CONSENSUS PAPER: STRENGTHS AND WEAKNESSES OF ANIMAL MODELS OF SPINOCEREBELLAR ATAXIAS AND THEIR CLINICAL IMPLICATIONS

Summary

Spinocerebellar ataxias (SCAs) represent a large group of hereditary degenerative diseases of the nervous and other systems, in particular the cerebellum that manifest with a variety of progressive motor, cognitive and behavioral deficits with the leading symptom of cerebellar ataxia. SCAs lead often to severe impairments of the patient’s functioning, quality of life and life expectancy. For SCAs, there are so far no proven effective pharmacotherapies that improve the symptoms or substantially delay disease progress, i.e. disease modifying therapies. To study SCA pathogenesis and potential therapies, animal models have been widely used and are an essential part of preclinical research. They mainly include mice, but also other vertebrates and invertebrates. Each animal model has its strengths and weaknesses arising from model animal species, type of genetic manipulation and similarity to human diseases. Types of murine and non-murine models of SCAs, their contribution to investigation of SCA pathogenesis, pathological phenotype, and therapeutic approaches including their advantages and disadvantages are reviewed in this paper. There is a consensus amongst the panel of experts that (1) animal models represent valuable tools to improve our understanding of SCAs, discover and assess novel therapies for this group of neurological disorders characterized by diverse mechanisms and differential degenerative progression, (2) thorough phenotypic assessment of individual animal models is required for studies addressing therapeutic approaches, (3) comparative studies are needed to bring pre-clinical research closer to clinical trials, and (4) animal models complement cellular and invertebrate models which remain limited in terms of clinical translation for complex neurological disorders as SCAs.

1. Introduction

Spinocerebellar ataxias (SCAs) represent a wide group of hereditary degenerative diseases of the nervous system particularly the cerebellum [1]. These diseases often progress to severe disability of the patient and even premature death [2, 3]. The pathogenesis most types of SCAs is so far not completely understood. Although some common features can be found especially in the later stages of disease, mechanisms of disease development are not the same in the different type of SCAs. Finally, there is neither a proven symptomatic nor a disease modifying pharmacotherapy for SCAs as of yet [4].

Animal models are intensively used for investigation of the pathogenetic mechanisms contributing to neuropathology development and particular manifestations of SCA as well as for development of new therapeutic approaches [5, 6]. Most animal models of SCAs are genetically modified mice allowing investigation of the diseases on several levels, but also other vertebrates and invertebrates are used. It is important to note that the spectrum of animal models reflects quite well the variability of human cerebellar degenerations. Nevertheless, new mutations associated with cerebellar neurodegenerative pathologies are still being discovered. On the other hand, for some human diseases, there are even several mouse models developed by different methods of gene engineering or carrying different variants of pathological alleles of the respective gene.

A crucial question is how exact and reliable transfer of knowledge from the animal study to a patient is and what aspects of the human diseases are really reflected in their animal models. Contribution of studies on animal models of SCAs to recognizing pathogenesis of SCAs on cellular, organ and functional level, various approaches of experimental therapy for SCAs, parallels and discrepancies between the models and human diseases are reviewed and discussed here. Finally, in terms of treatment, what ultimately counts are state-of-the-art randomized controlled clinical trials with clinically meaningful endpoints to evaluate the efficacy of treatment option, which arise from translational and back-translational research.

2. Definition of Spinocerebellar ataxias and problems of their therapy in humans (Mario Manto)

The terminology of SCAs describes a group of hereditary cerebellar ataxias with a dominant transmission. The prevalence is estimated between 1 to 5 cases per 100.000. The number of identified SCAs is continuously expanding and includes: SCA1–8, 10, 12–14, 15, 17–22, 25, 27, 28, 31, 32, 34–37, 38, 42–44, 46-48, ataxia with DNMT1 (DNA methyltransferase 1 mutation) and DRPLA (Dentatorubral-pallidoluysian atrophy). The phenotype of SCAs is highly heterogeneous. Patients exhibit features of cerebellar ataxia as a result of a progressive neurodegenerative process involving the cerebellum and its afferent/efferent pathways, in particular the brainstem [7]. Cerebellar symptoms often start during adulthood and occur in the motor domain (incoordination of speech, limbs, gait), oculomotor domain (in particular nystagmus, smooth pursuit deficits, and dysmetria of saccades) and/or cognitive/affective/emotional domain (Schmahmann’s syndrome) [8]. In addition, patients may exhibit pyramidal signs, extra-pyramidal signs, epilepsy, dementia, ophthalmoplegia, pigmentary retinal degeneration, signs of spinal cord involvement and peripheral neuropathy. Based on dominant neuropathology and symptomatology, there are cerebellar ataxias with a variety of associated signs (autosomal dominant cerebellar ataxias I, ADCAI) including SCA1-4,8,10,12-13,17-23, FGF14, 25, 27-28,32,35,36, cerebellar ataxia with macular degeneration (ADCAII) represented by SCA7, and a group of pure cerebellar ataxias (ADCAIII) consisting of SCA5-6, 11, 14-16, 20,22,26,19-31. Combination of symptoms and their intensity (including those arising directly from cerebellar degeneration) can be variable even in the frame of one SCA type.

The phenotype of SCAs may also overlap with the phenotype of hereditary spastic paraplegias (HSPs). Nevertheless, in some cases, the clinical features are highly suggestive of a SCA type:

• pigmentary retinal degeneration is suggestive of SCA7 (previously named ADCA type II)

• SCA14 is characterized by a myoclonus and dystonia

• upper limb postural tremor is common in SCA12, SCA15 and SCA27

• psychiatric symptoms are relatively common in SCA2 and SCA17

• ichthyosiform plaques are observed in SCA34

• facio-lingual fasciculations and sensorineural hearing loss occur in SCA36

• narcolepsy has been reported in DNMT1

Despite the highly heterogeneous clinical presentations, even within members of a SCA family, and the substantial progress in genetic tools, the phenotypic approach as a first step to identify the SCA type remains appropriate in 2020, complemented by neuroimaging and genetic assessment. A history of episodic attacks of ataxia should prompt the search for an episodic ataxia (EA), a group of disorders which (1) are also transmitted in an autosomal dominant manner, (2) may include a phenotype of migraine, seizure or crisis of dystonia, and (3) often start before the age of 25 years.

The global prevalence of SCAs is 3/100.000 with a large variation in prevalence of subtypes between regions of the world [9]. In particular, the prevalence of SCA2 is higher in Cuba (Holguin), SCA6 in the North of England, SCA7 in Venezuela, SCA36 in Spain (Galicia), and DRPLA in Japan. The variation is due to the different genetic make-up of the populations and founder effects in some regions of the world.

From the genetic standpoint, SCAs are also heterogeneous:

• A group of SCAs is caused by CAG repeat expansions encoding for polyglutamine (polyglutaminopathies) and is characterized by anticipation. The disease penetrance is high. This group accounts for about half of SCAs in Europe [7]. Non-repeat mutations are less common than repeat expansions [10].

• Pathogenic variants have been reported in ion channel genes such as CACNA1A, KCND3, KCNC3 and KCNA1 [11].

The list of mutations is expanding continuously. Thanks to next generation sequencing technique, the molecular causes of SCAs are better understood. As compared to a targeted panel of genes with a diagnostic rate of about 17%, the exome sequencing reaches a diagnostic rate of about 35 % [12].

Finally, SCAs can be also characterized or categorized based on predominant pathogenetic mechanism of the neuropathology. In normal situation, the CAG repeat is translated into polyglutamine (polyQ) tract in normal protein. This results in a protein showing normal folding. In case of expansion, the translation process generates an expanded polyQ, which results in protein aggregates (protein misfolding). These aggregates trigger neuronal dysfunction and death. The main mechanisms of SCAs can be summarized in (a) toxic gain of function for the RNA, (b) transcriptional dysregulation, (c) mitochondrial dysfunction, (d) channelopathies. In SCA10, the pentanucleotide repeat expansion ATTCT causes AUUCU RNA aggregates binding to the nuclear ribonucleoprotein K and triggering its sequestration leading to apoptosis [13]. The toxic gain of function for the RNA has also been documented in SCA8 (birectionally expressed CTG CAG repeat expansion), SCA36 (GGCCTG hexanucleotide repeat expansion) and SCA37 (upregulation of the reelin-DAB1 signaling). Degradation of misfolded proteins is under control of autophagy and the ubiquitin-proteasome system. Impaired autophagy has been shown in SCA3 and SCA7 [14]. Evidence for aberrant RNA metabolism is expanding in the SCAs [15]. In terms of transcriptional dysregulation, RNA interference and protein-DNA interactions have been reported [16]. Regarding mitochondria, there is evidence that reactive oxygen species (ROS) contribute to the pathogenesis of SCA2 [17]. Inherited channelopathies are involved in SCAs. A typical example is the mutation of the CACNA1A. These calcium voltage-gated channel subunits are highly expressed in cerebellar Purkinje and granular cells and cerebellar nuclei. The mutation impairs the physiology of Purkinje cells and results in impaired patterns of simple spikes/complex spikes. For further details on pathogenesis of SCAs shown by animal model studies, see part 5.

