Missense Mutation in CAPN1 Is Associated with Spinocerebellar Ataxia in the Parson Russell Terrier Dog Breed

Spinocerebellar ataxia (SCA) in the Parson Russell Terrier (PRT) dog breed is a disease of progressive incoordination of gait and loss of balance. Clinical signs usually become notable between 6 and 12 months of age with affected dogs presenting with symmetric spinocerebellar ataxia particularly evident in the pelvic limbs. The degree of truncal ataxia, pelvic limb hypermetria and impaired balance is progressive, particularly during the initial months of disease. A certain degree of stabilisation as well as intermittent worsening may occur. At the later stages of the disease ambulation often becomes difficult, with owners often electing to euthanise affected dogs on welfare grounds. Using a GWAS approach and target-enriched massively-parallel sequencing, a strongly associated non-synonymous SNP in the CAPN1 gene, encoding the calcium dependent cysteine protease calpain1 (mu-calpain), was identified. The SNP is a missense mutation causing a cysteine to tyrosine substitution at residue 115 of the CAPN1 protein. Cysteine 115 is a highly conserved residue and forms a key part of a catalytic triad of amino acids that are crucial to the enzymatic activity of cysteine proteases. The CAPN1 gene shows high levels of expression in the brain and nervous system and roles for the protein in both neuronal necrosis and maintenance have been suggested. Given the functional implications and high level of conservation observed across species, the CAPN1 variant represents a provocative candidate for the cause of SCA in the PRT and a novel potential cause of ataxia in humans.


Introduction
Spinocerebellar ataxia (SCA) also referred to as hereditary ataxia has been reported in several related fox terrier breeds including the Smooth-Haired Fox Terrier (SHFT) [1,2,3], the Jack Russell Terrier (JRT) [4,5,6] and the Parson Russell Terrier (PRT) [4].The JRT is similar, although shorter legged than the PRT, but is not a registered 'pure breed' with the UK Kennel Club.
Clinical signs of SCA are usually recognised by the owners when the dogs are 2 to 9 months of age [4,6].Initially the dogs' owners may notice incoordination, pelvic limb swaying when walking, and difficulty in climbing stairs and jumping.As the disease progresses, a characteristic 'prancing' or 'dancing' type of gait is observed, especially affecting the pelvic limbs.Severely affected animals frequently fall and have difficulty returning to a standing position.Neurological examination reveals symmetric spinocerebellar ataxia, characterised by hypermetria and spasticity particularly in the pelvic limbs.Postural reactions may be delayed and hypermetric.Spinal reflexes are normal to increased and cranial nerve examination is generally unremarkable [4,6].In one study, haematology, serum biochemistry, urinalysis, cerebrospinal fluid analysis, radiography, myelography, and spinal computed tomography did not identify any abnormalities in affected PRTs and JRTs [4].Brain stem auditory-evoked potentials (BAEPs) revealed abnormalities in some, but not all, of the affected dogs [4].The BAEP abnormalities were characterised by absence of waves III, IV and V. Progression of neurological dysfunction is variable.Generally clinical deterioration results in severe difficulty ambulating, with owners often electing to euthanise affected dogs on welfare grounds.However, one dog with a milder disease course survived for 15 years [4].
Post-mortem examination shows no macroscopic abnormalities of the central nervous system.Histopathological examination reveals bilateral symmetrical myelopathy, predominantly an axonopathy combined with myelin loss, in the dorsal and ventral or ventromedial funiculi.Swelling of axons and dilatation of myelin sheaths with loss of myelin adjacent to a mild astrogliosis were observed primarily in the spinocerebellar tracts of the cervical cord but were also observed in all parts of the brain in the PRT and in the JRT [4,6].Moderate diffuse gliosis, marked loss of myelinated nerve fibers, and argyrophilic axonal spheroids were detected in central auditory pathways, including superior olivary nuclei, cochlear nuclei, connecting nerve fibers between these nuclei and the trapezoid body, and the lateral lemniscus.These lesions correlate with the abnormalities reported on BAEPs as the neural generators of wave III and IV are the trapezoid body and lateral lemniscus, respectively.The SHFT was initially reported to have spinal cord involvement only [2,3], however, a more recent study reported histopathological evidence for degenerative changes in the brainstem and clinical evidence of brain involvement [1].
An autosomal recessive mode of inheritance was suspected in the Smooth-Haired Fox Terriers [2,3].In the study including PRTs and JRTs, complex segregation analysis across three families containing 115 individuals suggested a hereditary cause for the ataxia with a polygenic model most likely, although a major gene effect with additional polygenic factors was not excluded [4].Hereditary ataxia has been reported also in JRTs with concurrent other neurological disorders including behavioural changes, seizures, respiratory distress (Wessmann 2004), myokymia and neuromyotonia [5,7] suggesting phenotypic and genetic heterogeneity.
Genetic mechanisms for three types of inherited canine cerebellar disorders have been described in the literature to date.Retrotransposon insertion within the gene encoding glutamate receptor, metabotropic 1 (GRM1) has been associated with neonatal cerebellar ataxia in the Coton de Tulear dog [8,9].Cerebellar ataxia in the Finnish Hound was shown to be caused by a missense mutation in the sel-1 suppressor of lin-12-like (SEL1L) gene [10].Most recently neonatal cerebellar cortical degeneration in the Beagle was associated with an 8 bp deletion in the gene encoding beta-III spectrin (SPTBN2), which is known to caused spinocerebellar type 15 in humans [11].SCA in the PRT has a later onset and slower progression rate in comparison to the earlyonset canine cerebellar ataxias with known molecular mechanisms and is therefore likely to have a novel genetic cause.To improve our understanding of SCA in the PRT, we collected a set of SCA cases and controls to perform a genome-wide association study (GWAS) to elucidate the mode of inheritance and identify the causal mutation(s).Results of the GWAS and identification of a strongly associated and highly provocative potential causal mutation are described.

