A new humanized ataxin-3 knock-in mouse model ...

Neurobiology of Disease 73 (2015) 174–188

Contents lists available at ScienceDirect

Neurobiology of Disease journal homepage: www.elsevier.com/locate/ynbdi

A new humanized ataxin-3 knock-in mouse model combines the genetic features, pathogenesis of neurons and glia and late disease onset of SCA3/MJD Pawel M. Switonski, Wojciech J. Szlachcic, Wlodzimierz J. Krzyzosiak ⁎, Maciej Figiel ⁎ Institute of Bioorganic Chemistry, Polish Academy of Sciences, Noskowskiego 12/14, 61-704 Poznań, Poland

a r t i c l e

i n f o

Article history: Received 17 April 2014 Revised 6 September 2014 Accepted 24 September 2014 Available online 7 October 2014 Keywords: Ataxin-3 Mouse Knock-in Knockin SCA3 MJD Ataxia Spinocerebellar CAG repeats Serpina3n Polyglutamine

a b s t r a c t Spinocerebellar ataxia type 3 (SCA3/MJD) is a neurodegenerative disease triggered by the expansion of CAG repeats in the ATXN3 gene. Here, we report the generation of the first humanized ataxin-3 knock-in mouse model (Ki91), which provides insights into the neuronal and glial pathology of SCA3/MJD. First, mutant ataxin-3 accumulated in cell nuclei across the Ki91 brain, showing diffused immunostaining and forming intranuclear inclusions. The humanized allele revealed expansion and contraction of CAG repeats in intergenerational transmissions. CAG mutation also exhibited age-dependent tissue-specific expansion, which was most prominent in the cerebellum, pons and testes of Ki91 animals. Moreover, Ki91 mice displayed neuroinflammatory processes, showing astrogliosis in the cerebellar white matter and the substantia nigra that paralleled the transcriptional deregulation of Serpina3n, a molecular sign of neurodegeneration and brain damage. Simultaneously, the cerebellar Purkinje cells in Ki91 mice showed neurodegeneration, a pronounced decrease in Calbindin D-28 k immunoreactivity and a mild decrease in cell number, thereby modeling the degeneration of the cerebellum observed in SCA3. Moreover, these molecular and cellular neuropathologies were accompanied by late behavioral deficits in motor coordination observed in rotarod and static rod tests in heterozygous Ki91 animals. In summary, we created an ataxin-3 knock-in mouse model that combines the molecular and behavioral disease phenotypes with the genetic features of SCA3. This model will be very useful for studying the pathogenesis and responses to therapy of SCA3/MJD and other polyQ disorders. © 2014 The Authors. Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/3.0/).

Introduction Spinocerebellar ataxia 3 (SCA3), also called Machado-Joseph disease (MJD), is a dominantly inherited disease resulting from the expansion of CAG repeats in exon 10 of the ATXN3 gene (Kawaguchi et al., 1994) (MJD & ATXN3: OMIM 109150 & 607047). Healthy individuals present a non-pathogenic number of repeats, usually between 13 and 41 CAGs (Giunti et al., 1995), whereas SCA3 patients typically express 60–82 CAG repeats in one allele of ATXN3. The presence of this mutant allele evokes motor abnormalities, such as ataxia, parkinsonism, sensory loss, spasticity and ocular symptoms, which become evident in the third or fourth decade of life (Riess et al., 2008). The mechanisms of pathogenesis in SCA3 and other polyglutamine (polyQ) diseases have been thoroughly discussed (Paulson, 2012; Switonski et al., 2012). In brief, the relevant pathogenesis of SCA3 is

⁎ Corresponding authors at: Institute of Bioorganic Chemistry, Polish Academy of Sciences, Noskowskiego 12/14, 61-704, Poznan, Poland. E-mail addresses: mfi[email protected] (M. Figiel), [email protected] (W.J. Krzyzosiak). Available online on ScienceDirect (www.sciencedirect.com).

based on the toxic function of the mutant ataxin-3 protein (Riess et al., 2008), but the exact mechanism of the disease remains elusive. Mouse models of polyQ diseases have been very useful in exploring the pathogenesis and therapies of this disease (for review: (Figiel et al., 2012)). To date, 14 models and variants of SCA3 mouse models have been created, but the majority of them express cDNA driven by unrelated promoters, such as Purkinje-specific L7, PrP, rHTT and CMV promoters (Bichelmeier et al., 2007; Boy et al., 2010; Boy et al., 2009; Cemal et al., 2002; Goti et al., 2004; Ikeda et al., 1996; Silva-Fernandes et al., 2010). In addition, a C-terminal ataxin-3 cDNA genetrap model, without the N-terminal region of the protein and without a CAG repeat tract, has recently demonstrated ataxia-like changes (Hübener et al., 2011). All of these models have reproduced many features of SCA3 pathogenesis. However, the models show an unnatural expression pattern, both in tissues and during development, due to the use of unrelated promoters, the lack of regulatory flanking sequences or the presence of an excessive number of transgene copies. In addition, these models still express mouse wild-type (WT) ataxin-3, so it is difficult to precisely investigate its influence during disease. The only full gene model is the YAC transgenic (Cemal et al., 2002), which contains regulatory sequences of human origin and the mouse ataxin-3 gene. To create a

http://dx.doi.org/10.1016/j.nbd.2014.09.020 0969-9961/© 2014 The Authors. Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/3.0/).

P.M. Switonski et al. / Neurobiology of Disease 73 (2015) 174–188

truly valid SCA3 mouse, it is necessary to generate a modified mouse knock-in allele. Possible strategies involve the insertion of long CAG repeats into the Atxn3 gene or creating a modification of mouse Atxn3 that leads to the full humanization of the allele's coding sequence to produce a human ataxin-3 mutant protein. Here, we report the successful generation of a knock-in SCA3 mouse that contains a humanized version of the mouse Atxn3 gene and expresses the human ataxin-3 protein containing 91 glutamines. We reveal that the SCA3 Ki91 knock-in model shows molecular and cellular features of SCA3 pathogenesis. Moreover, Ki91 mice exhibit a late SCA3 disease onset, which manifests as deficits in coordination in the rotarod and static rod tests. These late cellular and behavioral phenotypes are consistent with the human SCA3 condition. Results Structure of the targeted Atxn3 gene and the expression of Ataxin-3 in targeted ES cells To generate SCA3 knock-in mice, we replaced the 3′ fragment of the mouse Atxn3 gene with the equivalent human coding sequence, which contains a CAG expansion in exon 10 (Fig. 1A). Analysis of the human and mouse ataxin-3 protein sequences showed that the N-terminus of the ataxin-3 protein, which is encoded by exons 1 through 6, is virtually identical in mice and humans (Suppl. Fig. S1). This homology indicates

175

that a possible strategy to obtain a humanized locus could involve exchanging the less homologous 14 kb of the Atxn3 genomic sequence containing exons 7 through 11 with human cDNA containing human exons 7 through 11 (for a description of the targeting cassette, see Materials and Methods and Fig. 1A). Human cDNA was derived from the GM06153 line of human fibroblasts (Coriell Cell Repository; Camden, NJ, USA) containing 69 CAG repeats, which expanded to 91 CAG repeats while processing the targeting vector. Furthermore, the modified mouse Atxn3 allele contained four SNP variants that are present in human GM06153 fibroblasts (Fig. 1A). Homologous recombination in the Atxn3 locus changed the PstI and EcoRV restriction fragment lengths, and targeted clones were identified by Southern blotting (Fig. 1B). Prior to blastocyst injection, the positive clone (1H8) was analyzed by RT-PCR and western blotting. These analyses revealed the presence of both WT and mutant ataxin-3 transcripts and WT and mutant ataxin-3 protein expression (Figs. 1C and D). Moreover, antibodies against the expanded polyglutamine stretch (1C2) detected an approximately 67-kDa band corresponding to the mutant ataxin-3 protein (Fig. 1D). The presence of this band indicates that CAG expansion was translated into the polyglutamine domain. Intergenerational instability of the CAG mutation in Ki91 mice We established a colony of Ki91 animals originating from four NEOR-free heterozygous animals. These mice were from the F2

Fig. 1. Targeted modification of the mouse ataxin-3 gene. (A) The targeting vector, containing the hybrid mouse/human exon 7, human exons 8–11 (along with 91 CAG repeats in exon 10) and a fragment of the human 3′ UTR, was used to replace the corresponding mouse sequence. The modified allele contains four SNP variants that are present in human fibroblasts and were used as a source for the cDNA sequence. (B) Modified ES cells showed variable lengths of the PstI and EcoRV restriction genomic fragments, indicating correct homologous recombination (clone 1H8). The clones were identified by Southern blot using two probes located outside the homology arms. (C) The mutant ataxin-3 transcript with 91 CAG repeats along with WT ataxin-3 transcripts was present in an RT-PCR analysis of the 1H8 clone. (D) Modified ES cells expressed both WT and mutant ataxin-3 proteins, which appear as immunoblot bands of 41 and 67 kDa, respectively, using anti-ataxin-3 and anti-polyQ antibodies. Abbreviations: restriction sites EcoRV (E), PstI (P), AvrII (A), BglI (B) and MfeI (M); A/N – sequence generated after AvrII and NheI ligation.

