In this article we have described NCCD in three Beagle dogs and report the use of genome-wide mRNA sequencing of a single case to identify an 8 bp deletion in the β-III spectrin gene, SPTBN2, that segregates consistently with the disorder. We have shown that the use of mRNA-seq for candidate gene analysis can potentially remove the need for a genome-wide association study (GWAS) stage, an approach commonly and successfully used in the dog even with small case-control sets, reducing cost and shortening study time-frames. Because the mRNA-seq approach requires far fewer samples than the GWAS approach, less time is needed in the sample collection phase, allowing projects to commence earlier, especially when cases are rare. To our knowledge this is the first report of this application for mRNA-seq. The approach has advantages over exome enrichment and sequencing methodology, in that the method can be performed in all species with reference genome sequence builds, without the need for a proprietary kit. The mRNA-seq approach is also not dependant on reference genome annotation, which may be inaccurate or incomplete in some species.
Initial clinical presentation and progression of the three affected Beagle puppies were highly suggestive of NCCD, with histopathological studies confirming Purkinje cell degeneration and necrosis in different stages according to disease progression in the two puppies that underwent necropsy. Histologically both the four-week old puppy and eight-week old puppy revealed Purkinje cell degeneration and loss. Lesions present in the eight-week old puppy demonstrated disease progression expected with the prolonged survival time, with increased proportions of lost Purkinje cells, and the presence of corresponding degeneration of the granular cell layer. Changes in both puppies were limited to Purkinje cell death and secondary changes associated with the loss of Purkinje cells (ie astrocytosis, granular cell degeneration, axonal swelling). Regional variation, as previously described by Yasuba et al., was not identified , but all the affected Beagles in that study were euthanized at 14 weeks of age, and the regional variation could just be related to the progression of the disease.
The identification of an 8 bp deletion in SPTBN2, a gene associated with spinocerebellar type 5 (SCA5) in humans , that fully segregates with the disease provides a strong candidate variant for NCCD in the Beagle. Spectrins are a family of cytoskeletal proteins, with tetrameric structures comprising two α and two β subunits, with diversity and specialisation of function. Spectrins are important structural components of the plasma membrane and play a significant role in restricting and stabilising membrane spanning proteins within specific subdomains of the plasma membrane. The spectrin cytoskeleton was first discovered in erythrocytes and has since been identified in a variety of cells . β-III spectrin is primarily expressed in the nervous system and the highest levels of expression are found in Purkinje cell soma and dendrites . β-III spectrin has been shown to stabilise the glutamate transporter EAAT4 at the plasma membrane of the Purkinje cells , facilitate protein trafficking by linking the microtubule motor to vesicle-bound cargo  and maintain a high density of sodium channels within the soma and dendrites of Purkinje cells . β-III spectrin is critical for development of Purkinje cells .
In humans, three mutations in SPTBN2 have been shown to cause autosomal dominant SCA5. The identified causal mutations include two in-frame deletions of 39 and 15 bp which alter the structure of the 3rd of 17 spectrin repeats, and a single base pair substitution causing an amino acid change (L253P) in a highly conserved region of the calponin homology domain . The consequence of the two in-frame deletions in β-III spectrin is predicted to be disruption of the highly ordered triple alpha helical structure of the spectrin repeat, causing conformational changes in the tetrameric α-β spectrin complex . Studies suggest the resulting mutant protein may affect the localisation of EAAT4 and GluRδ2, one possible outcome of which is glutamate signalling abnormalities and Purkinje cell death . The L253P missense mutation has been shown to result in loss of interaction with the Arp1 subunit of the dynactin-dynein complex, affecting the role of β-III spectrin in vesicle trafficking, preventing transport of both β-III spectrin and EAAT4 to the cell membrane from the Golgi apparatus in Purkinje cells causing cell dysfunction and death. 
