Bichelmeier et al., 2007). Compared to this line, the HDPromMJD148line has lower expression of ataxin-3 in the brain (Fig. 1C). In order tomake sure that this reduced signal is not just due to a weakerexpression in specific brain regions, we compared the expression inthe whole brain with the telencephalon, the cerebellum, and thebrainstem. These analyses revealed that the transgene expression inthe HDPromMJD148 line is evenly distributed throughout the brain(Fig. 1D). Immunohistochemical analyses confirmed this observation.In all analyzed brain regions, the transgene can be detected mainly inneuronal cells. In the cerebellum, the transgene is expressed in the

Fig. 2. Even distribution of transgene expression throughout the brain. Immunohistochemicmainly neuronal distribution of the ataxin-3 transgene throughout the brain. Shown are reprcerebellum (cerebellar cortex and cerebellar nuclei). In the cerebellar cortex, the transgene i148 polyglutamine repeats results in a predominantly nuclear localization of the transgene inanalyzed at the age of 14 months. No intranuclear inclusion bodies were observed at this timcerebellar cortex; GrCb, granular layer of the cerebellar cortex. Bar, 20 μm.

cerebellar nuclei as well as in all three layers of the cerebellar cortex,including Purkinje cells (Fig. 2). Interestingly, in homozygous mice,the expanded polyglutamine stretch within ataxin-3 gives rise to apredominantly nuclear localization, while in heterozygous mice, thetransgene is mainly located in the cytoplasm (Fig. 6).

CAG repeats within the HDPromMJD148 transgene behave unstably

Analyzing brain lysates from different HDPromMJD148 mice bywestern blotting revealed mutant transgenic ataxin-3 at slightlydifferent sizes (Fig. 3A, top). To investigate whether these sizedifferences were due to different CAG repeat numbers within therespective transgenes, fragment analyses were performed. Theseanalyses confirmed that the observed size differences of transgenicataxin-3 protein were due to different CAG repeat numbers: in theanalyzed mice, between 135 and 155 CAG repeats were counted(Fig. 3A, bottom).

We next wanted to find out whether the inheritance of thetransgene was linked to the variation in the CAG repeat number. Wetherefore analyzed parents and their offspring, and again deter-mined the number of CAG repeats using fragment analysis. In orderto follow the repeat number of certain specific transgenes, onlyheterozygous mice and only breedings between transgenic andwildtype mice were included in this study. In 86% of the analyzedtransmissions, the repeat length changed. Interestingly, this analysis

al staining using an antibody against ataxin-3 (1H9, Chemicon) confirmed an even andesentative examples from the cortex, the red nucleus (nucleus ruber), the pons, and thes expressed in all three layers, including Purkinje cells. Interestingly, the long stretch ofhomozygous mice. Both the wildtype and the homozygous HDPromMJD148mice wereepoint. MolCb, molecular layer of the cerebellar cortex; PuCb, Purkinje cell layer of the

Fig. 3. Instability of CAG repeats and selective changes after maternal and paternal transmission. Following the number of CAG repeats within the ataxin-3 transgene for severalgenerations, an instability of the CAG repeat number becomes apparent. (A)Western blot analysis showing examples of the unstable transgenic ataxin-3 protein in heterozygous andhomozygous animals, detected using the 1C2 antibody directed against expanded polyglutamine repeats. The CAG repeat numbers of the analyzed mice, as determined usingfragment analysis, are listed below each lane. TBP marks the TATA binding protein, which is usually co-detected when the 1C2 antibody is used. (B) Chromatograms obtained duringfragment analysis. Shown are examples of repeat number changes during paternal transmission (father with 147 CAG and his offspring with 153 CAG) as well as maternaltransmission (mother with 148 CAG and her offspring with 143 CAG). (C) Diagram showing the parental CAG repeat number (x axis) and the observed change of repeat number inthe offspring (y axis). Interestingly, an increase of CAG repeat number was only observed with paternal inheritance (black squares), while maternal inheritance led to a reduction ofthe number of CAG repeats (open rhombi). (D) Frequency of observed changes of repeat number. For each change of repeat number (x axis), the number of observed changes duringtransmission (y axis) is listed. Paternal transmission (black bars) led to no change or an increase of repeat number, while after maternal transmission (white bars) only reductions ofrepeat numbers were observed.