176


generation (F0 are chimeric mice) and exhibited 90, 91, 91 and 92 CAG repeats localized in the humanized Atxn3 gene. Heterozygous Ki91 mice were bred with WT C57BL/6 J mice to transfer the transgene onto the C57BL/6 J background, and the number of CAG repeats in each newborn heterozygous animal was determined. We observed a mild intergenerational instability of CAG repeats that was present in both maternal and paternal transmission. Upon maternal transmission, the CAG tract showed a tendency toward contraction (Fig. 2A, left histogram). In contrast, upon paternal transmission, the CAG tract showed a tendency toward expansion (Fig. 2A, right histogram). An example of the CAG repeat changes that took place after the mating of a heterozygous

male with a C57BL/6 J female is shown in Fig. 2B. In total, 67% of the Ki91 offspring contained a CAG number that was different than the parental CAG number (Table 1). The gender of the transmitting parent was significantly associated with intergenerational CAG repeat instability. The analysis revealed a striking difference in the percentage of contractions versus expansions between maternal and paternal transmissions (Fisher's exact test, p b 0.0001). When females transmitted the transgene, CAG repeats frequently contracted (58% of transmissions) but rarely expanded (11% of transmissions). Male transmissions resulted in sporadic contractions (5% of transmissions) and frequent expansions (59% of transmissions)

Fig. 2. Ki91 mice show intergenerational and somatic instability of the CAG mutation. (A) Intergenerational CAG instability in Ki91 animals showed a strong correlation between the gender of the transmitting parent and the instability pattern. Offspring usually inherit unchanged or contracted CAG repeats in maternal transmissions and unchanged or expanded CAG repeats in paternal transmissions. (B) An example of an electropherogram showing the expansion in the number of CAG repeats between the male transmitting parent and its offspring animal. (C) The gender of an offspring does not influence CAG repeat instability. The mean change of the CAG repeat number between female and male offspring in both paternal and maternal transmissions was not significantly different. (D) Western blot with the anti-ataxin-3 antibody demonstrates that the change in the CAG repeat number affects the size of the ataxin-3 protein in the brains of the Ki91 animals. The total protein cortical lysates were obtained from different heterozygous Ki91 animals. (E) The somatic instability of CAG repeats was visible in the form of expanded electrophoretic peaks with higher molecular weight in 40-week-old animals, which appeared in addition to the peaks observed in 10-week-old mice. Abbreviations: Po – pons, Cb – cerebellum, Kd – kidney, Te – testis, Sp – spleen; error bars: SEM.

P.M. Switonski et al. / Neurobiology of Disease 73 (2015) 174–188 Table 1 CAG repeats are unstable in Ki91 animals, and the instability pattern is dependent on the gender of the transmitting parent.

Number of transmissions Changed CAG number, n Expansions Contractions Mean CAG number change (SD, range) Mean male offspring CAG number change (SD, range) Mean female offspring CAG number change (SD, range)

Paternal

Maternal

Total

39 25 (64%) 23 (59%) 2 (5%) 0.87 (1.76, −6 to +5) 1.08 (1.18, 0 to +4) 0.53 (2.45, −6 to 5)

36 25 (69%) 4 (11%) 21 (58%) −0.61 (1.73, −3 to +5) −0.53 (1.71, −3 to +5) −0.71 (1.79, −3 to +5)

75 50 (67%) 27 (36%) 23 (31%) 0.16 (1.89, −6 to +5) 0.37 (1.63, −3 to +5) −0.13 (2.18, −6 to +5)

177

observed that the CAG-repeat profiles of several tissues from 40week-old animals showed additional expanded electrophoretic peaks with higher molecular weights than the peaks from 10-week-old animals (Fig. 2E and Suppl. Fig. S2). Significant expansion was observed in the pons and the striatum in the brain, the kidney and the testes. The most striking expansion was detected in the testes of 40-weekold males, where the highest peak was shifted several repeats from the main allele. Furthermore, the CAG mutation was stable in the cerebellum, the spleen, the lung and the muscle, where the electrophoretic patterns were similar in 10- and 40-week-old mice. In conclusion, the CAG mutation in the humanized ataxin-3 gene exhibited agedependent, tissue-specific expansion. The human mutant Ataxin-3 is expressed in brain regions, non-neuronal tissues and isolated astrocytes

of the CAG repeats (Table 1). In contrast, the gender of the offspring did not influence the CAG repeat instability (Student's t-test, p ≥ 0.3505, Fig. 2C). The change in CAG repeat number was reflected in the number of glutamine residues in the polyQ domain, which affected the size of the ataxin-3 protein in the brains of Ki91 mice. Mutant ataxin-3 isoforms of slightly different sizes could be detected by western blotting using brain lysates from different Ki91 mice (Fig. 2D). Somatic CAG instability in Ki91 mice The main mutant Atxn3 allele in all animals was deduced from the CAG-repeat profiling using tail-tip analysis, which was performed at 4 weeks of age. The molecular size of the main allele in each animal was later used as a reference to investigate the instability in tissues from the same older animal. Various animals and tissues were used for CAG-repeat profiling at the ages of 10 weeks and 40 weeks. We

To investigate the expression of ataxin-3 protein in transgenic mice, total protein was extracted from the brain structures that are primarily affected in SCA3 (cerebellum, pons, cerebral cortex and striatum, Fig. 3A) and from non-neuronal tissues (heart and skeletal muscle, lungs, liver, spleen, kidney and testis, Fig. 3C) and resolved by 12% SDS–PAGE. The expression levels of WT and human mutant ataxin-3 were examined by immunoblot using a polyclonal ataxin-3 antibody, in which the immunogen was the entire human molecule. In WT animals, the immunoblot analyses revealed a specific band representing mouse ataxin-3 with an estimated molecular weight of 41 kDa. A similar mouse ataxin-3 41-kDa band was detected in heterozygous Ki91 animals and was, as expected, weaker than that observed in the WT samples. In Ki91 lysates, we were also able to detect an approximately 67kDa band corresponding to the mutant ataxin-3 with 91 glutamines. The observed band was a polyglutamine protein because re-probing the blot with an anti-polyglutamine antibody also detected the expected 67-kDa immunoreactive protein band (Fig. 3B). Mutant ataxin-3 was

Fig. 3. Analysis of mutant and mouse ataxin-3 protein expression in the tissues of 10-week-old Ki91 mice. (A) In all examined brain structures, i.e., the cerebral cortex, the striatum, the pons and the cerebellum, western blot analysis revealed the presence of a mouse ataxin-3 41-kDa band that was weaker in Ki91 heterozygous (het) animals than WT mice. An additional 67-kDa band corresponding to the expected mutant ataxin-3 protein was present in the heterozygous Ki91 samples. (B) The mutant ataxin-3 expressed in the Ki91 brain contains a polyQ domain. The blots detected with anti-ataxin-3 antibody were re-probed with an anti-polyQ antibody, revealing a band of the same molecular weight. (C) Mutant ataxin-3 is expressed in peripheral tissues of Ki91 animals; however, the only tissue that does not reveal mutant ataxin-3 is the kidney. Some tissues exhibit WT and mutant ataxin-3 protein bands (41 and 67 kDa) and tissue-specific isoforms of the ataxin-3 protein.

178


present across brain regions such as the cerebral cortex, the cerebellum, the pons and the striatum and in non-neuronal tissues including the lung, testis, liver and spleen, whereas expression was hardly detectable in the skeletal muscles and the heart. Mutant ataxin-3 was undetectable in the kidneys of Ki91 animals. Mutant human ataxin-3 was also expressed in isolated Ki91 cortical and cerebellar astrocytes (Fig. 5C).

Mutant ataxin-3 accumulates in the nuclei of cells throughout the brain To investigate the cellular expression of mutant ataxin-3, parasagittal sections of 12-month-old mouse knock-in brains and WT brains were collected for fluorescent immunohistochemistry with anti-ataxin-3 antibodies. The sections revealed very intense staining in the knock-in mouse brain and less intense staining in WT brain sections (Figs. 4A and B).

Fig. 4. Mutant ataxin-3 accumulates in the nuclei of cells and nuclear inclusions in the 12-month-old Ki91 brain. (A) The micrograph shows a molecular layer of WT cerebellum stained with anti-ataxin-3 antibody (MAB5360). The ataxin-3 staining was of moderate intensity and was distributed evenly throughout the cell, showing some accumulation around the nuclei and in the nuclei. Similar staining was present throughout the cerebellum and in other brain regions. (B) Cells in the Ki91 cerebellum revealed strong ataxin-3-IR. The micrograph shows intensive accumulation of the ataxin-3-positive signal (MAB5360) in the nuclei of cells in the molecular layer. Similar nuclear accumulation was present in other Ki91 brain regions. (C–F) Ki91 brains stained with anti-polyQ antibody (MAB1574). (C) Cells in the molecular layer of Ki91 cerebella showing the nuclear accumulation of polyQ proteins and nuclear aggregates. (D) Purkinje cells showing intense polyQ-IR and aggregates. (E, F) Intranuclear inclusions were present in cells of the cerebral cortex and the hippocampus. The inclusions were also present in other brain regions.