There are currently no disease-modifying therapies to delay or even stop disease progression in SCAs. The elucidation of the pathogenesis of the SCAs is a critical step towards efficient therapies. Given their wide phenotypic/genotypic presentation, there is a huge need to obtain relevant models having a powerful translational impact. Findings obtained from animal models complement the findings resulting from progress in genetics and neuroimaging [4]. Developments are ongoing with antisense oligonucleotides and siRNA/shRNA, for instance in SCA1, SCA2, SCA3, and SCA7 [18–21]. Although most SCAs are associated with a prominent involvement of the cerebellar cortex, in other cases cerebellar cortex is relatively spared [10], highlighting that translational models need also to take into account the region of the cerebellum which is mainly affected. The recently described neuroanatomical links between cerebellum and basal ganglia have changed our view of the communication between cerebellum and cerebral cortex. Animal models may also contribute to a better understanding of the cerebello-basal ganglia-thalamo-cortical system which is often involved in SCAs. There is also a big gap in terms of unraveling of the cerebello-spinal communication in SCAs. Overall, the dissection the numerous mechanisms of cerebellar dysfunction in SCAs remains a major process for the assessment of therapeutics targeting deleterious pathway and for the screening of old or newly synthesized drugs [5]. Animal models are critical to reach this goal, keeping in mind their limitations and the prevailing opinion of the “3R” (Replacement, Reduction and Refinement) in animal research. Molecular biomarkers, such as neurofilament light chain (NfL) or ataxin proteins in blood, start to be accessible for clinical trials. Animal studies should also consider the evaluation of such biomarkers.

3. Development of murine models of spinocerebellar ataxias (Hirokazu Hirai)

There are several methods by which genetically modified animal models of human diseases can be generated. Consequently, there are more than a few types of animal models having various features specific to SCAs, which are advantageous as well as disadvantageous as tools to investigate SCAs. Particularly, level of similarities between the model and human disease on genetic, biochemical, cellular, neuropathological, and behavioral levels are of importance. On the other hand, some studies may require a simplified model highlighting only one component of the disease that has to be studied more specifically.

3.1. Conventional gene-modified mice

3.1.1. Transgenic mice

SCA transgenic (Tg) mice are produced by injection of linearized DNA carrying a promoter and mutant SCA gene into fertilized eggs. To express a mutant gene, previous studies used ubiquitous promoters like a cytomegalovirus (CMV) promoter or cell type-specific promoters such as the astrocyte-specific promoter and the Purkinje cell-specific L7/PCP2 promoter [22]. A major disadvantage of this method is difficulty in obtaining mice with suitable levels of mutant gene expression. High expression causes developmental defects, while weak expression requires long period before abnormal phenotypes emerge or results in loss of phenotype expression. Another drawback is that regions and cell types that are affected are determined by the exogenous promoter, and therefore, degenerative regions/cells may differ between the SCA transgenic mice and patients of corresponding SCA types. In addition, it is important to evaluate whether transgene insertion itself disrupts expression of genes at/around the insertion site.

3.1.2. Conditional Tg mice

Production of tetracycline-inducible conditional mice (Tet-on or Tet-off mice) is a good strategy to avoid defects that are caused by mutant gene expression during development. Application of doxycycline (Dox) to conditional Tet-on mice, which developed normally, initiates mutant gene expression. Conversely, Dox application in Tet-off mice suppresses mutant gene expression [23]. Generally, it takes a few weeks or more before manifestation of the ataxic phenotype. A primary concern is whether the mouse pathology is analogous to that of human SCA patients. Even in the absence of developmental defects, activated TRE (Tetracycline response element) promoter triggers high expression of mutant gene, likely leading to substantial endoplasmic reticulum stress in cerebellar neurons, whereas in SCA patients symptoms progress slowly with a more than 10-year time course. Thus, we cannot exclude the possibility that cellular mechanisms and extent of acute neuronal damage by massive production of mutant protein differs from chronic neurodegeneration seen in human SCA patients.

3.1.3. Knock-in mice

Knock-in mice express mutant SCA gene under the control of the endogenous promoter. Previously, knock-in mouse models of SCA1 [24, 25] and SCA3 [26, 27] have been produced. Generally, endogenous promoters show weaker promoter activity, compared with exogenous promoters used in Tg mice. Thus, homozygous knock-in mice expressing ATXN1 with moderately expanded 78 CAG repeats displayed only slightly poorer rotarod performance than their wild-type littermates at 9 months of age, but showed no ataxic phenotype in a home cage up to 18 months of age [25]. To facilitate expression of pathologic phenotypes, extremely long (154) CAG repeats were introduced in the SCA1 knock-in mice [24]. The SCA1 knock-in mice showed neuronal degeneration throughout the central nervous system, including cerebellar Purkinje cells, brainstem and spinal motor neurons [28, 29], resulting in progressive ataxia and premature death at around 50 weeks of age [24]. SCA3 knock-in mice that expressed ATXN3 carrying (relatively long) 91 CAG repeats showed slow progression of disease phenotype and only faint behavioral abnormality even at 1.5 years of age [27]. Similarly, another SCA3 knock-in mice that expressed ATXN3 comprising shorter 82 CAG repeats failed to exhibit motor defects even at one year of age, though intranuclear inclusions in various brain regions including Purkinje cells were present [26].

3.2. Production of model mice using viral vectors

Recent progress of viral vector technology allows us to express a transgene efficiently in neurons and glial cells in the CNS by direct injection as well as intravenous administration. To deliver a transgene to the CNS, lentiviral vectors and adeno-associated virus (AAV) vectors are used.

3.2.1. Overexpression models

We have previously produced SCA1 model mice [30] and SCA14 model mice [31] by cerebellar parenchymal injection of lentiviral vectors. SCA1 model mice virally expressed ATXN1[Q76] by the murine embryonic stem cell virus (MSCV) promoter, while SCA14 model mice virally expressed mutant protein kinase Cγ (S119P) by the MSCV promoter [31]. Both SCA1 and SCA14 model mice showed aggregation of mutant proteins in Purkinje cells and behavioral defects in rotarod test. Benefits of viral vector-mediated production of SCA mice are a simple procedure just by viral injection to mature mice and, accordingly, absence of developmental impairment. In addition, we can control expression levels of a mutant gene by adjusting the viral titers injected. Meanwhile, disadvantage of the direct virus injection is fluctuation of transduction efficacy, depending on the experimenters’ skill.

For instance, AAV-PHP.B, the blood-brain barrier (BBB)-penetrating AAV vectors have been reported [32]. Intravenous infusion of AAV-PHP.B in mice efficiently transduces the whole brain. Combination of AAV-PHP.B with a cell type-specific promoter permits us to express a transgene in a specific cell population [33]. Indeed, we have produced SCA3 mice by intravenous infusion of AAV-PHP.B expressing ATXN3 [Q89] under the control of the neuron-specific rSynI-minCMV promoter [34]. Thus, intravenous infusion of AAV-PHP.B enables efficient and reliable expression of a mutant gene throughout the brain or in a specific cell population when combined with a cell type-specific promoter.

3.2.2. Knock-down models

Previous studies have shown significant reduction of retinoid-related orphan receptor α (RORα) expression in Purkinje cells of SCA1 and SCA3 model mice [35–37]. To examine contribution of RORα reduction in the disease pathology, RORα expression was downregulated specifically in mature mouse Purkinje cells by intravenous infusion of AAV-PHP.B expressing a microRNA against RORα (miR-RORα) under the control of the Purkinje cell-specific L7-6 promoter [38]. The RORα knock-down caused degeneration of Purkinje cells. In behavioral experiments, mice expressing miR-RORα showed motor learning deficits, and later, overt cerebellar ataxia.

Previous studies showed that activity of chaperone-mediated autophagy (CMA) was reduced in the cellular model of SCA14 [39] and cellular and mouse models of SCA21 [40]. To clarify contribution of CMA defect to the SCA pathology, Seki’s group used AAV9 vectors to knock-down lysosome-associated protein 2A (LAMP2A), a protein essential for CMA [41]. Injection of AAV9 vectors expressing a microRNA against LAMP2A by the synapsin I promoter to mouse cerebellum caused progressive motor impairment and cerebellar neuronal loss, suggesting significant contribution of CMA impairment in the pathogenesis of various SCAs. Thus, AAV vectors work well also for production of knock-down models.

3.3. Production of model mice using CRISPR/Cas9

Clustered regularly interspaced short palindromic repeats (CRISPR)–CRISPR-associated protein (Cas)9 enables target gene modification [42]. By applying CRISPR/Cas9 to developmental engineering, we can rapidly produce various SCA model mice carrying a point gene mutation or abnormally expanded CAG stretch in genes responsible for SCAs. Since these CRISPR/Cas9-based model mice express mutant genes by endogenous promoters, they seem to precisely recapitulate the disease phenotype. Thus, this would be an ideal approach if clear disease phenotype emerges within the lifespan of modified mice.

3.4. Advantages and disadvantages of mouse models

Analogous methods can be basically used to develop also non-murine models of SCAs (see part 4). Nevertheless, there are many advantages in using mouse models. Chiefly, methods analyzing mouse phenotype have been well established. Ataxia can be assessed by rotarod, beam-walking test and foot print. Many excellent antibodies are available for mice, which allow us to examine morphology and intracellular changes in various molecules. In addition, synaptic function, synaptic plasticity and neuronal viability can be explored by patch clamp and Ca²⁺ imaging, which are difficult in larger animals like non-human primates. In contrast, sophisticated motor functions cannot be assessed in mice, which includes speech and fine/skillful muscle movements like usage of tools. It is possible, but less precise in mice than in non-human primates, to assess defects of higher brain functions such as intellectual and psychiatric disability.

4. Non-murine models of SCAs (Jan Tuma, Jan Cendelin)

Besides mice, several other animal species have been used to develop SCA models (Table 1). Among mammals, there are also ataxic rats and hamsters. Non-mammalian models of SCAs are fish (Danio rerio) and marginally frog (Xenopus laevis). Among invertebrates, Drosophila and Caenorhabditis elegans expressing gene alleles encoding SCA-determining pathological proteins have been developed. Vertebrates, particularly rodents can be used to investigate brain neuropathology, motor deficits as well as behavioral and cognitive impairments related to SCA. Invertebrates serve mainly to study subcellular and biochemical components of SCA pathophysiology. These aspects are important to understand the mechanisms of cell death and factors that could potentially delay the degeneration.