288 J. Boy et al. / Neurobiology of Disease 37 (2010) 284–293

revealed a clear link (pb0.001) between the change of the CAGrepeat number and the mode of inheritance: while an increase inCAG repeats was only observed with paternal inheritance, maternaltransmission was linked to a decreased number of CAG repeats (Fig.3B, C). No correlation between the original number of CAG repeatsand the increase or decrease, and no tendency of long repeatstowards a further extension was obvious. On the contrary, a largeincrease (+6) of CAG repeat numbers was observed in a mouse withan average repeat number, and reductions in CAG repeat numberwere also observed in mice with high numbers of CAG repeats (e.g.from 165 down to 155 CAG repeats). Analyzing the frequency ofchanges, we found that stable transmission and an increase of tworepeat units were most common outcomes of paternal transmission,while maternal transmission always led to a reduction in thenumber of repeat units, most frequently a reduction of two repeatunits (Fig. 3D).

Transgenic HDPromMJD148 mice show early hyperactivity, but reducedmotor coordination and impaired motor learning in late disease stages

In order to investigate whether transgenic HDPromMJD148 micedevelop a neurological phenotype, behavioral tests, including Rotarodanalysis, beam walking, pole test, and others were performed at theage of 4 and 7 months. These tests revealed no differences between

transgenic HDPromMJD148 mice and controls. Since we and othersdemonstrated before that mice transgenic for ataxin-3 with a normalrepeat length develop no phenotype and are indistinguishable fromwild type mice even at older age and irrespective of the appliedpromoter or ataxin-3 construct (Bichelmeier et al., 2007; Cemal et al.,2002; Chou et al., 2008; Goti et al., 2004; Ikeda et al., 1996), negative(wild type) littermates were used as controls for these analyses.

We then recorded the home cage activity of transgenicHDPromMJD148 mice for 23 h at the age of about 4 to 6 months,and observed different activity patterns of HDPromMJD148 micecompared to controls: During the first 2 h in the cage, theHDPromMJD148 mice were more active than the control mice(Fig. 4A). This hyperactivity of HDPromMJD148 mice compared tocontrols was not only characterized by an increased number of beambreaks, but also by a significantly increased percentage of time spentin the center of the cage, a significantly increased percentage ofambulatory movements, and a significantly increased number ofrearings (data not shown). In addition to hyperactivity, these data alsoindicate a reduced level of anxiety in HDPromMJD148 mice. Whenanalyzing the dark phase, we observed a higher activity of controlmice at the beginning of darkness and a higher activity ofHDPromMJD148mice at the end of 12 h of darkness. These differencesare due to a stronger decrease in the activity of control mice duringthe 12 h of darkness compared to HDPromMJD148 mice (Fig. 4B).

Fig. 4. Activity analyses. (A and B) Home cage activity analyses revealed hyperactivity of transgenic HDPromMJD148 mice at the age of 4 to 6 months. Mice were kept for 23 h inLabMaster cages (TSE Systems) and their activity during the light and the dark phase (hour 5 to hour 17) was recorded by the number of beam breaks. Shown are the total numbersof beam breaks in 15 min intervals (A) During the first 2 h in the LabMaster cages, HDPromMJD148 mice (black squares) were significantly more active than control mice (greytriangles). (⁎⁎pb0.05; ⁎⁎⁎p=0.001). Error bars, SEM. (B) At the beginning of the dark phase, control mice (gray line) were more active than HDPromMJD148 mice (black brokenline); however, their activity decreased more strongly (−141 beam breaks/15 min) during the 12 h in the dark phase than the activity of HDPromMJD148 mice (−88 beam breaks/15 min), resulting in a higher activity of HDPromMJD148 mice at the end of the dark phase (⁎⁎pb0.05). Shown is the mean of 12 transgenic and 10 control mice. For clarity, no errorbars are shown. (C–F) Open field analyses revealed hyperactivity and reduced anxiety in HDPromMJD148 mice at the age of 14 months (mean of 18 transgenic and 8 control mice).(C) Plot of the moved track in the arena during 15 min. Shown are two representative examples. While the wildtype mouse (control) spend most of the time in the margin area andavoided the center, the HDPromMJD148 mouse evenly moved throughout the arena without any preferences. (D) HDPromMJD148 mice moved longer distances in the open fieldarena during the first 8 min. The differences in interval/minute 2 and 7 are significant (pb0.05). (E) No difference between transgenic and control mice was observed regarding thedistance moved in the margin area (region 1) of the arena. (F) Distance mice moved in the transition area between the margin and the center (region 2). HDPromMJD148 micemoved (at four intervals significantly) longer distances in the transition area, indicating frequent changes between the center and the margin.

