Unraveling the genetic background of bovine milk fat composition

Identification of genomic regions, and preferably individual genes, responsible for genetic variation in bovine milk fat composition enhances the understanding of biological pathways involved in fatty acid synthesis and is expected to increase opportunities for changing bovine milk fat composition by means of selective breeding. This thesis aimed to unravel the genetic background of bovine milk fat composition by detection, confirmation and fine-mapping of quantitative trait loci (QTL) for milk fatty acids in Dutch Holstein Friesian cattle. In addition, causal relations between fatty acids were explored. For this study roughly 2,000 dairy cows were genotyped with 50,000 DNA markers and phenotyped for individual fatty acids in both winter and summer milk samples using gas chromatography. Genome-wide association studies (GWAS) showed that milk fat composition has a complex genetic background with three major QTL that explain a relatively large fraction of the genetic variation of several milk fatty acids, and many QTL that explain a relatively small fraction of the genetic variation. Results from the GWAS for summer milk fatty acids confirmed most associations that were detected in the winter milk samples. Moving from linkage analysis toward GWAS confirmed and refined the size of previously detected QTL regions and resulted in new QTL regions. Performing GWAS based on individual fatty acids resulted in additional QTL as compared to GWAS based on fat percentage or yield. This shows that refinement of complex phenotypes into underlying components results in better links between genes and phenotypes. By increasing the marker density, the QTL on BTA19 was refined to a linkage disequilibrium block that contained 2 genes: coiled-coil domain containing 57 and fatty acid synthase. A search for causal relations between fatty acids resulted in a pathway from C4:0 to C12:0, which resembled the de novo synthesis pathway. Causal relation between the QTL on BTA19 and de novo fatty acids showed that the QTL affects C4:0, C6:0, C8:0, C10:0 and C14:0 directly, while C12:0 was indirectly affected by the QTL through its effect on C10:0. The potential of GWAS based on MIR predicted fatty acids was explored but failed to detect some QTL and resulted in additional QTL that were not detected based on GC measurements. Therefore, MIR predicted phenotypes add complexity to the genotype-phenotype relationship, and renders MIR predicted phenotypes less appropriate to identify candidate genes and to infer the biological background of traits.