Degenerative Ataxia - also called Spinocerebellar Degeneration – represents a class of neurological diseases with progressive neurodegeneration primarily affecting neurons of the cerebellum. These diseases fall within the category of rare neurodegenerative diseases with an estimated prevalence of 15–20:100.000 [28], and encompasses genetic and sporadic types at a ratio of approximately 30–70%, respectively of all SCA cases according to a Japanese survey [29]. The most prominent clinical manifestation of degenerative ataxia is a compromised coordination of movements of the four limbs due to progressive cerebellar atrophy [3], most often caused by the loss of PCs, the most vulnerable neuronal population in the cerebellum [30, 31]. The patients further display additional major clinical signs attributed to cerebellar dysfunction, such as a poor coordination of eye movements and slurred speech [3]. Further neurological impairments including mental retardation and epilepsy accompany the main symptoms, which are specific for each SCA subtype. Most of the genetically caused degenerative ataxias are inherited in an autosomal manner. Autosomal dominant SCAs, typically named Spinocerebellar ataxias (SCAs) [32, 33], and autosomal ressesive SCAs, abbreviated SCARs [34, 35] appear with similar incidence rates (2.7:100,000 for SCA, and 3.3:100,000 for SCAR) [36]. The remaining small number of cases show X-linked dominant [37], or mitochondria-DNA inheritance [38].

Up to now, 39 genes that include 44 SCA loci, are registered in the Online Mendelian Inheritance in Men (OMIM) database [39]. These mutations are known to cause SCAs of which the encoding proteins fulfill diverse functions for example as transcriptional and translational regulators, ion-channels, kinases or phosphatases, protein homeostasis regulators or being involved in membrane trafficking [32, 33, 40]. Twelve out of these SCA-causing alleles carry expanded repeat sequences ranging from trinucleotide to pentanucleotide repeats found either in protein-coding or in non-coding sequences such as untranslated regions (UTR) or introns. Among them, CAG repeat expansions in protein coding sequences are translated to expanded polyglutamine (polyQ) stretches in the respective gene products, leading to the class of disorder known as polyQ disease which encompasses at least nine neurodegenerative diseases. These include Huntington’s disease, and spinal-bulbar muscular atrophy (SMBA), which selectively damage striatal neurons, and motoneurons, respectively, while the remaining seven polyQ diseases are referred to as polyQ-SCAs, characterized by cerebellar neurodegeneration among other pathologies [41]. PolyQ-SCAs includes SCA1, SCA2, SCA3/Machado-Joseph disease (MJD), SCA6, SCA7, SCA17, and Dentatorubral-pallidoluysian atrophy (DRPLA) [32, 33].

Neurodegenerative diseases are often accompanied by intracellular and/or extracellular deposits, such as Aβ42-plaques that accumulate in Alzheimer’s disese (AD) affected brains, Lewy-bodies associated with Parkinson’s disease (PD), and cytoplasmic aggregates involving TDP-43 in Amyotrophic Lateral Sclerosis (ALS) [42]. Similarly, nuclear and/or cytoplasmic inclusion bodies formed by polyQ proteins are a well-documented pathological hallmark manifested in degenerating neurons affected by polyQ diseases [30, 31]. As a specific feature of this class of disorder, polyQ diseases exhibit an inverse correlation between the age of onset and the number of expanded polyQ repeats, with a longer uninterrupted polyQ sequence resulting in an earlier age of onset and faster progression of disease symptoms [43]. Impariments in autophagy, a crucial system for maintaining neuronal protein quality, have been reported in animal models of cerebellar polyQ diseases, including SCA3/MJD, SCA6, SCA7, and DRPLA [44,45,46,47]. Among all neurons in the nervous system, cerebellar PCs are the most susceptible neuronal subpopulation affected by impaired autophagy processes, especially in juvenile mice [48, 49]. Thus, autophagy dysregulation impaired by polyQ proteins may affect PCs at younger stages, although the mechanisms underlying compromised autophagy differ among the different polyQ SCA diseases. Also, abundant expression of polyQ containing proteins in the cerebellum is considered critical for inducing cerebellar neurodegeneration pathologies, and at least for SCA6, the disease causing-gene is abundantly expressed in PCs [50]. These findings could explain the selective vulnerability of cerebellar neurons to polyQ-SCA proteins, as well as the broad temporal range of disease onset, which can initiate as early as childhood [51]. This contrasts with other neurodegenerative diseases that affect regions of the CNS other than the cerebellum, such as ALS showing a mid-to-late age onset, or late-onset degenerative disorders like AD and PD, where the initiation and progression of pathology are more closely associated with aging than in polyQ-SCAs.