289J. Boy et al. / Neurobiology of Disease 37 (2010) 284–293

290 J. Boy et al. / Neurobiology of Disease 37 (2010) 284–293

In order to confirm that these differences in activity persist at olderages, we analyzed mice at 14 months of age. Since in younger mice,significant differences were already observed at the beginning of thehome cage activity test, we confined this activity analysis to a shortertime period in a novel environment and performed open fieldanalyses. In this test, we again observed hyperactivity and reducedanxiety in HDPromMJD148 mice. During 15 min, control mice spendmost of the time in the margin area and only rarely visited the centerof the arena. HDPromMJD148 mice, however, moved through thewhole arena without any regional preference (Fig. 4C). Statisticalanalysis revealed an increased level of activity in HDPromMJD148mice, indicated by an increased distance moved in the arena (Fig. 4D).While no difference was observed regarding themargin area (Fig. 4E),HDPromMJD148 mice showed a significantly increased level ofactivity in the transition area between the center and the margin,indicating a reduced anxiety in these mice (Fig. 4F).

We then wanted to know whether older HDPromMJD148 micedevelop a motor phenotype and therefore performed Rotarodanalyses. These analyses revealed a significantly reduced Rotarodperformance in HDPromMJD148 mice at the age of about 1 year(Fig. 5A). The Rotarod performance of HDPromMJD148 mice wasgenerally impaired. Comparingmicewith shorter or longer CAG repeatlengths within the transgene, and comparing heterozygous andhomozygous mice, did not reveal any significant correlation betweenthe number of CAG repeats or the zygosity of the transgene and theseverity of symptoms (data not shown).

The ability of mice to learn novel motor skills can be studied byanalyzing the first two trials of naïve mice on a Rotarod (Buitrago etal., 2004).While control mice significantly improved during these firsttrials, no improvement was detectable in HDPromMJD148 mice,indicating a reduced motor learning capability of these mice (Fig. 5B).

Correlation of the neurological phenotype with neuropathological signs

We expected that the occurrence of neurological symptoms wouldbe reflected by the formation of intranuclear inclusion bodies inneurons of certain brain regions. However, our analysis at the age of14 months revealed no intranuclear aggregates in any of the brainregions examined (Fig. 2). In homozygous mice, intranuclearinclusion bodies were observed from the age of 18 months on but inheterozygousmice, inclusion bodies were not detected until the age of

Fig. 5.Motor andmotor learning deficits in HDPromMJD148mice. (A) Sensorimotor coordinaabout 1 year (57 weeks), a motor phenotype in transgenic mice was apparent in these miretention time on the rotating rod. (t-test: ⁎⁎pb0.05; ⁎⁎⁎pb0.001; two-way ANOVA: pb0.0restrictions of the ability of HDPromMJD148 mice to learn novel motor skills, the first twoWhile control mice significantly improved from the first to the second trial, no improveme

25 months. Intranuclear inclusions formed in certain brain regionslike the red nucleus, the pons, and the cerebellum including Purkinjecells (Fig. 6A).

We therefore wanted to know whether other signs of neuro-degeneration or dysfunction are present in phenotypical HDProm-MJD148 mice, and observed using electron microscopy darklystained Purkinje cells considered as a sign for degeneration (“darkcell degeneration”, Garthwaite and Garthwaite, 1991a,b; Petrasch-Parwez et al., 2007). No darkly stained cells were observed incontrol mice (Fig. 6B).

Taken together, while HDPromMJD148 mice show hyperactivityand reduced anxiety already at 4–6 months of age, motor symptomscan only be detected at the age of 12–14 months. At this age, nointranuclear aggregates could be detected, indicating the occurrenceof motor symptoms in this mouse model before the formation ofintranuclear aggregates (Fig. 7).

Discussion

Spinocerebellar ataxia type 3 is a neurodegenerative disease withlate onset, slow progression and a heterogeneous clinical phenotype(Riess et al., 2008). As previously generated mouse models of SCA3 donot reproduce all aspects of the disease in humans (Bichelmeier et al.,2007; Cemal et al., 2002; Chou et al., 2008; Goti et al., 2004; Ikeda etal., 1996) we generated a novel mouse model for SCA3 to reflect moreclosely the late manifesting and slowly progressing phenotype. Tocontrol the expression of a full-length ataxin-3 protein with 148 CAGrepeats, we used the well-characterized rat huntingtin promoter(Holzmann et al., 1998) with high similarity between the mouse andthe rat sequence (Lin et al., 1995).

As in human SCA3 patients, the expression of the transgene isevenly distributed throughout the whole brain without differencesbetween affected and non-affected brain regions (Schmidt et al.,1998; Schmitt et al., 1997).