Results
A GWAS was performed using 16 SCA cases and 16 controls.DNA samples were all successfully genotyped on the Illumina CanineHD array achieving call rates of .99.8%.Allelic association analysis was performed using the statistical analysis package PLINK which was implemented at the Linux command prompt.After exclusion of SNPs with a minor allele frequency of less than 0.05 and genotyping success rate of less than 0.95, 126,225 SNPs remained.The genomic inflation factor based on the median chisquared value was 0.8 indicating no stratification between the case and control populations, which was further confirmed by multidimensional-scaling (MDS) and quantile-quantile (QQ) plotting (Figure S1).The genomic inflation value of ,1 is most likely because of the use of closely related case-control pairs in the GWAS (two sibling pairs, six half-sibling pairs and four offspringparent pairs included -Table S1).Basic allelic association analysis on the filtered SNP set revealed a strong statistical signal on chromosome 18 (P raw = 7.04610 29 ) (Figure 1A).To correct for multiple testing, allelic association analysis was performed using 100,000 MaxT permutations in PLINK.A single signal on chromosome 18 of genome-wide significance remained (P ge- nome = 1.06610 23 ) (Figure 1B).Correction for population substructure and relatedness was performed using a mixed model, implemented in the statistical package R.The single strong statistical signal remained (P corrected = 9.89610 210 ) (Figure S2).Results were suggestive of a simple autosomal recessive mode of inheritance for the disorder.
Genotyping data from the region of genome-wide significance on chromosome 18 revealed that all but one of the 16 cases were homozygous for a shared haplotype (Figure 2).Recombination events in two individuals defined the disease-associated interval as chr18:53,533,360-55,418,743 based on the CanFam2 genome build.The disease-associated region for SCA contained 91 genes and was syntenic to chromosome 11 of the human genome (Figure S3).Genes within the interval were assessed for potential involvement in SCA.The gene encoding beta-III spectrin (SPTBN2) was a particularly strong candidate, as mutations in the gene have been shown to cause spinocerebellar ataxia type 5 (SCA5) in humans [12] as well as an inherited neurological disorder in the dog [11].Exon resequencing of the gene revealed no potentially causal coding polymorphisms.
The entire disease-associated interval was subsequently interrogated using a target-enriched massively parallel sequencing approach.Sequencing on a single lane of the Illumina HiSeq2000, produced a 19.64 Gb dataset of 51 bp paired-end reads, with 79% of bases in the region achieving 10x read coverage.In total 7,024 SNPs and 1,507 indels were identified across the diseaseassociated interval, with 541 variants segregating in accordance with disease status.Sequence data from the same region, generated from an additional 25 control individuals of the same/different breed, allowed 399 SNPs to be excluded from consideration, reducing the number of segregating SNPs to 142 (Table S2).Segregating variants were considered by assessing conservation levels and variant effect predictions.Two SNPs were considered as potentially causal.These were a missense mutation (C115Y) in the gene encoding the large subunit of calcium dependent cysteine protease, m-calpain (CAPN1) and a missense mutation (E24K) in the vacuolar protein sorting 51 homolog (VPS51).The two candidate SNPs were investigated by genotyping an additional cohort of 227 PRTs consisting of 27 cases and 200 control individuals.Results are summarised in Table 1.As four control individuals were homozygous for the disease associated VPS51 SNP allele, the variant could be excluded as potentially causal.
Additionally a multibreed panel consisting of 96 healthy individuals representing 32 dog breeds was genotyped for both the VPS51 and CAPN1 SNP variants.The non-reference VPS51 allele was found in four individuals (one Tibetan Spaniel, one Miniature Poodle, and two Doberman Pinschers), providing further evidence that the VPS51 variant was likely to be a common polymorphism.All 96 individuals were homozygous for the wild-type/reference CAPN1 allele.
Multispecies alignments revealed that cysteine residue 115 of CAPN1 is highly conserved (Figure 3).Of the 38 aligned species, all had a cysteine residue at the 115 position, apart from alpaca which has a leucine residue.A 1 bp insertion at residue 124 puts the alpaca transcript out of frame however, possibly suggesting that the gene is non-functional in this species.
To help assess whether the CAPN1 amino acid change could have an important functional effect, three computational methods of predicting mutation effects were used -SIFT (Sorting Intolerant from Tolerant), Polyphen (Polymorphism Phenotyping) and Mutation Taster [13,14,15].All three methods suggested the mutation to be damaging or disease causing, with high confidence level values, with the Mutation Taster predicting loss of a catalytic domain.
Genome-wide messenger RNA sequencing (mRNA-seq) of one SCA case and one control cerebellum tissue sample was used to assess gene expression levels across the disease-associated interval.On visual inspection of read alignments expression levels for the disease-associated region appeared to be equivalent for the case and control across the disease-associated region, enabling a change in gene expression to be excluded as the cause of SCA.
Five SCA affected JRTs were genotyped for the CAPN1 SNP.Of the five dogs three were homozygous wild-type, one was heterozygous and one was homozygous for the disease-associated allele.This may suggest that the identified locus is not a major cause of ataxia in the JRT breed.All eight outliers which were reported as clinically affected, but were not homozygous for the CAPN1 mutation were further investigated by genotyping 48 informative SNP markers distributed across the disease-associated interval.Results showed that none of the outliers were homozygous for any part of the disease-associated interval, which in turn suggested a different clinical or genetic cause for the clinical signs in these cases, rather than the possibility that the CAPN1 mutation was not causal.Genotyping results are summarised in Figure S4.