On the other hand, expanded nucleotide repeats of CTG, ATTCT, CAG, TGGAA, GAA, GGCCTG, or ATTTC causing SCA8, SCA10, SCA12, SCA27B, SCA31, SCA36, SCA37, respectively, are found in non-coding sequence, and are thought to cause RNA toxicity [32, 33, 52]. Interestingly, the antisense strand of the CTG expansion sequence at the SCA8 locus encodes for CAG repeats, which are also transcribed in the opposite orientation by repeat-associated non-ATG translation [53].These repeats generate not only a polyQ protein product, but also polySer, and polyLys containing proteins. These poly amino acid stretches likely result in combinatorial toxicity together with the CTG expanded RNA [54]. The remaining SCAs such as SCA5, SCA13, or SCA28, are caused by alleles carrying missense mutations, insertions, or deletions [55,56,57,58], while genomic duplications were identidied in SCA20 and SCA39 loci [59, 60]. Moreover, ATX-THAP11 has been reported recently as a new polyQ disease caused by an extended polyQ stretch in the coding sequence of the thap gene [61].

The ideal animal models, representing the genetic conditions of human SCAs, are generated by replacing a wildtype allele with a disease causing allele with the help of targeted knock-in strategies, such as the Crispr/Cas9 technique [8]. This strategy allows for generating a toxic gain-of-function allele whose expression is spatio-temporally controlled by the endogenous regulatory elements and therefore such alleles mimic the human SCA causing gene expression pattern and strength more precisely. In addition, a knock-in approach simultaneously generates a haploinsufficiency (loss-of-function) of the wildtype allele, thus lowering the expression of the wildtype transcript and protein. However, such knock-in mutant mice sometimes displayed only a limited SCA pathology even during late stages of their lives due to the much shorter life span of mice compared to humans [62]. Indeed, artificially expanded CAG stretches in SCA mutant proteins, which were signficantly longer than stretches found in human patients, were needed to induce an apparent progressive SCA neuropathology in polyQ knock-in mice [63, 64]. Since the toxic gain-of-function SCA variants are thought to be the main cause of dominantly inherited SCA pathogenesis, mutant SCA allele overexpression in zebrafish - that can be readily established with the help of mRNA-injection or Tol2 transposon transgenesis [65] - have been reported for some of the SCAs so far (Fig. 2; Supplementary Table 1).

Schematic presentation of pathogenic transgenes to generate zebrafish models for respective SCAs. For simplicity, protein coding exons are depicted as a single box. Zebrafish modeling for SCA37 was achieved by injection of pathological RNA carrying (AUUUC)57, which was synthesized from the intronic fragment including the pentanucleotide (ATTTC)57 repeat in the 5′ untranslated region (UTR) of the dab1 pathogenic allele, while RNAs containing (AUUUU) repeats from wild type alleles encoding (ATTTT)7 or (ATTTT)139 were used as controls [66]. The other zebrafish models for SCA1, 3, 13, 49 [23, 24, 67,68,69,70,71,72,73] were generated either by DNA, or mRNA injection to express protein coding sequences carrying CAG repeat expansions (for SCA1 and SCA3), or missense mutations (for SCA13 and SCA49), respectively. While most models were only generated as transient transgenic zebrafish, stable transgenic models were generated for SCA1 and SCA3 by raising DNA construct-injected embryos to adult zebrafish, screening the next generation for germline transmission and breeding the transgene through the following generations