Normal ataxin-3 is mainly localized in the cytoplasm (Paulson etal., 1997a; Schmidt et al., 1998), with nuclear localization in some cells(Tait et al., 1998; Trottier et al., 1998). Interestingly, in heterozygousHDPromMJD148 mice, the mutant transgene product is mainlylocalized in the cytoplasm, but in homozygous mice mainly in thenucleus. Nuclear localizations of the SCA3 transgene with expandedpolyglutamines were also described for other mouse models of SCA3

tion in HDPromMJD148 mice was analyzed using the Rotarod test. Starting at the age ofce. HDPromMJD148 mice (Tg) performed significantly weaker, as shown in a reduced05). Shown is the mean of 8 control and 20 transgenic mice, respectively. (B) To assesstrials of the first Rotarod analysis (see rectangle in figure A) were analyzed separately.nt was observed in HDPromMJD148 mice.

Fig. 6. Histopathological findings in HDPromMJD148 mice. (A) Formation of intranuclear aggregates in HDPromMJD148 mice. Brain sections of both heterozygous and homozygousmice at the ages of 7 months, 18 months, and 25 months were stained with antibodies against ataxin-3 and analyzed for the formation of intranuclear aggregates (see arrow heads).In heterozygous mice, the ataxin-3 transgene is mainly localized in the cytoplasm. No aggregates were detected at the age of 7 and 18 months. Only at the age of 25 months, a fewsmall aggregates were detected in restricted brain regions, like the brain stem (shown), the red nucleus, and pontine nuclei (not shown). About 10% of Purkinje cells containedintranuclear aggregates. In homozygous mice, the ataxin-3 transgene is mainly located in the nucleus. However, intranuclear aggregates were not detected at 7 months or14 months (see Fig. 2) of age. Intranuclear inclusion bodies were apparent from 18 months of age. No aggregates were observed in control mice. Bar, 20 μm. (B) Electronmicroscopical analysis of HDPromMJD148 mice. At the age of 20 months, dark cell degeneration of Purkinje cells was observed in HDPromMJD148 mice. No dark cells were noticedin controls. Bar, 2 μm or 5 μm.

291J. Boy et al. / Neurobiology of Disease 37 (2010) 284–293

(Cemal et al., 2002; Goti et al., 2004). It could be that the expandedpolyglutamine itself mainly contributes to the translocation oftransgenic ataxin-3 to the nucleus (Fujigasaki et al., 2000; Goti etal., 2004). In accordance with this nuclear localization, intranuclearinclusion bodies were observed much earlier (at 18 months of age) inhomozygousmice than in heterozygousmice (25months). In neuronswith aggregates, the nuclear staining was concentrated in the

inclusion body, while the remaining nucleus was mainly unstained,as we have previously observed in human SCA3 patients (Schmidt etal., 2002).

While mice at the age of about 1 year still lack intranuclearinclusion bodies, they develop a neurological phenotype, indicatingthat the formation of aggregates is no prerequisite for the develop-ment of symptoms also in transgenic mouse models for SCA3. While

Fig. 7. Time line of phenotype development in HDPromMJD148 mice. InHDPromMJD148 mice, hyperactivity was already observed with 4 months of age.However, no motor symptoms were present at this age. At 12 months of age, motorlearning deficits become apparent. Neuronal nuclear inclusions (NII), however, wereonly detected at 18 months of age in homozygous and 25 months of age inheterozygous mice, respectively, and dark cell degeneration of Purkinje cells wasobserved at 20 months of age.

292 J. Boy et al. / Neurobiology of Disease 37 (2010) 284–293

this is the first observation in mouse models of SCA3, it has previouslybeen reported that neurological symptoms, cellular dysfunction, orcell loss are independent of the formation of inclusion bodies intransgenic mouse and tissue culture models for other polyglutaminediseases as well (Cummings et al., 1999; Kim et al., 1999; Klement etal., 1998; Saudou et al., 1998; Takahashi et al., 2005).

Instability of CAG repeats

For our HDPromMJD148 mouse model, we observed an instabilityof the CAG repeats which is linked to the mode of inheritance. Whilepaternally transmitted CAG repeats tend to increase, maternaltransmission led to a reduction of the CAG repeat number.

In human cases of other polyglutamine diseases, a tendencytowards greater instability of the expanded allele during paternal anda tendency towards only slight changes or decrease during maternaltransmission has been reported in SCA1 (Chung et al., 1993), SCA7(Gouw et al., 1998), DRPLA (Koide et al., 1994; Komure et al., 1995),and HD (Trottier et al., 1994). In SCA3, however, the situation seems tobe not as clear as in other polyglutamine diseases (Cancel et al., 1995;Dürr et al., 1996; Maciel et al., 1995; Maruyama et al., 1995; Matilla etal., 1995; Takiyama et al., 1995). Inconsistent data between differentpublications are possibly explainable by the theory that the sex of thetransmitting parent has less influence on the repeat instability in SCA3than in other polyglutamine diseases (Dürr et al., 1996).