The staining was characteristically distributed in the cell nuclei of sections from knock-in mice, while the staining in WT brain sections was distributed more consistently throughout the entire cell. Such accumulation of the staining in the nuclei of cells from the knock-in animals was evident in the cerebellum (Figs. 4A and B) and in the cerebral cortex, the hippocampus, the striatum, the midbrain, the thalamus, the pons and the medulla (data not shown). In addition, we used an antibody that specifically binds polyQ domains to investigate whether the brain cells of Ki91 animals contained inclusions. As observed in sections stained with the anti–ataxin-3 antibody, the parasagittal sections from 12-month-old Ki91 animals showed nuclear accumulation of polyQ protein (Fig. 4C). The cerebellar Purkinje cells of Ki91 animals also showed a characteristic staining pattern in which the entire cell soma was positive when stained with the polyQ antibody (Fig. 4D). Moreover, we identified both intranuclear and perinuclear inclusions in various brain regions, including the cerebellum (Figs. 4C and D), the cerebral cortex (Fig. 4E) and the hippocampus (Fig. 4F). To further investigate ataxin-3 accumulation events, cerebellar protein samples from aging Ki91 animals (40–73 weeks) were subjected to western blotting using anti-ataxin-3 and anti-polyQ antibodies. The analysis revealed an age-dependent decrease in the mutant 67-kDa ataxin-3 band intensity and the concurrent presence of additional high-molecular-weight (approximately 80 kDa, 115 kDa and 125 kDa) bands (Suppl. Fig. S3). Ki91 animals exhibit early transcriptional changes in the brain, specifically elevated Serpina3n expression Transcriptional deregulation of genes is a characteristic hallmark of neurodegenerative diseases. Therefore, a range of genes including Bdnf, Slc17a6 (Vglut2), Anp32a (Lanp), Itpr1, Hspa1b (Hsp70), Syt1, Gabbr1, Grik2, Gabrb3 and Serpina3n (EB22/4) was selected, and their expression was measured by real-time PCR in 40-week-old Ki91 and WT animals in the cerebral cortex and cerebellum. The selected genes have been shown to be deregulated in other models of SCA3 and other neurodegenerative disorders (Chou et al., 2008; Evert et al., 2003; Lin et al., 2000). Among the selected genes, we identified transcriptional deregulation of Serpina3n (Suppl. Fig. S4). The mRNA and protein expression of Serpina3n was further assessed in the cerebral cortex, the pons and the cerebellum of 10-week-old animals by real-time PCR and western blotting, revealing a several-fold increase compared to the levels

179

present in the brain regions of WT animals (Figs. 5A and B). To identify the source of this high Serpina3n overexpression, separate astrocyte preparations were isolated from the cortex and cerebellum of newborn knock-in and WT animals. We found that Serpina3n protein was overexpressed in glia isolated from knock-in animals, while low levels were present in astrocytes from WT animals (Fig. 5C). This indicates that the induction of Serpina3n expression is an early transcriptional event already present in newborn Ki91 animals. Ki91 animals show increased astrogliosis in the cerebellar white matter and substantia nigra of the midbrain Increased expression of Serpina3n in the brain and astrocytes suggested the existence of an inflammatory process and/or reactive gliosis. To examine the possibility of reactive gliosis, we stained the parasagittal sections of 12-month-old knock-in and WT animals with anti-GFAP antibody. Strong GFAP signals were detected in many Ki91 brain regions; however, the signal in the cerebellum and substantia nigra revealed particular differences in intensity. In the cerebellar white matter of knock-in animals (Figs. 6B and D), blood vessels were readily visible as holes surrounded by intensive GFAP-IR, while such staining was not visible in the cerebellar white matter of WT sections (Figs. 6A and C). In the Ki91 substantia nigra, GFAP-positive astrocytes demonstrated thicker processes and more intense staining (Fig. 6F) compared to those in WT animals (Fig. 6E), which may indicate reactive gliosis. Interestingly, no changes in the overall level of GFAP protein content were detected by immunoblot analyses of whole cerebellar protein lysates from knock-in animals (Suppl. Fig. S5). Most likely, relatively large amounts of GFAP may mask more localized changes in GFAP expression when immunoblotting whole cerebellar lysates. Ki91 animals show cerebellar neurodegeneration, including a loss of Calbindin D-28 k–IR and mild loss of Purkinje cells Increased GFAP-IR in the brain often indicates that neurons in the vicinity may be affected by neurodegeneration. To examine the condition of Purkinje cells in the cerebellum, parasagittal sections from Ki91 and WT animals were stained using antibodies against Calbindin D-28 k. This Purkinje cell marker is useful in studying Purkinje cell soma morphology and dendritic tree structure. Moreover, its reduction is often associated with neurodegeneration in SCA3 and other polyQ diseases

Fig. 5. Serpina3n is a molecular marker expressed at high levels in Ki91 knock-in animals. (A) The brain structures involved in SCA3 pathogenesis revealed that the protein level of Serpina3n was increased in 10-week-old Ki91 brains. Serpina3n protein level was increased 4.9, 2.1 and 4.2 fold in cerebral cortex, the cerebellum and the pons respectively (Student's t-test, p b 0.001). Genotype of each sample is indicated. (B) The mRNA level of Serpina3n in the cerebral cortex of Ki91 mice was massively increased, as detected by real-time quantitative PCR (Student's t-test, p b 0.001). (C) Serpina3n protein levels were also increased in cortical and cerebellar astrocytes isolated from P1 Ki91 animals. Cell cultures expressed glia-specific GFAP protein. GAPDH was used as a loading control. n = 3, error bars: SEM.

180


Fig. 6. The Ki91 animals show astrogliosis in cerebellar white matter and the substantia nigra. (A) WT and (B) Ki91 tile scan micrographs of entire parasagittal sections from cerebellar hemispheres stained with an anti-GFAP antibody. The Ki91 sections revealed strong GFAP signals in the cerebellar white matter and around vessels, which appeared slightly broadened, delineated and readily visible (white arrows). The GFAP immunoreactive signal in cerebellar white matter was much less intense in WT animals. Further magnification of the (C) WT and (D) Ki91 posterior zone of the cerebellum. Again, the GFAP signal was much more intense in Ki91 animals, indicating reactive gliosis. (E) WT and (F) Ki91 parasagittal sections through the substantia nigra revealed increased GFAP staining and thicker processes in Ki91 sections compared to WT sections. DAPI nuclear staining was omitted for clarity. Picture data: oil 20×/0.75, Tile scan (each tile: 2048 × 2048 pixels); final picture: approximately 11000 × 8000 pixels. A fixed detector gain was used.

(Watase et al., 2002; Dougherty et al., 2012; Chang et al., 2011; Nóbrega et al., 2013). The cerebellar parasagittal sections through the vermis, either close to the midline or close to the vermal-hemispheral border, were examined. General examination of the cerebellar sections revealed that Calbindin D-28 k–IR was much weaker in 12-month-old knock-in animals (Fig. 7B) compared to the respective WT animals (Fig. 7A). Densitometric assessment (see Materials and Methods) revealed that the immunofluorescence of the molecular layer and the Purkinje cell bodies on Calbindin D-28 k-stained knock-in sections was 40% lower than in the respective WT animals (Fig. 7B, Graph inset). The loss of Calbindin D-28 k–IR was not uniform across the transverse zones in the Ki91 cerebellum. Although all zones across the parasagittal sections showed decreased Calbindin D-28 k staining, in the 12-month-old Ki91 animals, the staining in the nodular zone (lobule 10) was very weak, while the staining of Purkinje cell bodies remained visible in the anterior zone (Fig. 7B). The immunoreactivity of Calbindin D-28 k depends on the presence or absence of calcium ions in the molecule (Winsky and

Kuźnicki, 1996). Therefore, we investigated the expression of Calbindin D-28 k protein by immunoblotting and found that whole cerebellar protein extracts from WT and Ki91 animals revealed similar levels of Calbindin D-28 k in animals across a range of ages (10–64 weeks) (Suppl. Fig. S6). Subsequently, we examined the structures of the Purkinje cells and the Purkinje cell layer in more detail. Higher magnification also revealed a much weaker immunofluorescent Calbindin D28 k signal, resulting in the presence of an abundant population of Purkinje cells where primary dendrite was invisible in knock-in sections (Figs. 8B, C and D). Moreover, the Purkinje cell bodies in the knock-in mice were smaller and irregular compared to those in WT mice (Fig. 8A). Examination of single confocal planes from these image stacks also revealed that there was a relatively abundant population of Purkinje cells where the cell nuclei did not contain the Calbindin D28 k signal. In addition, Purkinje cells with more than one dendrite were readily found in knock-in sections (Figs. 8B and D), but they were very rarely observed in WT sections.


181

Fig. 7. The cerebella of KI91 animals show weak Calbindin D-28 k-IR. (A) WT and (B) Ki91 tile scan micrographs of entire parasagittal cerebellar sections (vermis close to the midline) stained with the anti-Calbindin D-28 k antibody. The micrographs (fixed detector gain) revealed a clear-cut decrease in Calbindin D-28 k–IR in Ki91 cerebellar sections. The Calbindin D-28 k–IR intensity differed across transverse zones in the Ki91 cerebellum. Cerebellar Ki91 Lobule I/II demonstrated Calbindin D-28 k–IR, while the staining was almost lost in lobule X. Graph inset: The densitometry quantification of the signal from the ROI, selecting the molecular layer and Purkinje cell bodies, revealed 40% lower immunofluorescence in 12month-old Ki91 animals than in respective WT animals. DAPI nuclear staining was omitted for clarity. Picture data: oil 20×/0.75, Tile scan (each tile: 2048 × 2048 pixels).