Discussion
In this investigation we have identified a CAPN1 mutation that is strongly associated with SCA in the PRT, using a GWAS approach followed by target-enriched massively parallel sequencing of the disease-associated interval.Mutations in CAPN1 have not previously been associated with ataxia in any species.The CAPN1 gene encodes an intracellular calcium dependent cysteine protease, which is a member of the calpain family and papain superfamily of cysteine proteases found throughout the plant, animal and fungal kingdoms [16,17].Conventional calpains consist of an N-terminal anchor helix region, two protease core domains (PC1 and PC2), a C2 like domain (C2L) and a penta EF-hand calcium binding domain (PEF) (Figure 4).The CAPN1 protein forms heterodimeric structures with the small regulatory subunit CAPNS1, interacting with the fifth EF-hand motif.On binding of calcium, conformational changes result in formation of a catalytic triad of cysteine, histidine and asparagine residues and activation of the enzyme [18].
The disease-associated CAPN1 mutation is a non-synonymous G to an A base substitution at position 344 of the CAPN1 transcript, resulting in substitution of a cytosine residue for a tyrosine (C115Y).This 115 cysteine residue corresponds to the catalytic cysteine residue that forms part of a catalytic triad with histidine and asparagine, and is therefore critical to the enzymatic properties of the protein (http://www.uniprot.org/uniprot/P07384).Consequently, substitution of this residue is likely to have a detrimental effect on enzyme activity, and could potentially result in loss of function.An extremely high level of conservation was observed across species for CAPN1 orthologues at the amino acid level for the 115 cysteine residue.High levels of conservation within species for calpain family members (paralogues of CAPN1) further suggest the critical importance of the residue.Predictive tools suggest the mutation to be potentially pathogenic.Collectively the evidence presents a strong case for the CAPN1 mutation to be deleterious.
Calpains are well associated with nervous function.Calpains have been associated with neuronal necrosis, with proteolytic activity increasing as cellular calcium levels rise due to loss of  homeostasis after trauma [19].Experimentally induced brain trauma by ischaemia (narrowing of the blood vessels) was shown to result in increased proteolysis of the calpain substrate frodin (alpha spectrin), a major cytoskeletal protein, an event hypothesised to be part of a cascade leading to neuronal cell death [20].Calpain inhibitors have therefore been suggested as potential therapeutics for traumatic brain injury [21,22].Calpains have also been linked to roles in long term potentiation and Alzheimer's disease [23].Despite links to neuronal death and disease, there is evidence that calpains may actually contribute to dendrite remodelling after neural injury, suggesting a maintenance role [24].Advances in the understanding of calpains have been made by analysing targeted gene knockouts.The importance of the ubiquitously expressed m/m calpains (CAPN1 and CAPN2) was shown by targeted knockout of the gene encoding the small regulatory subunit CAPNS1 (CAPN4), which resulted in embryonic lethality [25].Targeted deletion of mouse CAPN2 also resulted in embryonic lethality, but CAPN1 null mice appear phenotypically normal, apart from an observed reduction in platelet aggregation, although no effects on bleeding time were seen [26,27].Although genomic comparisons and predictive tools strongly suggest that the CAPN1 mutation is likely to be pathogenic, this contradicts the clinically normal phenotype observed in the CAPN1 null mouse.Conversely, there have been no suggestions of platelet disorders in dogs homozygous for the CAPN1 variant, although this potential defect has not been formally investigated.One possibility is that null mice do not have significant longevity or are euthanised before manifestation of clinical signs.Another possibility is that the CAPN1 gene has a slightly differing role in the mouse with the calpain family having a level of redundancy, allowing other family members to compensate for loss of calpain 1 activity.A third possibility is that the CAPN1 mutation is not the cause of the SCA but is in fact a marker in linkage disequilibrium with the disorder, although the exact residue changed by the CAPN1 variant is highly provocative.