It was shown that a certain haplotype of the C/G polymorphismadjacent to the CAG repeat favours the repeat instability (Igarashi etal., 1996). A comparable haplotype is also present in ourHDPromMJD148 mice (C in the transgenic and G in the endogenousataxin-3). No data concerning the respective polymorphism areavailable regarding the constructs used for the generation of othertransgenic SCA3 mouse models (Cemal et al., 2002; Chou et al., 2008;Goti et al., 2004; Ikeda et al., 1996). However, one cannot exclude thepossibility that this haplotype gave rise to the instability we observedin our mouse model.

No data regarding the in-/stability in other mouse models for SCA3are available yet (Bichelmeier et al., 2007; Cemal et al., 2002; Chou etal., 2008; Goti et al., 2004; Ikeda et al., 1996; Torashima et al., 2008).Therefore, this is thefirst report on inter-generational repeat instabilityin mouse models for SCA3 and our mouse model will therefore be animportant tool to study the mechanisms of repeat instability in SCA3.

Behavioral phenotype in HDPromMJD148 mice

In the HDPromMJD148 mouse model, we observed a mild and latemanifesting phenotype with slow progression. Since no developmen-tal defects or a phenotype was present in young mice, we can largelyexclude that other factors (e.g. integration site) except for thetransgene itself induce the phenotype of HDPromMJD148 mice.Hyperactivity was observed at 4 to 6 months of age without anyadditional motor phenotype. At the age of about 14 months, however,motor deficits as well as impaired motor learning were detected using

Rotarod analyses. For good Rotarod performance, an intact cerebellarfunction is required (Picciotto and Wickman, 1998). Histopatholog-ically, we indeed observed dark cell degeneration of Purkinje cells inolder mice. Degenerated Purkinje cells were also observed in humanSCA3 patients (Munoz et al., 2002; Rüb et al., 2002a,b) as well as inmouse models for SCA3 (Bichelmeier et al., 2007; Cemal et al., 2002;Chou et al., 2008). Intranuclear inclusion bodies, however, were notdetected in homozygous mice until the age of 18 months, and inheterozygous mice, not before 25 months. As motor coordination isknown to be impaired in aged rodents (Gage et al., 1984) no furtherRotarod analyses were performed at this age. However, our dataconfirm that the formation of intranuclear inclusion bodies is notrequired for the manifestation of behavioral symptoms.

The mild progression of the phenotype in the HDPromMJD148mouse model might be due to the lower expression of the mutanttransgene protein compared to the previous SCA3 mouse modelgenerated by us using the Prp promoter (Bichelmeier et al., 2007). Incontrast to the rapid progression of symptoms and premature deathin other mouse models of SCA3 (Bichelmeier et al., 2007; Goti et al.,2004), the slow progression in the HDPromMJD148 mice thereforebetter reproduces late manifesting forms of the disease in humans(Riess et al., 2008) and allows us to study pathological symptomsbefore the onset of measurable motor symptoms. In addition, inmouse models of SCA3 with stronger transgene expression, inclusionbodies were observed in multiple brain regions (Bichelmeier et al.,2007; Goti et al., 2004) including regions that are typically sparedfrom the formation of inclusion bodies in humans. However, in theHDPromMJD148 mouse model, although the transgene is ubiquitous-ly expressed, inclusion bodies only occur in restricted regions of thebrain stem (red nucleus, pontine nuclei) and in Purkinje cells, regionsalso affected in SCA3 patients (Munoz et al., 2002; Rüb et al., 2002a,b)implying a more “native” situation in HDPromMJD148 mice.

In both home cage analyses and open field tests, we observedhyperactivity of HDPromMJD148 mice compared to controls. Hyper-activitywas also observed in amousemodel for DRPLA (Schilling et al.,2001) and can be an indication of reduced emotionality (van der Staayet al., 1990) as observed in human SCA3patients (Zawacki et al., 2002).

In other mouse models of SCA3, rather a hypoactivity than ahyperactivity was reported (Bichelmeier et al., 2007; Cemal et al.,2002; Goti et al., 2004). However, in late disease stages,HDPromMJD148 mice also develop hypoactivity, indicating that theearly onset of symptoms in the other mouse models mentioned abovemight have covered the hyperactivity in earlier disease stages. Acomparable biphasic course of the disease was also reported for ratand mouse models of Huntington's disease (Lüesse et al., 2001;Menalled et al., 2003; Nguyen et al., 2006; Reddy et al., 1999; Slow etal., 2003), reflecting comparable observations in human HD patients(Kirkwood et al., 2001). It remains to be studied whether this biphasicdisease course is also relevant for human SCA3 patients.