To characterize neuronal degeneration in the cerebellum in greater detail, the number of Purkinje cells in the nodular zone (lobule 10) and the anterior region (lobule 4/5) of Ki91 and WT mice was investigated. We found that the loss of Purkinje cells in both zones of the Ki91 cerebellum was mild (15 % loss in lobule 4/5 and 17 % loss in lobule 10, Student's t-test, p b 0.006 and b0.02 respectively, Fig. 8E). In addition, we investigated the number of Purkinje cells with weak immunoreactivity for Calbindin D-28 k. A subpopulation of Purkinje cells with invisible primary dendrites showed more than 3.5 fold increase in Ki91 cerebella compared to WT animals (Student's t-test, p b 0.001,

Fig. 8F). Moreover, a subpopulation of Purkinje cells with an unstained nuclear compartment was about two times overrepresented in lobule 4/5 of Ki91 animals compared to WT littermates (Student's t-test, p b 0.004, Fig. 8G). Ki91 mice demonstrate a late disease onset with motor incoordination To assess motor deficits in Ki91 mice, we used the accelerating rotarod and parallel rod-floor assays. The rotarod test was performed longitudinally in 15-, 30-, 45-, 60- and 90-week-old mice. Their

182


Fig. 8. Cerebellar Purkinje cells in Ki91 mice show neurodegeneration. (B–D) The Purkinje cells cell bodies and dendrites from Ki91 animals showed much weaker Calbindin D-28 k–IR compared to (A) the Purkinje cells from WT animals. The Ki91 Purkinje cell bodies were smaller and irregular. (B, D) Purkinje cells with more than one dendrite (white arrows) were readily visible in Ki91 animals. DAPI nuclear staining was omitted for clarity. (E – G) Quantification of Calbindin D-28 k immunoreactive Purkinje cells in Ki91 and WT mice. (E) Loss of Purkinje cells in both nodular zone (lobule 10) and the anterior region (lobule 4/5) of the Ki91 cerebellum was mild. (F) Weakly immunoreactive subpopulation of Purkinje cells with invisible primary dendrites and (G) a subpopulation with an unstained nuclear compartment were significantly overrepresented in Ki91 cerebella compared to WT animals (Student's t-test, ***p b 0.01, **p b 0.01 and *p b 0.05).

performance on the rotarod was significantly affected by their age and declined in both Ki91 and WT mice (Fig. 9A, main effect of time, p b 0.0001, two-way ANOVA). There was a significant difference between the two genotypes (main effect of genotype, p = 0.0465, twoway ANOVA), indicating greater rotarod impairment in Ki91 compared to WT mice, although no time point showed a significant difference in the Bonferroni post hoc test. Subsequently, we analyzed the rotarod data from four consecutive trial days for every time point of the rotarod test. We found that the difference between the two genotypes in rotarod performance was significant for 90-week-old mice, indicating rotarod impairment in Ki91 mice at 90 weeks (Fig. 9B; main effect of genotype, p = 0.004, two-way ANOVA; Bonferroni posttest, **p b 0.01 and

*p b 0.05). This change in rotarod performance was not significant for animals at the ages of 15, 30, 45 and 60 weeks (Suppl. Fig. S6). The parallel rod floor test was used to examine both locomotor activity and motor incoordination. Ki91 and WT animals at the ages of 30 and 45 weeks showed no difference in the distance traveled. In contrast, 60-week-old Ki91 animals traveled longer distances than WT animals, although the difference was not statistically significant (Fig. 9C). The Ki91 and WT mice also made a similar number of foot slips, corrected for the distance traveled (Fig. 9D; main effect of genotype, p = 0.9487, two-way ANOVA). Body weight was measured from 4 to 90 weeks of age, and no differences in weight gain were observed between male Ki91 mice and their WT male littermates (Fig. 9E).


183

Fig. 9. Ki91 mice demonstrate rotarod impairment. (A) Rotarod performance of Ki91 and WT males tested from 15 to 90 weeks of age. There were significant differences in the mean time spent on the accelerating rod between the two genotypes (genotype p b 0.05, two-way ANOVA). (B) Rotarod performance in 4 consecutive trial days of the rotarod test of 90-week-old Ki91 and WT animals. The difference in rotarod performance between the two genotypes was significant, indicating rotarod impairment for Ki91 animals at 90 weeks of age (p b 0.05, twoway ANOVA; Bonferroni posttest **p b 0.01 and *p b 0.05). (C) Parallel rod floor tests at the ages of 30 and 45 weeks showed that similar distances were traveled by both Ki91 and WT animals during the 10-min trial. Although 60-week-old Ki91 animals traveled longer distances than WT animals, the difference was not statistically significant. (D) The ataxic phenotype was assessed as the number of foot slips corrected to the travelled distance. The data were assessed with two-way ANOVA for genotype and age [n = 13 (WT) and 14 (Ki91) for rotarod assay, n = 9 (WT) and 10 (Ki91) for the parallel rod floor test; error bars: SEM]. (E) Body weight of Ki91 and WT males. There was a similar weight gain in both genotypes.

To further evaluate and confirm the incoordination phenotype observed in 90-week-old animals in the rotarod test, we tested a cohort of Ki91 and WT mice (mean age: 96.6 weeks) on the static rod apparatus. The performance of mice to orient (turn 180°) and traverse several oval wooden rods (35 mm, 28 mm, 21 mm, 17 mm and 9 mm in diameter) was investigated. The Ki91 animals showed an increased latency to turn 180°, and this effect was statistically significant for the 28-mm and 21-mm rods compared to the WT animals (Fig. 10A; main effect of genotype, p = 0.0088 and p = 0.013 for the 28-mm and 21-mm rods, respectively, two-way ANOVA; Bonferroni posttest, *p b 0.05). The last 9-mm rod was too narrow for the Ki91 and WT animals, and mostly maximum penalty scores were received for this rod (data not shown). In addition, statistical assessment using ANOVA for repeated measures did not reveal a significant difference between the Ki91 and WT animals in terms of their ability to traverse the rod after turning around (Fig. 10B).

Discussion In a previous attempt to generate the knock-in animals, we replaced a fragment of the mouse Atxn3 gene with human cDNA encoding exons 2–11. Unfortunately, splicing events led to the lack of both mutant and endogenous ataxin-3 protein expression (Switonski et al., 2011). Therefore, we established a new targeting strategy, replacing the nonhomologous exons 7–11 in the mouse Atxn3 locus (Suppl. Fig. S1) and generating new SCA3 Ki91 knock-in animals. The first SCA3 phenotype detected while breeding the animals was the intergenerational instability of the CAG tract; in particular, the tract is prone to paternal expansion and maternal contraction. In human polyQ diseases, the gender of the transmitting parent has a significant effect on the intergenerational instability of the CAG tract. Moreover, expansion bias in paternal transmissions and contraction bias in maternal transmissions have been observed in patients with

184


Fig. 10. Ki91 mice show impaired coordination in the static rods test. The experimental paradigm included a training day (T), followed by trial days (1-3) on rods with an increasing difficulty level, including 35 mm, 28 mm, 21 mm, 17 mm and 9 mm rods. (A) Performance of mice in turning 180° on the rod, as measured in seconds. Compared to WT mice, Ki91 animals required significantly more time to turn around on 28 and 21 mm rods (genotype p = 0.009 and p = 0.013, respectively, two-way ANOVA for trial days 1–3; Bonferroni posttest *p b 0.05). (B) Performance in traversing the rods, as measured in seconds. ANOVA for repeated measures did not reveal significant differences between the genotypes in terms of the time needed to traverse the rods. [n = 15 (Ki91) and 8 (WT); mean age of animals in weeks: 97 (Ki91) and 96 (WT); error bars: SEM]. Data for the 9mm rod are not shown because both genotypes mostly failed the test on this rod.

HD, SCA1, SCA7 and DRPLA (Aziz et al., 2011; Chung et al., 1993; Gouw et al., 1998; Koide et al., 1994). In the case of SCA3, several publications have reported larger expansions in paternal than in maternal transmissions; however, the differences are not significant. Instead, male transmissions are associated with larger variations of inherited repeat length than maternal ones (Dürr et al., 1996; Maciel et al., 1995). In the Ki91 model, we observed a strong correlation between parental sex and the instability pattern. When males transmitted the transgene, expansions accounted for 92% of unstable transmissions, whereas 84% of all unstable events in female transmissions were contractions. Moreover, there was no clear difference in the variation of the size of the transmitted CAG length according to the parental sex (Table 1). These mouse-human dissimilarities in transmitted repeat instability might be explained by inter-species factors, such as different cellular and DNA metabolism during gametogenesis or the different chromosomal context surrounding the CAG mutation. Nonetheless, the Ki91 mouse model is similar to human SCA3 patients because it shows intergenerational CAG repeat instability, and upon the completion of this manuscript, we were able to generate animals containing a tract as long as 104 CAGs by selectively breeding males with longer repeats. The CAG mutation in the humanized ataxin-3 gene exhibits agedependent tissue-specific expansion. The pattern of somatic instability observed in Ki91 animals is comparable to the instability observed in SCA3 patients. Similar to the human condition, relatively high instability was observed in the pons, striatum, liver and kidney, and the CAG tract was stable in tissues such as the cerebral cortex, the cerebellum, the