Experimentally induced β-III spectrin deficiency in mice from two independent studies resulted in phenotypes that resemble NCCD in Beagle dogs [31, 34]. One β-III spectrin deficient strain was produced by targeting replacement of exon 3 to 6 of SPTBN2 with the neomycin-resistance gene, resulting in a frameshift and a premature stop codon in exon 7. As a result no full length β-III spectrin is produced in -β-III−/− mice, although a low level of near full length protein is produced due to novel exon 1 (rather than exon 2) to exon 7 splicing . Homozygous β-III spectrin deficient mice develop characteristics of progressive cerebellar ataxia from a few weeks of age with cerebellar atrophy and Purkinje cell loss. In the parallel study the β-III spectrin deficient mouse strain is the result of βgeo insertion between exons 25 and 26 resulting in premature termination in spectrin repeat 14, which is closer to the position to the Beagle mutation, although results in the loss of the ankyrin binding domain . The β-III spectrin deficient mice from this study display a mild non-progressive ataxia by 6 months and a myoclonic seizure disorder by one year . It is apparent that onset of ataxia is later for the β-III spectrin deficient mouse in comparison to Beagle NCCD cases, with mice not showing significant signs of ataxia until six months of age (past sexual maturity). This is more comparable to the human disease, though the differences in the modes of inheritance suggest different mutational effects. Deficient mice also show only a mild ataxia and remain ambulatory, while the dogs described are more severe both in terms of degree of ataxia (astasia) and Purkinje cell loss. In the study by Stankewich et al., no Purkinje cell loss was documented by 18 months of age, only atrophy of the dendritic arbor. Disparity in phenotype between species may suggest differences in cerebellar development, function, and potentially the involvement of β-III spectrin. Further understanding of canine cerebellar function would be required to shed light on the described differences and common principles.
It has been shown that heterozygous mice, generated by exon 2–6 replacement, do not display any characteristics of cerebellar ataxia , in common with heterozygous dogs in the Beagle population, suggesting that SCA5 in heterozygous humans is caused by dominant negative effects of mutant β-III spectrin, rather than haploinsufficiency. Histopathologic examination in heterozygous mice revealed normal size and morphology of the cerebellum and immunostaining studies showed no changes on Purkinje cell morphology. These histopathological findings cannot be correlated with heterozygous Beagle dogs as none underwent post-mortem examination and all of them are currently alive and clinically unaffected. Interestingly, slight motor impairments were reported for heterozygous mice generated by βgeo insertion between exons 25 and 26, perhaps indicating that the truncated protein is having a slight dominant negative effect, and illustrates how disease progression is dependent on the positioning of SPTBN2 mutations.
The 8 bp deletion in the dog is located at a tandem repeat sequence, suggesting homologous recombination as the deletion mechanism. The position of a SNP (c.5580 T > C) 18 bp downstream of the deleted sequence removes a possible termination site for the mutant protein and extends the sequence of potential aberrant amino acids from 6 to 27. Although expression analysis was limited, due to the availability of only one case and one control, results are suggestive of a 68 x reduction in the relative levels of SPTBN2 in the NCCD case cerebellum, which may be due to nonsense mediated decay. Even though SPTBN2 expression is greatly reduced in NCCD affected cerebellum tissue, sufficient read depth from the mRNA-seq expression was still achieved, because of the high levels of SPTBN2 expression normally seen in cerebellum tissue. Further to a reduction in mRNA levels, no full length or truncated β-III spectrin was detectable in NCCD affected cerebellum tissue by Western blot analysis. This may indicate that the 8 bp deletion results in a full knock-out of SPTBN2. A full gene knock-out eliminates the possibility of a dominant negative effect that could be caused by a truncated form of the β-III spectrin protein, which is in agreement with heterozygous dogs showing no clinical signs.
Although NCCD is likely to be heterogeneous in different canine breeds, screening for the SPTBN2 deletion in non-beagle cases has not been investigated to confirm this. It is possible that the mutation could exist at very low frequencies in other breed populations, especially those closely related to the beagle, but extensive screening of large numbers of individuals would be required to fully investigate this possibility.