Taken together, the new HDPromMJD148 mouse model of SCA3reflects many features of the disease in humans, in particular, someaspects that were not reproduced by previous mouse models of SCA3.We observe a biphasic course of the disease with hyper- andhypoactivity, reduced anxiety, the formation of intranuclear inclusionbodies in restricted brain regions and a measurable motor phenotype,which seems to be at least partly due to Purkinje cell dysfunction ordegeneration. Due to the late onset of symptoms in this model, theHDPromMJD148 mice allow the study of pathological processesoccurring before the actual onset of symptoms. In addition,HDPromMJD148 mice allow the study of factors affecting the repeatinstability in SCA3. We therefore recommend that the use of the Hun-tingtin promoter instead of the commonly used Prion protein promoterbe considered for the generation ofmousemodels for diseaseswith lateonset, since the Huntingtin promoter results in an evenly distributedbut weaker expression of the transgene, and therefore in a transgenicmouse model that more closely resembles the phenotype in man.

293J. Boy et al. / Neurobiology of Disease 37 (2010) 284–293

Acknowledgments

The technical help of Gabi Frommer-Kästle in electron microscopyis greatly appreciated. This study was supported by the GermanResearch Foundation (DFG) to OR, and by a grant from the EuropeanUnion (6th Framework Programme, EUROSCA).

References

Bichelmeier, U., et al., 2007. Nuclear localization of ataxin-3 is required for themanifestation of symptoms in SCA3: in vivo evidence. J. Neurosci 27, 7418–7428.

Bradford, M.M., 1976. A rapid and sensitive method for the quantitation of microgramquantities of protein utilizing the principle of protein-dye binding. Anal. Biochem.72, 248–254.

Brown, S.D., et al., 2005. EMPReSS: standardized phenotype screens for functionalannotation of the mouse genome. Nat. Genet. 37, 1155.

Buitrago, M.M., et al., 2004. Short and long-term motor skill learning in an acceleratedrotarod training paradigm. Neurobiol. Learn. Mem. 81, 211–216.

Cancel, G., et al., 1995. Marked phenotypic heterogeneity associated with expansion of aCAG repeat sequence at the spinocerebellar ataxia 3/Machado–Joseph diseaselocus. Am. J. Hum. Genet. 57, 809–816.

Cemal, C.K., et al., 2002. YAC transgenic mice carrying pathological alleles of the MJD1locus exhibit a mild and slowly progressive cerebellar deficit. Hum. Mol. Genet. 11,1075–1094.

Chou, A.H., et al., 2008. Polyglutamine-expanded ataxin-3 causes cerebellar dysfunctionof SCA3 transgenic mice by inducing transcriptional dysregulation. Neurobiol. Dis.31, 89–101.

Chung, M.Y., et al., 1993. Evidence for a mechanism predisposing to intergenerationalCAG repeat instability in spinocerebellar ataxia type I. Nat. Genet. 5, 254–258.

Cummings, C.J., et al., 1999. Mutation of the E6-AP ubiquitin ligase reduces nuclearinclusion frequency while accelerating polyglutamine-induced pathology in SCA1mice. Neuron 24, 879–892.

Dürr, A., et al., 1996. Spinocerebellar ataxia 3 and Machado–Joseph disease: clinical,molecular, and neuropathological features. Ann. Neurol. 39, 490–499.

Fujigasaki, H., et al., 2000. Ataxin-3 is translocated into the nucleus for the formation ofintranuclear inclusions in normal and Machado–Joseph disease brains. Exp. Neurol165, 248–256.

Gage, F.H., et al., 1984. Spatial learning and motor deficits in aged rats. Neurobiol. Aging5, 43–48.

Garthwaite, G., Garthwaite, J., 1991a. AMPA neurotoxicity in rat cerebellar andhippocampal slices: histological evidence for three mechanisms. Eur. J. Neurosci.3, 715–728.

Garthwaite, G., Garthwaite, J., 1991b. Mechanisms of AMPA neurotoxicity in rat brainslices. Eur. J. Neurosci 3, 729–736.

Gatchel, J.R., Zoghbi, H.Y., 2005. Diseases of unstable repeat expansion: mechanismsand common principles. Nat. Rev. Genet. 6, 743–755.

Goti, D., et al., 2004. A mutant ataxin-3 putative-cleavage fragment in brains ofMachado–Joseph disease patients and transgenic mice is cytotoxic above a criticalconcentration. J. Neurosci. 24, 10266–10279.

Goto, J., et al., 1997. Machado-Joseph disease gene products carrying different carboxyltermini. Neurosci. Res. 28, 373–377.

Gouw, L.G., et al., 1998. Analysis of the dynamic mutation in the SCA7 gene showsmarked parental effects on CAG repeat transmission. Hum. Mol. Genet. 7, 525–532.