muscle, the lung and the spleen (Cancel et al., 1998; Tanaka et al., 1996). Testicular tissue was exceptional among the tissues tested because it showed a relatively high CAG repeat instability. The expansion of CAG repeats in the testis of 40-week-old, but not 10-week-old, mice suggests that expansions accumulate with age in diploid mitotic germ cells. The notion that expansions arise in diploid germ cells agrees with the results obtained in HD patients and DM1 mice (Savouret et al., 2003; Zhang et al., 2002). In addition, significantly larger expansions in testicular tissue point to the existence of important tissuespecific modifiers of CAG stability. Protein expression analysis with knock-in animals showed that human ataxin-3 from the targeted allele is ubiquitous throughout neuronal and non-neuronal tissues. In particular, human ataxin-3 was expressed in all of the brain regions tested, and the level of this expression was similar to or greater than the level of expression from the nontargeted WT allele. In addition, we found that the lung, testis, liver and spleen tissues contain high levels of mutant ataxin-3, while the heart and muscle contain a relatively low level of this protein. We also examined the expression of mutant ataxin-3 in brain cells. The immunostaining showed that mutant ataxin-3 accumulated in the cell nuclei of various brain regions, and such nuclear staining was much more prominent in Ki91 compared to WT animals. In addition, our knock-in model demonstrates that the polyQ protein forms inclusions that are both intranuclear and outside the nucleus. The inclusions were rather infrequent, but they showed very intensive staining that never overlapped with other autofluorescent granules, such as lipofuscin. In addition, protein lysates from aging brain tissue of Ki91 animals revealed a decreased level of mutant ataxin-3 protein and the presence of extra protein bands in blots stained with anti-Atxn3 and anti-polyQ antibodies (Suppl. Fig. S3). This finding demonstrates that the mutant protein in Ki91 animals may be subjected to an aggregation process that intensifies with age. In addition, the role of other processes that may be involved in the formation of additional bands in aging animals, e.g., ubiquitination or protein interaction, cannot be excluded. Ki91 animals also revealed transcriptional changes in the brain. The data indicated that the induction of Serpina3n expression was an early transcriptional event that had already occurred in newborn Ki91 animals, long before the onset of behavioral indicators of disease. In astrocytes, a high expression of Serpina3n is sustained without a trophic brain environment in cell culture conditions, indicating that an intrinsic factor, such as the expression of human mutant ataxin-3, is the direct cause of elevated Serpina3n expression. Serpina3n and SERPINA3 are also increased in other polyQ conditions, including the SCA1 mouse model and in Huntington's disease (HD) patients (Lin et al., 2000; Hodges et al., 2006). Interestingly, the SCA3 brain expresses multiple markers of inflammation and shows amyloid depositions (Evert et al., 2003; Evert et al., 2001). Serpina3n is expressed during the acute inflammatory phase and was identified as a highly expressed molecule that marks populations of activated astroglia (Zamanian et al., 2012). In our model, whole cerebellar protein extracts did not show increased expression of GFAP, which is not unusual because the overall GFAP-positive signal in Bergmann glia is relatively high. Like SCA3 patients (Scherzed et al., 2012), the Ki91 animals demonstrated increases in GFAP-positive glia in cerebellar white matter, indicating the local activation of astrocytes. These increases are often concentrated around vessels, which makes them delineated and readily visible in Ki91 but not WT animals. Such local changes may be a sign of the degeneration of cerebellar neurons, nerve fibers and white matter. In fact, cerebellar white matter and peduncles are degenerated in SCA3 patients, as detected by magnetic resonance imaging (MRI) examination and histopathology of the cerebellum (Scherzed et al., 2012; Guimarães et al., 2013). The presence of increased GFAP-IR is one of the hallmarks of SCA3 in humans (Scherzed et al., 2012), and in Ki91 mice, increased GFAP-IR accompanies Purkinje cell degeneration in 12-month-old cerebella. First, whole cerebellar scans revealed that the intensity of fluorescence


resulting from Calbindin D-28 k–IR was strongly decreased in Ki91 versus WT animals. In neurodegenerative conditions, Purkinje cells often follow a complicated gradient pattern of degeneration that is dependent on the cerebellar zone (Apps and Hawkes, 2009; Duffin et al., 2010). Calbindin D-28 k–IR was decreased in Ki91 mice throughout the entire parasagittal cerebellar section, but some differences across transverse zones were observed. However, Purkinje cell bodies in the anterior zone of the cerebellum still showed Calbindin D-28 k–IR in 12-monthold Ki91 animals, while such staining was almost absent in the nodular zone (lobule 10). Pronounced degeneration of the posterior part of the cerebellum is often observed in SCA3 patients (Scherzed et al., 2012; Eichler et al., 2011). We also found that there was a mild loss of Purkinje cells in 12-month-old Ki91 cerebella, while the same animals revealed several times as many cells with weak Calbindin D-28 k staining compared to WT animals; this indicates that dysfunction rather than death of Purkinje cells may contribute to SCA3 pathogenesis in Ki91 animals. Decreased Calbindin D-28 k–IR in the presence of similar protein expression in Ki91 and WT animals (Suppl. Fig. S6) may be explained by the fact that the conformation and the immunoreactivity of Calbindin D-28 k and other calcium-binding proteins are dependent on the presence or absence of calcium ions in the molecule (Winsky and Kuźnicki, 1996; Kojetin et al., 2006; Zimmermann and Schwaller, 2002). It remains to be investigated whether Ki91 animals harbor changes in intracellular calcium homeostasis, however, such a finding would be in agreement with the altered calcium homeostasis already identified in SCA3 and other polyQ mouse models (Chen et al., 2008; Panov et al., 2002). In addition to weak Calbindin D-28 k–IR, Ki91 Purkinje cells viewed at higher magnifications revealed obvious neurodegenerative changes in morphology, including an irregular shape, small cell bodies, invisible primary dendrites and decreased Calbindin D-28 k–IR in cell nuclei. Purkinje cell degeneration and the loss of Calbindin D-28 k–IR in Ki91 mice are highly relevant because the same changes have been observed in histopathological examinations of SCA3 cerebella, with regions almost completely devoid of Calbindin D-28 k–IR (Scherzed et al., 2012). Ki91 animals at 90 weeks of age showed a significant decrease in rotarod performance, and similar motor incoordination was confirmed in the static rod test. A comparable late-developing behavioral phenotype has been reported in the full-length hemizygous SCA3 YAC model (Cemal et al., 2002; Rodríguez-Lebrón et al., 2013). In general, the manifestation of behavioral changes is usually late in polyQ knock-in animals (Figiel et al., 2012). Moreover, the late motor incoordination phenotype in Ki91 is consistent with the human SCA3 condition, where disease onset is usually late and becomes evident in the third or fourth decade of life. The knock-in Ki91 SCA3 mice express a single copy of full-length (not truncated) human mutant ataxin-3 (compared to the multiple copies in random transgenic models). Therefore, it is not surprising that Ki91 animals do not develop an aggressive phenotype. While a rapid phenotype is useful for quick screening of drugs, the slower phenotype present in Ki91 animals may be useful for fine tuning and observing long-term consequences of therapy. In addition, in the knock-in model, the expression of transgenes and endogenes can be strictly defined, resulting in a better approximation of patient conditions compared to simple transgenic models, which often contain two endogenous mouse copies and several transgene copies. Additionally, the genetic vicinity of the transgene is important for the disease modeling, which is a clear advantage of our knock-in model. Conclusions We present an original humanized Atxn3 knock-in mouse model that is currently the most faithful genetic and pathogenic representation of SCA3 disease. Our knock-in model reveals the expansion of mutations in the SCA3 gene, not only in offspring mice but also in somatic tissues of the brain. In Ki91 animals, the degeneration of neurons in the cerebellum is accompanied by excessive activation of glial cells in the white

185

matter of the cerebellum, which corresponds to the induction of Serpina3n expression, a marker of brain damage and neuroinflammation. Ki91 animals further demonstrate cerebellar degeneration with a pronounced loss of Calbindin D-28 k–IR and a mild loss of Purkinje cells. Moreover, the late neurodegeneration observed in SCA3 patients is in agreement with the late motor incoordination phenotype of Ki91 animals. Thus, our knock-in model will be very useful in understanding the pathogenesis of SCA3. Methods Generation of SCA3 knock-in mice The SCA3 knock-in transgenic mouse model was generated at PolyGene AG (Rümlang, Switzerland). A targeting vector backbone containing the 3′ fragment of mouse intron 6, hybrid mouse/human exon 7 and the 5′ fragment of human exon 8 was generated using DNA synthesis and cloned into the pBlueskript vector. The BglII-BglI fragment of human ATXN3 cDNA containing the 3′ fragment of exon 8, exons 9–11 (along with 91 CAG repeats in exon 10) and a fragment of the human 3′ UTR was cloned from human fibroblasts (GM06153) and subcloned into the targeting vector. A 7-kb 5′ homology arm (AvrII-AvrII fragment spanning a portion of intron 5, exon 6 and intron 6) and a 2-kb 3′ homology arm (BglI-MfeI fragment spanning the portion of 3′ UTR) were generated from the murine Sv129 BAC bMQ-421 J03 clone. The FRTflanked neomycin resistance gene was cloned into the BamHI site, which was previously created at the end of intron 6 in the targeting vector backbone. This construct was electroporated into 129/Ola ES cells, and clones displaying G418 resistance were examined for proper homologous recombination by PCR and Southern blotting. Confirmed ES clones were microinjected into C57BL/6 blastocysts. Germline transmission was obtained by crossing chimeric males with C57BL/6 females. The NEO cassette was excised by crossing heterozygous mice with Flp deleter animals. Mice backcrossed to C57BL/6 J for four generations were used for experiments. Animals Animals were maintained under standard conditions with an 18/6-h light/dark cycle and provided water and food ad libitum. The animals were marked using numerical ear tags (National Band & Tag Company, Newport, USA). The genotype and the number of CAG repeats were determined by PCR using tail-tip DNA. The animals were sacrificed by placing them in a 70% CO2 atmosphere. The stress level of the animals was minimized throughout all procedures and animal handling, which were approved and monitored by the Local Ethical Commission for Animal Experiments in Poznan. Genotyping DNA was prepared from tail-tip biopsies using the Spin Column Genomic DNA Kit (Bio Basic Inc., Markham, Canada). Genotypes were determined by performing multiplex PCR using GoTaq Flexi DNA Polymerase (Promega, Madison, USA). The following set of four primers was used: forward GGACCTATCAGGACAGAGTTCACATCCA and reverse CACATTACCAAAGTGGACCCTATGCTGT, targeting the humanized region of the Atxn3 gene, and forward TCCTCTCTAGGATGGCTTTG and reverse TAAACGCCCCTACCTATTTG, targeting the endogenous region of the Atxn3 gene. The PCR cycling conditions were as follows: 3 min at 94 °C, 12 × (35 s at 94 °C, [45 s at 64 °C - 0.5 °C/cycle], 45 s at 72 °C), 25 × (35 s at 94 °C, 30 s at 58 °C, 45 s at 72 °C) and 2 min at 72 °C. The reaction products were separated on 1.5% agarose gels in TBE buffer and stained with ethidium bromide. The number of CAG repeats was determined by performing PCR using GoTaq Flexi DNA Polymerase with the 6-FAM-labeled forward primer GGAAGAGACGAGAAGCCTAC and reverse primer TCACCTAGAT