Table 1:

Overview of the main non-murine animal models of SCAs

SCA	Disease human gene	Mutational mechanism	Animal model	Gene manipulation	Reference
SCA 1	ATXN1	CAG repeat	Drosophila melanogaster	mutant	[43]
SCA 2	ATXN2	CAG repeat	Drosophila melanogaster	mutant	[44]
SCA 3 (MJD)	ATXN3	CAG repeat	Drosophila melanogaster	mutant	[45–48]
			Danio rerio	mutant	[49, 50]
			Caenorhabditis elegans	mutant	[51–54]
			Marmoset lines	mutant	[55]
SCA 6	CACNA1A	CAG repeat	Drosophila melanogaster	mutant	[56]
SCA 7	ATXN7	CAG repeat	Drosophila melanogaster	mutant, con-mutant	[57, 58]
			Danio rerio	morphant	[59]
SCA 8	ATXN8OS (opposite strand)	non-coding CTG repeat	Drosophila melanogaster	mutant	[60]
SCA 13	KCNC3	point mutation	Danio rerio	mutant	[61, 62]
SCA 17	TBP	CAG/CAA repeat	Drosophila melanogaster	mutant	[63]
			Rattus norvegicus	mutant	[64]
SCA 31	BEAN1, TK2	TGGAA repeat	Drosophila melanogaster	mutant	[65, 66]

MJD…Machado-Joseph disease

con-mutant… conditional mutant

Rattus norvergicus, is a relatively new and spare but very promising animal model of SCAs that can be successfully employed in their investigation. Rats as mammals share many characteristics with mice; however, some phenotypic traits of particular SCA’s are difficult to fully mimic with mouse models. Rats are excellent animal model for learning and behavior generally. Also, their larger brain facilitates direct non-invasive in vivo procedures such as positron emission tomography and diffusion tensor imaging [64].

Another rodent used for cerebellar degeneration and ataxia investigation is the ataxic Syrian hamster (Mesocricetus auratus) [67]. It is not a model of one of the human SCA, but a spontaneous mutant homologous to Purkinje cell degeneration (known as pcd) mice [68]. In the hamster, the mutation leads to milder pathological phenotype than in Purkinje cell degeneration mice suggesting the importance of the genetic background [68].

Zebrafish (Danio rerio) is getting more popular in last few years. Its cerebellar cytoarchitecture and functions are highly conserved compared with mammals and it offers a useful platform to demonstrate causal relation between the mutation and resulting defect [49, 62]. Zebrafish animal model can be easily knocked-down using antisense oligonucleotides (morpholinos), in order to validate the rescue effect through morphant phenotype after the injection of human, or zebrafish, wild-type or mutated gene variant transcript [59]. Culturing neurons from transgenic zebrafish embryos is also a useful tool for investigating cell biology and proteinopathy signatures of mutant proteins related to development of SCA [50]. Furthermore, optical transparency of zebrafish embryos allows for longitudinal in vivo high-resolution fluorescence imaging of neurodegeneration [61].

Xenopus laevis constitutes only minor animal model organism that is involved in the study of SCAs. Nevertheless, this animal model has a potential to clarify our understanding of some transcriptional mechanisms in the etiology of SCAs [69] or in finding of their pharmacological treatments [70].

Drosophila melanogaster is a significant non-murine animal model that provides a spectrum of particular models in order to study SCAs (see Table 1). In the case of nucleotide repeat ataxias, Drosophila melanogaster very often offers several variants based on the number of nucleotide sequence repeats (for details see review [71]). This animal model is predominantly used in the study of mutation-related neurotoxicity of pathological protein isoforms [72] or pathogenic role of RNA transcripts [46]. However, it has been also employed as a model of particular SCAs behavioral phenotypes, such as motor behavior [73] or learning and memory [44].

Caenorhabditis elegans has been employed to investigate the role of normal ataxins, which is not completely known [74–76]. Several studies have used genetically modified Caenorhabditis elegans, particularly a model of SCA3 to study consequences of mutant ataxin expression and modes of its toxicity modulation that could have therapeutic potential [51–54].

Among non-murine models, rats would have a potential to substitute mice in the SCA research having advantages for translational research compared to mice. Nevertheless, genetically modified mouse strains still dominate over rats for practical and technical reasons. Advantage of invertebrate models includes a short reproducing cycle, relatively easy and cheap breeding, less ethical and legislative limitations for their use in experimental studies. Their disadvantage is different arrangement of the organism complicating translation of the findings to humans and excluding their use for more complex studies.

Diseases analogous to SCAs occur also e.g. in dogs, cats, and horses. Nevertheless, these species are not really used as models to investigate human SCAs.

5. Lessons about SCA pathogenesis from model systems (Mandi Gandelman, Stefan M. Pulst)

Understanding disease mechanisms at the molecular level is key to developing effective treatments targeted towards each disease specifically or across multiple disease entities. Recognizing causative mechanisms that directly trigger the death of neurons and that are maladaptive versus secondary changes that emerge as an adaptive response remain a challenge even with the use of animal models. Here, we provide an overview of the contribution of animal models to our knowledge of 5 particular SCAs that exemplify both converging and distinctive mechanisms of disease.

5.1. SCA1

Studies of SCA1 animal models have provided valuable insight into the pathogenic mechanisms of expanded ATXN1, suggesting they lie at the intersection between gain of toxic function and loss of function changes. Whereas complete loss of function by deletion of ATXN1 does not cause neurodegeneration or ataxia-like symptoms [77], partial loss of function by removal of the non-expanded Atxn1 allele in Atxn1^154Q/2Q KI mice, with the KI allele remaining untouched aggravates the disease [78, 79].

The aberrant interactome and mislocalization of expanded ATXN1 informs on multiple potentially pathogenic gain-of-function mechanisms. Nuclear localization of ATXN1 is necessary for disease development in Purkinje neurons, as Pcp2-ATXN1(82Q) mice with a K772T mutation that disrupted the nuclear localization signal and prohibited nuclear import of ATXN1 failed to develop ataxia symptoms [80]. The interactome of expanded ATXN1 is greatly enriched with nuclear transporter proteins, causing their mislocalization and that of its cargoes to ATXN1 nuclear inclusions in both cellular models and the ATXN1(Q82) mice [80–82].

Phosphorylation of Ser residue 776 is also required for disease to develop. Reduction of phospho-ATXN1- S776 in mice decreased ATXN1 inclusion formation and improved cerebellar motor performance phenotypes [83, 84]. Interaction of ATXN1 with the 14-3-3 chaperone was enhanced with polyQ expansion, and binding to 14-3-3 prevents ATXN1 dephosphorylation at serine 776, stabilizing it and increasing its levels, thus promoting the formation of pathogenic nuclear inclusions [85, 86].

It was first shown in Drosophila that interactions of expanded ATXN1 through its AXH domain had a key role in pathogenesis [87]. Deletion of this domain abolishes the interaction of expanded ATXN1 with Senseless, causing this protein to induce a pathological phenotype [87]. Analogously, in mice, loss of Gfi-1, Senseless mammalian ortholog, leads to Purkinje cell degeneration [87]. The AXH domain of ATXN1 also mediates its interaction with the transcription repressor Capicua, modulating its activity. Reducing levels of Capicua in the Atxn1^154Q/2Q mice by genetic interaction improved learning, memory, life span and motor phenotypes, demonstrating the importance of the ATXN1-capicua interaction for disease [88].

5.2. SCA2

Both polyQ expanded ATXN2 and wild type ATXN2 localize to stress granules in response to stress. Stress granule formation is initiated by phosphorylation of the eukaryotic translation initiation factor 2 subunit 1 (eIF2α), which is hyperphosphorylated in the cerebellum of the ATXN2-Q127 mouse [89]. Aggregation of ATXN2 within stress granules and defective autophagy clearance of stress granules were described in this mouse model, with cytoplasmic aggregates of ATXN2 and its interacting protein Staufen-1 (STAU1) in Purkinje cells [15]. STAU1 is a key mediator of ATXN2 toxicity, as it can regulate autophagy, p-eif2a and endoplasmic reticulum stress levels in response to expanded ATXN2. Notably, reducing STAU1 levels by genetic interaction improves SCA2 molecular phenotypes and ataxia-like symptoms in ATXN2-Q127 mice, constituting a novel target to treat SCA2 [15, 89]. STAU1 levels are also increased across multiple neurodegeneration models, including amyotrophic lateral sclerosis (ALS), possibly constituting the missing link that makes ATXN2 intermediate expansion increase the risk for ALS [90, 91]. Remarkably, an antisense oligonucleotide targeting ATXN2 developed with the aid of the ATXN2-Q22 mouse model, which expresses human ATXN2 without a pathological expansion, has recently entered phase 1 clinical trials for ALS patients [92].

Transcriptome studies in ATXN2-42KI mice and electrophysiological experiments in ATXN2-Q127 mice have established a key role for calcium dysregulation in the pathology of SCA2. Cytoplasmic calcium levels were elevated in Purkinje cells of ATXN2-Q127 mice, associated with synaptic mGluR1 dysfunction [93]. Indeed, administering dantrolene, antagonist of ryanodine receptors that inhibits calcium release from endoplasmic reticulum, ATXN2-Q58 mice attenuated Purkinje cell loss and improved motor deficits [94].