Green, E.C., et al., 2005. EMPReSS: European mouse phenotyping resource forstandardized screens. Bioinformatics 21, 2930–2931.

Holzmann, C., et al., 1998. Isolation and characterization of the rat huntingtin promoter.Biochem. J. 336 (Pt. 1), 227–234.

Holzmann, C., et al., 2001. Functional characterization of the human Huntington'sdisease gene promoter. Brain Res. Mol. Brain Res. 92, 85–97.

Igarashi, S., et al., 1996. Intergenerational instability of the CAG repeat of the gene forMachado-Joseph disease (MJD1) is affected by the genotype of the normalchromosome: implications for the molecular mechanisms of the instability of theCAG repeat. Hum. Mol. Genet. 5, 923–932.

Ikeda, H., et al., 1996. Expanded polyglutamine in the Machado–Joseph disease proteininduces cell death in vitro and in vivo. Nat. Genet. 13, 196–202.

Kawaguchi, Y., et al., 1994. CAG expansions in a novel gene for Machado–Joseph diseaseat chromosome 14q32.1. Nat. Genet. 8, 221–228.

Kim,M., et al., 1999. Mutant huntingtin expression in clonal striatal cells: dissociation ofinclusion formation and neuronal survival by caspase inhibition. J. Neurosci. 19,964–973.

Kirkwood, S.C., et al., 2001. Progression of symptoms in the early and middle stages ofHuntington disease. Arch. Neurol. 58, 273–278.

Klement, I.A., et al., 1998. Ataxin-1 nuclear localization and aggregation: role inpolyglutamine-induced disease in SCA1 transgenic mice. Cell 95, 41–53.

Koide, R., et al., 1994. Unstable expansion of CAG repeat in hereditary dentatorubral-pallidoluysian atrophy (DRPLA). Nat. Genet. 6, 9–13.

Komure, O., et al., 1995. DNA analysis in hereditary dentatorubral-pallidoluysianatrophy: correlation between CAG repeat length and phenotypic variation and themolecular basis of anticipation. Neurology 45, 143–149.

Laccone, F., 2002. A fast polymerase chain reaction-mediated strategy for introducingrepeat expansions intoCAG-repeat containinggenes.MethodsMol.Biol. 192, 217–223.

Laemmli, U.K., 1970. Cleavage of structural proteins during the assembly of the head ofbacteriophage T4. Nature 227, 680–685.

Lin, B., et al., 1995. Structural analysis of the 5′ region of mouse and human Huntingtondisease genes reveals conservation of putative promoter region and di- andtrinucleotide polymorphisms. Genomics 25, 707–715.

Lüesse, H.G., et al., 2001. Evaluation of R6/2 HD transgenic mice for therapeutic studiesin Huntington's disease: behavioral testing and impact of diabetes mellitus. Behav.Brain Res. 126, 185–195.

Lundkvist, A., et al., 2004. Under stress, the absence of intermediate filaments fromMuller cells in the retina has structural and functional consequences. J. Cell Sci. 117,3481–3488.

Maciel, P., et al., 1995. Correlation between CAG repeat length and clinical features inMachado–Joseph disease. Am. J. Hum. Genet. 57, 54–61.

Maruyama, H., et al., 1995. Molecular features of the CAG repeats and clinicalmanifestation of Machado–Joseph disease. Hum. Mol. Genet. 4, 807–812.

Matilla, T., et al., 1995. Molecular and clinical correlations in spinocerebellar ataxia type3 and Machado-Joseph disease. Ann. Neurol. 38, 68–72.

Menalled, L.B., et al., 2003. Time course of early motor and neuropathological anomaliesin a knock-in mouse model of Huntington's disease with 140 CAG repeats. J. Comp.Neurol. 465, 11–26.

Michalik, A., Van Broeckhoven, C., 2003. Pathogenesis of polyglutamine disorders:aggregation revisited. Hum. Mol. Genet. 12 (Spec No 2), R173–R186.

Munoz, E., et al., 2002. Intranuclear inclusions, neuronal loss and CAGmosaicism in twopatients with Machado–Joseph disease. J. Neurol. Sci. 200, 19–25.

Nguyen, H.P., et al., 2006. Behavioral abnormalities precede neuropathological markersin rats transgenic for Huntington's disease. Hum. Mol. Genet. 15, 3177–3194.

Padiath, Q.S., et al., 2005. Identification of a novel 45 repeat unstable allele associatedwith a disease phenotype at the MJD1/SCA3 locus. Am. J. Med. Genet. BNeuropsychiatr. Genet. 133B, 124–126.

Paulson, H.L., et al., 1997a. Machado–Joseph disease gene product is a cytoplasmicprotein widely expressed in brain. Ann. Neurol. 41, 453–462.