186


CACTCCCAAGT; these primers flank the CAG repeat region in the humanized Atxn3 gene. The cycling conditions were as follows: 5 min at 94 °C, 30 × (20 s at 94 °C, 10 s at 60 °C, 30 s at 72 °C) and 7 min at 72 °C. The PCR fragments were sized using the ABI 3130xl Analyzer and Peak Scanner 1.0 software (Life Technologies, Carlsbad, USA). Because the presence of CAG repeats slows the electrophoretic migration of PCR fragments, a calibration curve composed of sequenceverified fragments with 21, 69, 81, 91 and 97 CAGs was prepared and used to assess the exact number of CAG repeats in the analyzed samples. RNA isolation and real-time expression analysis Tissue samples were harvested and immediately submerged in RNAlater solution (QIAGEN, Venlo, Netherlands). The tissue samples were homogenized in TRI REAGENT solution (MRC; Cincinnati, USA) using a rotor-stator homogenizer, and the total RNA was isolated according to the manufacturer's instructions. The RNA concentration was measured with a NanoDrop spectrophotometer. A total of 500 ng of RNA was reverse-transcribed at 42 °C for 50 min using 200 U of Superscript II (Life Technologies) and 125 ng of random hexamer primers (Promega) in a final volume of 20 μl. The light cycler 480 II (Roche) was used for real-time PCR analysis. Immunoblotting analyses Tissue samples were harvested, immediately frozen in liquid nitrogen and homogenized in a buffer containing 60 mM TRIS-base, 2% SDS, 10% sucrose and 2 mM PMSF using a rotor-stator homogenizer. The protein concentration was estimated using a NanoDrop spectrophotometer, and 20 μg of total protein was diluted in a sample buffer containing 2mercaptoethanol and boiled for 5 min. The proteins were separated by SDS–polyacrylamide gel electrophoresis (5% stacking/12% resolving gel), transferred to nitrocellulose and stained with Ponceau S solution. The blots were blocked with 5% nonfat milk in PBS/0.05% Tween 20 and subsequently incubated for 24 h at 4 °C with the following primary antibodies: rat anti-ataxin-3 (1:3,000; ProteinTech Chicago, IL, USA), mouse anti-polyQ (1:3,000; Millipore, Billerica, MA, USA), mouse antiGAPDH (1:20,000; Millipore), mouse anti-GFAP (1:3,000; BD Pharmingen, Warsaw, Poland) and goat anti-Serpina3n (1:3,000, R&D Systems, Minneapolis, MN, USA). The blots were probed with the respective HRP-conjugated secondary antibody (anti-goat and anti-rat, 1:3,000; Jackson Immuno Research, Suffolk, UK), (anti-mouse, 1:3,000, Sigma–Aldrich, St. Louis, MO, USA). The immunoreaction was detected using the ECL substrate (ThermoFisher Scientific, Waltham, MA, USA). Immunohistochemistry The animals were deeply anesthetized and transcardially perfused using saline followed by 4% PFA. The brains were removed, post-fixed in 4% PFA for 48 h and cryopreserved with graded sucrose (10–20– 30%) over 72 h. The 30-μm parasagittal mouse brain sections were cut using a cryostat at − 20 °C, collected on silanized glass slides and dried in a microwave oven for 4 min. The sections were stored at − 80 °C or processed immediately. The HIER procedure was applied by boiling the sections in citrate buffer (pH 6.0) for 30 min, and the sections were subsequently reduced in 1% borohydrite for 1 h. The sections were blocked via incubation in 2% normal goat serum in PBS for 30 min. For immunofluorescence staining, the sections were incubated overnight at 4 °C with the following primary mouse antibodies: antiCalbindin D-28 k (1:500; Sigma–Aldrich), anti-GFAP (1:1,000; BD Pharmingen), anti-polyQ (1:500; Millipore) and anti-ataxin-3 (1:500; Millipore). To detect inclusions and to differentiate them from background lipofuscin aggregates, we used secondary antibodies labeled with a hybrid fluorophore that was excitable in far red (650 nm), exhibited a very long Stokes shift and emitted in infrared (750 nm) (FRET effect). The respective Alexa Fluor 750-Allophycocyanin (Invitrogen),

Cy-3, Cy2, anti-rabbit and anti-mouse fluorescent secondary antibodies (Jackson ImmunoResearch; Suffolk, UK) were used at a concentration of 1:500 for 2 h at RT in PBS-Tween 20. Fluorescent confocal images were acquired using the TCS SP5 II (Leica Microsystems; Poland) or LSM 510 META system (Zeiss; Poznan, Poland). Densitometry assessment of immunofluorescence and Purkinje cell counts The immunofluorescent signal was assessed by reading the entire immunofluorescence intensity from 20-μm-thick confocal sections using fixed detector gain values. Micrographs of the entire cerebellar section area were visualized using the tile scan function. Singlechannel, large micrographs representing entire cerebellar sections were then processed by selecting the stripe of Calbindin D-28 k immunoreactive dendrites and Purkinje cell bodies from entire cerebellar sections from three 12-month-old knock-in mice and three respective WT mice using the “ROI” and “mean gray value” functions of ImageJ. The Purkinje cell numbers were estimated separately for lobule 10 and lobule 4/5 of the cerebellum. Sections used for Calbindin staining were randomly selected from series of sections through the vermis. Five stacks (area of 387.5 μm × 387.5 μm and approximately 10 μm thickness) composed of 10 confocal images were randomly acquired using a 40× objective from each investigated lobule from each mouse. Purkinje cells were then counted from each stack of confocal images by marking the cells with the “multi-point” tool of ImageJ to count each cell only once. Subsequently, the length of the counted Purkinje cell layer was estimated using the “segmented line” and “measure” tools of ImageJ. Cell culture Primary glial cultures were initiated from the cerebral hemispheres or cerebella of mouse pups, and separate glial cultures were derived from individual animals. Tissue pieces were carefully freed of meninges and stored in ice-cold medium. Collected tissue pieces were subsequently chopped, incubated for 20 min in PBS containing 0.1% trypsin and 0.02% EDTA and transferred into HBSS with 10% FBS. The tissues were gently dissociated with plastic pipettes and directly suspended in DMEM/F12 supplemented with 15% FBS. Cells obtained from the cerebral hemispheres or the cerebellum of one mouse pup were plated into 100 mm poly-D-ornithine (0.1 mg/ml)-coated culture dishes in 10 ml of culture medium. The cultures were incubated at 37 °C with 5% CO2. The culture medium was replaced 24 h after plating and every third day thereafter. All experiments were performed with glial cultures that had been sub-plated two times. Such cultures have been shown to be highly enriched in astrocytes (Figiel and Engele, 2000). Rotarod analysis The rotarod protocol was adapted from that described in reference (Hockly et al., 2003). Mice were placed on a smooth non-slip rod (model 47600, Ugo Basile, Italy), which accelerated from 4 to 40 rpm over a period of 570 s. The mice were tested on 4 consecutive days, for 4 trials per day, with a rest period of 20 min between the trials. The final rotarod performance score for each mouse was the average from 4 consecutive days. Parallel rod floor test The parallel rod floor apparatus together with the ANY-maze software (Stoelting Europe, Dublin, Ireland) was used to measure locomotor activity and incoordination in the mice (Kamens et al., 2005). The animals were placed in a 20 × 20 cm acrylic box with a floor made of steel rods that were spaced by 8 mm and raised approximately 1 cm above a base steel plate. Each time a mouse's paw slipped through the parallel rods and touched the steel plate, an error was recorded and


the number of errors was used as a measure of incoordination. The locomotor activity was recorded by video camera, and the distance traveled by each mouse was determined. One day before the data collection, the mice were acclimatized to the apparatus for a period of 20 min. During the testing session, exploratory parameters and the number of errors were recorded for a 10-min period. An ataxic phenotype was assessed as a number of paw slip errors normalized to the distance traveled during the test. Static rod test The setup and protocol of the static rods test was adapted from that of Deacon (Deacon, 2013), with modifications. In brief, 5 wooden 60-cm long rods of 35 mm, 28 mm, 21 mm, 17 mm and 9 mm in diameter representing varying grades of difficulty were fixed to a dedicated shelf 60 cm above a floor that was padded to avoid any harm to the mice. The shelf was supplied with nesting material to encourage the mice to go towards the shelf and intensive light was spotted on the starting point of the rod to encourage the mice to turn around and traverse the rod. The time until the mice turned around and the time to traverse the rod were recorded. The mice received a maximum score of 30 s if they failed to turn around, fell off the rod, turned upside down on the rod or failed to traverse, or any combination of these behaviors in this time. The mice were subjected to one day of training and then were tested for 3 consecutive days with a 2–3 min rest between rods. Statistics Two-group comparisons of gene expression data were compared using Student's t-test. Behavioral data were subjected to the repeated measures two-way ANOVA with age and genotype or trial day and genotype as independent variables, followed by Bonferroni post-tests. P values less than 0.05 were considered significant. Abbreviations human Ataxin-3 gene (ATXN3); Cytomegalovirus (CMV); Myotonic Dystrophy type 1 (DM1); Duchenne Muscular Dystrophy (DMD); Glial fibrillary acidic protein (GFAP); Huntington disease (HD); − immunoreactivity (− IR); Common name for heterozygous knock-in SCA3 mouse with 91 CAG repeats (Ki91); Machado-Joseph disease (MJD); polyglutamine (polyQ); prion protein (PrP); rat Huntingtin (rHTT); Spinal-Bulbar Muscular Atrophy (SBMA); Spinocerebellar ataxia type 3 (SCA3) Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.nbd.2014.09.020. Competing interests The authors declare that they have no competing interests. Authors' contributions MF and WJK are joint senior authors. MF and PMS designed and performed the experiments, analyzed the data, and wrote the paper. MF and PMS planned and executed the live animal experiments (rotarod and PRF). MF planned and performed the immunohistochemistry experiments. MF planned and conceived the strategy of mouse generation. MF and WJSz planed and performed static rod experiments. WJK advised on mouse generation and the manuscript. Acknowledgements We would like to express our thanks to Prof. Karl Schilling (Institute of Anatomy, University of Bonn) for helpful and kind discussions on the

187

cerebellar deficits identified in Ki91 animals. We would like to thank Prof. Eliza Wyszko (Laboratory of Subcellular Structures Analysis; Institute of Bioorganic Chemistry, PAS) for granting access to the TCS SP5 II confocal microscope and the Light Cycler 480 II. This work was supported by the European Regional Development Fund within the Innovative Economy Programme (grant numbers POIG.01.03.01-30-049/09 and POIG.01.03.01-30-098/08), a doctoral fellowship for PMS from the National Science Centre (grant number DEC-2013/08/T/NZ4/00712) and a grant from the Polish Ministry of Science and Higher Education (grant number: N N302 299536).