The first reported genetic model for human SCAs in zebrafish aimed at modelling the SCA3/MJD polyQ disease that is caused by an expanded CAG containing allele of ATAXIN-3 (ATXN3), which encodes for a deubiquitination enzyme. Injection of mutant ATXN3 mRNA containing a CAG80 repeat expansion encoding ATXN3[80Q] caused p53-dependent apoptosis that was mainly detected in the head of the zebrafish at 24hpf, possibly indicating neuronal degeneration [71]. To better illustrate a cell-autonomous neuropathology attributed to mutated ATXN3-polyQ expression in neurons at later larval and adult stages, a subsequent study described a stable transgenic line carrying an ATXN3[84Q] allele fused to EGFP that was expressed together with a cytoplasmic red fluorescent reporter protein [73]. Both transgenes were expressed under control of tandem repeats of upstream activating sequences (UAS), which serve as consensus binding site for the transcription factor Gal4VP16 (Gal4) [74] (Fig. 3A, B). When crossed with an elavl3:Gal4 zebrafish strain, expressing Gal4 in a pan-neuronal manner [75], double transgenic offspring carrying both Gal4 and UAS-transgenes induced ATXN3[84Q] expression together with the red fluorescent reporter in neurons throughout the entire nervous system (Fig. 3A) [73]. Compared to uninjected wildtype or ATXN3[23Q] expressing transgenic fish, such SCA3 zebrafish larvae displayed shortened spinal motor axons. With respect to behavior these SCA3 larvae swam shorter distances starting at 6dpf, a locomotor defect that lasted into adult stages. Furthermore, the average lifespan of these SCA3 zebrafish were reduced by 45 days compared to control fish expressing a wildtype ATXN3[23Q] allele [73]. When a Gal4 driver line miR218:Gal4 with a motoneuron specific miR218 enhancer [76] was used for breeding, ATXN3[84Q] expression in the offspring was restricted to motoneurons (Fig. 3B). These specimens exhibited a similar deficit in swimming as observed in zebrafish of the pan-neuronal SCA3 model (hereafter SCA3 model or SCA3 zebrafish) both at larval (6dpf) and adult (3 months) stages  [73]. Again, these motoneuron specific SCA3 zebrafish displayed shortened axons of spinal motoneurons, suggesting that the compromised movement observed in the SCA3 model is mainly caused by axonal dysfunction of motoneurons. Of note, one year old SCA3 zebrafish displayed accumulated polyQ protein as aggreates in neurites, especially in the medulla  [73], and such aggregates are a histological hallmark of polyQ-SCAs [30, 31, 77]. A subsequent study using flow cytometry revealed an increased number of fluorecent aggregates identified as detergent-insoluble ATXN3 particles in SCA3 larvae at 2dpf. These findings indicated an early initiation of polyQ protein related-pathology in SCA3 zebrafish even at early larval stages [78].

In addition, SCA3 zebrafish recapitulated the generation of toxic fragments of the ATXN3[84Q] protein generated by calpain-mediated proteolytic processing [73]. These proteolytic fragments are another pathological hallmark of SCA3 known from neurons derived from human SCA3 patients [79]. Of note, the impaired swimming of SCA3 zebrafish was restored when larvae were treated with the Calpain inhibitor Calpeptin, further supporting the impact of Calpain activation on SCA3-induced motor decificts. Noteworthy, SCA3 zebrafish treated with Calpeptin not only showed a decrease in ATXN3[84Q] cleavage fragments, but also a reduced level of full length ATXN3[84Q] protein expression [73]. More interestingly, Calpeptin treated SCA3 larvae displayed an enhanced autophagic activity [73], likely contributing to the removal of both full length and cleaved toxic fragment of ATXN3[84Q] in addition to the inhibition of Calpain-mediated ATXN3[84Q] proteolysis. Indeed, autophagy inhibition by active Calpain at multiple autophagy steps has been found in various pathological conditions [80], which also likely occur in zebrafish neurons affected by ATXN3[84Q]. Taken together, these results from the zebrafish SCA3 model strongly support the hypothesis that a toxic gain-of-function caused by the mutant ATXN3 protein, especially the appearance of toxic ATXN3 fragments generated by Calpain, is the central pathogenic event in SCA3-induced neuronal degeneration.