Paulson, H.L., et al., 1997b. Intranuclear inclusions of expanded polyglutamine proteinin spinocerebellar ataxia type 3. Neuron 19, 333–344.

Petrasch-Parwez, E., et al., 2007. Cellular and subcellular localization of Huntingtin[corrected] aggregates in the brain of a rat transgenic forHuntington disease. J. Comp.Neurol. 501, 716–730.

Picciotto, M.R., Wickman, K., 1998. Using knockout and transgenic mice to studyneurophysiology and behavior. Physiol. Rev. 78, 1131–1163.

Reddy, P.H., et al., 1999. Transgenic mice expressing mutated full-length HD cDNA: aparadigm for locomotor changes and selective neuronal loss in Huntington'sdisease. Philos. Trans. R Soc. Lond. B Biol. Sci. 354, 1035–1045.

Riess, O., et al., 2008. SCA3: Neurological features, pathogenesis and animal models.Cerebellum 125–137.

Rüb, U., et al., 2002a. Degeneration of the external cuneate nucleus in spinocerebellarataxia type 3 (Machado–Joseph disease). Brain Res. 953, 126–134.

Rüb, U., et al., 2002b. Spinocerebellar ataxia type 3 (Machado–Joseph disease): severedestruction of the lateral reticular nucleus. Brain 125, 2115–2124.

Saudou, F., et al., 1998. Huntingtin acts in the nucleus to induce apoptosis but deathdoes not correlate with the formation of intranuclear inclusions. Cell 95, 55–66.

Scherzinger, E., et al., 1999. Self-assembly of polyglutamine-containing huntingtinfragments into amyloid-like fibrils: implications for Huntington's disease pathology.Proc. Natl. Acad. Sci. U. S. A. 96, 4604–4609.

Schilling, G., et al., 2001. Distinct behavioral and neuropathological abnormalities intransgenic mouse models of HD and DRPLA. Neurobiol. Dis. 8, 405–418.

Schmidt, T., et al., 1998. An isoform of ataxin-3 accumulates in the nucleus of neuronalcells in affected brain regions of SCA3 patients. Brain Pathol. 8, 669–679.

Schmidt, T., et al., 2002. Protein surveillance machinery in brains with spinocerebellarataxia type 3: redistribution and differential recruitment of 26S proteasome subunitsand chaperones to neuronal intranuclear inclusions. Ann. Neurol. 51, 302–310.

Schmitt, I., et al., 1997. Characterization of the rat spinocerebellar ataxia type 3 gene.Neurogenetics 1, 103–112.

Schöls, L., et al., 1997. Autosomal dominant cerebellar ataxia: phenotypic differences ingenetically defined subtypes? Ann. Neurol. 42, 924–932.

Sisodia, S.S., 1998. Nuclear inclusions in glutamine repeat disorders: are theypernicious, coincidental, or beneficial? Cell 95, 1–4.

Slow, E.J., et al., 2003. Selective striatal neuronal loss in a YAC128 mouse model ofHuntington disease. Hum. Mol. Genet. 12, 1555–1567.

Tait, D., et al., 1998. Ataxin-3 is transported into the nucleus and associates with thenuclear matrix. Hum. Mol. Genet. 7, 991–997.

Takahashi, T., et al., 2005. Polyglutamine represses cAMP-responsive-element-mediated transcription without aggregate formation. Neuroreport 16, 295–299.

Takiyama, Y., et al., 1995. Evidence for inter-generational instability in the CAG repeat in theMJD1 gene and for conserved haplotypes at flanking markers amongst Japanese andCaucasian subjects with Machado–Joseph disease. Hum. Mol. Genet. 4, 1137–1146.

Torashima, T., et al., 2008. Lentivector-mediated rescue from cerebellar ataxia in amouse model of spinocerebellar ataxia. EMBO Rep. 9, 393–399.

Trottier, Y., et al., 1994. Instability of CAG repeats in Huntington's disease: relation toparental transmission and age of onset. J. Med. Genet. 31, 377–382.

Trottier, Y., et al., 1995. Polyglutamine expansion as a pathological epitope inHuntington's disease and four dominant cerebellar ataxias. Nature 378, 403–406.

Trottier, Y., et al., 1998. Heterogeneous intracellular localization and expression ofataxin-3. Neurobiol. Dis. 5, 335–347.

van der Staay, F.J., et al., 1990. Genetic correlations in validating emotionality. Behav.Genet. 20, 51–62.

von Hörsten, S., et al., 2003. Transgenic rat model of Huntington's disease. Hum. Mol.Genet. 12, 617–624.

Zawacki, T.M., et al., 2002. Executive and emotional dysfunction in Machado–Josephdisease. Mov. Disord. 17, 1004–1010.