References Apps, R., Hawkes, R., 2009. Cerebellar cortical organization: a one-map hypothesis. Nat. Rev. Neurosci. 10, 670–681. http://dx.doi.org/10.1038/nrn2698. Aziz, N.A., van Belzen, M.J., Coops, I.D., Belfroid, R.D.M., Roos, R.A.C., 2011. Parent-of-origin differences of mutant HTT CAG repeat instability in Huntington's disease. Eur. J. Med. Genet. 54, e413–e418. http://dx.doi.org/10.1016/j.ejmg.2011.04.002. Bichelmeier, U., Schmidt, T., Hübener, J., Boy, J., Rüttiger, L., Häbig, K., Poths, S., Bonin, M., Knipper, M., Schmidt, W.J., Wilbertz, J., Wolburg, H., Laccone, F., Riess, O., 2007. Nuclear localization of ataxin-3 is required for the manifestation of symptoms in SCA3: in vivo evidence. J. Neurosci. 27, 7418–7428. http://dx.doi.org/10.1523/JNEUROSCI. 4540-06.2007. Boy, J., Schmidt, T., Wolburg, H., Mack, A., Nuber, S., Böttcher, M., Schmitt, I., Holzmann, C., Zimmermann, F., Servadio, A., Riess, O., 2009. Reversibility of symptoms in a conditional mouse model of spinocerebellar ataxia type 3. Hum. Mol. Genet. 18, 4282–4295. http://dx.doi.org/10.1093/hmg/ddp381. Boy, J., Schmidt, T., Schumann, U., Grasshoff, U., Unser, S., Holzmann, C., Schmitt, I., Karl, T., Laccone, F., Wolburg, H., Ibrahim, S., Riess, O., 2010. A transgenic mouse model of spinocerebellar ataxia type 3 resembling late disease onset and gender-specific instability of CAG repeats. Neurobiol. Dis. 37, 284–293. http://dx.doi.org/10.1016/j.nbd. 2009.08.002. Cancel, G., Gourfinkel-An, I., Stevanin, G., Didierjean, O., Abbas, N., Hirsch, E., Agid, Y., Brice, A., 1998. Somatic mosaicism of the CAG repeat expansion in spinocerebellar ataxia type 3/Machado-Joseph disease. Hum. Mutat. 11, 23–27. http://dx.doi.org/10. 1002/(SICI)1098-1004(1998)11:1b23::AID-HUMU4N3.0.CO;2-M. Cemal, C.K., Carroll, C.J., Lawrence, L., Lowrie, M.B., Ruddle, P., Al-Mahdawi, S., King, R.H.M. , Pook, M.A., Huxley, C., Chamberlain, S., 2002. YAC transgenic mice carrying pathological alleles of the MJD1 locus exhibit a mild and slowly progressive cerebellar deficit. Hum. Mol. Genet. 11, 1075–1094. Chang, Y.-C., Lin, C.-Y., Hsu, C.-M., Lin, H.-C., Chen, Y.-H., Lee-Chen, G.-J., Su, M.-T., Ro, L.-S., Chen, C.-M., Hsieh-Li, H.M., 2011. Neuroprotective effects of granulocyte-colony stimulating factor in a novel transgenic mouse model of SCA17. J. Neurochem. 118, 288–303. http://dx.doi.org/10.1111/j.1471-4159.2011.07304.x. Chen, X., Tang, T.-S., Tu, H., Nelson, O., Pook, M., Hammer, R., Nukina, N., Bezprozvanny, I., 2008. Deranged calcium signaling and neurodegeneration in spinocerebellar ataxia type 3. J. Neurosci. Off. J. Soc. Neurosci. 28, 12713–12724. http://dx.doi.org/10.1523/ JNEUROSCI.3909-08.2008. Chou, A.-H., Yeh, T.-H., Ouyang, P., Chen, Y.-L., Chen, S.-Y., Wang, H.-L., 2008. Polyglutamine-expanded ataxin-3 causes cerebellar dysfunction of SCA3 transgenic mice by inducing transcriptional dysregulation. Neurobiol. Dis. 31, 89–101. http:// dx.doi.org/10.1016/j.nbd.2008.03.011. Chung, M.Y., Ranum, L.P., Duvick, L.A., Servadio, A., Zoghbi, H.Y., Orr, H.T., 1993. Evidence for a mechanism predisposing to intergenerational CAG repeat instability in spinocerebellar ataxia type I. Nat. Genet. 5, 254–258. http://dx.doi.org/10.1038/ ng1193-254. Deacon, R.M.J., 2013. Measuring motor coordination in mice. J. Vis. Exp. http://dx.doi.org/ 10.3791/2609. Dougherty, S.E., Reeves, J.L., Lucas, E.K., Gamble, K.L., Lesort, M., Cowell, R.M., 2012. Disruption of Purkinje cell function prior to huntingtin accumulation and cell loss in an animal model of Huntington Disease. Exp. Neurol. 236, 171–178. http://dx.doi. org/10.1016/j.expneurol.2012.04.015. Duffin, C.A., McFarland, R., Sarna, J.R., Vogel, M.W., Armstrong, C.L., 2010. Heat shock protein 25 expression and preferential Purkinje cell survival in the lurcher mutant mouse cerebellum. J. Comp. Neurol. 518, 1892–1907. http://dx.doi.org/10.1002/cne. 22309. Dürr, A., Stevanin, G., Cancel, G., Duyckaerts, C., Abbas, N., Didierjean, O., Chneiweiss, H., Benomar, A., Lyon-Caen, O., Julien, J., Serdaru, M., Penet, C., Agid, Y., Brice, A., 1996. Spinocerebellar ataxia 3 and Machado-Joseph disease: clinical, molecular, and neuropathological features. Ann. Neurol. 39, 490–499. http://dx.doi.org/10.1002/ana. 410390411. Eichler, L., Bellenberg, B., Hahn, H.K., Köster, O., Schöls, L., Lukas, C., 2011. Quantitative assessment of brain stem and cerebellar atrophy in spinocerebellar ataxia types 3 and 6: impact on clinical status. Am. J. Neuroradiol. 32, 890–897. http://dx.doi.org/10. 3174/ajnr.A2387. Evert, B.O., Vogt, I.R., Kindermann, C., Ozimek, L., de Vos, R.A., Brunt, E.R., Schmitt, I., Klockgether, T., Wüllner, U., 2001. Inflammatory genes are upregulated in expanded ataxin-3-expressing cell lines and spinocerebellar ataxia type 3 brains. J. Neurosci. Off. J. Soc. Neurosci. 21, 5389–5396. Evert, B.O., Vogt, I.R., Vieira-Saecker, A.M., Ozimek, L., de Vos, R.A., Brunt, E.R., Klockgether, T., Wüllner, U., 2003. Gene expression profiling in ataxin-3 expressing cell lines