5.3. SCA3

ATXN3 is a nucleo-cytoplasmic deubiquitinase, with roles in protein quality control and regulation of protein degradation by the ubiquitin-proteasome system. ATXN3 knock-out mice present with an increase in total ubiquitinated protein but are phenotypically normal [95]. As a result of polyQ expansion in ATXN3, the protein forms pathological aggregates in cellular nuclei and its interaction with his normal binding partners, which includes multiple transcription factors, is altered, leading to changes in gene expression (reviewed in [96, 97]). The consequences of ATXN3 aggregation in its gene regulation functions was evidenced in a recent study that found overexpression of wild type ATXN3 in the YAC15Q mouse model results in only minor transcriptomic changes, however polyQ expansion, its level of expression and degree of aggregation correlate with a greater transcriptional deregulation [98]. Key experiments in the ATXN3-Q148 mouse showed that deletion of the nuclear export signal resulted in a more benign disease phenotype, while targeting ATXN3 to the nucleus with a nuclear localization signal resulted in earlier and more severe disease [99].

5.4. SCA6

The CACNA1A gene has an internal ribosomal entry site (IRES), leading to the production of two unrelated proteins, the a1A subunit of P/Q-type voltage gated calcium channel and a1ACT, a transcription factor, which bears the polyglutamine repeat at its C-terminal end [100, 101]. The distinctive roles of these two proteins support the view that SCA6 pathological changes are brought about by an overlap of ion channelopathy, transcription dysregulation and polyglutamine disease. SCA6 KI mouse variants have been generated, however splice variants and altered expression of the KI gene have confounded some of the results. In spite of this, they have provided evidence against a pure channelopathy mechanism [102, 103]. An AAV KI mouse expressing a P/Q-type channel with a SCA6-related long C-terminal in Purkinje cells showed an SCA6-like phenotype, including ataxia, long-term depression (LTD) and long-term potentiation (LTP) dysfunction with loss of eyeblink conditioning, Purkinje cell firing alterations and degeneration and nuclear and cytoplasmic aggregates [104]. In a different AAV model, delivering a1ACT to neonatal mice causes severe ataxia and developmental Purkinje cell abnormalities [105]. These studies point at the a1A C-terminal protein as the pathogenic product of CACNA1A and suggest targeting the IRES could constitute a successful disease therapy.

5.5. SCA7

Multiple pathogenic mechanisms have been described for SCA7. Early studies in the Sca7^266Q/5Q and ataxin-7[Q128] mouse models show large intranuclear aggregates of ATXN7 forming faster in vulnerable neuron populations [106, 107]. Proteolytic modifications of ATXN7 can modulate its toxicity, most prominently shown in the SCA7-D266N mouse, which prevents caspase-7 proteolysis of expanded ATXN7 and exhibits decreased neurodegeneration, improved motor phenotype and extended lifespan [108].

ATXN7 is a subunit of the Spt-Ada-Gcn5 acetyltransferase (SAGA) protein complex, a transcriptional coactivator with chromatin remodeling activities. In the nucleus, SAGA components are sequestered into aggregates, resulting in transcriptional alterations. In ATXN7-Q52 mice multiple gene groups related to motor dysfunction and neuronal degeneration are deregulated, including neuronal differentiation, synaptic transmission, axonal transport, glial homeostasis and myelin sheath proteins [109].

6. Functional impairments and their neuropathological correlates in SCA mouse models (Filip Tichanek)

To investigate functional impairments and gross neuropathology, mouse models are predominantly used among animal models of SCAs. Mouse models of SCA show more or less characteristic morphological and functional features of the disease adapted to a mouse organism. Nevertheless, they are not always precise analogy to the human disease and not all aspects of the disease have been examined in detail yet. Functional impairments and gross neuropathology in representative SCA mouse models are summarized in Table 2 and already published reviews [6, 160–164]. The mouse models expressing the mutated gene specifically outside the brain (e.g. muscle-specific models of SCA17 [153]) were not included. Similarly, for the SCA types with expanded repeats (e.g. polyglutamine SCAs, CTG-repeats disorder SCA8 or ATTCT expansion in SCA10), we did not include models without an analogous expansion in the affected gene (e.g. knock-out and other [84] mice).

Table 2: Representative mouse models of spinocerebellar ataxias, their gross neuropathology and functional impairments.

Knock-in and knock-out mice models include ‘KI’ or ‘KO’ respectively at the end of the name of given model. ‘PC arbor atrophy’ indicates either measurably reduced volume of the cerebellar molecular layer or decreased complexity of the Purkinje cells dendritic arbors. ‘Motor deficit’ indicate reduced performance on rotarod or in walk beam test. ‘Ataxia’ indicates either altered gait using specific gait examinations or apparently uncoordinated and imbalanced movement. ‘Hypoactivity’ and ‘hyperactivity’ indicate altered distance walked in open field, immobility time in the open field or changes in home cage activity.

Disease (gene)	Mouse model (type)	Cerebellar pathology	Non-cerebellar neuropathology	Behavioural deficits	Other impairments
SCA1 (ATXN1)	Pcp2-ATXN1[82Q] [110]	Cerebellar gliosis [111] PC loss [112]		Motor deficit and ataxia [112] Hyperactivity, impaired cognition [112, 113]
	Atxn1-78Q/78Q KI [25]			Mild motor deficit
	Atxn1-154Q/2Q KI [24]	Cerebellar gliosis [111] PC shrinkage, loss and arbor atrophy [114	General brain atrophy [24, 114] Brainstem and spinal pathology [115] Hippocampal neuronal loss [116]	Motor deficit, ataxia, clasping [24, 114, 117] Impaired cognition [24, 113, 114, 118] Reduced activity, behavioral despair [114]	Muscle wasting, reduced weight [24, 114] Respiratory dysfunction, early death [115]
SCA2 (ATXN2)	BAC-Q72 [119]	PCs arbor atrophy		Motor deficit	Reduced body weight
	Pcp2-ATXN2-127Q [120]	PC shrinkage, loss and arbor atrophy		Motor deficit.
	Pcp2-ATXN2-58Q [121]	PC loss		Motor deficit, ataxia and clasping [94, 121]
	ATXN2-75Q [122]	PC arbor atrophy		Motor deficit
	Q42 KI [123]			Motor deficit	Reduced body weight
SCA3 (ATXN3)	YAC ATXN3-84Q [124]	Cerebellar gliosis and neuronal loss	NL loss in brainstem and SN [124, 125]	Motor deficit, ataxia, clasping, hypoactivity	Reduced body weight, premature death
	Q71C [126]		Neuronal loss in SN	Motor deficit, ataxia, hypoactivity	Reduced body weight, premature death
	70.61 [99]	PC shrinkage and loss		Motor deficit, ataxia, clasping, hypoactivity	Reduced body weight, premature death
	Q79HA [127]	PC shrinkage and arbor atrophy		Motor deficit, ataxia, clasping, hypoactivity	Reduced body weight
	PrP/MJD77 cond. [128]	PC shrinkage.		Motor deficit, ataxia, clasping, hyperactivity
	HDProm-MJD148 [129]	PC shrinkage		Motor deficit and abnormal activity
	hemi-CMVMJD94 [130]	Altered CN morphology	Brainstem and thalamic neuronal pathology	Motor deficit
	Ki91KI [27]	PC loss and cerebellar gliosis	Astrogliosis in SN	Motor deficit
	Atxn-304Q KI [131]	PC loss		Motor deficit and ataxia	Reduced body weight
SCA5 (SPTBN2)	ẞ-NISI KO [132]	PC shrinkage, loss and arbor atrophy		Motor deficit and ataxia
	SCA5 184-2 and 645 [133]	PC arbor atrophy		Motor deficit and ataxia
SCA6 (CACNA1A)	SCA6-84Q KI [103]	PC shrinkage, loss and arbor atrophy [134]		Motor deficit and impaired balance
	MPI-118Q KI [135]	PC shrinkage, loss and arbor atrophy		Motor deficit and ataxia
SCA7 (ATXN7)	PrP-SCA7-c92Q [136, 137]	PC shrinkage and arbor atrophy	Retinal degeneration	Motor deficit, ataxia and clasping Reduced exploratory behavior	Impaired vision, tremor Reduced body weight, premature death
	Sca7-266Q/5Q KI [106]	PC shrinkage	Retinal degeneration General brain atrophy Brainstem degeneration [138]	Motor deficit and ataxia Reduced activity	Tremor and myoclonus, impaired vision Reduced body weight, muscle wasting Respiratory dysfunction, early death [138]
	Ataxin-7-Q52 [109]	PC shrinkage and arbor atrophy		Motor deficit, ataxia, clasping, hypoactivity
SCA8 (ATXN8)	BAC-Exp [139]			Motor deficit and ataxia Reduced activity	Muscle stiffness, wasting, kyphosis Reduced body weight, premature death
SCA10 (ATXN10)	pcp2-LacZ [13]		NL in hippocampus, NP in cortex and pons	Ataxia, clasping and reduced activity	Susceptibility to seizures
SCA13 (Kcnc3)	Kv3.1 KO [140]			Motor deficit	Reduced body weight
	Kv3.3 KO [141]			Ataxia [142]
	Kv3.1/Kv3.3 double KO [141]			Motor deficit and ataxia Increased activity and lower thigmotaxis	Reduced body weight Tremor and myoclonus
SCA14 (PKCγ)	PKCγ H101Y [143]	PC shrinkage and loss.		Clasping
	S361G-PKCγ [144]	PC shrinkage, loss and arbor atrophy		Motor deficit
SCA15 (ITPR1)	ITPR1-null mice [145]		General brain atrophy	Ataxia	Reduced weight, seizures, early death
	Opt mice [146]			Impaired locomotion	Reduced weight, epileptic seizures
	Itpr1-δ18/δ18 [147]			Motor deficit
SCA17 (TBP)	TBP-71Q [148]	Cerebellar gliosis PC loss and arbor atrophy		Motor deficit and clasping	Abnormal body weight, kyphosis [149] Tremor, seizures, premature death
	TBP-105Q [148]	Loss of granular neurons and PC Cerebellar gliosis	NP in cerebral cortex	Motor deficits and clasping	Reduced body weight, kyphosis Tremor, seizures, premature death
	L7-hTBP [150]	Cerebellar neuronal loss and gliosis	NL in brainstem, thalamus and striatum	Motor deficit, ataxia and clasping
	Nestin- TBP cond. KI [151]	PC loss		Motor deficit and hypoactivity	Low weight, muscle wasting, kyphosis
	Inducible TBP KI [152]	PC loss		Motor deficit	Reduced body weight. kyphosis, early death
	Germline-TBP KI [153]	PC loss		Motor deficit	Muscle wasting, low weight, early death
SCA23 (PDYN)	PDYNR212W mice [154]	PC shrinkage, loss and arbor atrophy		Ataxia
SCA27 (FGF14)	Fgf14-deficient mice [155]			Motor deficit, ataxia, clasping, hypoactivity Spatial memory deficit [156]	Hyperkinetic movements, tremor, seizures Reduced grip strength
SCA28 (AFGF3L2)	Afg3l2 [+/Emv66] [157, 158]	PC and granular neuronal loss		Motor deficit, ataxia and clasping	Pelvic elevation
SCA42 (CACNA1G)	Cacna1g KI [159]	PC loss and arbor atrophy		Ataxia

