Across-breed analyses of genome-wide association studies for stature and mammary gland morphology in cattle reveal pleiotropic effects of the Friesian POLLED haplotype

Background Genome-wide association studies (GWAS) in cattle populations have traditionally relied on progeny-derived phenotypes such as estimated breeding values (EBVs) as input phenotypes to identify additive quantitative trait loci (QTL) for complex traits. Increasing availability of cow genotype data now enables GWAS using own performance records to detect both additive and non-additive QTL. Results Sequence-variant genotypes were imputed for 57,863 cows from the Holstein, Brown Swiss, Original Braunvieh, and Simmental cattle populations that had own performance records for stature and three mammary gland morphology traits (fore udder position, central ligament, front teat position). Genomic heritability ranged from 0.25 to 0.33 for fore udder position, 0.27 to 0.43 for udder central ligament, 0.49 to 0.59 for front teat position and 0.61 to 0.73 for stature. Additive genetic effects explained most of the SNP-based heritability for all traits and breeds. Within-breed genome-wide association studies identified 118 additive and 29 non-additive QTL for the four traits. Non-additive associations were only detected for stature. Although the majority of lead variants were in non-coding regions, we prioritized four missense variants in HMGA2 (rs385670251), ZBTB20 (rs470925681), ARSI (rs447362502) and CRAMP1 (rs445465383) as plausible causal variants for four stature QTL. Meta-analysis of the additive GWAS identified 63 mammary gland morphology and 43 stature QTL. A Holstein-specific mammary gland morphology QTL (Chr1:2748715, p=7.59e-35) colocalized with the POLLED locus on chromosome 1. Fine-mapping of this region revealed undesired effects of the Friesian POLLED haplotype on mammary gland morphology traits. Conclusion Direct phenotypes for a large cohort of genotyped cows provide high statistical power for additive and non-additive association testing. Sequence-based association studies revealed QTL and candidate causal variants for stature and mammary gland morphology traits. Pleiotropic effects of the Friesian POLLED haplotype highlight the need for careful monitoring of potential unintended consequences when selecting for polledness in cattle.