Schematic representation of cell type-specific expression cassettes for generating respective SCA (A–E) and acute cell ablation (F) models. A and B SCA3 zebrafish express Gal4VP16, or its variant KalTA4, in a pan-neuronal (A) or motoneuron specific (B) manner under the control of the elavl3 promoter or mir218 enhancer fused to the gata2 basal promoter. Translated Gal4VP16, or KalTA4 activates the expression of dsRed together with human SCA3[84Q] fused to EGFP from a bidirectional UAS responder unit [73]. C SCA1 zebrafish express humanATXN1[82Q] together with a membrane-targeted fluorescent protein mScarlet specifically in zebrafish cerebellar Purkinje cells (PCs) [23]. This transgene expression is mediated by a bidirectional expression construct using 8 copies in tandem of a PC specific regulatory element, cpce, flanked on both sides by E1b minimal promoters. To further restrict transgene expression to cerebellar PCs, a 4x miRNA181a target sequence (4xmiR181aT) was inserted between the coding sequence and the polyA (pA) sequence to eliminate ectopic transgene expression in tectal and hindbrain neurons. D The zebrafish kcnc3R335H and kcnc3R363H alleles, which are equivalent to the human SCA13 causing mutant alleles KCNC3R420H and KCNC3L456H, respectively, were fused to EGFP and expressed under the control of a motoneuron specific regulatory element composed of 3 tandem repeats of a mnx1 (hb9) enhancer element (3x mnx1) linked to the gata2 basal promoter [70]. E The same approach as for SCA1-modelling was used for generating a PC specific zebrafish SCA13 model, this time using 4 copies in tandem of the PC specific regulatory element cpce. A zebrafish kcnc3R335H mutant   equivalent to the human SCA13 causing mutant allele KCNC3R420H, was coexpressed either with membrane-targeted mScarlet or a membrane targeted TagRFP-T (mem-TagRFP-T) together with nuclear localized localized EGFP (Nuc-EGFP) mediated by a self-cleaving T2A peptide [24]. F In an alternative approach, the zebrafish pathogenic mutant alleles zkcnc3aR335H and zkcnc3aR338H, equivalent to human KCNC3R420H and KCNC3R423H, respectively, were fused to P2A-mCherry to induce co-expression of each SCA13 mutant together with a red fluorescent reporter protein mCherry under the control of a PC specific aldoca promoter [68]. G A zebrafish PC-ATTAC line - allowing for drug inducible PC ablation - expresses the red fluorescent reporter protein TagRFP fused to T2A-Caspase8 (Casp8)-ERT2 (Tamoxifen-interacting estrogen receptor ligand-binding domain) [81]. This transgene allows for visualizing red fluorescent-labeled PCs, which can be ablated by the dimerization and activation of Casp8-ERT2, leading to apoptosis upon Tamoxifen treatment. 4-Hydroxytamoxifen (4-OHT), or Endoxifen (Endox), both active metabolites of Tamoxifen, were used for PC ablation in larval and adult zebrafish, respectively. The PC specific promoter is composed of the cpce regulatory element fused to a CMV minimal promoter

Despite such successful use of SCA3 zebrafish, broad expression of the ATXN3 mutant protein throughout the CNS or its restricted expression in motoneurons still raises the question of whether such models could recapitulate the brain region-specific pathology commonly observed in individuals with neurodegenerative diseases [82]. Indeed, approximately half of the SCA3 patients display peripheral neuropathy [83], which has been successfully replicated in the pan-neuronal and motoneuron specific SCA3 zebrafish. However, it remains largely unknown whether the SCA3 mutant protein damages specific brain regions in addition to the spinal motoneurons in zebrafish and whether these affected neuronal populations in the brain contribute to zebrafish behavioral impairment corresponding to human ataxia.

Specific distribution patterns of disease-associated proteins are considered to be strongly associated with brain region- and/or neuronal population-specific damage, leading to individual disease symptoms. Thus, for genetic disease modeling it is preferable to target vulnerable neurons specifically affected in the respective disease to be studied using neuronal cell type specific regulatory elements. Since cerebellar PCs are regarded as a neuronal population primarily affected in SCA diseases [30, 31], PCs have been targeted when SCA transgenic mice models were generated, which successfully displayed ataxic-like behavior [84]. We therefore sought to develop SCA zebrafish coexpressing a SCA causative mutant together with flu