188


reveals distinct effects of normal and mutant ataxin-3. J. Neuropathol. Exp. Neurol. 62, 1006–1018. Figiel, M., Engele, J., 2000. Pituitary adenylate cyclase-activating polypeptide (PACAP), a neuron-derived peptide regulating glial glutamate transport and metabolism. J. Neurosci. Off. J. Soc. Neurosci. 20, 3596–3605. Figiel, M., Szlachcic, W.J., Switonski, P.M., Gabka, A., Krzyzosiak, W.J., 2012. Mouse models of polyglutamine diseases: review and data table. Part I. Mol. Neurobiol. 46, 393–429. http://dx.doi.org/10.1007/s12035-012-8315-4. Giunti, P., Sweeney, M.G., Harding, A.E., 1995. Detection of the Machado-Joseph disease/ spinocerebellar ataxia three trinucleotide repeat expansion in families with autosomal dominant motor disorders, including the Drew family of Walworth. Brain J. Neurol. 118 (Pt 5), 1077–1085. Goti, D., Katzen, S.M., Mez, J., Kurtis, N., Kiluk, J., Ben-Haïem, L., Jenkins, N.A., Copeland, N.G., Kakizuka, A., Sharp, A.H., Ross, C.A., Mouton, P.R., Colomer, V., 2004. A mutant ataxin-3 putative-cleavage fragment in brains of Machado-Joseph disease patients and transgenic mice is cytotoxic above a critical concentration. J. Neurosci. Off. J. Soc. Neurosci. 24, 10266–10279. http://dx.doi.org/10.1523/JNEUROSCI.2734-04.2004. Gouw, L.G., Castañeda, M.A., McKenna, C.K., Digre, K.B., Pulst, S.M., Perlman, S., Lee, M.S., Gomez, C., Fischbeck, K., Gagnon, D., Storey, E., Bird, T., Jeri, F.R., Ptácek, L.J., 1998. Analysis of the dynamic mutation in the SCA7 gene shows marked parental effects on CAG repeat transmission. Hum. Mol. Genet. 7, 525–532. Guimarães, R.P., D'Abreu, A., Yasuda, C.L., França Jr., M.C., Silva, B.H.B., Cappabianco, F.A.M., Bergo, F.P.G., Lopes-Cendes, I.T., Cendes, F., 2013. A multimodal evaluation of microstructural white matter damage in spinocerebellar ataxia type 3. Mov. Disord. Off. J. Mov. Disord. Soc. 28, 1125–1132. http://dx.doi.org/10.1002/mds.25451. Hockly, E., Woodman, B., Mahal, A., Lewis, C.M., Bates, G., 2003. Standardization and statistical approaches to therapeutic trials in the R6/2 mouse. Brain Res. Bull. 61, 469–479. Hodges, A., Strand, A.D., Aragaki, A.K., Kuhn, A., Sengstag, T., Hughes, G., Elliston, L.A., Hartog, C., Goldstein, D.R., Thu, D., Hollingsworth, Z.R., Collin, F., Synek, B., Holmans, P.A., Young, A.B., Wexler, N.S., Delorenzi, M., Kooperberg, C., Augood, S.J., Faull, R.L.M., Olson, J.M., Jones, L., Luthi-Carter, R., 2006. Regional and cellular gene expression changes in human Huntington's disease brain. Hum. Mol. Genet. 15, 965–977. http://dx.doi.org/10.1093/hmg/ddl013. Hübener, J., Vauti, F., Funke, C., Wolburg, H., Ye, Y., Schmidt, T., Wolburg-Buchholz, K., Schmitt, I., Gardyan, A., Driessen, S., Arnold, H.-H., Nguyen, H.P., Riess, O., 2011. Nterminal ataxin-3 causes neurological symptoms with inclusions, endoplasmic reticulum stress and ribosomal dislocation. Brain J. Neurol. 134, 1925–1942. http://dx.doi. org/10.1093/brain/awr118. Ikeda, H., Yamaguchi, M., Sugai, S., Aze, Y., Narumiya, S., Kakizuka, A., 1996. Expanded polyglutamine in the Machado-Joseph disease protein induces cell death in vitro and in vivo. Nat. Genet. 13, 196–202. http://dx.doi.org/10.1038/ng0696-196. Kamens, H.M., Phillips, T.J., Holstein, S.E., Crabbe, J.C., 2005. Characterization of the parallel rod floor apparatus to test motor incoordination in mice. Genes Brain Behav. 4, 253–266. http://dx.doi.org/10.1111/j.1601-183X.2004.00100.x. Kawaguchi, Y., Okamoto, T., Taniwaki, M., Aizawa, M., Inoue, M., Katayama, S., Kawakami, H., Nakamura, S., Nishimura, M., Akiguchi, I., 1994. CAG expansions in a novel gene for Machado-Joseph disease at chromosome 14q32.1. Nat. Genet. 8, 221–228 (doi: 7874163). Koide, R., Ikeuchi, T., Onodera, O., Tanaka, H., Igarashi, S., Endo, K., Takahashi, H., Kondo, R., Ishikawa, A., Hayashi, T., 1994. Unstable expansion of CAG repeat in hereditary dentatorubral-pallidoluysian atrophy (DRPLA). Nat. Genet. 6, 9–13. http://dx.doi. org/10.1038/ng0194-9. Kojetin, D.J., Venters, R.A., Kordys, D.R., Thompson, R.J., Kumar, R., Cavanagh, J., 2006. Structure, binding interface and hydrophobic transitions of Ca2 + -loaded calbindin-D(28 K). Nat. Struct. Mol. Biol. 13, 641–647. http://dx.doi.org/10.1038/ nsmb1112. Lin, X., Antalffy, B., Kang, D., Orr, H.T., Zoghbi, H.Y., 2000. Polyglutamine expansion downregulates specific neuronal genes before pathologic changes in SCA1. Nat. Neurosci. 3, 157–163. http://dx.doi.org/10.1038/72101.

Maciel, P., Gaspar, C., DeStefano, A.L., Silveira, I., Coutinho, P., Radvany, J., Dawson, D.M., Sudarsky, L., Guimarães, J., Loureiro, J.E., 1995. Correlation between CAG repeat length and clinical features in Machado-Joseph disease. Am. J. Hum. Genet. 57, 54–61. Nóbrega, C., Nascimento-Ferreira, I., Onofre, I., Albuquerque, D., Hirai, H., Déglon, N., de Almeida, L.P., 2013. Silencing mutant ataxin-3 rescues motor deficits and neuropathology in machado-joseph disease transgenic mice. PLoS One 8, e52396. http://dx. doi.org/10.1371/journal.pone.0052396. Panov, A.V., Gutekunst, C.-A., Leavitt, B.R., Hayden, M.R., Burke, J.R., Strittmatter, W.J., Greenamyre, J.T., 2002. Early mitochondrial calcium defects in Huntington's disease are a direct effect of polyglutamines. Nat. Neurosci. 5, 731–736. http://dx.doi.org/ 10.1038/nn884. Paulson, H., 2012. Chapter 27 – Machado–Joseph disease/spinocerebellar ataxia type 3. In: Subramony, Sankara H., Dürr, Alexandra (Eds.), Handb. Clin. Neurol. Elsevier, pp. 437–449. Riess, O., Rüb, U., Pastore, A., Bauer, P., Schöls, L., 2008. SCA3: Neurological features, pathogenesis and animal models. Cerebellum 7, 125–137. http://dx.doi.org/10.1007/ s12311-008-0013-4. Rodríguez-Lebrón, E., Costa, M. doCarmo, Luna-Cancalon, K., Peron, T.M., Fischer, S., Boudreau, R.L., Davidson, B.L., Paulson, H.L., 2013. Silencing mutant ATXN3 expression resolves molecular phenotypes in SCA3 transgenic mice. Mol. Ther. J. Am. Soc. Gene. Ther. 21, 1909–1918. http://dx.doi.org/10.1038/mt.2013.152. Savouret, C., Brisson, E., Essers, J., Kanaar, R., Pastink, A., te Riele, H., Junien, C., Gourdon, G., 2003. CTG repeat instability and size variation timing in DNA repair-deficient mice. EMBO J. 22, 2264–2273. http://dx.doi.org/10.1093/emboj/cdg202. Scherzed, W., Brunt, E.R., Heinsen, H., de Vos, R.A., Seidel, K., Bürk, K., Schöls, L., Auburger, G., Del Turco, D., Deller, T., Korf, H.W., den Dunnen, W.F., Rüb, U., 2012. Pathoanatomy of cerebellar degeneration in spinocerebellar ataxia type 2 (SCA2) and type 3 (SCA3). Cerebellum 11, 749–760. http://dx.doi.org/10.1007/s12311-011-0340-8. Silva-Fernandes, A., Costa, M. do C., Duarte-Silva, S., Oliveira, P., Botelho, C.M., Martins, L., Mariz, J.A., Ferreira, T., Ribeiro, F., Correia-Neves, M., Costa, C., Maciel, P., 2010. Motor uncoordination and neuropathology in a transgenic mouse model of Machado-Joseph disease lacking intranuclear inclusions and ataxin-3 cleavage products. Neurobiol. Dis. 40, 163–176. http://dx.doi.org/10.1016/j.nbd.2010.05.021. Switonski, P.M., Fiszer, A., Kazmierska, K., Kurpisz, M., Krzyzosiak, W.J., Figiel, M., 2011. Mouse ataxin-3 functional knock-out model. Neruomol. Med. 13, 54–65. http://dx. doi.org/10.1007/s12017-010-8137-3. Switonski, P.M., Szlachcic, W.J., Gabka, A., Krzyzosiak, W.J., Figiel, M., 2012. Mouse models of polyglutamine diseases in therapeutic approaches: review and data table. Part II. Mol. Neurobiol. 46, 430–466. http://dx.doi.org/10.1007/s12035-012-8316-3. Tanaka, F., Sobue, G., Doyu, M., Ito, Y., Yamamoto, M., Shimada, N., Yamamoto, K., Riku, S., Hshizume, Y., Mitsuma, T., 1996. Differential pattern in tissue-specific somatic mosaicism of expanded CAG trinucleotide repeats in dentatorubral-pallidoluysian atrophy, Machado-Joseph disease, and X-linked recessive spinal and bulbar muscular atrophy. J. Neurol. Sci. 135, 43–50. Watase, K., Weeber, E.J., Xu, B., Antalffy, B., Yuva-Paylor, L., Hashimoto, K., Kano, M., Atkinson, R., Sun, Y., Armstrong, D.L., Sweatt, J.D., Orr, H.T., Paylor, R., Zoghbi, H.Y., 2002. A long CAG repeat in the mouse Sca1 locus replicates SCA1 features and reveals the impact of protein solubility on selective neurodegeneration. Neuron 34, 905–919. Winsky, L., Kuźnicki, J., 1996. Antibody recognition of calcium-binding proteins depends on their calcium-binding status. J. Neurochem. 66, 764–771. Zamanian, J.L., Xu, L., Foo, L.C., Nouri, N., Zhou, L., Giffard, R.G., Barres, B.A., 2012. Genomic analysis of reactive astrogliosis. J. Neurosci. Off. J. Soc. Neurosci. 32, 6391–6410. http://dx.doi.org/10.1523/JNEUROSCI.6221-11.2012. Zhang, Y., Monckton, D.G., Siciliano, M.J., Connor, T.H., Meistrich, M.L., 2002. Age and insertion site dependence of repeat number instability of a human DM1 transgene in individual mouse sperm. Hum. Mol. Genet. 11, 791–798. Zimmermann, L., Schwaller, B., 2002. Monoclonal antibodies recognizing epitopes of calretinins: dependence on Ca2 + -binding status and differences in antigen accessibility in colon cancer cells. Cell Calcium 31, 13–25. http://dx.doi.org/10.1054/ceca. 2001.0255.