Canine genome-wide association study identifies DENND1B as an obesity gene in dogs and humans

Obesity is a far-reaching, heritable disease, but its genetic basis is incompletely understood. Canine population history facilitates trait mapping. We performed the first successful canine genome-wide association study for body condition score, a measure of obesity, in 241 Labrador retrievers, with a polygenic score replicated in 350 more. Using an innovative cross-species approach, we showed new canine obesity genes are also associated with rare and common forms of obesity in humans. The lead association in dogs was with rs24430444 within the gene DENN domain containing 1B (DENND1B). Each copy of the alternate allele was associated with ~7% greater body fat per allele. We demonstrate a novel role for this gene in regulating signaling and trafficking of melanocortin 4 receptor, a critical controller of energy homeostasis. Thus, canine genetics identified novel obesity genes and mechanisms relevant to both dogs and humans.

Methods
To summarise, we studied Labrador retriever dogs kept as pets or working assistance dogs. Only adult dogs (age 1-10 years, mean 6 years) were included, free of known or suspected systemic illness and not being treated with medications likely to affect obesity status. Body fat mass was assessed using a well-validated measure of adiposity, Body Condition Score (BCS) which uses a combination of haptic and visual cues to assign dogs to BCS categories 1-9 according to standardized descriptors. DNA was extracted from blood or saliva samples and direct genotyping was performed on the CanineHD Genotyping BeadChip (Illumina) array and data were then imputed to 9.4 million single nucleotide polymorphisms (SNPs). For the GWAS, we retained SNPs called with 70% confidence and which were called in >95% of dogs with an allele frequency > 5% and a Hardy Weinberg equilibrium test p >0.001%. There were 4.5 million SNPs included in the GWAS. We performed a GWAS for BCS in 241 Labrador retriever dogs applying a linear mixed effects model (GCTA MLMA-LOCO). Regression modeling was used to identify factors significantly affecting BCS in the population which were then included as covariates for the GWAS. These included sex, neuter status, and sex: neuter status interaction term. We constructed a PRS comprising 16 SNPs weighted for GWAS effect size on BCS using the ‘clumping and thresholding’ technique and applied the PRS to determine its utility as a predictor of BCS in Labradors and other breeds (FCR, pugs, golden retrievers). 

This data could not have been collected, nor research conducted without several collaborative efforts and. We are indebted to the owners who volunteered for the study and to the dogs who took part, the Northern England Flatcoated Retriever Association and Kennel Club for helping with recruitment, Guide Dogs UK for sample contributions, Nai-Cheih Liu for gathering phenotypes in pugs.

We also used data from the Golden Retriever Lifetime Study in this research, data for which is available via their 'Data Commons' website.