A possible role for calpain 1 in neuronal maintenance and remodelling role best fits the potential for a defective calpain 1 protein being the cause of SCA in the PRT.SCA is largely a disease of motor neurone degeneration in the spinocerebellar tract, with Wallerian type degeneration observed on histopathological examination.Defective maintenance mechanisms due to lack of calpain 1 leading to neurite degeneration and necrosis would explain these observations, although processing of substrates by calpain 1 have been implicated in neurite degeneration confusing calpain's role and potentially suggesting involvement in multiple molecular processes and pathways [28,29].
The gene encoding beta-III spectrin (SPTBN2) provided a strong candidate gene within the disease associated region.Mutations in SPTBN2 have been shown to cause SCA5 in humans [12] and recently neonatal cerebellar cortical degeneration in the Beagle dog [30].Beta-III spectrin is primarily expressed in the nervous system and the highest levels of expression are found in  Purkinje cell soma and dendrites [31].Beta-III spectrin has been shown to stabilise the glutamate transporter EAAT4 at the plasma membrane of the Purkinje cells [32], facilitate protein trafficking by linking the microtubule motor to vesicle-bound cargo [33] and maintain a high density of sodium channels within the soma and dendrites of Purkinje cells [34].Beta-III spectrin is critical for development of Purkinje cells [35].On exon resequencing of SPTBN2 no non-synonymous SNPs, frameshift or splice site variants were identified excluding coding changes as a potential cause of SCA in the PRT.Targeted resequencing of the SCA disease-associated region enabled both coding and non-coding regions of the SPTBN2 to be investigated.No potentially causal variants were identified.Data from mRNA-seq experiments provided further evidence regarding the SPTBN2 transcript and enabled mis-splicing and changes in expression levels to be excluded as potential causes of SCA.The SPTBN2 gene was therefore excluded from further consideration.
Although similar clinical signs of ataxia are shared in both SCA and SCA5, other clinical features suggest potentially different disease mechanisms.Clinical features of SCA5 suggest a predominately cerebellar disease [36], whereas histopathological examination of SCA cases suggested limited cerebellum pathology, with degeneration of the brain stem and spinocerebellar tract involved in disease progression.This evidence suggests a different genetic cause for SCA and SCA5, although it is possible for different mutations in the same gene to result in variable phenotypes.
Four PRT and four JRT which displayed clinical signs of SCA, but were not homozygous for the CAPN1 disease-associated allele were identified in the study.Efforts were made to follow up these discordant cases by contacting owners and veterinarians.Seven of the cases had clinical signs consistent with SCA, although full neurological examinations had not been performed to rule out other causes of spinocerebellar disease.On review by a veterinary neurologist (LDR) one outlying case (the outlier from the GWAS) was deemed not to fit the case definition because typical pelvic limb hypermetric and spastic gait with truncal ataxia was not reported.Of the eight discordant cases, six were homozygous for the wild-type CAPN1 allele and two were heterozygous.Given the available information it is impossible to rule out a second genetic cause of a clinically similar form of hereditary ataxia in the PRT and JRT.Anecdotal evidence suggests a neonatal onset form of ataxia also segregates within the JRT and PRT dog breeds.