Abbreviations: PC= Purkinje cells; CN= cerebellar nuclei; SN= substantia nigra; NL/NP= neuronal loss/pathology. If specific reference is not provided, the basic reference for the mouse strain is applicable - see the Mouse model (type) field.

Most of the mouse models listed in Table 2 display cerebellar dysfunctions and related motor impairments, corresponding with the fact that these impairments constitute the core hallmarks of the SCAs. An exception, heterozygous SCA11 mice (TTBK2-KI mice [165]) do not show any impairments (in contrast to heterozygous human patients), while homozygous mice die in the embryonal stage. Some models of SCA, e.g. fibroblast growth factor (FGF) [155] and Kcn3.3 [141] deficient mice, tend to display robust motor impairments but no structural alternation of the cerebellum. In these mice, a detailed electrophysiological examination revealed abnormalities in Purkinje cells firing frequency or altered excitability of the excitatory synapses between parallel fibers and the Purkinje cells [166], suggesting that cerebellar synaptic deficits are sufficient to induce the motor deficits even in absence of gross neuroanatomical pathology. Similarly, the mouse model of SCA10 with expanded ATTCT repeat within LacZ gene [13] showed locomotor impairments and abnormal gait along with neuronal loss in several regions (pons, cerebral cortex, and hippocampus) but no alterations in the cerebellum. This also contrasts with the situation in human patients, which display severe cerebellar degeneration [167]. Matsuura and colleagues [167] attributed this contrast to the different expression pattern of the mutant gene in the mouse model (driven by PrP promoter) vs. in human patients (driven by native ATXN10 promoter). We also speculate that the focus on neuronal loss may be incomplete as loss of cerebellar neurons often occurs in late disease stages after the onset of atrophy in Purkinje cell dendrites. In some SCA models [24], there are also incongruences between the documented onsets of motor dysfunction, possibly explainable by the different settings of the tests, as well as inherent instability of expanded repeats.

In humans, premature death accompanies many SCA types [10], most often due to brainstem pathology and related impairments in swallowing, breathing and coughing, altogether resulting in pulmonary infection and respiratory failure [168]. Premature death often occurs in SCA mouse models although the causes seem to be more complex than in human patients and include malnutrition and dehydration due to disability of self-care. The fact that mice with purely cerebellar phenotype usually do not die prematurely suggests that the premature death in SCA mice stems from complex neuropathology, often including the brainstem (Table 2). However, specific brainstem-related impairments analogous to the human bulbar and respiratory symptoms have rarely been studied in mouse models. An exception are recent studies in mouse knock-in models of SCA1 [115] and SCA7 [138] reporting respiratory dysfunction correlating with pathology in hypoglossal neurons and nerves in these mouse models.

Many SCA patients also suffer from psychiatric issues, including cognitive deficits and impaired emotional regulation [169, 170]. These psychiatric aspects have been increasingly recognized as the factor significantly contributing to reduced life quality in SCA patients and affecting overall health condition [171]. However, analogous deficits have been studied sparsely in mouse models. The exception represents mouse models of SCA1, where the cognition has been studied repeatedly [24, 114, 118] and has been compared across different SCA1 models [113]. Particularly Atxn1^154Q/2Q mice suffer from a robust cognitive deficit from an early age (~8 weeks of age). The deficits include worse spatial learning in Morris water maze and Barnes maze, impaired contextual conditioning, reduced prepulse inhibition and impaired learning and cognitive flexibility in water T-maze test [24, 113, 114, 118]. SCA1 model with expression limited to the cerebellum shows fewer impairments, suggesting the cerebellar pathology contributes to these deficits partially [113]. Besides the cerebellar pathology, Atxn1^154Q/2Q knock-in mice display pathology in other brain regions known to be related to cognition, including the hippocampus [114, 116, 118, 172, 173] and cerebral cortex [174]. The hippocampal and cortical pathology occur early and precede cerebellar degeneration [114, 174]. Some of these impairments, e.g. impaired hippocampal neurogenesis, occur even in Atxn1 knock-out mice [116], suggesting that cognitive dysfunction may be caused by loss of ATXN1 function. Beside the SCA1, cognitive dysfunction occurs, along with hippocampal and prefrontal-cortical synaptic dysfunctions [175], also in a mouse model of SCA27 [156], mirroring severe psychiatric issues in SCA27 patients [175]. As cognitive deficits are relatively common in many SCAs [176], we expect that cognitive deficits might be more widespread in SCA mouse models. However, low focus on these dysfunctions and their difficult dissociation from other deficits in the mouse models might result in their low detectability [106].

Many mouse SCA models exhibit reduced activity and locomotion, usually assessed as decreased distance walked in the open field. Although reduced locomotor activity may reflect fatigue or lower motivation to explore the arena, it is also significantly confounded with ataxia and its presence is thus not surprising in SCA mice. More interestingly, few SCA models show opposite tendencies, such as hyperactivity [141], increased explorative behavior [112] or decreased thigmotaxis [141]. Since these abnormalities resemble impulsivity and hyperactivity seen in cerebellar patients [177–179] and are similar to the phenotype of other mice with cerebellar-specific degeneration [180, 181], we speculate that they may be partially attributed to cerebellar-related behavioral disinhibition.

Many SCA patients suffer from anxiety and depression [171] but analogous phenotypes have not been reported until recent studies. One of the studies documented increased anxiety- and depressive-like phenotype in Atxn1^154Q/2Q mice, partially correlating with hippocampal atrophy [114]. Another study replicated the anxiety-like behavior but contradicted the depressive-like phenotype in the same mouse model [182]. Interestingly, the pcp2-specific transgenic SCA1 mice showed reduced anxiety-like behavior in the study of Asher et al. [182]. It suggests that the anxiety-like tendencies may be due to the non-cerebellar dysfunctions whereas cerebellar dysfunction alone may act in a rather opposite direction. However, as the cerebellum affects emotional regulation [183] and cerebellar inflammation may induce depressive-like behavior in mice [184], we speculate that the emotional-related impairments may be more common in the SCA mouse models and the partial cerebellar contribution to these abnormalities still could not be fully excluded.

In human SCAs caused by expanded triplet repeats, disease severity, onset and progression generally correlate with number of pathological triplet repetitions. This phenomenon can also be seen in SCA mice if comparing two models with different number of repetitions [113]. Mouse models also show that homozygotes usually have earlier symptom onset and more rapid disease progress than heterozygotes [25, 110]. In humans, homozygous combination usually does not appear since the diseases have relatively low population frequencies. Homozygous mice can be used to assess the effect of gene dose if compared with heterozygotes and provide a model of more aggressive SCA pathology, though it does not correspond with a situation seen in humans.

7. SCA therapy in pre-clinical studies, clinical trials and translation research (Michael Strupp and Mario Manto)

Although therapy of SCs is problematic, there are several approaches that are used or tested in human patients. It is important to note that pre-clinical studies and clinical trials should follow methodological standards and are often connected with methodological difficulties, for instance, because we are dealing with rare and ultra-rare diseases.

7.1. Types and aims of clinical trials

There are basically three types of clinical trials with different primary aims:

1. Symptomatic therapy to improve ataxia and other associated symptoms and signs

2. Disease modifying therapy to delay the progression of the disease

3. Combination of both effects, which would be ideal.

7.2. How to design a clinical trial in cerebellar ataxias

The advantage of clinical trials in SCAs is that the diagnosis can be precisely made by genetic testing and that the same mutations can be induced in animals to obtain adequate models for pre-clinical phase. Nevertheless, in patients there is often (1) a high heterogeneity of the phenotype in terms of symptoms, (2) the individual condition of the patients depends on the time of manifestation (from early post-natal to late onset types with the same mutation), and (3) individual progression of the disease may be variable, even within one family. This has several practical implications for the design of clinical trials. For instance, patients who have almost no symptoms and those who are very severely impaired are often not included. In the first group, one would not expect to see any benefit. In the latter, patients are often unable to participate because they cannot perform the tasks anymore. These limitations prevent drawing conclusions of an intervention on these two subgroups [185]. In animal models, use of individuals in advanced disease stages is problematic for the same reason like in humans. Experimental treatment in animals in pre-symptomatic stage is, on the other hand, reasonable since it can delay or prevent disease symptom onset that would be expected within few weeks (faster disease progress in animals) in untreated control group.

