A preliminary genome‐wide association study of paroxysmal dyskinesia in the Norwich Terrier

Paroxysmal dyskinesias (PxDs) are typified by recurring episodes of involuntary abnormal movement without loss of consciousness (Bhatia, 2011; Cerda-Gonzalez et al., 2021). PxD is phenotypically heterogeneous in dogs, but characteristics are often shared between breeds and with human PxD (Cerda-Gonzalez et al., 2021; Lowrie & Garosi, 2017). A pedigree analysis of a group of Norwich Terriers in a previous study of PxD in the UK showed clustering of cases and suggested an inherited component, although mode of inheritance could not be ascertained (De Risio et al., 2016). A PxD prevalence of 13% (95% confidence interval 9–18%) was estimated in this study, although as it was questionnaire-based it was likely to have had a bias towards affected dogs. We conducted a preliminary genome-wide association study (GWAS) aiming to identify PxD-associated loci in Norwich Terriers and help to assess the mode of inheritance of the disorder. GWAS individuals were 24 Norwich Terrier PxD cases and 24 PxD-unaffected Norwich Terriers aged over 6 years (Appendix S1). Genotype data were generated for the 48 dogs using the Axiom Canine HD array. All individuals had a genotype call rate >90%. Population-corrected association analysis (using a linear mixed model) was carried out on the autosomes using GEMMA (Zhou & Stephens, 2012), including only the 230 972 variants with call rate >97%, minor allele frequency >0.05, and Hardy–Weinberg equilibrium p-value >5 × 10−5. No loci reached genome-wide statistical association (p ≤ 2 × 10−7; Figure 1). As this was a small preliminary study lacking statistical power to detect associations for a non-Mendelian disease, we investigated if the associations of any of the top tier of SNPs were augmented by adding data from an independent Norwich Terrier SNP genotype dataset (Illumina canineHD array) (Marchant et al., 2019). PxD case–control status was defined for a subset of these dogs and the dataset was imputed to obtain genotypes for Axiom array SNPs (Appendix S1). To determine if population stratification was present within the GWAS sample set, or between the GWAS set and the upper airway syndrome set, two-dimensional multidimensional scaling plots based on a matrix of pairwise identity-by-state distances were generated. This did not indicate material population stratification (Figure S1). The addition of the 10 cases and 45 controls augmented statistical associations that were directionally consistent for two of the eight GWAS SNPs; within the SIK3 and FCHSD2 genes (Table 1). However, the risk allele for the SNP in FCHSD2 was not present in any of the additional cases and controls and was extremely rare (0.007) in the 232 dogs of the upper airway syndrome set. A meta-analysis of the two study sets was subsequently carried out (Appendix S1, Figure S2). As expected, due to the small PxD case number in the upper airway syndrome set, no loci achieved genome-wide significance in this analysis. In conclusion, our preliminary GWAS suggests that PxD in the Norwich Terrier is not likely to be a single-gene disorder, and may be multigenic or complex. Future research will build on these findings through meta-analysis with a larger GWAS dataset, or by investigating the condition using whole genome sequencing of multiple cases. Christopher A. Jenkins: Data curation; formal analysis; funding acquisition; investigation; methodology; writing – original draft; writing – review and editing. Luisa De Risio: Data curation; funding acquisition; writing – review and editing. Elisabeth Dietschi: Data curation; resources; writing – review and editing. Tosso Leeb: Resources; writing – review and editing. Ulrich Rytz: Resources; writing – review and editing. Peter Schawalder: Resources; writing – review and editing. Jeffrey J. Schoenebeck: Resources; writing – review and editing. Cathryn S. Mellersh: Resources; writing – review and editing. Sally L. Ricketts: Conceptualization; funding acquisition; investigation; methodology; project administration; supervision; writing – review and editing. This study was funded by a Dogs Trust Canine Welfare Grant. CAJ, SLR and CSM were supported during this study by the Kennel Club Charitable Trust. JJS receives institutional strategic programme support from the Biotechnology and Biological Sciences Research Council (BBS/E/D/10002070, BBS/E/D/10002071, BBS/E/D/30002275, and BBS/E/D/30002276). We thank the Oxford Genomics Centre at the Wellcome Centre for Human Genetics (funded by Wellcome Trust grant reference 203141/Z/16/Z) for the generation and initial processing of the Axiom array data. We thank the Vetsuisse Biobank for making samples available. We thank Janine Lamb, University of Manchester, for critical review of the manuscript; Donna Foster for support with sample collection; and Ellen Schofield for bioinformatics assistance. We would like to sincerely thank all owners and breeders for providing samples from their dogs for this study and the UK Norwich Terrier Club for their support. The authors have declared that no conflict of interests exist. The study design and sample collection protocols were approved by the Animal Health Trust Clinical Research Ethics Committee (Project No. 35–2017) and University of Cambridge Department of Veterinary Medicine Ethics and Welfare Committee (No. CR527), and samples were obtained following owner informed consent. Axiom array data and phenotypic data for the 48 Norwich Terriers will be available at https://osf.io on publication. Figure S1. Figure S2. Table S1. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.