Conclusions
Using a GWAS approach and target-enriched massively parallel sequencing a disease-associated SNP in CAPN1 has been identified.The SNP is a missense mutation causing a cysteine to tyrosine substitution at residue 115 of the calpain protein.Cysteine 115 is a highly conserved residue and forms a key part of a catalytic triad of amino acids that are crucial to the enzymatic activity of cysteine proteases.Given the function and high level of conservation, substitution of the cysteine residue is highly likely to have a negative effect on the activity of the enzyme, although functional studies would be required to provide formal evidence to support this hypothesis and to assess the extent of activity loss.Loss of CAPN1 activity as a cause of SCA is difficult to prove, although a suggested role for CAPN1 in neuronal maintenance fits with the pathogenesis of the disease based on histopathological evidence.The finding represents the first association of a mutation in CAPN1 with spinocerebellar ataxia and may represent a novel candidate gene for ataxia in human patients.

Ethics Statement
Cerebellum tissue samples were collected post-mortem after dogs had been euthanised on welfare grounds.Euthanasia was carried out solely to alleviate suffering and no healthy individuals were sacrificed for use in this study.Euthanasia was carried out in accordance with the Veterinary Surgeons Act 1966 and under the auspices of the RCVS.DNA was collected by buccal swabbing, which is a relatively non-invasive procedure that does not require a license.All samples used in this study were collected after permission had been granted by PRT dog owners.This study did not require ethics committee approval as euthanasia was carried out on welfare grounds only and DNA was collected using a relatively non-invasive procedure which did not require a license.