Second, the rate of natural progression of symptoms and signs often vary inter-individually in patients with the same genotype. This high variability theoretically implies a high number of patients to be included to have sufficient statistical power. On the other hand, SCAs are rare or ultra-rare diseases. In this context, another important aspect is sample-size calculation because there is evidently so far no general agreement what a “clinically meaningful difference” is. Therefore, sample-size calculations are often not correct and not precise. In animal models of SCAs kept in standardized laboratory conditions, the phenotype is more homogenous and relatively small samples can show significant effects.

7.3. Primary and secondary endpoints for symptomatic therapy and disease modifying therapy

The aim is to use endpoints, which are functionally relevant, clinically meaningful and sensitive to detect changes for symptomatic and/or disease-modifying effects. These endpoints evidently differ from endpoints used in pre-clinical studies with animal models. In addition, one has to be aware that in most clinical trials so far only scales have been applied which are only validated for long-term studies. This means they have been used to evaluate the natural progression of the disease and may therefore not be useful or sensible to detect short-term symptomatic effects. The methodological limitations are also reflected by the high number of scores, scales and functional tests available and recommended [186–191]. For a summary of their concrete application in clinical studies, strengths, and weaknesses, see [4, 185, 192, 193].

7.4. Aim of the study and observational periods

For the evaluation of symptomatic effects, a short-term treatment period of 6 to 8 weeks is usually used in humans (e.g., [194, 195]. The duration can be estimated from animal studies and “individual-case of off-label-use” if the drug is approved for other indications (e.g., [196, 197]. For the evaluation of disease-modifying effects, a treatment period of at least 12 months is recommended (also by the EMA). At the end of this treatment period, there has to be a wash-out period of the drug to discriminate symptomatic from neuroprotective effects. In animal models, these principles are the same, just the time of treatment is usually shorter respecting animal’s life span and time-course of the neurodegeneration in each particular model.

7.5. Transfer from animal models to clinical trials: Pharmacokinetics, toxicology, safety, dosing, way of drug administration, duration of treatment and re-purposing of drugs

The transfer from an agent, which shows efficacy in an animal model to a clinical trial, is very demanding: according to the regulators it requires for a non-approved drug pharmacokinetic, toxicological, cancerogenic and teratogenetic studies in at least two animal species before it can be tested in a phase I trial. Drug dosing can be much higher in animal per kg and day than could be used in humans. A transfer of the daily dosage from animal studies to clinical trials can often not directly be made because of the different pharmacokinetics between different species due to different absorption, trans-membraneous transporters, metabolism, target structures, interactions and toxicology. Further, one has to find an appropriate way of application of the drug to the patient (for instance, intrathecal or intraperitoneal applications, which may work in animal models, are not ideal for patients).

A methodologically much easier way is “re-purposing of drugs” which has become important, powerful and successful: based on the mode of action and the approved use and safety in other diseases, its effects in ataxias can be much easier examined in clinical trials.

7.6. Examples of therapies used or tested in human patients and related animal model studies

Classical component of cerebellar ataxia management in humans is physical activity and rehabilitation [198, 199]. Animal model studies provided experimental evidence of physical activity efficiency, showed potential mechanisms and particular effects on the molecular, morphological or functional levels. In a mouse model of SCA1, treadmill training enhanced markers of neuronal plasticity, improved motor coordination and preserved Purkinje cells via mechanisms associated with induction of rpS6 phosphorylation [200]. Physical activity can also increase life span of SCA1 mice [88]. On the other hand, only mild effects of forced activity or enriched environment were observed in cerebellar mutant mouse Lurcher [201].

Riluzole (a benzothiazole drug, market approved for amyotrophic lateral sclerosis) was found to be effective in improving function in patients with hereditary cerebellar ataxia (SCA or Friedreich’s ataxia) as measured by the SARA score [188] in a multi-center, double-blind placebo-controlled trial with treatment duration of 12 months [202]. Riluzole decreased SARA by only one point (on a 40-point scale) in 50% of the treated patients versus 11% in the placebo group [202]. A study in SCA1 mice showed that acute (2 days) riluzole administration had no effect on motor performance [203]. Furthermore, long-lasting therapy did not improved motor and behavioral symptoms and even promoted loss of calbindin positivity of Purkinje cells in SCA3 mice [204]. Thus, the results are rather controversial and the drug is not widely used in human patients.

In a double-blind, randomized, placebo-controlled crossover trial, varenicline (a drug already approved for another indication) improved axial symptoms and rapid alternating movements in 20 SCA3 patients at 8 weeks and was fairly well tolerated, although a high placebo dropout rate led to trial termination after the first period [205]. The methodology of this study was criticized and these findings in terms of safety and efficacy could not be confirmed by others [206, 207] so that there is no convincing evidence for its efficacy [208]. In rats with 3-acetylpyridine-induced cerebellar ataxia, one week of varenicline administration improved motor performance [209] but data from specific SCA mouse models are lacking. Again, few groups administer the medication in routine ataxiology and some doubts and controversies remain [208].

Rovatirelin, a synthetic thyrotropin-releasing hormone analog, had no significant effect in patients with cerebellar ataxia (including SCA6 and SCA31) in two multicenter, parallel-group, double-blind, randomized, placebo-controlled phase 3 trials [210, 211] However, when tested selectively only in patients with a more severe baseline ataxia, rovatirelin was superior over placebo-treated group [212]. A positive effect of rovatirelin and taltirelin, another thyrotropin-releasing hormone analog, was found in rolling mouse Nagoya [213, 214].

Lithium is supposed to have a neuroprotective effect [215] and influences metabolic pathways in the cerebellum of wild type and SCA1 mice [216]. However, pre-clinical and clinical studies showed controversial results. A single-center randomized, placebo-controlled clinical trial in SCA3 patients showed no effect of lithium carbonate treatment on disease progression after 48 weeks [217]. In SCA3 mice, chronic lithium administration did not have beneficial effects on motor performance [218]. Furthermore, combination of lithium with another autophagy inducer had a neurotoxic effect in both wild type and SCA3 mice [219]. On the other hand, lithium carbonate supplementation attenuated motor symptoms and reduction of dendritic branching in hippocampal pyramidal neurons in SCA1 mice [118].

Baclofen is used to manage muscle tone and spastic disorders in some neurological diseases symptoms of which include ataxia [220]. Studies in mouse models of SCAs showed that baclofen alone or in combination with chlorzoxazone could improve motor performance, modulate dendritic dysfunction and abnormal firing pattern of Purkinje cells [30, 221–223].

Overall, re-purposing of drugs is a promising way to initiate clinical trials within a reasonable period of time and lower costs. This implies for experiments using animal models that such drugs are good candidates for pre-clinical research, which could have an immediate impact on patients. Nevertheless, in many of the tested substances sufficient effect has not been confirmed and none of the molecules tested in SCAs so far was capable to significantly neurodegenerative process or restore cerebellar function and the results of pre-clinical and clinical studies are often controversial (for review see [224]). Thus, searching for completely new therapeutic approaches is still necessary (see parts 8 and 9).

8. Experimental therapy based on specific gene and pathogenic mechanisms (Harry T. Orr, Marija Cvetanovic)

Animal models, notably SCA mouse models, have been instrumental in identifying critical steps in pathogenesis with potential of being therapeutic targets. Conditional mouse models, in which expression of mutant proteins can be turned off, demonstrated that reducing production of the pathogenic proteins is a promising approach for treating the SCAs. Studies revealed that the earlier in the process of disease that expression of the mutant protein was blocked, the greater the therapeutic benefit. When expression of mutant ATXN1 and ATXN3 were stopped at an early stage of disease, Purkinje cell pathology and motor dysfunction were completely reversible [23, 128]. Yet, halting expression of mutant protein even at later stages of disease led to a partial recovery. SCA animal models studies using siRNA or antisense oligonucleotides targeting expression of SCA-causing genes, the first step in pathogenesis, verify that nucleotide-based gene-silencing approaches are very promising therapeutic means for treating SCAs [20, 21, 225–229].

Another potential therapeutic target revealed by SCA model genetic studies is the interaction of SCA mutant proteins with other cellular proteins where the interaction is critical for pathogenesis [230]. For instance, ATXN1’s interaction with a transcriptional repressor Capicua (CIC) is the primary driver of SCA1 pathogenesis in Purkinje cells. This was indicated by ameliorated disease phenotype of SCA1 transgenic mice that express polyQ ATXN1 with mutation in the CIC binding site [231]. Reduction of Staufen1, an ATXN2 interactor ameliorated disease pathogenesis in a SCA2 mouse model [15].

Genetic studies also indicated therapeutic potential of targeting pathways deregulated in SCAs. Deregulation of brain cholesterol metabolism is associated with several neurodegenerative diseases, including SCA3. Expression of cytochrome P450 family 46 subfamily A member 1 (CYP46A1), the key enzyme involved in the efflux of brain cholesterol is decreased in SCA3 mice. Moreover, rescuing CYP46A1 expression improved disease phenotype in SCA3 mice and was also effective in reducing pathology caused by ATXN3 [232]. Intriguingly, both Staufen1 and CYP46A1 are linked to impairments in autophagy, implicating deficient autophagy as an important pathogenesis target in SCAs [233, 234].

