Frequencies of milk protein variants and haplotypes estimated from genotypes of more than 1 million bulls and cows of 12 French cattle breeds

Due to their major effects on milk composition and cheese-making properties and their putative effects on human health, there is a great deal of interest in bovine milk protein variants. The objectives of this study were to estimate frequencies of milk protein variants and haplotypes in 12 cattle breeds as well as their trends over time to assess the effect of selection on milk traits. Milk protein variants and haplotypes were identified from SNP genotype data from more than 1 million animals from 12 dairy, beef, or dual-purpose cattle breeds that had been genotyped for genomic selection. We examined a total of 15 loci in the genes that encode β-lactoglobulin (β-LG) and 3 caseins (αS₁-CN, β-CN, and κ-CN); genotypes were directly called from customized SNP chips (50.6%) or imputed (49.4%). Variants A and B of β-LG were frequent in the 12 breeds. For the caseins, we found 3 variants for αS₁-CN (B, C, and D), 6 for β-CN (A1, A2, A3, B, C, and I), and 5 for κ-CN (A, B, C, D, and E). For αS₁-CN, the B variant was the most frequent in all breeds except Jersey. For β-CN, the A2 variant was the most abundant in all breeds except Tarentaise, although in Normande animals, the I variant (30.9%) was almost as common as A2 (39.7%). The C variant was very rare except in the Tarentaise sample (4.8%). The most frequent variant for κ-CN was A in 5 breeds (including Holstein), and B in the 7 other breeds. The B variant was present at a particularly high frequency in Jersey (82.6%) and Normande (85.5%) animals. The C and E variants of κ-CN appeared to be particularly frequent in the Tarentaise (12.7%) and Holstein (9%) breeds, respectively. We found 20 haplotype combinations of αS₁–β–κ CN that were present at a frequency >0.1% in at least one breed; however, only 6 to 9 haplotypes were found in any given breed, demonstrating a strong degree of linkage disequilibrium. The most frequent haplotypes were B-A1-A, B-A2-A, B-A2-B, B-I-B, C-A2-A, and C-A2-B. Some alleles were predominantly found in only one haplotype, such as the E and C variants of κ-CN and the I variant of β-CN, which were mainly found in the B-A1-E, B-A1-C, and B-I-B haplotypes, respectively. We observed changes in the frequency of certain variants over time in several breeds, such as an increase in the frequency of variants A of β-LG, I of β-CN, and B of κ-CN. With these results, we update and complete frequency data that were first estimated 30 to 50 yr ago, and, for the first time in these breeds, we assess the effect of selection on milk protein variants.