Genome-Wide Association Study of Golden Retrievers Identifies Germ-Line Risk Factors Predisposing to Mast Cell Tumours

Abstract

Canine mast cell tumours (CMCT) are one of the most common skin tumours in dogs with a major impact on canine health. Certain breeds have a higher risk of developing mast cell tumours, suggesting that underlying predisposing germ-line genetic factors play a role in the development of this disease. The genetic risk factors are largely unknown, although somatic mutations in the oncogene C-KIT have been detected in a proportion of CMCT, making CMCT a comparative model for mastocytosis in humans where C-KIT mutations are frequent. We have performed a genome wide association study in golden retrievers from two continents and identified separate regions in the genome associated with risk of CMCT in the two populations. Sequence capture of associated regions and subsequent fine mapping in a larger cohort of dogs identified a SNP associated with development of CMCT in the GNAI2 gene (p = 2.2x10-16), introducing an alternative splice form of this gene resulting in a truncated protein. In addition, disease associated haplotypes harbouring the hyaluronidase genes HYAL1, HYAL2 and HYAL3 on cfa20 and HYAL4, SPAM1 and HYALP1 on cfa14 were identified as separate risk factors in European and US golden retrievers, respectively, suggesting that turnover of hyaluronan plays an important role in the development of CMCT.

Fig 1. Multiple dimensional scaling plot visualising the genetic similarity between the individuals within the population using the two first principal components as calculated in PLINK.

a) American and European data combined b) American data c) European data.

Fig 2. Manhattan plot showing the–log 10 p-values in relation to the chromosomes.

a) American data b) European data c) American and European data combined. Q-Q plots showing the expected p-value in relation to the observed p-value for each GWAS analysis. Shaded area indicates 95% confidence interval. Stippled line marks nominal significance threshold.

Fig 3. a) Close up view of the most associated region from the American GWAS analysis. b) Minor allele frequency of the associated region. c) Further close up of the associated peak showing the LD structure of the SNPs in the area relative to the most associated SNP. d) Close up view of the genes located in the area of the associated haplotype.

Red arrow indicates the location of the most associated SNP.

Fig 4. a) Close up view of the most associated region from the European GWAS analysis. b) Minor allele frequency of the associated region. c) Further close up of the associated region showing the LD structure of the SNPs in the 48 MB locus relative to the most associated SNP in the locus. d) Further close up of the associated peak showing the LD structure of the SNPs in the 42 MB locus relative to the most associated SNP in the locus. e) Close up view of the genes located on the associated haplotype in the 42MB region.

Red arrow indicates the location of the most associated SNP in the region.

Fig 5. PCR products confirming the alternative splice site in the individuals carrying the A allele on CFA20: 42080147.

PCR was performed using primers specific for the alternative splicing, excluding exon 3.