Sequencing a short, 280 repeat SCA10 allele of reduced penetrance revealed ATGCT sequence interruptions within the ATTCT expansion and prompted attempts at sequencing into the interior of longer SCA10 expansions [3]. These experiments revealed ATATTCT and ATTTTCT heptanucleotide interruptions at the 5' end of the repeat from one Mexican SCA10 family with a high incidence of epilepsy while a second family with a lower incidence of epilepsy lacked these interruptions [3]. These data lead to the hypothesis that repeat purity plays a significant role in phenotypic expression of SCA10. However, the limitations of current sequencing technologies, including next generation sequencing, combined with the intrinsic nature of the repetitive sequence prevents us from sequencing across the entirety of the expansion.

ATXN10 mRNA and protein are both widely expressed throughout the body as well as the brain [2,13,14]. ATXN10 encodes a 52 kDa protein whose features are rather unremarkable with the exception of two C-terminal armadillo repeats and stretches of proposed alpha-helices [13].

ATXN10 protein is required for neuronal survival [15] and can induce neurite outgrowth when overexpressed [14]. ATXN10 interaction partners include the G-protein beta 2 subunit (Gbeta2) involved in signaling cascades [16] and O-linked N-acetylglucosamine transferase [17,18].

At the DNA structural level, the ATTCT repeats cause an unpaired DNA structure when present within a supercoiled plasmid and acts as a DNA unwinding element (DUE) in E. coli [19] which may help to explain how the expansion can function as a replication origin initiator in mammalian cell extracts [20]. Errors during the replication process or aberrant replication starting may play a role in repeat instability and fragility of the SCA10 expansion as demonstrated in a yeast model system expressing ATTCT repeats within the context of an artificially constructed intron within the URA3 gene [21]. Additionally, in vitro assays of nucleosome formation suggest that SCA10 repeats sequences can readily form nucleosomes [22]. Combined, these studies further suggest that the SCA10 expansion itself may have an effect on the local chromosomal structure and may assert effects on the local chromosomal environment in cis.

However, SCA10 patient phenotypes are not likely to result from a loss of function of the ATXN10 gene product. ATXN10 transcript levels from SCA10 patients are similar to normal controls and Atxn10 heterozygous knockout mice are morphologically and behaviorally normal despite reduced Atxn10 levels [14]. Furthermore, there are no signs of ataxia or epilepsy in the members of a family carrying the translocation t(2,22) (p25.3; q13.31) which disrupts one ATXN10 copy [23].

During pre-mRNA processing, the intron bearing expanded SCA10 repeats is spliced out and accumulates as AUUCU-containing cytoplasmic and nuclear foci [24,25]. Short, non-pathogenic lengths of AUUCU repeats take on a hairpin structure although pathogenic lengths of SCA10 expansions have not been tested [26]. The RNA binding protein, hnRNP K, was identified as an AUUCU interacting protein by pull-down assay using protein extracts from mouse brains and was later found colocalized with AUUCU RNA foci within SCA10 patient fibroblasts and transgenic mouse brains [24,25].

Questions regarding phenotypic variability in SCA10 can be modeled in transgenic mice. However, the transgenic mouse lines studied in the past are informative yet limiting for reasons described below. Because of these limitations, we have taken this opportunity to reexamine these models and how they might be improved. Additionally, both SCA10 transgenic models were made using SCA10 expansions from patients carrying repeat interruptions and as the corresponding transgenic model with pure SCA10 repeats was not constructed, a number of questions still remain regarding the impact of repeat purity on disease progression, ataxia and epileptic phenotypes.

Both the intron and the 3'UTR models clearly demonstrate the role of the non-coding repeat expansion as the toxic agent in the pathogenic progression of SCA10. However, due to the loss of one model and the poor reproductive capacity of the other, new SCA10 models need to be made. At this time, we need to reassess what is required in order to make the best model possible.

First, in order to understand the role of repeat interruption and repeat purity on phenotypic expression of SCA10, transgenic models that expression interrupted and non-interrupted versions of the SCA10 expansion need to be developed and compared. As transgenic constructs randomly integrate into the genome, transgenic lines will need to be carefully matched for transgene copy number and expression levels in order to have comparable lines. Alternatively, targeted transgenic lines where the transgene is selectively targeted into a known genomic locus, such as the ROSA26 locus [28] could be an option; however, the stability of the SCA10 expansion in embryonic stem cells is unknown at this time.