Growing evidence demonstrates that a hallmark of SCA pathology is cerebellar Purkinje cell dysfunction and consequent dysfunction of associated downstream circuitry [235–239]. The substantial dendritic arbors of Purkinje cells receive extensive excitatory input from parallel and climbing fiber synapses. Disruption of the input from these synapses, critical for an appropriate inhibitory Purkinje cell signaling to the deep cerebellar nuclei, is found in mouse models of the SCAs [237]. Mutations in the gene KCNC3, encoding the voltage-gated potassium channels Kv3.3 and Kv4.3, cause SCA13 and SCA19/22, respectively [240–242]. SCA13 can manifest with onset either in infancy or adulthood. Recent evidence using a SCA13 zebrafish model showed that infant onset mutations impacted Purkinje cell excitability differently from Purkinje cells expressing adult-onset mutation [243]. In Purkinje cells expressing an infant onset KCNC3 mutation excitability was dramatically altered with Purkinje cell degeneration during development. In contrast, Purkinje cells expressing an adult onset KCNC3 mutation resulted in latent hypoexcitability and normal development. Additionally, in mouse models of polyglutamine SCAs genetic studies identified a critical role of ion-channel dysfunction in pathogenesis. In a mouse model of SCA1, rescuing reduced expression of potassium channels BK and the G-protein coupled inwardly-rectifying potassium channel ameliorated Purkinje cell dysfunction and disease pathogenesis [244]. In mouse model of SCA2 reducing calcium inositol-phosphate mediated signaling rescued Purkinje cell dysfunction and motor deficits [120, 245, 246].

Consistent with a high metabolic activity and the corresponding high consumption of ATP by Purkinje cells, mouse genetic studies demonstrate that impairments in mitochondrial function contribute to SCA pathogenesis. Application of the mitochondria-targeted antioxidant MitoQ or succinic acid, a complex II (succinate dehydrogenase) electron donor, ameliorated SCA1 cerebellar mitochondrial dysfunction, pathology, and motor deficits [247, 248]. Mitochondrial dysfunction was also described in other SCAs, including SCA3 [249], and SCA7 where recent evidence supports its important role in disease pathogenesis [250].

Finally, contribution of astrocytes and microglia in pathogenesis of SCAs were revealed using mouse genetics. This is an important factor that could become one of the targets of new SCA therapy. It is intriguing that in physiological conditions cerebellar astrocyte and microglia have unique pro-inflammatory transcriptional properties compared to glia in other brain regions [251, 252]. An initial indication that astrocytes may contribute to SCA pathogenesis came from a mouse genetic study in which expression of mutant SCA7 only in Bergmann glia, a type of cerebellar astrocyte, was sufficient to induce Purkinje cell degeneration and motor deficits [136, 253]. Similarly, transient activation of pro-inflammatory NF-κB/IKK2 only in astrocytes was sufficient to induce loss of Purkinje cells [254]. In mouse models of SCA1, SCA6 and SCA21, glial activation precedes motor dysfunction implicating that glia may actively contribute to the pathogenesis [40, 111, 255]. Selective genetic reduction of astroglial NF-κB signaling indicated a biphasic role of astrogliosis in SCA1 pathogenesis: beneficial early but harmful later in disease [256]. These results imply that anti-inflammatory astroglia-based therapeutic approaches may need to consider disease progression to achieve therapeutic efficacy [256]. On the other hand, microglial inflammation was harmful in SCA1 mice [257]. The extent to which glial disease responses are due to a direct effect of glial expression of SCA causing genes versus an indirect response to neuronal pathology remains to be determined.

9. Neurotransplantation therapy in SCA models (Jan Cendelin)

Neurotransplantation represents another potential therapeutic approach for cerebellar degenerations [258]. However, there are only few clinical studies on human patients [259, 260] and long-lasting efficiency and safety for routine clinical use has not been confirmed yet. Thus, further investigation on animal models is needed. Cerebellar transplantation research was performed first on spontaneous mutant mice that are not always models of particular human diseases [261–267]. Later, genetically engineered mouse models of human SCAs have been used in few studies investigating neurotransplantation [268–272].

Neurotransplantation would be important particularly for replacement of lost neurons and thereby recovery of cerebellar reserve [2, 273]. The question, indeed, is whether such goal requiring adequate differentiation, long-term survival and precise functional integration of grafted cells can be routinely achieved. Nevertheless, there are more potential mechanisms of the graft’s therapeutic effect (for review see [258]).

In one of the classical studies providing evidence that cerebellar transplantation could alleviate symptoms of SCA, Kaemmerer and Low described that the embryonic cerebellar grafts survived in three quarters of engrafted SCA1 mice and these animals performed better on a rotarod, had narrower gait and were more mobile than sham-operated controls [268]. The effect persisted for several weeks but then a progressive decline of cerebellar functions followed [268].

The most commonly used type of graft in SCA mice are mesenchymal stem cells. Although they usually do not transdifferentiate into neurons, they have been shown to alleviate neurologic symptoms of SCAs [269, 270, 274]. Their effect is rather an enhancement of residual cerebellar tissue plasticity, rescue of degenerating Purkinje cells or delay of their death and reduction of their anomalies [269, 270, 274]. Mechanisms that are likely to mediate these effects include production of neurotrophic factors by the engrafted cells [274, 275] and fusion with the host’s Purkinje cells [276, 277]. Delay of Purkinje cell death has been observed also after neural precursor transplantation in SCA1 mice [278]. Interestingly, Chang et al. [269] reported that intravenous but not intracerebellar (efficient in other studies) injection of human mesenchymal stem cells delayed the onset of motor function impairment in SCA2 mice.

It has been shown that subventricular zone-derived neural precursor cells injected into the cerebellar white matter of SCA1 transgenic mice migrated into the cerebellar cortex only in individuals with already advanced Purkinje cell loss suggesting that diseased tissue might provide signals attracting grafted cells [278]. In these animals, transplantation led to enhancement of motor functions although grafted cells did not differentiate into the Purkinje cells [278]. Similarly, Purkinje cell fusion with grafted mesenchymal stem cells was more frequent in symptomatic SCA1 mice than in asymptomatic individuals [277]. It seems that SCA-related neuropathology can in some cases stimulate phenomena involved in graft effect mechanisms.

Although there are some contradictions in the neurotransplantation studies employing mouse models of SCAs summarized in Table 3 most of them are in principle consistent in reporting positive effects of transplantation. On the other hand, in spontaneous (not genetically engineered) cerebellar mutants (Lurcher, Purkinje cell degeneration, Tambaleante mice), the results of transplantation studies are more variable. While some studies showed improvement after transplantation [265, 266, 275], others report no or very limited functional improvement after transplantation [280–283]. Furthermore, diseased cerebellar cortex does not seem to provide always a permissive niche for grafted cells, e.g. in cerebellar mutant mouse Lurcher [282–284] while in other spontaneous cerebellar mutants lacking infiltration of the host cerebellum by the graft was not observed [281, 283]. In SCA mice, cerebellar degeneration has a relatively late onset, slow progression and Purkinje cell dysfunctions often precede for a longer period before cell death [94, 110, 121]. On the other hand, in Lurcher or Purkinje cell degeneration mice, the most frequently used spontaneous mutants, Purkinje cell degeneration is rapid, complete and the cell pathology is quite aggressive [285–288]. In these mutants, therapy might require different approach such as early transplantation in newborn age as used by Jones et al. [275].

Table 3: Mouse models of SCAs used for investigation of neurotransplantation therapy –

graft type used, grafting site, and observed effect of the treatment.

SCA model	Graft	Grafting site	Effect	References
SCA1	Embryonic cerebellar cells	Cerebellar nuclei	Reduction of motor impairment	[268]
	Neural precursors	Cerebellar white matter	Purkinje cell rescue Reduction of motor impairment	[278]
	Mesenchymal stem cells	Cerebellar cortex	Fusion with degenerating cell Neural function not tested	[277]
		Intrathecal injection over the superior colliculus	Mitigation of cerebellar disorganization Suppression of Purkinje cell dendrite atrophy Reduction of motor impairment	[270]
SCA2	Mesenchymal stem cells	Intravenous injection	Differentiation of grafted cells into neurons, astrocytes and oligodendrocytes Purkinje cell rescue Reduction of motor impairment Reduction of mutant ATXN3 aggregates	[269]
		Cerebellum	Mild Purkinje cell rescue No motor improvement	[269]
SCA3	Cerebellar neural stem cells	Cerebellum	Purkinje cell rescue Reduction of motor impairment	[271]
	Mesenchymal stem cells	repetitive systemic administration	Purkinje cell rescue Reduction of motor impairment	[274]
	Human olfactory ensheathing cells	Dorsal raphe nucleus	Purkinje cell rescue Reduction of motor impairment Reduction of cerebellar inflammation	[279]

The problem is that neurotransplantation therapy has been investigated only in few models of SCAs, mainly those based on polyglutamine tract expansions. Therefore, it is still not clear if cerebellar transplantation would be effective in other SCA types and heterogeneity of results similar to that in spontaneous mutants cannot be ruled out without any doubts prior investigating more SCA models.

Our recent knowledge of cerebellar neurotransplantation is not systematic. Existing studies used models of different diseases, different types of grafts and graft administration techniques, examined not always using the same morphological, biochemical or functional parameters. That is why it is difficult to compare results of individual studies and make general conclusions.

It would be necessary to test neurotransplantation in models of most SCAs because heterogeneity of these diseases could lead to different outcomes of the therapy. Furthermore, comparative studies under identical experimental conditions would be of importance. Finding differences in graft development and functional integration in animals with different cerebellar neuropathology, degeneration progress and extent might help to identify factors that are responsible for good or poor success of neurotransplantation therapy (e.g. specific local tissue niche component, character of neural circuitry damage).

The task for future research is to answer the question whether neurotransplantation, as an invasive and risky method, 1) could be a universal therapeutic approach for SCAs, or 2) would have acceptable risk/benefit ratio only in some of the SCAs, and/or 3) only under specific conditions (e.g. stage or form of the disease), or 4) would be even not reasonable at all. In addition, identifying the optimum graft type for particular diseases and their stages would be necessary. Respecting pathogenesis of the disease, its onset and progress speed, treatment of different diseases could require employment of different mechanisms of the graft effects provided by individual types of grafts in a different extent (for review see [258]). Plans for SCA treatment by neurotransplantation should also reflect degeneration of extracerebellar structures that contribute to complex neurological phenotype seen in the patients or mouse models and that varies not only between individual SCA types but also in patients depending on disease stage (for review see [1]. Number of existing mouse models of SCAs would allow such research direction, though it would be a long-lasting process.