Sample Collection and Nucleic Acid Extraction
All DNA samples were collected from privately owned pet dogs by buccal swabbing and extracted using the QIAamp midi DNA extraction kit (Qiagen).Cases were owner reported with video evidence provided for four individuals.Case details of individuals included in the GWAS stage were reviewed by a veterinary neurologist (LDR) (Table S1).To provide a better description of the phenotype, LDR (a specialist in veterinary neurology) performed a neurological examination and measured BAEPs in one of the affected dogs which was homozygous for the CAPN1 mutation.Video footage and clinic-diagnostic details of this dog have been provided (Video S1).The neurological examination of this PRT revealed the typical gait and neurological deficits of SCA and the BAEP revealed absence of waves III and IV, suggestive of dysfunction of the the trapezoid body and lateral lemniscus, respectively, as previously described in PRT with hereditary ataxia (Figure 5 in Wessmann 2004) [4].Controls were dogs that were reported as clinically healthy with no signs of ataxia at or above 4 years of age.The additional panel of 96 dogs from 32 other breeds genotyped for both the CAPN1 and VPS51 SNPs were all reported by their owners to be healthy and had been recruited to participate in other unrelated studies.Cerebellum tissue samples were collected post-mortem and total RNA extracted using the RNeasy midi kit (Qiagen).

GWAS Analysis
DNA samples from 16 PRT cases and 16 PRT controls were genotyped using the Illumina CanineHD SNP genotyping array that comprises 173,662 SNPs.The SNP genotyping dataset was analysed for association using the statistical package PLINK [37].Sample call rates for all individuals were .99%.SNPs with a genotyping call rate of ,95% and/or minor allele frequency of ,5% were discarded.The strongest statistical signal from the unadjusted association analysis was termed P raw .Correction for multiple testing was performed using 100,000 MaxT permutations in PLINK.The strongest statistical signal after permutations was termed P genome .Correction for genomic inflation was performed using a mixed model approach implemented in the statistical package R [38].The strongest statistical signal after correction was termed P corrected.

SPTBN2 Resequencing
The SPTBN2 gene was exon resequenced using Sanger sequencing methodology.Templates for sequencing were created by PCR.Reaction mixes consisted of 1.5 mM dNTPs, 16PCR buffer, 0.8 mM forward and reverse primer, 1.2 U HotStarTaq polymerase (Qiagen), 2 ml genomic DNA and ultrapure water to give a final volume of 12 ml.Sequencing was performed using BigDye v3.1 terminator chemistry (Applied Biosystems) for capillary electrophoresis on ABI31306l genetic analysers.PCR primers are listed in Table S3.

Figure 1 .
Figure 1.Allelic association analysis plots for SCA in the PRT.Allelic association analysis on 16 SCA cases and 16 controls.Each dot represents a single SNP, with -log10 (p) values on the y-axis plotted against genome position (split into chromosomes) on the x-axis.(A) Raw unadjusted log(p) values with a strong statistical signal indicated on chromosome 18 (P raw = 7.04610 29 ).(B) Plot of -log10(p) values after 100,000 maxT permutations analysis to correct for multiple testing, showing a single peak reaching genome-wide significance on chromosome 18 (P genome = 1.06610 23 ).doi:10.1371/journal.pone.0064627.g001

Figure 2 .
Figure 2. Raw genotyping data across the disease associated region for SCA.Graphical overview of the raw genotyping data across the SCA critical region.Each column represents one individual, and SNP markers are listed in the rows with the approximate position indicated.Major alleles are highlighted in blue and minor alleles in red.Interval defining cases are marked by a circle at the top of individual columns.The outlying case which does not share the disease associated haplotype across the interval is marked with an asterisk.doi:10.1371/journal.pone.0064627.g002

Figure 3 .
Figure 3. Multispecies alignment across residue 115 of the calpain peptide.Alignment across 46 vertebrate species visualised using the UCSC genome browser (human genome build hg19) to assess the level of cross-species conservation for the CAPN1 gene.doi:10.1371/journal.pone.0064627.g003