A second consideration is the genetic background used for making transgenic mice. This can matter a great deal for phenotypic expression, particularly in our prior intronic SCA10 model. As these mice breed poorly and were made on a C57/Bl6 background, it leaves us to wonder whether there is a connection between the two. The choice in genetic background is one of great impact as sensitivity to epileptic phenotypes are known to vary across the different background strains [29] and some strains are more sensitive to drug-induced seizures, such as the DBA/2J strain [30], while the C57Bl/6 strain are relatively more resistant and display little to no excitotoxic cell death [31].

Another point to consider lies in the choice of reporter gene used in the expression cassette with the SCA10 expansions. The reporter gene ideally should be one that is robust and easy to assay. Obvious choices include green fluorescent protein (GFP), or a fluorescent protein of another variety, luciferase or LacZ. GFP carries the obvious advantage as it can be visualized in the live animal; however, luciferase is more useful for sensitive, quantifiable applications. In addition, there should be a method in place to easily assess correct splicing of the intron containing the SCA10 expansion.

Another important consideration is the choice in promoter used to drive expression of the transgene construct. Ideally the endogenous ATXN10 promoter would give the most faithful recapitulation of the native ATXN10 expression pattern. However, the minimal ATXN10 promoter, from any species, has yet to be defined. Alternatively, a BAC construct carrying the SCA10 expansions could be made from human SCA10 patients. Such a construct would contain endogenous human promoter elements making it an attractive alternative. However, after attempts with two BAC libraries made from SCA10 patient genomic DNA, we were unable to isolate such a BAC clone (unpublished data). This leaves us with the choice of using neuronal-specific promoters to drive expression within the brain or certain subsets of neuronal population or using an inducible promoter allowing us to control the timing and location of expression.

The choice of which promoter to use may prove to be a crucial one as we are currently unsure of the extent of neuronal populations that are affected in SCA10 pathology given the relative dearth of autopsy tissue available for analysis. We do know that the cerebellum and sometimes the brainstem show signs of atrophy by imaging analysis of SCA10 patients [6,7,12] therefore the transgene should be expressed in these regions at a minimum. Thus, promoters to consider should be either pan-neuronal, such as neuronal enolase (Eno2) promoter [32] as was used in the intronic SCA10 model [24] or a cell-type restrictive, such as Purkinje cell protein-2 (Pcp2), promoter which drives expression in the Purkinje cells of the cerebellum [33,34].

However, an inducible expression system, such as the bi-transgenic tet-operon/repressor system (reviewed in [35]), might have an advantage over conventional transgenic models as precise control of tissue- and timing-specific expression is gained with this system. The tetracycline transactivator (tTA) protein in the transactivator transgene, is made up of the tet-repressor protein from the E. coli Tn10 transposon fused to the VP16 transactivation domain from herpes simplex virus [36]. The responder construct contains the expression of the transgene of interest downstream of the Tc response element (TRE) consisting of the minimal cytomegalovirus (CMV) promoter with seven tandem repeats of the 17 bp tet operator (tetO). Binding of the dimerized tTA protein at the tetO sites causes expression of the transgene in the absence of a tetracycline analog and turns off expression in the presence of a tetracycline analog (Tet-Off). Conversely, an engineered reverse tetracycline transactivator (rtTA) protein allows for expression of the transgene in the presence of tetracycline (Tet-On) for easier control in the timing of the transgene expression in the mouse [37].

Whatever the mode of transgenic model production, the resulting lines of mice will need to be rigorously assessed for deficiencies in behavioral, movement and seizures susceptibility. Well-known batteries of behavioral assays have been described by the Irwin [38] and SHIRPA [39] protocols as well as in numerous behavioral and neurological phenotyping guides [40,41,42,43]. The chosen assays should be simple and easy enough to rapidly perform without the need for specialized equipment and yet also allow for quantification of results. Results of an initial battery of tests can then be used to guide more detailed, intensive behavior phenotyping.

The process of developing a transgenic model involves multiple choices to consider at various steps of the process. Our past experiences with previous transgenic mouse models of SCA10 can guide some of these choices. The value of any transgenic mouse model to mimic the pathological mechanisms of a human disease rests upon its ability to faithfully mimic aspects of the human disease in a lower vertebrate organism.