10. Discussion and conclusions

Studies employing animal models of SCAs have provided large amount of information on the pathogenesis of SCAs and other cerebellar degenerations, generated promising results in experimental therapy but also helped to highlight limitations, unclear issues or ambiguous effects of experimental approaches. Furthermore, back-translation research – based on positive clinical findings by using drugs approved for other indications - is an elegant way to elucidate the mechanisms of action of the treatment and then to even modify the pharmacological agent to further improve it. Animal models are essential and irreplaceable tools to investigate cerebellar degenerative diseases complementary to clinical studies and observation and in vitro studies.

In vitro experiments using cells caring SCA determining mutant alleles have different features, applications, advantages, goals and limitations than animal studies. SCA human patient-derived induced pluripotent stem cells (iPSCs) allow generation of various cell types including neurons that can be used to investigate cellular level of SCA pathogenesis [289–291]. Human cell cultures are important for validation of biochemical, molecular, and cellular findings from animal models before translation to human organism as whole. Cell cultures do not include whole complexity of animal or human organism but can provide at some level defined experimental conditions with specific local niche and eliminate factors and influences by the whole organism allowing specific focusing on certain processes.

Animal-based studies have special importance since they permit (maintaining in vivo conditions and complex whole organism point of view): 1. Performing experiments that would be risky and ethically unacceptable in humans and 2. Performing mechanistic/causal studies under specific and well-defined experimental conditions that are not possible in human patients. Compared to clinical studies and observations, such experimental studies allow more direct interpretation thanks to their goal-directed design and (partial) elimination of (some) disruptive factors. Although verification of the findings in human patients is necessary and translation to humans will have limitations, data obtained from animal model studies allow more targeted designing and reduction of risks of subsequent clinical studies (e.g. via identifying potential negative side effects of therapies under development and eliminating therapies showing no or poor benefits before testing them in humans).

When planning an animal-based study, selection of an appropriate model (or models) is crucial. One must be aware of the advantages and limitations of a model in regards to the aim and methodology of the experiments. Since for these purposes, different models (types of genetic modification – see part 3) or different organisms (vertebrates, invertebrates, isolated organs or tissues – see part 4) might be the choice, one should consider following questions:

1) Do we need a complex SCA model with high similarity to human disease with respect to complexity of neuropathology and pathogenesis of a particular SCA type? This is relevant for most of the studies testing complex therapeutic effects (“does the therapy improve overall health state, functioning and quality of life and is therefore clinically meaningful or does it just modify a laboratory parameter or a biomarker of the disease?”).

2) Are we focused on a specific component of the disease or certain level of the pathological process (organ, cell, metabolic pathway, signaling pathway)?

Animals carrying human mutant allele causing one of the SCAs serve as models of specific disease. Nevertheless, not only the specific SCA-related gene mutation but also its expression level, spatio-temporal pattern of expression in various cell populations determine the pathological phenotype. To establish a precise model of a disease, adequate expression levels of the pathological protein in adequate cell types are required. If the mutant protein is produced only in certain brain structure in the animal model, while in humans its expression is wider, the model does not correspond to the human disease in all its features. It could be, however, used for studies focused only on neuropathology of a specific structure and to filter effects of more complex CNS dysfunction. Such limitation of neuropathology should, indeed, be taken into consideration when planning the studies and interpreting their results. Furthermore, the effect of pathological alleles can be modified by other genes present in the genetic background of the animal. Via this factor, species and strain of the model organism can play a role. From this point of view, organisms that are closer to humans have a more direct translation of the findings. Detailed genetic characterization of the model organism taking into consideration not only the gene of interest and its mutation but also control of its expression and genetic background as well as potential epigenetic phenomena is of importance. Due to potential genetic instability (e.g. changing number of repetitions in CAG repeats) genetic characterization of the models should be repeated from time to time to detect changes in the gene of interest or its expression when maintaining the animal model for more generations. Furthermore, detailed phenotypic characterization on morphological as well as functional level is needed to verify the model. Comparing two different models having different genetic background but carrying the same human pathological disease-determining allele can be used in genetic studies to identify disease-modifying genes or their specific alleles. Respective gene products (proteins) could be of interest for potential therapy development. For this purpose, transfer of the pathological allele on the background of several consomic strains is used. Suppression of pathological phenotype in one of the consomic strains allows identification of a chromosome on which the disease-modifying gene is located.

Species of the model organism is also important for similarities or dissimilarities of their CNS and cerebellum with humans and also similarities and dissimilarities of behavioral patterns, cognitive functions or mode of locomotion and its neural control. It determines whether given organism can be used (always with some level of limitation) to study particular neuropathology in specific neural structures and changes of complex neural functions due to SCA. As discussed above, in many mouse models of SCAs, analogy of neuropathological changes and motor, behavioral, and cognitive impairments to human patients have been documented.

Complex neuropathology affecting several CNS structures leads to a complex pathological behavioral phenotype. For compatibility of the model and the human disease and translation of the research, similarity of pathogenesis, neuropathology and functional deficits is important. On the other hand, it could complicate interpretation of the observations. Cerebellar mutant mice, rather spontaneous mutants, are also used to describe functional deficits related to cerebellar degeneration and thereby to identify cerebellar functions (the function that is impaired in a cerebellar mutant should have something to do with the cerebellum). However, it is not so easy and there is always a question: could observed behavioral changes be attributed directly to cerebellar degeneration or is it a product of pathology elsewhere? Furthermore, the cerebellum is well known to be involved in many functions, its damage alone has a wide spectrum of motor as well as non-motor symptoms [8, 170, 292–296] and there are interactions between motor and sensory deficits and emotional changes [297]. It is not surprising that results of behavioral tests in laboratory rodents depend often on several factors and it is usually difficult to attribute poor performance in a test purely to one function or one neural structure disorder not only in animal models of cerebellar degenerations [181] but also in human cerebellar patients. For instance, performance in the Morris water maze, a classical spatial learning and orientation test, depends not only on spatial learning and orientation abilities itself but also on motor performance (swimming speed, direction maintenance), vision (to detect landmarks) and stress and anxiety levels [180, 181, 298] providing motivation to escape but also triggering maladaptive phenomena and behaviors such as behavioral disinhibition modifying animal’s behavior in various tests [298, 299]. All these factors can be influenced by dysfunction of the cerebellum (directly or indirectly) as well as of other neural structures. To clarify this issue, specific models having restricted and well-defined neuropathology would be helpful.

11. Perspectives and future directions

Development of models for all known SCAs and other types of cerebellar degeneration are needed that show tight phenotypic and genotypic analogy to human diseases or allow highlighting specific disease features for their selective investigation. For specific research, double mutants, conditioned mutants and other advanced genetically modified organisms are necessary. Detailed phenotypic characterization of individual animal models on neuropathological, behavioral, cellular, and biochemical level is also needed for planning and interpretation of studies testing experimentally therapeutic approaches.

Comparative studies between several models of the disease and in models of several diseases should be done. They could highlight specific features of individual SCA types or their models or, on the other hand, identify some common features and generally applicable facts. This has specific importance for therapy development because we should know whether a therapeutic method or drug would be effective in more cerebellar degenerative diseases (or even in cerebellar diseases of other origins) or only in one specific disease. Knowledge of therapeutic efficiency in individual SCA types helps to assess risk/benefit ratio and to follow principles of personalized medicine. Testing in several different models (different background strains or mutants generated via different methods) would also help to avoid giving too much importance to phenomena that are strictly strain- or model-specific and have no wider validity. For clinical trials, we need clinically meaningful and validated endpoints to evaluate short-term symptomatic and long-term disease modifying effects to transfer the knowledge from animal models to our patients.

Gene-editing technologies such as CRISPR-Cas9, zinc finger nuclease and virus-mediated technologies should be included in experimental treatment research. Especially, CRISPR-Cas9 and the relevant improved technologies would be the promising therapeutic approaches, since successful deletion of polyglutamine in SCA3-derived iPSC has already been demonstrated [300].

Abbreviations:

AAV

adeno-associated virus

ADCA

autosomal dominant cerebellar ataxia

ALS

amyotrophic lateral sclerosis

ATXN

ataxin

BBB

blood-brain barrier

CIC

Capicua

CMA

chaperone-mediated autophagy

CMV

cytomegalovirus

CN

cerebellar nuclei

CNS

central nervous system

CRISPR

clustered regularly interspaced short palindromic repeats

DNMT1

DNA methyltransferase 1 mutation

Dox

doxycycline

DRPLA

dentatorubral-pallidoluysian atrophy

eIF2α

eukaryotic translation initiation factor 2 subunit 1

FGF

fibroblast growth factor

HSPs

hereditary spastic paraplegias

IRES

internal ribosomal entry site

KI

knock-in

LAMP2A

lysosome-associated protein 2A

LTD

long-term depression

LTP

long-term potentiation

miR-RORα

microRNA against RORα

MSCV

murine stem cell virus

NfL

neurofilament light chain

NL/NP

neuronal loss/pathology

NLS

nuclear localization signal

PC

Purkinje cell

RORα

retinoid-related orphan receptor α

ROS

reactive oxygen species

SAGA

Spt-Ada-GCN5-acetyltransferase

SCA

spinocerebellar ataxia

SN

substantia nigra

STAU

Staufen-1 protein

Tg

transgenic

TRE

Tetracycline response element